<%BANNER%>

Sputter Deposition of Rare Earth Doped Zinc Sulfide for Near Infrared Electroluminescence


PAGE 1

SPUTTER DEPOSITION OF RARE EARTH DOPED ZINC SULFIDE FOR NEAR INFRARED ELECTROLUMINESCENCE By WILLIAM ROBERT GLASS III A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2003

PAGE 2

Copyright 2003 by William Robert Glass III

PAGE 3

ACKNOWLEDGMENTS First of all I would like to thank my advisor, Dr. Holloway. He has been the best advisor I have known. It was an honor to work with him. I would also like to thank Dr. Mark Davidson. Without his help, both mentally and physically, I would not have been able to reach my goals. It was also a pleasure to work with all of the people out at Microfabritech including Barbara, Diane, Scott, Chuck, Andreas, and Maggie. Ludie, of course, deserves a huge thank you. Ludie is the best secretary a group could ever have. Ludie was never without a smile and made things go smoother than I could ever have imagined. I appreciate all of the members of Dr. Holloways group including Ajay, Nigel, Jie, Dave, etc for all of their help and support. I, of course, want to thank my parents for their support and love. Without them I would never have been able to make it to where I am today. Finally, I want to thank my wife Jackie. Without her I would have been lost. She is the best thing that has ever happened to me. Words are not enough to express my love to her. iii

PAGE 4

TABLE OF CONTENTS Page ACKNOWLEDGMENTS.................................................................................................iii LIST OF TABLES............................................................................................................vii LIST OF FIGURES.........................................................................................................viii ABSTRACT.....................................................................................................................xiii CHAPTER 1 INTRODUCTION........................................................................................................1 2 BACKGROUND..........................................................................................................3 2.1 Introduction.............................................................................................................3 2.2 Infrared Emitters.....................................................................................................6 2.3 Electroluminescent Device Structure.....................................................................8 2.4 Device Physics......................................................................................................11 2.5 ACTFELD Materials............................................................................................17 2.5.1 Substrates....................................................................................................17 2.5.2 Insulators....................................................................................................19 2.5.3 Conductors..................................................................................................22 2.6 Phosphor Luminescence.......................................................................................24 2.6.1 Host Materials............................................................................................24 2.6.2 Luminescent Centers..................................................................................27 2.6.3 Rare Earth Doped ZnS................................................................................29 2.6.3.1 ZnS:Tm.............................................................................................30 2.6.3.2 ZnS:Er..............................................................................................31 2.6.3.3 ZnS:Nd.............................................................................................32 2.7 Electrical and Optical Characterization................................................................32 2.7.1 Brightness versus Voltage..........................................................................33 2.7.2 Threshold Voltage......................................................................................35 2.7.3 Efficiency versus Voltage...........................................................................35 2.7.4 Electrical Testing........................................................................................37 2.7.5 Charge versus Voltage (Q-V).....................................................................38 2.7.6 Capacitance versus Voltage........................................................................42 2.7.7 Internal Charge versus Phosphor Field.......................................................43 2.7.8 Maximum Charge versus Maximum Voltage............................................46 iv

PAGE 5

3 EXPERIMENTAL PROCEDURE.............................................................................48 3.1 Substrate and Target Preparation..........................................................................48 3.2 Sulfide Sputter Deposition System.......................................................................48 3.3 Top Contact Deposition........................................................................................52 3.4 Sample Handling and Storage..............................................................................53 3.5 Sputtered Film Characterization...........................................................................53 3.5.1 Thickness Measurements............................................................................54 3.5.2 X-ray Diffraction (XRD)............................................................................54 3.5.3 Electroluminescence...................................................................................56 3.5.4 Photoluminescence and Photoluminescent Excitation...............................59 3.5.5 Electron Microprobe...................................................................................60 3.5.6 Energy Dispersive X-ray Spectroscopy (EDS) on a Scanning Electron Microscope (SEM)...........................................................................................61 3.5.7 Time Resolved Electroluminescence.........................................................62 3.5.8 Electrical Measurements............................................................................63 4 PHYSICAL EFFECTS OF CHANGING TARGET DUTY CYCLES AND SUBSTRATE TEMPERATURE DURING RF MAGNETRON SPUTTER DEPOSITION.............................................................................................................65 4.1 Introduction...........................................................................................................65 4.2 Spectra..................................................................................................................65 4.3 Target Duty Cycle Alteration...............................................................................69 4.3.1 Concentration.............................................................................................69 4.3.2 Crystallinity................................................................................................69 4.3.3 Thickness....................................................................................................71 4.4.4 Threshold Voltage......................................................................................72 4.4.5 Infrared Emission.......................................................................................73 4.5 Deposition Temperature Effects...........................................................................74 4.5.1 Concentration.............................................................................................74 4.5.2 Crystallinity................................................................................................75 4.5.3 Thickness....................................................................................................76 4.5.4 Threshold Voltage......................................................................................77 4.5.5 Infrared Emission.......................................................................................78 4.6 Discussion.............................................................................................................79 4.7 Comparison of Infrared to Visible Emission........................................................85 5 ELECTRICAL EFFECTS OF CHANGING TARGET DUTY CYCLES AND SUBSTRATE TEMPERATURE DURING RF MAGNETRON SPUTTER DEPOSITION.............................................................................................................89 5.1 Introduction...........................................................................................................89 5.2 Charge-Voltage (Q-V) Data.................................................................................89 5.3 C-V Data...............................................................................................................97 5.4 Qint-Fp Data.........................................................................................................100 5.5 Time Resolved Electroluminescence..................................................................107 v

PAGE 6

5.5.1 Discussion of TREL Data.........................................................................108 5.6 Discussion...........................................................................................................117 5.6.1 Q-V Analysis............................................................................................117 5.6.2 C-V Analysis............................................................................................125 5.6.4 Interface Layer Discussion.......................................................................132 6 CONCLUSIONS......................................................................................................138 6.1 Deposition Effects on the Physical Properties and Optical Properties of ZnS:RE Phosphors.............................................................................................................138 6.2 Electrical Properties of ZnS:RE Phosphors........................................................139 LIST OF REFERENCES.................................................................................................141 BIOGRAPHICAL SKETCH...........................................................................................149 vi

PAGE 7

LIST OF TABLES Table page 2-1 List of insulators used in ACTFEL devices and their properties of interest...............22 2-2 Properties of ZnS and SrS...........................................................................................25 2-3 Optical properties of common sulfide based EL materials..........................................28 2-4 Physical properties of ZnS...........................................................................................30 vii

PAGE 8

LIST OF FIGURES Figure page 2-1 Sketch of phosphor-LEP lamp.......................................................................................7 2-2 Cross-sectional view of (a) normal and (b) inverted ACTFELD structure...................9 2-3 Equivalent circuit for an ACTFEL device...................................................................11 2-4 Energy band diagram illustrating the five primary physical processes responsible for ACTFEL device operation.......................................................................................12 2-5 Energy band diagram of an ACTFEL device with and without space charge in the phosphor layer..........................................................................................................18 2-6 Energy level diagrams and radiant transitions of Tm3+, Nd3+, and Er3+......................26 2-7 Impact cross sections of the 3F4 and 1G4 levels in Tm3+ [78]......................................31 2-8 Brightness vs. voltage curve showing the threshold voltage.......................................34 2-9 ACTFELD efficiency versus drive voltage.................................................................36 2-10 Schematic of a Sawyer-Tower test setup...................................................................37 2-11 Trapezoidal waveform with important points marked for reference.........................38 2-12 Typical Q-V plot........................................................................................................40 2-13 Typical C-V plot........................................................................................................43 2-14 Typical Qint-Fp plot....................................................................................................45 2-15 Typical Qmax-Vmax plot..............................................................................................47 2-16 Typical Qemax-Vmax plot.............................................................................................47 3-1 Schematic of the sputter system used for RF magnetron sputtering...........................50 3-2 View of sample platter showing substrate positions and spaces for additional substrates..................................................................................................................51 viii

PAGE 9

3-3 Schematic of the heating system in the sputtering system..........................................52 3-4 Back view of the sample on the test stage...................................................................57 3-5 Spectral sensitivity of the Ocean Optics #13 grating..................................................58 3-6 Side view of the sample stage and fiber optic detection system.................................59 3-7 System to measure time resolved luminescence and electrical data...........................63 4-1 Electroluminescent spectrum of ZnS:TmF3................................................................66 4-2 Electroluminescent spectrum of ZnS:NdF3.................................................................67 4-3 Electroluminescent spectrum of ZnS:ErF3..................................................................67 4-4 Energy levels of rare earth ions and transitions luminescence producing transitions observed in Figs. 4-1, 4-2 and 4-3............................................................................68 4-5 Effect of target duty cycle on the Tm, Nd, and Er concentrations in the ZnS films measured by EDS and EPMA..................................................................................70 4-6 Effect of duty cycle ratio on the full width at half maximum of the 28.5o x-ray diffraction peak of ZnS............................................................................................71 4-7 Normalized thickness of the rare earth doped ZnS films. Deposition times were changed to attempt to achieve the same thickness for each rare earth film.............72 4-8 NIR threshold voltages of the doped ZnS films with varying deposition duty cycles73 4-9 Effect of target duty cycle on the near infrared emission of each rare earth...............74 4-10 Concentration of each rare earth in the ZnS films as a function of substrate temperature during deposition measured by EDS....................................................75 4-11 Increasing FWHM of the ZnS 28.5o diffraction peak as the deposition temperature is increased...................................................................................................................76 4-12 Decreasing phosphor thickness with increasing deposition temperature..................77 4-13 Optical turn on voltage variation with increasing deposition temperature for each material.....................................................................................................................78 4-14 Decrease of near infrared irradiance with increasing deposition temperature..........79 4-15 Comparison of NIR turn on voltage and phosphor thickness as deposition temperature is varied................................................................................................81 ix

PAGE 10

4-16 Comparison of NIR turn on voltage and phosphor thickness as duty cycle and deposition time is varied..........................................................................................82 4-17 NIR irradiance as a function of rare earth concentration. Note that the maximum occurs near 1 at% for each rare earth.......................................................................84 4-18 B40 (left ordinate) for the NIR emission and B40 (right ordinate) for the visible emission in ZnS:TmF3 for various Tm concentrations............................................86 4-19 B40 (left ordinate) for the NIR emission and B40 (right ordinate) for the visible emission in ZnS:NdF3 for various Nd concentrations..............................................87 4-20 B40 (left ordinate) for the NIR emission and B40 (right ordinate) for the visible emission in ZnS:ErF3 for various Er concentrations................................................88 5-1 Typical Q-V data for ZnS:TmF3 at drive voltages between 100 and 150 volts..........91 5-2 Typical Q-V data for ZnS:NdF3 at drive voltages between 80 and 170 volts.............92 5-3 Typical Q-V data for ZnS:ErF3 at drive voltages between 80 and 150 volts..............93 5-4 Electrical threshold voltages for each phosphor as a function of duty cycle..............94 5-5 Electrical threshold voltages for each phosphor as a function of deposition temperature...............................................................................................................95 5-6 Plot of Q-V of ZnS:TmF3 at B40 with increasing deposition temperature (140-180oC)96 5-7 Plot of Q-V of ZnS:NdF3 at B40 with increasing deposition temperature..................97 5-8 Typical C-V data for ZnS:TmF3 at drive voltages between 100 and 150 volts...........98 5-9 Typical C-V data for ZnS:NdF3 at drive voltages between 80 and 170 volts.............99 5-10 Typical C-V data for ZnS:ErF3 at drive voltages between 80 and 150 volts..........100 5-11 Internal Charge vs. phosphor field for increasing voltage in ZnS:TmF3.................102 5-12 Internal Charge vs. phosphor field for increasing voltage in ZnS:NdF3.................103 5-13 Internal Charge vs. phosphor field for increasing voltage in ZnS:ErF3..................104 5-14 Internal charge vs. phosphor field for ZnS:TmF3 as the deposition temperature is changed...................................................................................................................105 5-15 Internal charge vs. phosphor field for ZnS:NdF3 as the deposition temperature is changed...................................................................................................................106 x

PAGE 11

5-16 Internal charge vs. phosphor field for ZnS:ErF3 as the deposition temperature is changed...................................................................................................................107 5-17 Time resolved electroluminescence of the NIR and blue emission from ZnS:TmF3110 5-18 Time resolved electroluminescence of the visible emission from ZnS:NdF3 for voltage pulse durations of 5 and 30 s...................................................................111 5-19 Time resolved electroluminescence of the visible emission from ZnS:ErF3...........112 5-20 Log plot of TREL decay of the 480 nm emission from ZnS:TmF3.........................113 5-21 Log plot of TREL decay of the 800 nm emission from ZnS:TmF3.........................114 5-22 Log plot of TREL decay of the 600 nm emission from ZnS:NdF3 for a 30 s voltage pulse.......................................................................................................................115 5-23 Log plot of TREL decay of the 600 nm emission from ZnS:NdF3 for a 5 s voltage pulse.......................................................................................................................116 5-24 Log plot of TREL decay of the 530 nm emission from ZnS:ErF3..........................117 5-25 Energy band diagram of an ACTFEL device showing how the distribution of interface states can affect the electric field necessary for tunnel injection............118 2-26 Transferred charge versus maximum applied voltage showing the electrical threshold for a typical ZnS:TmF3 device...............................................................120 5-27 Irradiance from ZnS:Tm versus applied voltage showing the optical threshold is the same for NIR and visible emission........................................................................121 5-28 Irradiance from ZnS:Nd versus applied voltage showing the optical threshold is the same for NIR and visible emission........................................................................122 5-29 Irradiance from ZnS:Er versus applied voltage showing the optical threshold is the same for NIR and visible emission........................................................................123 5-30 Comparison of optical and electrical threshold voltages with changing duty cycle ratios for each dopant.............................................................................................124 5-31 Comparison of optical and electrical threshold voltages versus deposition temperature for each dopant...................................................................................125 5-32 Normalized internal charge, phosphor field and NIR brightness versus Tm concentrations in ZnS:TmF3. Note that while the average of internal charge is nearly constant, the trend for both B40 and Fp is down as the temperature increases. This correlation is discussed in the text.................................................128 xi

PAGE 12

5-33 Relation of internal charge and phosphor field with NIR brightness for ZnS:NdF3 with changing deposition temperature...................................................................129 5-34 Relation of internal charge and phosphor field with NIR brightness for ZnS:NdF3 with changing target duty cycle.............................................................................130 5-35 Relation of internal charge with NIR brightness for various Nd concentrations in ZnS:NdF3................................................................................................................131 5-36 Relation of internal charge and phosphor field with NIR brightness for ZnS:ErF3 with changing deposition temperature. Note that the brightness correlates with Fp and not with the internal charge.............................................................................132 5-37 Calculated interface layer thicknesses for ZnS:TmF3 as a function of deposition temperature and duty cycle (Duty cycle data plotted at the duty cycle + 100 i.e. a duty cycle ratio of 50 is plotted at 150)..................................................................136 5-38 Calculated interface layer thicknesses for ZnS:NdF3 as a function of deposition temperature and duty cycle (Duty cycle data plotted at the duty cycle + 100 i.e. a duty cycle of 50 is plotted at 150)..........................................................................137 xii

PAGE 13

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy SPUTTER DEPOSITION OF RARE EARTH DOPED ZINC SULFIDE FOR NEAR INFRARED ELECTROLUMINESCENCE By William Robert Glass III December, 2003 Chair: Dr. Paul H. Holloway Major Department: Materials Science and Engineering Near infrared emitting alternating current thin film electroluminescent (ACTFEL) phosphors were fabricated by simultaneous R.F. magnetron sputtering from both a target of doped ZnS and an undoped ZnS target. The intensities of both near infrared (NIR) and visible emission from ZnS doped with thulium (Tm), neodymium (Nd), or erbium (Er) fluorides were dependent on deposition parameters such as target duty cycle (varied from 25 to 100% independently for the two targets) and substrate temperature (140-180oC), with lower temperatures giving 400% better NIR brightness. By optimizing the rare earth concentration between 0.8 and 1.1 at%, the near infrared irradiance was improved by 400% for each dopant. The increase in brightness and optimal concentrations are attributed to decreased crystallinity and increased dopant interaction at higher rare earth concentrations. The brightness increase with decreasing deposition temperature was attributed to a reduction of thermal desorption of the ZnS during deposition, and xiii

PAGE 14

consequently thicker films and optimized rare earth concentration. Luminescent decay lifetimes were short (20-40 sec) because of a high concentration of non-radiative pathways due to defects from the strain of the large rare earth ions on the ZnS lattice. The threshold voltage for visible and near infrared emission was identical despite emission of NIR and visible light resulting from electrons relaxing from low and high energy excited levels, respectively. The optical threshold voltages were identical to the electrical threshold voltages, and it was concluded that at the voltages necessary for electrical breakdown, the accelerated electrons had enough energy to excite either the visible or NIR emitting levels. Phosphors doped with Nd exhibited increased internal charge at higher dopant concentrations despite a reduction in phosphor field (i.e. reduced applied voltage) In contrast; the charge did not change appreciably for Er and decreased for Tm doped films at reduced fields. The charge differences were attributed to dopant effects on the distribution of states near the interfaces. It was postulated that Nd doped devices have a shallower state distribution, while the majority of states in Tm doped devices are deeper and require higher fields for tunnel injection. The electrical behavior of all of the devices also demonstrated that field clamping occurred despite non-ideal phosphor breakdown during device operation. It is postulated that a high breakdown strength, low dielectric constant, interface layer is formed during deposition, and reduces capacitance before and after phosphor breakdown and results in field clamping. The thickness calculated for the interface layer decreases with increasing deposition temperature implying that the layer is formed during deposition, and this decreasing thickness results from increased atomic mobility at higher temperatures. xiv

PAGE 15

CHAPTER 1 INTRODUCTION Recently, much interest has been given to technologies for emitting visible light for use in flat panel displays. One of these technologies is the alternating current thin film electroluminescent device (ACTFELD) [1]. While visible emitting ACTFEL devices have garnered much attention, little attention has been given to infrared emitting devices. Near infrared emitting ACTFEL devices are suitable for applications that require mechanically robust, thermally stable devices than have lower power consumption than infrared emitting resistive devices. ZnS doped with various rare earths ions are promising materials for the development of infrared emitting ACTFEL phosphors [2]. While phosphors such as ZnS doped with Tm, Nd or Er emit blue, orange, and green visible light, they also emit strongly in the near infrared region (0.7-2 um). However, infrared emission from these phosphors is undesirable when used for their visible output. In this study the relationship between visible and infrared emission and the determination of the deposition conditions necessary for maximizing the infrared output of these devices has been performed. Chapter 2 will present background information on infrared emitting devices as well as a review of ACTFELD structures and operation. In chapter 3 the experimental and characterization methods and equipment used in this study will be presented. In chapter 4, results will be presented that show a dramatic increase in the infrared output of the rare earth doped ZnS by alteration of deposition conditions during R.F. magnetron sputtering. It will be shown that the rare earth concentration is a critical parameter determining the 1

PAGE 16

2 intensity of infrared and visible emission. Chapter 5 will present electrical characterization data and a discussion of the factors limiting the output of these materials and devices. Finally, conclusions will be presented in chapter 6.

PAGE 17

CHAPTER 2 BACKGROUND 2.1 Introduction Much work has recently been done on the development of visible thin film phosphors for use in flat panel displays. Thin film phosphors which emit in the infrared have often been overlooked. While infrared phosphors do not have the same markets as their visible counterparts, there are applications in which infrared alternating current thin film electroluminescent devices (ACTFELDs) are desirable. Industry can use infrared emitting devices for absorption based gas sensors or production of thermal bandages and the auto industry has investigated infrared systems for improving safety during night driving [3]. Military applications include night vision, friend/foe identification, scene projectors for night mission training, and infrared portable computer screens for night operations. Industrial gas sensors operate by light absorption of a gas through the vibration-rotation bands of polar molecules. When these bands are centered at wavelengths characteristic of the bending and stretching of the molecules, the absorption depends on the number of molecules in the light path [4]. For example, devices with emission at 761 nm can be used to detect oxygen [5]. Currently thermal sources, such as tungsten lamps, are the light sources for most gas sensors. However, advances in semiconductor technology and a decrease in component costs can lead to the replacement of filtered thermal sources in gas sensors. 3

PAGE 18

4 Automobile companies such as Daimler Chrysler are testing infrared illumination systems to make night driving safer. Daimler Chrysler has fitted active night vision systems onto its Jeep Grand Cherokee and had tested the system on a bus. The buss night vision system allows the driver to see further than with conventional headlights without blinding oncoming drivers [6]. Other auto companies currently investigating night vision include Acura, Cadillac, and Volvo [7]. The United States military has wanted to engage its enemies under cover of night since the revolutionary war. Such attacks proved to be extremely dangerous until effective night vision equipment was developed. The first true night vision systems were developed during World War II in the form of infrared sniper scopes. These scopes emitted an infrared light that the scope could detect and turn into a visible picture. While current military practice focuses on passive night vision (the amplification of existing light), active night vision may be a more effective tool. During desert storm military helicopters had infrared aiming lights installed on their landing skids to avoid sand dunes. The helicopters were in no danger of being seen because the Iraqi army did not have near infrared detection devices [8]. IFF (identification friend or foe) has concerned the military since World War II. IFF was developed in England to avoid shooting down their own planes when they returned home. IFF is a concern whenever aircraft are in the sky [9]. Infrared emitting phosphors can be used in each of these applications. A phosphor is a material that emits light when excited by an energy source. Emission that ceases within 10 nanoseconds of the excitation is known as fluorescence [10]. Longer lasting luminescence, known as phosphorescence, can last hours [11]. The exciting

PAGE 19

5 energy can be photonic, electronic, ionic, or thermal. Thin film phosphor devices usually operate in one of several ways. Photoluminescent devices are excited by higher energy photons from sources such as ultraviolet lamps or lasers [12]. Cathodoluminescent devices, such as televisions, operate by the emission of electrons from a tip or electron gun that strike the film [13]. Electroluminescent devices use an applied electric field across the phosphor to induce luminescence [14]. Research into rare earth doped zinc sulfide has been concentrated on the search for efficient red, green, and blue phosphors; infrared emission from these materials was overlooked or actively discouraged to improve the efficiency of visible emission. Zinc sulfide doped with thulium is a blue emitting phosphor whose emission is generally too weak for use as a display phosphor however; it exhibits significant near infrared (NIR) emission [15]. Neodymium and erbium doped zinc sulfide also emit in both the visible and infrared regions. Neodymium emits in the orange and erbium emission is stronger in the green regions, with weaker emission in the red. Unlike thulium and neodymium, the infrared emission from erbium has been of interest, mainly for telecommunications [16]. Strontium sulfide also has been studied as a host for rare earth phosphors. While SrS is a better host for blue devices due to its superior electron high-field transport properties [17], ZnS is better for infrared. Hot electrons (the excitation source in electroluminescent ZnS doped with TmF3 or other rare earths, as shown below) in ZnS do not appear to have enough energy to excite shorter wavelength luminescent centers [18]. This leads to decreased blue emission compared to SrS, but these electrons can stimulate infrared emission. As discussed below, the ratio of infrared to visible emission is dependent on deposition conditions.

PAGE 20

6 Modification of the phosphors, including codoping with alkalis such as lithium, has been tested to improve the visible brightness of ZnS:RE films by lowering the symmetry around the rare earth [19]. These alterations succeeded in decreasing the infrared to visible ratio. In addition, others have introduced oxygen into the phosphor films in an effort to increase the visible luminescence. While this was effective in increasing the blue emission in ZnS:TmF3, it also increased the infrared output. These increases are thought to result from reduced non-radiative transitions at sulfur vacancies [20]. The non-radiative transitions are caused by the defects produced at the sulfur vacancies. It is possible to improve the crystallinity of the ZnS by annealing etc. without needing to add oxygen. Finally, because of the decrease in infrared emission, doping with alkalis should be avoided if an infrared emitter is desired. For these reasons rare earth doped ZnS phosphors used for infrared emission are often deposited simply as fluorides. 2.2 Infrared Emitters There are several sources of infrared light other than thin film devices. The most common are light emitting diodes (LEDs), lasers, and thermal emitters. Infrared LEDs are the analog of the common visible light LEDs. One of the possible drawbacks with LEDs is that they are limited to a fairly large size compared to the possible pixel size of an electroluminescent thin film. This makes LEDs undesirable for screen applications such as scene projectors or more flexible applications such as thermal bandages. However, rare earth ACTFLED phosphors can be used in LEDs for other applications by depositing the phosphor on a blue emitting GaN chip and using the blue light to photo excite the phosphor (Figure 2-1). A major drawback of such a design is a loss of efficiency [21].

PAGE 21

7 Figure 2-1 Sketch of phosphor-LEP lamp Infrared lasers can be much more intense than infrared ACTFEL devices but they are usually limited to applications that an ACTFELD would not be suited for. Infrared lasers are useful for directional applications such as target identification but fail when a more omni-directional device is needed. In addition a lasers emission wavelength is unstable with temperature, drifting several nanometers as the temperature changes [22]. Applications such as gas sensors need stable light sources to function properly. Infrared lasers can also have problems with long-term stability due to amplitude variations when wavelength modulated [23]. Thermal emitters are similar to the filament of an incandescent light bulb. The main differences are the material used and the temperature of the glower. A common

PAGE 22

8 type of thermal emitter is the Globar. Globars are silicon carbide rods that are heated a desired temperature. The emission of the globar approximates that of a blackbody source at the same temperature [24]. Two of the drawbacks of thermal emitters are that they need to be heated to elevated temperatures to emit strongly in the near infrared and because of their blackbody character they do not emit at distinct wavelengths but instead over a wide spectrum. 2.3 Electroluminescent Device Structure Electroluminescent devices are flat electrically driven light emitters that use an electric field to produce luminescence without heat generation. The structure of an ACTFELD is essentially that of a dielectric-phosphor-dielectric sandwich. A complete device consists of a conductor, insulator, phosphor, insulator, conductor stack deposited on a substrate [25, 26]. Thin film electroluminescent devices have two basic designs based on the same structure. Typically, a normal device is deposited on a transparent substrate with a transparent conductor and insulator between the phosphor and the substrate. The top dielectric may be transparent or opaque and the top conductor is often reflective. A so-called inverted structure is the same layer sequence deposited on an opaque substrate with a transparent top insulator and conductor. An inverted structure is viewed through the top electrode while a regular device is viewed through the substrate (Figure 2-2)[15].

PAGE 23

9 Figure 2-2 Cross-sectional view of (a) normal and (b) inverted ACTFELD structure

PAGE 24

10 Both standard and inverted ACTFEL devices are commercially used. The choice of which structure to use depends on the application and processing requirements. The typical transparent substrate structure has several advantages over the inverted structure. One advantage is that if a suitable top conductor, such as aluminum, is used then the device experiences self-healing breakdown [27]. Self-healing causes the top electrode to pull back from short circuit paths such as pinholes and defects preventing catastrophic device failure. The electrode maintains effective contact to the rest of the device while isolating the short. Another advantage of this structure is its inherent durability. Since this device is viewed through the substrate, the films are protected. An advantage of the inverted structure is higher processing temperatures. At about 600oC the glass substrate commonly used for visible emitting normal structures begins to buckle and melt. Using an inverted structure, a higher melting temperature material, such as silicon, can be used. A disadvantage of the inverted structure is that self-healing top electrodes are not possible with transparent conductors. This means that the phosphor must have a very low defect density for the device to be reliable. Another ACTFEL device structure, commonly used for testing, is the one-insulator or half stack structure. As the name implies, a half stack device is the same as either a standard or inverted device except that one of the insulators is missing, while a full stack device has both insulators. The removal of this insulating layer from the device reduces the time needed to produce a device by eliminating one of the processing steps. Another advantage of half stack devices is that they tend to run at lower voltages than a comparable full stack device. However, half stack devices leave the phosphor layer more exposed than full stack devices and therefore exhibit poor long term reliability.

PAGE 25

11 2.4 Device Physics Understanding the basic physics of ACTFEL devices give insight into how they may be improved. An ACTFEL device can be modeled as circuit in which the phosphor is represented as a capacitor shunted by back-to-back Zener diodes with the insulators represented as capacitors [28] (Figure 2-3). Operation of an ACTFEL device follows five basic steps. They are (1) electron injection from interface states, (2) electron transport across the phosphor, (3) excitation of luminescent centers, (4) photon emission from radiative recombination, and (5) electron trapping [29]. These steps are shown in figure 2-4. Figure 2-3 Equivalent circuit for an ACTFEL device

PAGE 26

12 Figure 2-4 Energy band diagram illustrating the five primary physical processes responsible for ACTFEL device operation When the applied voltage is below the threshold voltage, the electrical circuit characteristics are that the Zener diodes are below their breakdown voltage. Hence, an ACTFELD below electrical threshold can be modeled simply as three capacitors. The capacitance for each of the layers can be modeled as parallel plate capacitors using the following equation.

PAGE 27

13 tACr0 where C is the capacitance of the layer, r is the relative permittivity, 0 is permittivity of free space, A is the area, and t is the thickness of the layer [30]. The equation for the whole device is simply that of three (or two in the case of a half stack) capacitors in series, bottomitopibottomiptopipbottomitopipCCCCCCCCCC where Cp is the capacitance of the phosphor and Citop and Cibottom are the capacitances of the top insulator and bottom insulator respectively. For the half stack device this equation simplifies to ipipCCCCC When the applied voltage becomes high enough, the phosphor reaches its threshold voltage; the circuit behaves as though the Zener diodes have reached their breakdown voltage; and the capacitance of the device is now just that of the insulating layers. Therefore, during device operation, injection of electrons from the insulator-phosphor interface into the phosphor occurs when a voltage large enough to breakdown the phosphor is applied to the device. When threshold is reached, the electrons trapped in interface states can tunnel into the conduction band of the phosphor [31]. The large electric field in the phosphor layer accelerates the electrons to ballistic energies and they travel across the phosphor. Sufficiently hot electrons may excite the host or non-luminescent centers which then transfer energy to the luminescent dopant, or the electrons may directly strike the luminescent centers causing impact excitation or impact

PAGE 28

14 ionization. After this collision, the electrons are again accelerated and the process continues. Once an electron travels across the phosphor from either the interface or from impact ionization, it will be captured by interface states on the other side of the phosphor. It is possible that electrons can be trapped in bulk states creating a space charge on the other side of or throughout the phosphor. Once the next voltage pulse arrives, the polarities of the electrodes are switched and the process begins again in the opposite direction. The interface between the insulator and the phosphor can be modeled after a Schottky barrier. The tunnel emission for a Schottky barrier is given by [32] qhEBqmEJ328exp23*2 where E is the electric field, m* is the electron effective mass, q is the charge of an electron, B is the barrier height, and h is Plancks constant. For interface state emission, the equation must be modified by replacing the barrier height with the interface trap depth. While the tunneling is temperature independent, the device current is temperature dependent. Thermionic emission has been suggested as an additional mechanism for charge injection. The Richard-Dushman equation for thermionic emission is [33] rWeemqJ3222 where Je is the electric charge flux, is the temperature multiplied by Boltzmanns constant, m is the mass of an electron, q is the charge of an electron, is Plancks

PAGE 29

15 constant divided by 2, and W is the work function. This equation is for the metal-vacuum interface, so in the ACTFELD case the equation must be modified to take the phosphors electron affinity into account. Roughening of the insulator-phosphor interface creates a wider interface region resulting in a broader distribution of interface trap energy. This can lower the field necessary to turn on the EL device [34]. Once the electrons have been injected into the phosphors conduction band, they must be accelerated to high enough energies to induce luminescence (typically >2eV). The electric field in the phosphor can be calculated by rearranging Maxwells equations for a series of capacitors yielding totippiipVddE where Ep is the phosphor electric field, is the dielectric constant, d is the thickness of the layer, and the subscripts i and p are for the insulator and phosphor, respectively. Inserting typical values for the dielectric constants and the thickness yield electric fields of about 2 to 2.5 MV/cm. Electrons accelerate very quickly in this high field. Their energies are limited by scattering, which can occur by several mechanisms, including low-energy quantum states [35]. Interface roughening, as mentioned earlier, broadens the energy distribution of traps at the interface. A broad energy distribution will allow tunneling of electrons in higher energy states to occur at lower electric fields. The acceleration due to the weaker field will result in lower energy ballistic electrons. The lower fields will not accelerate the electrons to as high an energy as would a large field. The energy levels necessary for infrared radiative transitions lie lower than those for visible emission, so it would appear that the lower energy electrons would result in

PAGE 30

16 increased infrared emission at lower voltages. However, this has not been tested, so it is unknown how the relative emission from visible and infrared emitting transitions will be affected. Energetic electrons may cause excitation of the host material or directly excite the luminescent centers in the phosphor. As the excited host ions return to a lower energy, the excited electrons may transfer energy through exciton states to the luminescent centers in the device or lose the energy to phonons, plasmons or Auger transitions [36]. With high enough energies the hot electrons can interact with the luminescent centers promoting ground state electrons to higher energy levels. As previously mentioned, the electrons can either be promoted to the conduction band of the host or to a higher level within the atom through impact ionization and impact excitation respectively [37]. The probability of an interaction is related to the impact cross section which will be discussed in the phosphor luminescence section. An electron that is impact excited to a higher energy level can then de-excite radiatively or non-radiatively. Non-radiative de-excitation usually occurs through phonon generation. Phonon energies are small compared to photon energies, usually about 20 meV [38]. Radiative de-excitation occurs through photon generation with the photon energy matching the energy level transition of the electron [39]. When the electron promoted into the conduction band of the host material is carried away by the electric field, it will either impact an ion in the phosphor or be carried to the interface. A luminescent center can only emit light when it captures another electron through a non-radiative transition from the conduction band into one of the atoms excited states. If the band gap of the host is a lower energy than the excited state of the luminescent center, visible or near IR emission is greatly reduced [40].

PAGE 31

17 The previous description does not take into account space charge, a very common occurrence in ACTFEL devices [41]. Some of the electrons or holes in the phosphor may be trapped in bulk trap states and create a space charge. The space charge will produce bending of the bands near the interface causing the field across the phosphor to be non-uniform. If holes are concentrated near the cathode then the field will have an increased strength near the cathodic interface and lower strength as it approaches the anode (Figure 2-5). Space charge is presumed to result from ionization of deep traps at the interface, field emission from bulk traps, or band to band impact ionization and subsequent hole trapping [42,43,44]. Space charge in SrS phosphors has been photo-induced [45]. Space charge generation in ZnS:MnCl has been attributed to the impact ionization of zinc vacancies that are part of chlorine-zinc complexes [46]. Zinc-fluorine complexes formed when using fluorides instead of chlorides as the starting compounds could lead to similar states. 2.5 ACTFELD Materials 2.5.1 Substrates The substrate for a standard ACTFELD needs to be transparent, smooth, robust, and preferably inexpensive. The substrate of a visible ACTFELD is often Corning 7059 soda-lime glass. Corning 7059 glass has a softening temperature of about 600oC so rapid thermal annealing below 650oC is possible but anything higher will deform the substrate [47]. Smaller samples, up to 2 inches square, may be annealed up to 850oC for short times. In addition, Corning 7059 glass is free of alkalis; so alkali diffusion into the device is avoided [48]. For phosphors requiring higher temperature anneals or for mid

PAGE 32

18 Figure 2-5 Energy band diagram of an ACTFEL device with and without space charge in the phosphor layer

PAGE 33

19 infrared applications, Corning 7059 glass is an unsuitable choice. High temperature glass is often too expensive to be a viable option, but silicon is a suitable choice for use with inverted structures or mid-infrared applications. Silicon is readily available and inexpensive and, with proper doping, can be used as the bottom contact for the inverted structure. A silicon substrate will withstand annealing up to 1400oC before melting, so high temperature processing is limited by the robustness of the deposited layers. Silicon has already been used for active matrix displays where each pixel was controlled using a circuit array on the wafer [49]. 2.5.2 Insulators In a full stack device the phosphor is sandwiched between two dielectric layers and in a half stack device the phosphor is deposited onto a dielectric layer. The insulator affects the phosphor-insulator interface that determines the interface states that play a large role in the production of the current necessary for light generation [50]. More importantly, these layers contribute to the stability of the device by preventing large currents from flowing through the phosphor when the device is driven at the large voltages, typically 2 Mv/cm, needed for electrical breakdown. Because of the high electric fields present during device operation, the insulator needs high dielectric breakdown and needs to be as defect free as possible. The insulator should also prevent charge leakage into the phosphor layer. In addition, the dielectric layers need high thermal stability to withstand heat treatments and the insulators also need to adhere well to the phosphor and the contacts. Also, in order to prevent the diffusion of foreign species into the phosphor layer, the insulator should be chemically stable. Finally, as with the bottom contact in a standard structure, the dielectric layer should be as

PAGE 34

20 transparent as possible to the emission wavelengths of the device. So, the essential insulator requirements for use in ACTFEL devices are as follows [51]: 1. Sufficient dielectric breakdown electric field, FBD 2. High relative dielectric constant, r 3. Small number of defects and pinholes 4. Good adhesion to phosphor and contacts 5. Transparency 6. Good thermal and chemical stability 7. Small dielectric loss factor, tan In order to have efficient device operation, as much of the applied voltage as possible should be dropped across the phosphor layer. The proportions of the voltage dropped across the phosphor and insulators are determined by the capacitance of the phosphor, Cp, and the capacitance of the insulator, Ci. As discussed in section 2.4, the capacitances of the layers are determined using tCro where o is the permittivity of free space, r is the relative permittivity, and t is the thickness of the layer. In order to maximize the voltage drop across the phosphor, the capacitance of the insulator should be much larger than the capacitance of the phosphor. Using the above equation, either the insulator should be very thin or the relative dielectric constant of the insulator should be large. Unfortunately, charge leakage has been shown to occur in insulators that are thinner than 50 nm [52]. As noted above, high dielectric breakdown strength is necessary for insulators because if the phosphor becomes a virtual short after

PAGE 35

21 breakdown then the additional voltage will be dropped across the insulators increasing the electric field they experience. The thinner the insulator the larger the field; however, most insulators with high dielectric constants have low breakdown strengths. In addition, insulators with high dielectric constants often exhibit propagation breakdown, which occurs when a small portion of the insulator breaks down forming a short that heats up the insulator leading to catastrophic failure. On the other hand, many insulators with lower dielectric constants experience self-healing breakdown in which the breakdown areas become an open instead of a short circuit so they do not exhibit catastrophic breakdown. See Table 2-1 for a list of insulators and their properties [49]. Pinholes and defects in the insulator should be minimized to prevent device failure. If the insulator experiences propagating breakdown, a pinhole or defect can lead to failure of the entire device. Stability of the device also requires that the insulating layers adhere well to the contacts and the phosphor. Insulators with poor adhesion will cause the device lifetime to be short. Obviously, the bottom insulator of a standard ACTFEL device has the same requirement as the bottom contact in that it needs to be transparent to the emitted light. Again, like the bottom conductor, the insulators need to be able to withstand the thermal processing of the device. The bottom insulator must also be chemically stable so that it does not affect the conductivity of contacts such as ITO, or modify the composition of the phosphor layer. Finally, the insulator must be able to maintain the charge balance in the device. The insulator can cause charge loss or leakage disrupting the proper function of the ACTFELD. Because of this it is believed that leakage charge, as can occur with thin layers, negatively affects device operation [53]. For this reason, the loss factor of the insulator should be kept small.

PAGE 36

22 Table 2-1 List of insulators used in ACTFEL devices and their properties of interest InsulatorDepositionrFBDorFBDBreakdownMethod*(MV/cm)(C/cm2)Mode**SiO2Sputtering462SHBSiOxNySputtering674SHBSiOxNyPVCD674SHBSi3N4Sputtering88 to 94 to 6SHBAl2O3Sputtering853.5SHBAl2O3ALE886SHBSiAlONSputtering88 to 94 to 6SHBY2O3Sputtering123 to 53 to 5SHBY2O3EBE123 to 53 to 5SHBBaTiO3Sputtering143.34SHBSmO3EBE152 to 43 to 5SHBHfO2Sputtering160.17 to 40.3 to 6SHBTa2O5-TiO2ALE20712SHBBaTa2O6Sputtering223.57SHBTa2O5Sputtering23-251.5 to 33 to 7SHBPbNb2O6Sputtering411.55SHBTiO2ALE600.21PBSr(Zr,Ti)O3Sputtering100326PBSrTiO3Sputtering1401.5 to 219 to 25PBPbTiO3Sputtering1500.57PBBaTiO3****Press/sinter5000???Westaim proprietary***Press/sinter1700??? 2.5.3 Conductors In a normal ACTFEL structure, the emitted light must be able to pass through the bottom insulator and the bottom contact. This means that the most important property of the bottom contact is that it is transparent at the desired wavelengths. In addition, the bottom contact must have an electrical resistivity low enough not to affect the capacitance of the device or cause resistive heating. Finally, the contact must also be able to withstand the thermal processing of the device. For wavelengths longer than 1150 nm it is possible to use doped silicon as a bottom conductor in addition to being the substrate. This has the benefits of reducing the number of deposition steps needed as well as the advantage of silicons tolerance for higher temperature processing.

PAGE 37

23 For visible applications and those in the near IR the most common material for a bottom conductor is ITO, indium tin oxide, an alloy of ~90 wt% In2O3 and 10 wt% SnO3. ITO can be deposited in several ways including RF magnetron sputtering, plasma ion-assisted deposition, focused ion beam, and pulsed laser deposition [54-58]. The ITO layer is typically 200 nm thick with a resistivity of ~1x10-4 -cm. This provides a sheet resistance of 5-10 /. ITO is transparent over the visible and near infrared range because the bandgaps of In2O3 and SnO3 are both about 3.5eV. ITO is conductive due to oxygen vacancies and Sn4+ ions occupying In3+ sites creating shallow donors a few meV below the conduction band [59]. So the conductivity of the layer can be reduced during annealing if not suitable protected. In addition, transparent conductors may cause reliability problems for inverted structures because they do not exhibit self healing like some opaque conductors. The final layer of a normal ACTFEL device is typically an opaque top contact. Like the bottom contact, the top contact must be highly conductive but does not need to survive high temperature processing since annealing can be completed before the top contact is deposited. In addition, this layer must be able to adhere well to the insulating layer in the case of a full stack device or the phosphor in the case of a half stack device. Aluminum is almost always the material of choice for the top contact. Aluminum has many advantages including low cost and low resistivity. In addition, aluminum adheres well to most insulators which makes it good at self-healing around short circuits. Finally, aluminum is easy to deposit using either evaporation or sputtering. Aluminums low melting temperature of 660oC is both an advantage and a disadvantage. The low melting temperature makes aluminum easy to thermally evaporate, but is undesirable for inverted

PAGE 38

24 structures when high temperature processing is needed. For inverted structures other metals with higher temperature tolerances, such as tantalum, molybdenum, or tungsten, can be used [49]. 2.6 Phosphor Luminescence 2.6.1 Host Materials Determining the host material is an important first step in designing an electroluminescent device. Thulium, erbium, neodymium, and dysprosium all have excited state energy levels between 6666 and 14000 cm-1 above the ground state at zero (corresponding to photons with wavelengths of 1500 nm to 715 nm in the NIR) [60] (Figure 2-6). ZnS provides a suitable host for these infrared emitters ions for several reasons. The typical phosphor fields during ACTFEL device operation exceed 1 MV/cm and depend on the thickness and dielectric constant of the phosphor [61]. The dielectric breakdown strength of ZnS is ~1.5 MV/cm [62]. This dielectric strength is sufficient for ZnS to act as an insulator below threshold and act as a conductor at high fields. For near infrared devices the bandgap of the host must be at least 1.6eV while visible phosphors need a bandgap of 3.1eV. ZnS has a bandgap of ~3.7eV at room temperature and is transparent from below 400 nm to past 10 m [63]. The glass substrates typically used for this type of device are only transparent to about 4 m; however, a silicon or chalcogenide substrate could be used for longer wavelengths [10]. The ITO often used as a transparent conductor has a plasma edge, the long wavelength cutoff for transparency dependent on carrier concentration, at ~1.8 m. As mentioned above, silicon can be used as substrate and bottom conductor for longer wavelength applications solving this problem.

PAGE 39

25 In addition, hot electron distribution in ZnS is less energetic than in SrS, another common ACTFELD host material [18]. Low energy (cooler) electrons can only excite ground state electrons to the lower energy states on the luminescent centers. This lower energy electron distribution shifts the strong luminescent emission for dopants such as Tm and Nd from the visible to the near infrared. An example of this is that SrS:Nd produces an orange-white light while ZnS:Nd produces only orange. This is because the neodymium doped ZnS phosphor has no emission shorter than 530 nm. The emission at 600 nm due to relaxation from the 2H11/2 to the 4I9/2 ground state is active in both hosts, but the higher 4G7/2 and 4G9/2 levels that produce shorter wavelength emission are more active in the SrS phosphor [64]. The difference is unlikely to be an effect of the host lattice symmetry due to the shielded nature of the 4f transitions in rare earths but the lack of high energy electrons would produce weaker visible emission compared to infrared emission [65]. Table 2-2 shows a comparison of the properties of ZnS and SrS [49]. Table 2-2 Properties of ZnS and SrS Itemllb-Vlb compoundlla-Vlb compoundMaterialZnSSrSMelting point (C)1800-1900>2000Band Gap (eV)3.64.3Transition typeDirectIndirectCrystal StructureCubic zinc blende orRock SaltHexagonal wurtzite(NaCl type)Dielectric constant8.39.4Lattice constant ()5.4096.019Ionic Radius ()0.741.13Ionicity0.623>0.785

PAGE 40

26 Figure 2-6 Energy level diagrams and radiant transitions of Tm3+, Nd3+, and Er3+

PAGE 41

27 The host material must also be insulating below its threshold voltage. As discussed in the insulator section, this is to ensure a sufficient voltage drop across the phosphor. In order to maximize the electric field across the phosphor layer, the capacitance of the host should be low. Finally, electroluminescent phosphors are often annealed at temperatures in excess of 500oC so the phosphor host needs to have a melting temperature well above this. In summary, ZnS is an excellent choice for the host material for rare-earth doped NIR emitting EL phosphors. 2.6.2 Luminescent Centers When choosing a host and a luminescent dopant combination care must be taken to ensure that the two are compatible. The size and charge of the dopant will affect its performance in each host. The luminescent dopant should be incorporated into the lattice without creating too many defects, since defects can act as non-radiant relaxation sites and reduce the luminance from the device. Also, the charge of an incorporated ion must be accounted for. If the charge of an ion in a substitutional site is different than the displaced ion, the charge difference must be compensated to maintain charge neutrality for the solid. For this reason charge compensators, such as interstitial F1and Cl1compensating the 3+ rare earth ions substituting for Zn2+ ions, are introduced. Table 2-3 shows a list of common host and luminescent dopant ions for a sulfide based system [49]. The emitted light in ACTFEL devices comes from the luminescent dopant not host material. These centers in phosphors luminesce by one of two mechanisms. The first is through the recombination of electrons trapped in deep donor states and holes trapped in deep acceptor states. The recombination energy depends on the trap depths of the donors and acceptors the as well as band gap of the host because of its effect on the

PAGE 42

28 Table 2-3 Optical properties of common sulfide based EL materials PhosphorEmissionLuminouslayercolorefficiencymaterial1 kHz60 Hz(lm/W)(1 kHz)ZnS:MnYellow50003002-4ZnS:Sm,FReddish-orange12080.05ZnS:Sm,ClRed 200120.08CaS:EuRed200120.05ZnS:Mn/FilterRed1250750.8ZnS:Tb,FGreen21001250.5-1ZnS:Mn/FilterYellow-green130080-----CaS:CeGreen150100.1ZnS:Tm,FBlue2<1<0.01SrS:CeBlue-green900650.44ZnS/SrS:CeBluish-green1500961.3ZnS/SrS:Ce/FilterGreenish-blue220140.2CaGa2S4:CeBlue 21013-----SrS:Ce,EuEggshell-white540320.4SrS:Ce/CaS:EuPaper-white28017-----ZnS:Mn/SrS:CeYellowish-white24502251.3LuminanceL (cd/m2) trap depth. Examples of this type of phosphor are ZnS doped with Al or Cl as donors and Ag, Au, or Cu as acceptors. This type of phosphor is not used in ACTFEL devices because the high electric fields in the phosphor destabilize the traps and sweep the electrons and holes toward opposite sides of the film [66]. The second type of radiative relaxation operates through the electronic transitions of the luminescent ions. This type of phosphor depends on the energies of the ground state and excited states of the individual ions and not the host. The quantum mechanical selection rules governing electronic transitions are important to emission from this type of relaxation. The spin and parity selection rules, governing transitions between states depending on the electronic spin or the symmetry of the stationary state wave function respectively, determine if a particular transition is allowed or disallowed. Essentially, these rules state that transitions are allowed only if they are between states with the same

PAGE 43

29 spin and disallowed for electronic shells with the same reflection property of the waveform, also called parity. In other words transitions between the p and s shells or the f and d shells are permitted, but transitions within a particular shell or transitions between the d and s or f and p shells are forbidden [67]. Luminescent ions in this category include the 3+ ionized rare earths and transition metals ions such as Mn2+, Ag1+, and Cu1+. Trivalent rare earth ions have filled 6s shells, incompletely filled 4f shells, and empty 5d shells. This leads to the two types of excited state to ground state recombination that are observed in rare earths. The first occurs when there are electrons excited from the 4f into the 5d shell, as in Ce3+ and Eu2+. Because the transition is between the d and f orbitals the parity rule is not broken. However, the transition in Eu2+ is spin forbidden while the transition in Ce3+ is not. This means that the decay time of Ce3+ is much faster (several ns) than the decay time of Eu2+ (several s) [39,68]. Since the 5d orbital has a higher energy than the 6s shell, 5d-4f transitions can be strongly affected by the crystal field around the luminescent center and can shift in wavelength depending on the host. The other type of transition is the intra-shell 4f-4f transitions. This type of transition occurs in rare earths such as Tm3+, Nd3+, and Er3+ [39,68]. Because these are intra-shell transitions they are forbidden by the parity selection rule and thus have longer decay times. Since electrons in the 4f shell are shielded by those in the 6s shell, these transitions are relatively well shielded from crystal field effects and characterized by sharper transition lines. 2.6.3 Rare Earth Doped ZnS A variety of techniques have been used to deposit rare earth doped zinc sulfide films, including CVD (chemical vapor deposition), MOCVD (metal oxide chemical vapor deposition), and thermal evaporation [69-71]. At room temperature ZnS exhibits

PAGE 44

30 two crystal structures, a zincblende cubic structure called sphalerite and the hexagonal wurtzite phase. The properties of these phases are given in Table 2-4 [72]. The effect of phase on the luminescent centers is minimal because they have similar properties and symmetry. As mentioned above; Tm, Nd, and Er emit by parity forbidden 4f intrashell transitions. Because of this the spectra from each of these dopants exhibits little change due to crystal field effects and each one typically has luminescent decay times in the ms range [73]. Table 2-4 Physical properties of ZnS ParameterZincblendeWurtziteLattice Constant ( )5.409a=3.814; c=6.258Mass Density (g/cm3)4.084.1Melting Point (K)21002100Heat of Formation (kJ/mol(300K))477-206Specific Heat (J/kg-K(300K))472Debye Temperature (K)530Value 2.6.3.1 ZnS:Tm ZnS doped with Tm exhibits both visible and infrared emission. The NIR peak is near 800 nm while the visible emission is largely in the blue at 480 nm with weaker emission possible in the red region at 650 nm [74]. Photoluminescent excitation of thulium doped zinc sulfide takes place through efficient energy transfer from the ZnS host to the luminescent centers [75]. Sputter deposited films have exhibited the highest photoluminescent infrared to blue intensity ratios [76]. Infrared electroluminescence is often achieved by direct impact excitation of the rare earth by hot electrons [77]. The low hot electron energy in ZnS, as mentioned above, makes it less likely that direct impact excitation will produce higher energy blue light resulting in weaker visible emission. In addition, in a ZnS host, the impact cross section of the 1G4 level, from which blue light is produced, is much smaller than that of the 3F4 energy level, from

PAGE 45

31 which the 800 nm infrared light is produced [78] (Figure 2-7). Due to this difference in excitation mechanisms, ZnS:Tm excited by PL tends to have increased blue luminescence while EL favors the infrared emission. Figure 2-7 Impact cross sections of the 3F4 and 1G4 levels in Tm3+ [78] 2.6.3.2 ZnS:Er Erbium is the rare earth most people think of when considering infrared. This is because of the proliferation of erbium-based infrared telecommunications equipment. Erbium is doped into fiber optic cables for transmission in the fibers absorption minimum at 1550 nm [79]. In addition to the 1350 and 1550 nm lines used for telecommunications, ZnS doped with erbium also emits in the NIR at 990 nm, weakly in the red (660 nm) and strongly in the green (530 nm). Unlike ZnS:Tm in which the infrared emission originates

PAGE 46

32 from a lower energy state than the visible, ZnS:ErF3 green emission at 530nm originates from 4S3/2 (18900 cm-1) with decay to the 4I15/2 ground state while the near infrared at 990nm starts at from the higher 2F7/2 (20100 cm-1) and decays to the 4I11/2 (10100 cm-1) excited state [80]. The emission of NIR light from the higher energy state implies that ZnS is a poor host choice for infrared emitting Er however emission from the 4I11/2 state to the ground level also emits at ~1000nm. 2.6.3.3 ZnS:Nd Neodymium is a rare earth element with four transitions in the near infrared. These transitions have wavelengths of 900 nm, 1060 nm, 1365 nm, and 1800 nm [81]. The transitions responsible for the 900 and 1060 nm emission are from the 4F3/2 to the 4I9/2 and 4I11/2 levels respectively, and the excited state is lower in energy than the 2F11/2 and 4G7/2 states from which the visible emission originate. (Figure 2.6) In previous studies ZnS:Nd was found to have the highest near infrared electroluminescence of any ZnS:RE phosphors [82]. Direct current electroluminescence of ZnS:Nd films has also been found to be more efficient than ZnS:Tm films under similar conditions [83]. 2.7 Electrical and Optical Characterization Optical characterization is useful because the ultimate goal of most ACTFELD research is light output of a specific color at the lowest possible input power. The characterization properties of most interest are brightness versus voltage, power efficiency, and the emission spectrum of these devices. Because the electrical properties of an ACTFEL device are critical to its EL performance, electrical characterization is useful for understanding the fundamental materials properties. Four types of electrical data will be discussed: charge versus voltage (Q-V), capacitance versus voltage (C-V),

PAGE 47

33 internal charge versus phosphor field (Qint-F), and maximum charge versus maximum applied voltage (Qmax-Vmax). 2.7.1 Brightness versus Voltage The luminance of an ACTFEL device is very sensitive to the voltage waveform as well as the drive frequency and amplitude. The most common types of waveforms are sinusoidal or trapezoidal. If a trapezoidal waveform is used the luminescence is dependent on the amplitude of the pulse and the pulse width as well as the rise and fall times of the pulses. Drive frequency has a large effect on the radiant output of an ACTFELD. The emission intensity increases markedly (often nearly linearly) as the drive frequency increases. As the frequency increases there are more pulses per given time yielding more excitation of the luminescent centers and higher brightness. However, the luminescent centers need time to de-excite. If the drive frequency is faster than the luminescent relaxation time, EL output will be saturated and further increases in frequency do not produce increased luminance. Also, the device may heat up at these higher frequencies leading to decreased luminescence due to thermal quenching, i.e. increasing probability of non-radiative recombination [84]. Frequencies of up to 7 kHz have been used but typical drive frequencies are 60 Hz to 2.5 kHz. Visible emitting phosphor brightness is discussed in terms of luminous flux, i.e. the photon flux convoluted with the wavelength response function of the human eye, with units of lumens, nits or candela/m2 [85]. However, infrared phosphor brightness must be specified using the irradiance (W/cm2) of the emitted light, since the human eye response is zero. Brightness versus voltage (B-V) is a typical measure of ACTFEL devices, i.e. the brightness at the wavelength of interest measured at steadily increasing applied voltages. A typical B-V curve is shown in Figure 2-8.

PAGE 48

34 00.0020.0040.0060.0080.010.0120.0140.016050100150200250Voltage (volts)Irradiance (mW/m2nm) Figure 2-8 Brightness vs. voltage curve showing the threshold voltage Of course, the physical properties of the phosphor, and not just the voltage pulse, affect the luminescence of the device. Three properties that have a large effect on the ACTFELD operation are the thickness of the phosphor film, the insulator capacitance (see above), and the concentration of the luminescent dopant. As the phosphor becomes thicker there are more luminescent centers for an accelerated electron to impact. In addition, a thicker phosphor will have a lower capacitance yielding a larger voltage drop across the phosphor layer. Both of these effects cause thicker phosphor devices to be brighter than thinner ones. However, as the thickness increases the operating voltage also increases, which is undesirable because of the size and expense of high voltage supplies. The choice of insulator affects how the voltage is dropped across the device as discussed above. Recall that as the insulator capacitance increases with respect to that of the phosphor, the threshold voltage of the device will decrease because more of the

PAGE 49

35 voltage will be dropped across the phosphor instead of the insulator. Therefore a thin dielectric layer with a high dielectric constant and high breakdown strength is desired. Finally, the doping concentration of the luminescent impurity has a large effect on the emission characteristics of an ACTFEL device. At low concentrations, typically < 1mol% for most materials, the luminance increases steadily with increasing dopant concentration. Once a maximum emission is reached at an optimum concentration of activator, the luminance will decrease with further concentration increases. This decrease has generally been attributed to concentration quenching, which results from interactions between neighboring centers that lead to non-radiative relaxation through self quenching or contact with killer centers such as defects or impurities [86]. 2.7.2 Threshold Voltage The optical threshold voltage, Vth, is the voltage at which the device begins to emit light, and is dependent on several physical properties including the capacitance of the insulator and the thickness of the phosphor. There are several definitions of threshold voltage, but the most common is the voltage axis intercept of the extrapolation of the maximum slope portion of a B-V curve (Figure 2-8). Another common definition of EL threshold is that voltage at which a certain brightness value is achieved, such as 1cd/m2 for visible emitters. For this study the optical threshold voltage will be determined using the first method. A second type of threshold voltage is that where current is transferred across the phosphor in charge versus voltage (Q-V) tests. This is a measure of electrical threshold, and is commonly different from the optical threshold voltage. 2.7.3 Efficiency versus Voltage In addition to the brightness, the power efficiency of a device is an important quantity. Power consumption is a critical concern for any device that needs to use

PAGE 50

36 batteries, e.g. portable displays or sensors. Hence it is desirable to know the light produced per unit power input. For visible emission this is termed the luminous efficiency and is described in lumens per watt. As discussed above, for infrared emitters lumens are inapplicable so power efficiency is expressed as watts of optical output per watt of electrical input. The input power can be determined using ttdttitvAP')'()'(1 where A is the area of the device, is the period of the driving waveform, v(t) is the applied voltage, and i(t) is the current. The output power can be determined by knowing the sensitivity of the detector (determined by using a calibrating light source with a known power output). This output power efficiency can be plotted versus the input voltage of the device yielding an efficiency versus voltage plot (Figure 2-9). Figure 2-9 ACTFELD efficiency versus drive voltage

PAGE 51

37 2.7.4 Electrical Testing The typical circuit used for electrical testing (Figure 2-10) employs a Sawyer-Tower arrangement with either a sense capacitor or a sense resistor [29]. An arbitrary waveform generator is used to produce a voltage pulse that is then amplified and used to drive the circuit consisting of a series resistor, the ACTFEL device and a sense element in series. The series resistor is used to limit the current to the ACTFELD in the case of catastrophic failure. If the sense element is a capacitor then the external charge of the device is measured, while external current can be measured by using a resistor as the sense element. If the capacitor is used its capacitance value must be much larger than that of the device. If a resistor is used as the sense element, its resistance is typically near 100. In the case of a resistor, if the resistance is too large, the dynamic response of the device will be delayed as the RC time constant of the circuit increases. Figure 2-10 Schematic of a Sawyer-Tower test setup

PAGE 52

38 The typical waveform used for testing is a bipolar trapezoidal waveform as discussed above and shown in figure 2-11. The pulses have a rise and fall time of ~5s, a 30s plateau, and a frequency between 60 Hz and 2.5 kHz. The labels A-J are used to designate important points during the cycle. Most of the points are self explanatory except for points B and G. These points are to designate the electrical threshold voltages, i.e. the voltages at which the phosphor begins to conduct charge. This labeling scheme is common in the literature and will be used for throughout this document in discussions of electrical properties and the matching of points on Q-V, Q-F data curves with points on the driving waveform [29]. Figure 2-11 Trapezoidal waveform with important points marked for reference 2.7.5 Charge versus Voltage (Q-V) The most basic measure of the electrical characteristics of ACTFEL devices is the charge versus voltage (Q-V). A Q-V plot displays the charge stored between the external terminals of the capacitive ACTFEL device versus the voltage across the terminals.

PAGE 53

39 When the device is driven below the electrical and optical threshold voltages, the plot is simply a straight line with a slope equal to the total capacitance of the device (assuming leakage current is negligible). When the device is driven above electrical and/or optical threshold, the Q-V plot becomes a hysteresis loop due to the dissipative charge conduction through the device. Hence, the electrical threshold can be determined as the voltage at which the plot becomes hysteretic. The voltage drop across the ACTFEL device can be found from V2 and V3 by using )()()(32tVtVtVEL When using a sense capacitor, the external charge can be determined using )()(3tVCtqsexr where Cs is the sense capacitance. When using a sense resistor, it is first necessary to calculate the current passing through the device. Since the current through each element will be the same, the current through the ACTFEL device can be determined using seriesRtVtVti)()()(21 The external charge can then be found by integrating this current over time, such that textdttitq0)()( An example Q-V plot using the labeling scheme from above is shown in figure 2-12. The voltage labeled Vto in figure 2-12 is the turn on voltage of the device, which is different from the electrical threshold voltage due to the polarization charge, Qpol. The plus and minus superscripts are to signify which occurred during the positive waveform pulse and which occurred during the negative pulse. Positive is defined as when the voltage pulse is applied to the Al electrode and negative is when the pulse is applied to

PAGE 54

40 the transparent electrode. The polarization charge is the result of charge buildup at the phosphor-insulator interface creating a charge imbalance in the phosphor at the conclusion of a pulse. The polarization charge may help the next pulse because the built-up charge at the phosphor-insulator interface creates an electrical field that is the same polarity as that of the pulse. The threshold voltage may be defined as )(lim0poltoQthQVVpol The polarization charge may be reduced by the leakage charge, Qleak, over the time between voltage pulses (during segments EF and JA). Leakage charge results from electrons escaping relatively shallow states due to the polarization field. Figure 2-12 Typical Q-V plot The other types of charge occur during the voltage pulses. The conduction charge, Qcond, is the charge conducted through the ACTFEL device from device turn-on until the

PAGE 55

41 end of the voltage plateau (segments BD and GI). The conduction charge is essentially the charge flow responsible for impact excitation or ionization of the luminescent centers, assuming the charge conduction between electrical and optical thresholds is small. Qrelax is the relaxation charge, the charge that flows during the plateau portion of the voltage pulses (segments CD and HI). The term relaxation charge is used because the flow of this charge, at a constant voltage, sets up an electric field opposed to the total field across the phosphor. Finally, the maximum charge, Qmax, is the charge at the maximum voltage measured across the sense element for a given applied voltage. Qmax differs from the other charges discussed in that it is taken in reference to the zero point of the charge axis instead of being referenced to a specific point of the driving voltage waveform. Qmax is a useful term for evaluating transferred charge, as will be shown later. The Q-V curve is useful for determining several parameters of an ACTFEL device. First, as discussed above, below the turn-on voltage the slope of the Q-V curve is a measure of the total device capacitance. Second, above the turn-on voltage the phosphor is assumed to be a conductor, which means that the Q-V slope can be used as a measure of the capacitance of the devices insulator layer(s). Care must be taken when determining this because several factors may skew this measurement in a real device. If the phosphor is not completely shorted then the slope will be less than the insulator capacitance due to some remaining phosphor capacitance and/or resistance. If there is a build-up of space charge in the phosphor, the slope may be larger than that of the insulator capacitance [28]. Third, the area inside the Q-V curve is proportional to the input electrical power density delivered per pulse [87].

PAGE 56

42 2.7.6 Capacitance versus Voltage A capacitance versus voltage (C-V) plot allows the measurement of the dynamic capacitance of an ACTFEL device against the voltage across the terminal during the rising edge (segments AC and FH) of the voltage pulse. C-V plots are derived from the data obtained during measurements for Q-V; the slope of the Q-V plot during the rising edges of both pulses is plotted. The capacitance is calculated using dttdvtivC)()()( where i(t) is the current derived using the inverse of the previous equation for dqext(t), dttdqtiext)()( These reduce to the following formula, )()()(tdvtdqvCext An example of a C-V plot is shown in figure 2-13. The total physical capacitance, CtCV, is the capacitance of the whole device below the turn-on voltage and is usually in good agreement with the capacitance calculated from dielectric constant and film thickness measurements, Ctphys. Ciphys, the capacitance above turn on is simply that of the insulators, since the phosphor is shorted (broken down) at higher voltages. CiCV, is the measurement of the capacitance above turn-on and, in the ideal case, should equal the calculated insulator capacitance. Usually, however, the value of CiCV is either above or below the value of the ideal case. CiCV less than the calculated value occurs when the phosphor does not completely short, resulting in some remnant capacitance. CiCV can be larger than Ciphys if there is dynamic space charge built up in the phosphor that decreases

PAGE 57

43 the total phosphor field, as discussed previously. Space charge can also lead to CV overshoot, a sudden increase followed by a decrease at higher voltages in the capacitance when the device turns on. Figure 2-13 Typical C-V plot 2.7.7 Internal Charge versus Phosphor Field Internal charge versus phosphor field (Qint-Fp) is a technique used to provide information field clamping or charge relaxation, which are difficult properties to determine from a Q-V plot [88]. In a Q-V plot the total charge (capacitive plus phosphor charge) and total device voltage are studied, not just the charge and voltage across the phosphor. The Qint-Fp data are the charge transported in the phosphor layer and the electric field in the phosphor layer only. The internal charge of the phosphor can be determined from

PAGE 58

44 )()()()()(32tvtvCtqCCCtqpextipi where q(t) is the internal charge, Ci and Cp are the insulator and phosphor capacitances, respectively, qext is the external charge, and v2 and v3 are the voltages measured on each side of the device. The phosphor field is obtained using )()()(132tvtvCtqdfiextpp where dp is the phosphor thickness. These equations basically use the raw data from a Q-V curve, remove the capacitive displacement charge, and remove the voltage drop across the insulators to calculate the field across the phosphor. The equations are developed from the equations used to describe an ACTFELD with a phosphor layer free from space charge [89]. A typical graph of Qint-Fp is shown in figure 2-14. Unlike a Q-V plot the Qint-Fp loop goes in a clockwise direction. A Qint-Fp plot shows several of the same quantities as in a Q-V plot, but these are only charges and fields in the phosphor. The charge information shown is Qcond, the conduction charge transported across the phosphor, Qpol, the polarization charge stored at the phosphor/insulator interface, Qleak, the leakage charge between the voltage pulses, Qrelax, the relaxation charge flowing during the voltage pulse plateau, and Qmax, the maximum charge across the phosphor (Figure 2-14). The other information available is the steady state field, Fss. Field clamping, when the charge flow through the device is sufficient to counteract the increasing field generated by increasing the applied voltage, can be determined by comparing Fss at different voltages above the threshold voltage. If there is field clamping, then Fss will be independent of voltage above threshold. Some devices

PAGE 59

45 demonstrate field overshoot because of dynamic space charge effects that are usually manifested around points B or G in the Qint-Fp plot [90]. Figure 2-14 Typical Qint-Fp plot Reduction of Q-V data to Qint-Fp data depends on knowing the capacitance of the phosphor and insulator(s) as well as the thickness of the films. Uncertainty in the capacitances results in distortion in the Q-Fp plots. The capacitance values can be refined using their effects on the shape of these plots [91]. Often the capacitance values are adjusted to obtain a vertical slope for the BC and GH portions and a horizontal slope for the ED and IJ segments of the plot (Figure 2-14). If there is a large amount of space charge then the capacitance values will be larger than the physically measured values. If the phosphor thickness value is incorrect, there is inaccuracy in the phosphor field. So, great care must be taken when using a Qint-Fp plot versus Q-V or C-V plots to make sure that the data are meaningful.

PAGE 60

46 2.7.8 Maximum Charge versus Maximum Voltage The final electrical characterization technique, maximum charge-maximum voltage (Qmax-Vmax), measures the transferred charge at the maximum pulse amplitude for several applied voltages [92]. Typically there are two types of charge measured, the internal charge, Qmax, and the external charge, Qemax. The internal charge values can be taken from the maximum charge points (points D and I) in the Q-Fp plot, while the external charge values, known as a Qemax Vmax or Q-V plot, can be taken from the maximum charge values of the Q-V plot [87]. Figure 2-15 shows a typical Qmax-Vmax plot for an ACTFEL device. Often charge values from both the positive and negative pulses are plotted simultaneously to determine if the charge transfer in the device is symmetric. Care must be taken because the accuracy of the Qmax values is affected by the accuracy of the phosphor and insulator capacitances. The Qmax-Vmax plot looks similar to a B-V curve and is a measure of the internal charge needed for a desired brightness. Qemax-Vmax data are more reliable than Qmax-Vmax data because it is directly measured so capacitance inaccuracies are unimportant. Above threshold, the slope of the Qemax-Vmax plot is proportional to the insulator capacitance. If the slope is too small then there is insufficient transferred charge. If the slope is too large then there is more than the expected transferred charge, possibly caused by dynamic space charge. The voltage derivative of Qemax-Vmax (Figure 2-16) is a direct measure of the capacitance and can give information about how the capacitance changes with voltage. Dynamic space charge can lead to C-V overshoot which is easily seen in this type of plot.

PAGE 61

47 Figure 2-15 Typical Qmax-Vmax plot Figure 2-16 Typical Qemax-Vmax plot

PAGE 62

CHAPTER 3 EXPERIMENTAL PROCEDURE 3.1 Substrate and Target Preparation Zinc Sulfide doped with rare earth fluoride thin films were deposited onto 2.5 x 5 cm 7059 glass coated with 360 nm of a polycrystalline indium tin oxide (ITO) transparent conducting electrode and 160 nm of amorphous aluminum titanium oxide (ATO) transparent dielectric layer obtained from Planar Systems. The substrates were cleaned in a UVOCS Inc. ultraviolet light ozone cleaner for six minutes in air to remove organic contaminants. They were then blown clean with dry nitrogen to remove any particles. In addition, substrates (2.5 x 5 cm) of bare 7059 glass were prepared by the same methods for simultaneous coating, the used for film thickness measurements and destructive testing. Several targets were used for deposition of the doped ZnS films. All of the doped targets were pressed powder, were 5 cm in diameter, 0.65 cm thick, and were manufactured by Target Materials Incorporated. The targets included ZnS doped with 1.5mol% of either 99.9% pure TmF3, NdF3, or ErF3. In addition, a CVD grown plate of pure, undoped, dense ZnS from Morton Thiokol was cut with a diamond saw and used as an undoped target. All targets were conditioned for one hour after any break in the vacuum of the deposition system. 3.2 Sulfide Sputter Deposition System Films were deposited by RF planar magnetron sputtering in a high vacuum chamber using a Leybold Trivac rotary vane pump for backing and roughing, and a Leybold 1600W magnetic levitation turbomolecular pump with a Leybold Mag.DriveL 48

PAGE 63

49 controller. The ultimate pressure of the system varied between 6 x 10-7 and 2 x 10-6 Torr as measured by a hot filament ionization gauge. The system is designed to run as many as three sputter sources simultaneously. An Angstrom Science Onyx 2 magnetron sputtering gun was used for the undoped target and an AJA A300 magnetron sputtering gun was used for all of the doped targets. During dual rare earth depositions the Angstrom Science gun held the thulium doped target. The target face of the Angstrom Science gun was 10 cm from the substrates while the target face of the AJA gun was 5 cm away. Power to each gun was supplied by an RFPP RF5S radio frequency controller with an RFPP matching network. Duty cycles of 25, 50, 75, and 100% were used with a consistent pulse width of 40 milliseconds and varying delays between each pulse.. RF power was set to 120 watts in all cases. By independently varying the duty cycles used for two targets (e.g. undoped ZnS and ZnS:RE), the concentration of the rare earth (RE) fluoride in the thin films can be varied. A schematic of the deposition chamber is shown in figure 3-1. The substrates were held on a multi-sample platter consisting of four 2 x 2 sample mounting positions that was rotated at 11 seconds per cycle to ensure that the film deposited at each sample position was identical. A schematic of the sample holder and sample positions is shown in figure 3-2. Deposition rates varied from 4.0 to 12.5 nm/min depending on the duty cycles of the targets and the substrate temperature. Deposition times were varied to maintain film thicknesses between 0.2 and 0.6 m, depending on the experimental run. These parameters resulted in deposition times from 50 minutes to 220 minutes, depending on the target materials and duty cycles.

PAGE 64

50 Figure 3-1 Schematic of the sputter system used for RF magnetron sputtering

PAGE 65

51 Figure 3-2 View of sample platter showing substrate positions and spaces for additional substrates Ultra high purity argon was used as the sputter deposition gas. The gas was introduced into the chamber using Unit UFC 1100A 20, 50, and 100 sccm mass flow controllers for three inlet lines. The argon pressure was regulated using the flow controller and a throttle valve before the turbo pump. Using this method, the argon pressure was maintained at 2x10-2 Torr measured by a baritron capacitance gauge. The substrates were radiatively heated by an array of resistive carbon cloth filaments. A graphite plate was situated above the heater clothes and used as the seat for

PAGE 66

52 the sample platter. The platter holding the samples rested on four 1.25 cm high ceramic feet to reduce conductive heating. The sample positions were square holes in the platter with small ledges for the samples to rest on two sides. See figure 3-3 for a schematic of the heating system. Deposition temperatures were measured by a thermocouple positioned just above, but not contacting, the platter surface. Figure 3-3 Schematic of the heating system in the sputtering system 3.3 Top Contact Deposition All of the devices tested were of the half stack configuration, i.e. no top dielectric layer was deposited. A stainless steel shadow mask was used to create an array of aluminum contacts directly on the phosphor surface. The aluminum was thermally

PAGE 67

53 evaporated onto the phosphor using an Edwards Coating System E306 thermal evaporator (base pressure of 1x10-5 Torr). The aluminum contacts were 0.3 cm diameter circles between 190 and 250 nm thick. The bottom conductor was the ITO layer. The ITO was buried under the ATO and phosphor layers so these layers needed to be removed before the bottom contact can be connected. To achieve contact, the phosphor layer and the ATO layer were removed by scratching with a diamond scribe. Once the layers were removed, a multimeter was used to test conductivity in the scratched area to ensure that the ITO layer was exposed. Once ITO contact was confirmed, an indium wire was melted with a soldering iron into the scratched area to create a contact to the ITO. 3.4 Sample Handling and Storage Substrates and incomplete devices (devices with no top contact) were stored in a nitrogen cabinet under a steady dry nitrogen flow. Completed devices were stored either in the nitrogen cabinet or in a standard cabinet under normal room humidity. Time delays from a day to over a year occurred between deposition of the phosphor and deposition of the top contact. Varying time delays also occurred between sample completion and device testing. All samples were handled with latex or nitrile gloves and/or with tweezers. Note that storage in a humid environment after device completion and the various time lags during device construction did not appreciably affect device performance. 3.5 Sputtered Film Characterization The sputtered films were characterized using a variety of techniques including optical interferometry, x-ray diffraction (XRD), electron microprobe (EMP), energy dispersive x-ray spectroscopy (EDS), photoluminescence (PL), photoluminescent

PAGE 68

54 excitation (PLE), electroluminescence (EL), time resolved electroluminescence, and electrical measurements. The details are provided below. 3.5.1 Thickness Measurements Optical interferometry [93] was used to measure the thickness of each deposited film. The films deposited on the bare 7059 glass substrates were used to avoid interference from the ITO/ATO layers. The index of refraction of the film (2.5) and the substrate (1.5) is known. Upon shining a beam of light onto the sample, interference patterns will be created from reflection at the air-film and film-substrate interfaces. The frequency of the interference fringes is dependent on the thickness of the film and the optical index. Using an in-house developed Excel macro, the film thickness can be determined by curve matching a calculated pattern to the experimental pattern. 3.5.2 X-ray Diffraction (XRD) X-ray diffraction [94] was used to evaluate the ZnS crystallinity. The diffractometer was a Phillips model APD 3720 operated at 40 kV and 20 mA. The wavelengths used were from Cu K lines at 0.15406 and 0.15444 nm. The Cu K was blocked using a nickel filter. The diffractometer was scanned over the range of 26.5o to 31.5o to encompass the primary emission peak of both cubic and hexagonal ZnS at 28.5o. The goniometer scanned 0.01o per second with a step size of 0.01o. X-ray diffraction is used primarily to determine phase of a material but it may also be to determine crystal size, strain of the lattice, film thickness, and semi-quantitative composition analysis [95]. These parameters can be extracted from the diffraction peak intensity, width and position. Atoms can scatter x-rays, other photons, and electrons. Diffraction consists of the constructive and destructive interference of the scattered wave. Constructive interference

PAGE 69

55 results in a diffraction signal causing an intensity peak while destructive interference results in no signal. Constructive and destructive interference is the result of the periodically arranged atoms in a crystalline solid. The atomic alignment necessary to cause constructive interference is defined by Braggs law n = 2dhkl sin where n is the order of the diffraction (typically 1), is the wavelength of the incident radiation, dhkl is the spacing between the atomic layers with Miller indices of (hkl), and is the angle between the beam of the incoming radiation and the normal of the plane of atoms [96]. ZnS has two crystal structures, a cubic structure commonly called sphalerite and a hexagonal structure called wurtzite. The crystal planes that can produce constructive interference vary with each crystal structure. For example, face centered cubic lattices, such as sphalerite, can only produce reflections if the indices are all even or all odd [97]. Sphalerite has an intense diffraction signal from the (111) plane at 28.58o. Wurtzite has an intense diffraction signal from the (100) plane at 26.94o and another intense peak at 28.53o from the (002) plane. If the films are thinner than the penetration depth of the x-rays (typically a few microns for ZnS) the peak heights will be artificially adjusted if the films are not all the same thickness. Due to the thinness of the deposited films in this study (<1 m) and the penetration depth of the x-rays, diffraction scans of films deposited on ATO/ITO substrates also exhibit diffraction peaks from ITO. Since there is variation in the film thickness from sample to sample the full width at half maximum (FWHM) of the peaks is used to compare the crystallinity of the films. As crystallinity decreases the FWHM of the peaks increases until, in the case of an amorphous material,

PAGE 70

56 the XRD pattern appears as a series of low broad undulations. In addition, the peak position can be used to determine if the film is strained because strain will cause an increase or decrease in the interatomic distance which, using Braggs law, will affect the value of [98]. 3.5.3 Electroluminescence Electroluminescent brightness was measured using various detectors depending on the wavelength range. The excitation source was a custom built driver based on a design by Planar Inc. The EL driver produced trapezoidal voltage pulses that had a rise time of 5 microseconds, a plateau width of 5, 30, or 800 microseconds (typically used at 30 microseconds), <5 microsecond fall time, and a frequency of 2.5 kilohertz. The high voltage for the driver was supplied by a Sorensen DCS 600-1.7 high voltage power supply. The current from the supply was limited to 0.025 amps and the voltage to 300 volts. The input pulses traveled through a 125 ohm resistor positioned before each terminal of the device. The sample to be measured was placed on the sample holder as shown in figure 3-4. The sample was placed on a glass slide attached to a mounting card and held in position by pogo pins that also acted as leads to the device. The pogo pins were connected to terminals on the card that was then placed into a card holder attached to an x-z translation stage for alignment with the detector. The detector for 350 to 1200 nm was an Ocean Optics S2000 silicon CCD with Ocean Optics spectroscopic grating #13 installed. (See Figure 3-5 for the response of grating #13.) The data were processed by computer using OOIBase32. OOIbase32 is a program written by Ocean Optics Inc. to gather and process data received by the Ocean Optics detectors. OOIbase32 collects and displays spectral data in real time over a range from 200 nm to 1600 nm with integration times as short a 5 ms. Other detectors, used

PAGE 71

57 mainly for time resolved electroluminescence and described below, included an Oriel 77341 photomultiplier for visible emission and an Oriel 71654 germanium detector for near infrared emission. Calibration of the silicon CCD and photomultiplier tube was done using an Oriel 63358 45W tungsten halogen calibrated lamp. Calibration of the germanium detector was done using a 99.9+% efficient blackbody source. Figure 3-4 Back view of the sample on the test stage The light path from the sample to spectrometer was an Ocean Optics VIS-IR optical fiber with an attached 74-VIS collimating lens. The card and translation stage assembly were installed in a test housing designed to minimize stray light. For a

PAGE 72

58 schematic of the test stage assembly see figure 3-6. Other detectors are described in the following section. Figure 3-5 Spectral sensitivity of the Ocean Optics #13 grating

PAGE 73

59 Figure 3-6 Side view of the sample stage and fiber optic detection system 3.5.4 Photoluminescence and Photoluminescent Excitation Photoluminescent brightness was measured using the same detectors used for electroluminescence [99]. The excitation source was an Oriel model 66902 lamp with a 300W xenon bulb. Broadband light from the xenon lamp was monochromatized by an Oriel Cornerstone 74100 spectrometer with 3 mm slits. Emitted light was focused on the entrance slits of an Oriel MS257 monochromater. An Oriel 77265 photomultiplier tube was used for detecting visible and near ultraviolet emission from 300 to 800 nm. The detector used from 800 nm to 2m was a germanium detector. The detector for 2 to 5 um was a thermoelectrically cooled lead selenide detector. Signal detection and chopping for noise reduction was controlled by an Oriel Merlin control unit. Traq32, a program

PAGE 74

60 created by Oriel Inc., controlled the MS257 and Cornerstone spectrometers. Traq32 was written specifically to control Oriel spectrometers and to collect and process data. Using Traq32, all spectrometer functions and data acquisition parameters can be specified. Unlike like silicon detector discussed above, data are collected by Traq32 by scanning the wavelength range, not at all wavelengths simultaneously. Data from Traq32 and OOIbase32 can be easily imported into Microsoft Excel for data processing and analysis. 3.5.5 Electron Microprobe The electron microprobe [100] was one method used to determine film composition. A JEOL Superprobe 733 was used. Primary electrons were generated by thermionic emission from a tungsten filament. The operating voltage was 8 kV. Since the samples were on nonconductive bare glass substrates or on ITO/ATO substrates. Because high beam currents are used during microprobe analysis (~20 nA) all of the samples including the samples with ITO were evaporation coated with carbon to prevent charging. For electron microprobe analysis (EMPA), characteristic x-rays generated from the inelastic ionizing collisions of electrons in the sample are used to quantitatively determine elemental concentrations. The X-rays may be energy analyzed using dispersion by wavelength (wavelength dispersive spectrometry-WDA) or energy dispersion (EDS). For this study energy dispersive analysis was used but at higher currents, as discussed above, than the EDS analysis discussed below. The microprobe data are quantitated based upon materials standards for the desired elements. Film compositions were also determined with a x-ray spectrometer on an SEM as detailed below.

PAGE 75

61 3.5.6 Energy Dispersive X-ray Spectroscopy (EDS) on a Scanning Electron Microscope (SEM) EDS was used to verify film composition. A Hitachi S450 SEM with a Princeton Gamma-Tech Prism digital spectrometer as the x-ray energy analyzer was used. Primary electrons were generated by thermionic emission from a tungsten filament. The operating voltage was 20 KV. The minimum usable voltage (10 KV was set by the fact that the L line emission from the rare earths require this energy to be excited. 20 KV was used, even with a greater penetration and excitation depth, because of reduced analysis errors as compared to those found when using the lower accelerating voltage with rare earths. The samples measured were on ATO/ITO substrates and the ITO was sufficiently conductive to not require surface coating but the samples were daubed along their edge with carbon paint to make electrical contact with the sample holder and reduce charging from the sample. Collection time was twenty minutes to ensure high enough signal to noise. Rare earth and rare earth fluoride standards were used as references for determining the rare earth and fluorine concentration in each sample. For the other elements, standardless quantification was used. The high current electron microprobe analysis and the low current EDS use the x-rays produced from atomic ionization induced by high energy electron bombardment. Inelastic scattering of the energetic electron causes an inner shell electron to be ejected from the atom. When an outer shell electron de-excited to fill the inner shell hole either an Auger electron or a characteristic x-ray will be emitted. For EDS, the emitted x-rays are collected by a silicon diode producing a charge pulse proportional to the energy of the incident x-ray. These pulses are then amplified and processed to produce an energy spectrum of the incoming x-rays [101].

PAGE 76

62 3.5.7 Time Resolved Electroluminescence Time resolved electroluminescence [102] was performed with an experimental setup similar to that of photoluminescence measurements. The sample was placed in the same position used for photoluminescence; however the sample was excited using the EL driver and sample holder described in the electroluminescence section. A Tektronics 2024 digital oscilloscope or Tektronics TDS 3014 B digital oscilloscope was added to the setup in the following manner. Channel one, called V1, of the oscilloscope was connected before the resistor to the positive input terminal of the sample holder. Channel two, called V2, was connected to the positive side of the holder after the resistor. Channel three, called V3, was connected to the negative side of the device after the resistor. Channel four of the scope was connected to the detector that was required using a splitter and BNC cable to connect the detector to the oscilloscope and Merlin detection system simultaneously (Figure 3-7).

PAGE 77

63 Figure 3-7 System to measure time resolved luminescence and electrical data 3.5.8 Electrical Measurements Electrical data were taken with the samples in position for electroluminescence measurements. Leads from the oscilloscopes were connected in the same manner as for time resolved electroluminescence measurements. Using V=IR and the known resistance, the current through the device can be determined by subtracting the value of V2 from V1. Using the setup shown in figures 3-7 and 2-10, V3 corresponds to the current through the sample when divided by the value of the sense resistor and this was verified by subtracting the signal of V2 from that of V1. The sense resistor was a 125% ohms. The PMT was connected directly to channel four of the oscilloscope when time resolved measurements were made. The horizontal resolution of the scopes was set to either 40 or 50 microseconds per division. This resolution provided information on either

PAGE 78

64 a positive or negative pulse. The trigger value was 20 volts on the positive pulse edge. The vertical resolution was dependent on the voltage of the pulses or the signal from the PMT. The data was either sent to a computer via a GPIB cable or saved directly to disk in the oscilloscope. The data was processed using Excel. The processing included determining the external charge of the device during operation. The charge was determined by integrating the current through the device over time. In addition, the capacitance and electric field in the device were determined by further processing of the data as detailed in section 2.7.

PAGE 79

CHAPTER 4 PHYSICAL EFFECTS OF CHANGING TARGET DUTY CYCLES AND SUBSTRATE TEMPERATURE DURING RF MAGNETRON SPUTTER DEPOSITION 4.1 Introduction In this chapter, the data on the effects of deposition conditions of ZnS:[RE]F3, where RE is Tm, Nd, and Er, are presented. The objective of this study was to determine the effects of sputter deposition parameter changes on infrared electroluminescent intensity and to compare results from various rare earth dopants to draw trends to apply to other lanthanides. It was found that changing the substrate temperature and the sputtering target duty cycles modified several structural properties of the phosphors that affect the infrared and visible emission. Duty cycle changes are listed as 100 multiplied by the ratio between the duty cycle of the target doped with 1.5% rare earth fluoride to the total duty cycles of the doped target and the undoped target. So a ratio of 50 means that each of the targets was sputtering 100% of the time (100/(100+100) = .5 x 100 = 50) while a ratio of 33 means that the doped target was sputtered 50% of the time while the undoped target was sputtered 100% of the time (50/(100+50) = .33 x 100 = 33). The substrates were heated so that the thermocouple described in chapter 3 measured temperatures ranging from 130 oC to 190 oC. 4.2 Spectra None of the as-deposited phosphors exhibited photoluminescence. The xenon lamp used as an excitation source was not intense enough to produce luminescence from this condition. However, typical electroluminescence spectra obtained for as-deposited ZnS 65

PAGE 80

66 doped with Tm, Nd, or Er are shown in figures 4-1 to 4-3. The spectrum from ZnS:TmF3 has two major peaks at 480 nm and 800 nm and one minor peak at 650 nm. These correspond to the 1G4 3H6, 3F4 3H6, and 3F3 3H6 transitions, respectively. The ZnS:NdF3 spectrum exhibits one major visible peak at 600 nm and two major NIR peaks 890 nm and 1080 nm as well as several minor peaks. The major peaks are from the 2H11/2 4I9/2 for the visible emission and the 4F3/2 4I9/2 and 4F3/2 4I11/2 transitions for the NIR emission. The ZnS:ErF3 phosphor has several major peaks. The emission at 530, 550, 660, and 1000 nm correspond to the 2H11/2 4I15/2, 4S3/2 4I15/2, 4F9/2 4I15/2, and 4I11/2 4I15/2 transitions respectively. The energy levels and transitions are shown in figure 4-4 00.00050.0010.00150.0020.00250.003300400500600700800900100011001200Wavelength (nm)Irradiance (mW/m2nm) Figure 4-1 Electroluminescent spectrum of ZnS:TmF3

PAGE 81

67 00.0020.0040.0060.0080.010.012300400500600700800900100011001200Wavelength (nm)Irradiance (mW/m2nm) Figure 4-2 Electroluminescent spectrum of ZnS:NdF3 00.00020.00040.00060.00080.0010.00120.00140.0016300400500600700800900100011001200Wavelength (nm)Irradiance (mW/m2nm) Figure 4-3 Electroluminescent spectrum of ZnS:ErF3

PAGE 82

68 Figure 4-4 Energy levels of rare earth ions and transitions luminescence producing transitions observed in Figs. 4-1, 4-2 and 4-3.

PAGE 83

69 4.3 Target Duty Cycle Alteration Changes in the duty cycles of the sputtering targets affect the infrared emission intensity of the ACTFEL devices. The possible duty cycles for both the undoped target and the rare earth doped target were 100%, 75%, 50%, or 25%. If one target was set to a duty cycle below 100% then the other was set to be on 100% of the time. The duty cycles are listed as the ratio of doped target on time divided by on times of the doped and undoped targets. The concentration of rare earth corresponding to each of the duty cycle ratios is different for each rare earth and is discussed in the following section. 4.3.1 Concentration The effect of duty cycle on the concentration of the individual rare earths, as tested by EDS on the SEM and EPMA, is shown in figure 4-5. The trend was for the concentration of each of the rare earths to increase as the relative duty cycle on the doped target increased. As the duty cycle was changed the concentration of thulium in the phosphor increased from 0.6 at% to 1.4 at%. As with the thulium doped samples, increasing the duty cycle increased the neodymium concentration in the phosphor. The Nd concentration rose from 0.55 at% to over 2.0 at%, while the concentration of Er in the ZnS film exhibited the least change with changing duty cycle. 4.3.2 Crystallinity The full width at half maximum (FWHM) of the 28.5o x-ray diffraction peak of ZnS, which is observed from both the sphalerite (from the 111 plane) and wurtzite (from the 002 plane) phases of ZnS, was used to characterize the crystallinity of the ZnS:[RE]F3 films. The FWHM increased for all of the films as the rare earth doped targets duty cycle increased indicating that the host became less crystalline with increasing rare earth concentration. The Tm and Er doped films experienced an increase in the FWHM of the

PAGE 84

70 ZnS peak of over 30% while the data for the Nd doped films are too sparse to detect a trend (Figure 4-6). 00.511.522.520304050607080Duty Cycle Ratio (doped/total)Rare Earth Concentration (at%) Tm Nd Er Figure 4-5 Effect of target duty cycle on the Tm, Nd, and Er concentrations in the ZnS films measured by EDS and EPMA

PAGE 85

71 0.20.30.40.50.60.70.80.9120304050607080Duty Cycle ratio (doped/total)FWHM (deg.) Nd Tm Er Figure 4-6 Effect of duty cycle ratio on the full width at half maximum of the 28.5o x-ray diffraction peak of ZnS 4.3.3 Thickness The undoped target was further away from the substrate (8 cm) than the doped targets (6 cm) yield resulting in a slower deposition rate for the pure material. In addition, the sputter process changes the surface morphology of the targets as material is sputtered causing the deposition rate to change slightly (~10%) from one deposition to the next. For each film, the deposition time was changed in an effort to maintain a uniform thickness between the samples of the same material. This effort was successful for the Tm and Er doped films, however there was a large difference in thickness for the Nd doped phosphors. Figure 4-7 shows the film thicknesses normalized to the thickest film for each material and shows that the film thicknesses were usually within 5% of the

PAGE 86

72 average for ZnS:Tm and ZnS:Er however, there was a large discrepancy in ZnS:Nd thicknesses. 0.30.40.50.60.70.80.9120304050607080Duty Cycle Ratio (doped/total)Normalized Thickness Nd Tm Er Figure 4-7 Normalized thickness of the rare earth doped ZnS films. Deposition times were changed to attempt to achieve the same thickness for each rare earth film. 4.4.4 Threshold Voltage The NIR optical threshold voltage of each of the materials is shown in figure 4-8. As will be shown in Chapter 5, the turn on voltage for infrared and visible emission is identical. The turn on voltage for the Tm doped samples rose slightly as the Tm target duty cycle increased but the majority of samples maintained a turn on voltage of approximately 100 volts. The Nd doped films exhibited a turn on voltage near 200 volts for the lower duty cycle ratios, but decreased to 130 volts for the higher duty cycle ratios.

PAGE 87

73 The turn on voltage for the Er doped devices was consistently 110 volts except for the lowest duty cycle ratio. 50709011013015017019021023020304050607080Duty Cycle Ratio (doped/total)Turn On Voltage (volts) Tm Nd Er Figure 4-8 NIR threshold voltages of the doped ZnS films with varying deposition duty cycles 4.4.5 Infrared Emission Alteration of the target duty cycles had a large effect on the emission intensity of the near infrared emission. The effect of duty cycle on the different materials is shown in figure 4-9. The brightness of the near infrared peak was highest for each of the rare earths near the 50 ratio. The Tm emission maximum was at a duty cycle ratio of 57 and the intensity decreased as the duty cycle ratio decreased. In contrast, the maximum Nd and Er doped phosphor brightness were at lower duty cycle ratios and exhibited rapid declines in infrared emission as the duty cycle ratio increased. There were similar trends for the visible emission from each phosphor.

PAGE 88

74 00.10.20.30.40.50.60.70.80.9120304050607080Doped to total ratio (x100)Normalized Intensity Tm Nd Er Figure 4-9 Effect of target duty cycle on the near infrared emission of each rare earth 4.5 Deposition Temperature Effects The substrates were radiatively heated by resistive carbon cloth heaters located below the sample stage to temperatures between 130oC and 190oC. The duty cycle ratio that produced the brightest infrared emission at a substrate temperature of 160 C was used for each of the rare earth dopants to study the effects of varying the substrate temperature. In addition, the deposition time for each material was the same (50 min for Tm and Er and 120 min for Nd) at each of the deposition temperatures. 4.5.1 Concentration The effect of deposition temperature on the concentration of the different rare earth dopants is shown in figure 4-10. As the temperature of the substrate was increased the concentration of thulium, as tested by EDS and EPMA, in the deposited phosphor film increased from below 0.5 at% to over 2 at%. The concentration of Tm rose steadily

PAGE 89

75 between 130oC and 170oC with a sharp increase at 180oC. As with the thulium doped samples, increasing the deposition temperature increased the neodymium and erbium concentrations in the phosphors. The Nd concentration rose from below 1 at% to 1.5 at%. The concentrations of Er rose from 0.5 at% to 1.5 at% between 140oC and 190 oC. The Nd and Er concentrations experienced sharp rises at the higher tested temperature, similar to the thulium doped films. 00.511.522.53130140150160170180190Deposition Temperature (Deg. C)Rare Earth Concentration (at%) Tm Nd Er Figure 4-10 Concentration of each rare earth in the ZnS films as a function of substrate temperature during deposition measured by EDS 4.5.2 Crystallinity The full width at half maximum (FWHM) of the 28.5o x-ray diffraction peak of ZnS, observed from both the sphalerite and wurtzite structures, increased for all of the films as the deposition temperature increased indicating that the host became less

PAGE 90

76 crystalline at higher temperatures. The Tm and Nd doped phosphors experienced an increase in the FWHM of the ZnS peak of 30% while the FWHM of the Er doped phosphor increased by 50% (Figure 4-11). 0.30.40.50.60.70.80.911.11.2130140150160170180190Deposition Temperature (Deg. C)FWHM (deg.) Nd Er Tm Figure 4-11 Increasing FWHM of the ZnS 28.5o diffraction peak as the deposition temperature is increased 4.5.3 Thickness As the deposition temperature was increased the thicknesses of each of the phosphor layers decreased as shown in figure 4-12. The reduction in the thickness of the films ranged from 55 to 30% of the maximum thicknesses obtained between 140oC and 150oC.

PAGE 91

77 0.60.650.70.750.80.850.90.951130140150160170180190Deposition Temperature (Deg. C)Normalized Thickness Er Tm Nd Figure 4-12 Decreasing phosphor thickness with increasing deposition temperature 4.5.4 Threshold Voltage The turn on voltage also decreased as the deposition temperatures increased, presumably due to the reduced film thickness (Figure 4-13). For the Tm doped films the turn on voltage decreased from the maximum voltage of 130 volts at the lowest tested temperatures (140 C) to 90 volts at the 180oC deposition temperature. The effects of deposition temperature on the turn on voltages of the Nd based phosphor were similar to those of the thulium doped sample. The turn on voltage was at a maximum at the lowest deposition temperatures and then fell with increasing temperature. For ZnS:ErF3 the turn on voltage dependence on deposition temperature was smaller than for the other materials, but higher deposition temperatures produced the lowest turn on voltages.

PAGE 92

78 60708090100110120130140130140150160170180190Deposition Temperature (deg. C)Turn On Voltage (volts) Tm Nd Er Figure 4-13 Optical turn on voltage variation with increasing deposition temperature for each material 4.5.5 Infrared Emission Deposition temperature had a distinct effect on the emission intensity of the near infrared and visible emission as shown in figure 4-14. The near infrared brightness was highest at the 140 C deposition temperature for each of the rare earth dopants. Increasing deposition temperature steadily reduced the infrared emission in each case. The overall intensity loss was close to 80% in all cases.

PAGE 93

79 00.10.20.30.40.50.60.70.80.91130140150160170180190Deposition Temperature (Deg. C)Relative NIR intensity Tm Nd Er Figure 4-14 Decrease of near infrared irradiance with increasing deposition temperature 4.6 Discussion It is clear that changing the RE concentration and substrate temperature critically affected the properties of the phosphors. The reason behind most, if not all, of the deposition temperature effects is because of Zn and S thermal desorption during deposition. As the deposition temperature was raised the rare earth concentrations for each of the phosphors increased. This is attributed to faster thermal desorption of the host species than the rare earth dopants. This desorption is based on a lower sticking coefficient for Zn and S at elevated temperatures. Thermal desorption has been used previously to affect zinc and sulfur concentrations in materials such as ZnSxSex-1 [103] and decreasing thickness with increasing deposition temperature in ZnS films deposited

PAGE 94

80 by spray pyrolysis has been attributed to re-evaporation [104]. The rate of desorption is given by the Arrhenius type equation [105] ndesnndesdesRTEkRexp where Rdes is the rate of desorption, kdes is a desorption rate constant, is the coverage, Edes is the desorption activation energy, and vn is the frequency factor of desorption. The changes in concentration due to duty cycle variations are simply explained by the increase in the amount of time the doped target was sputtered compared to the undoped target. The variations from the expected trend for each material are the result of changing sputtering target morphologies affecting the sputtering rates. In addition to and because of the changing the rare earth concentrations, the higher desorption rates at higher deposition temperatures modified the thickness and crystallinity of the films. Since the deposition times for the temperature series films were the same, the increased desorption of the host material as the temperature was increased resulted in thinner films, as was shown in figure 4-12. Because the thickness was decreased, a lower electric field was necessary to breakdown the phosphors resulting in lower threshold voltages. The decrease in threshold voltage with increasing substrate temperature correlates with the decrease in thickness, observed by the normalized values for each shown in figure 4-15. The correlation between film thickness and turn on voltage is supported by the duty cycle series (Figure 4-16).

PAGE 95

81 0.50.550.60.650.70.750.80.850.90.951120130140150160170180190Deposition Temperature (Deg. C)Normalized Values Tm turn on Nd turn on Er turn on Tm thickness Nd thickness Er thickness Figure 4-15 Comparison of NIR turn on voltage and phosphor thickness as deposition temperature is varied

PAGE 96

82 0.40.50.60.70.80.9120304050607080Duty Cycle Ratio (doped/total)Normalized values Tm turn on Nd turn on Er turn on Tm thickness Nd thickness Er thickness Figure 4-16 Comparison of NIR turn on voltage and phosphor thickness as duty cycle and deposition time is varied Because the ZnS phosphors are crystalline as deposited, the change in crystallinity with changing deposition conditions could be measured using XRD. It could be expected that the crystallinity of the films would increase with increasing deposition temperature due to the increased mobility of the sputtered species caused by increased thermal energy. However, the crystallinity of the phosphors decreased as the deposition temperature increased, as indicated by the increasing FWHM of the ZnS 28.5o diffraction peak. The decrease in crystallinity is the result of increasing amounts of rare earths being incorporated into the ZnS matrix. The rare earths incorporate substitutionally on the zinc sites. The ionic radius of Zn2+ is 88 pm while the ionic radii of Nd3+, Er3+, and Tm3+ are 112, 103, and 102 pm respectively [106,107]. The rare earths have an average radius that is 20% larger than the Zn ion resulting in the dopants creating more strain in the crystal

PAGE 97

83 lattice. The duty cycle data support this interpretation since samples were deposited at the same temperature and there is a general increase in rare earth concentrations correlating with a decrease in crystallinity. As the rare earth concentration increases defect formation results in an increasingly poorer crystallinity and an increasing number of defects. The infrared emission was affected by each of the changes in the properties of the devices. Decreasing the thickness of the film yields a smaller volume of phosphor to produce photons. Because of this, the changing thicknesses of the phosphor films can mask how the concentration affects the NIR irradiance of the luminescent centers. Normalizing the thicknesses of the films gives a more accurate view of how the crystallinity and concentration affect the infrared output. This procedure is supported by the nearly linear correlation between brightness and film thickness shown in Figure 4-16. The data in Figure 4-17 show that the optimal concentration for all of the rare earths is near 1 at%. The number of data points in Figure 4-17 is large because the data is from both the duty cycle series and the deposition temperature series. This number compares well with reports in the literature for the maximum visible luminescence from these materials [74, 108,109]. Changes in irradiance are controlled by two competing processes as the rare earth concentration increases. Higher rare earth concentrations mean that there are more luminescent centers available to radiate. This would result in a steadily increasing brightness as the luminescent center concentration was increased. However, higher rare earth concentrations result in poorer crystallinity and possible non radiative interaction between neighboring luminescent centers. It has been shown that ZnS doped with Eu has poorer crystallinity than ZnS doped with an element closer in size

PAGE 98

84 to Zn, such as Mn [110]. As the crystallinity of the phosphor decreases there are more defects in the film. The defects can act as non radiative relaxation sites decreasing the radiative efficiency of the devices. Finally, dopant to dopant interaction is another increasingly prominent method of non radiative relaxation. The increasing concentration of the rare earths can lead to concentration quenching [111-113, 86] caused by dipole-dipole interactions or other non-radiative relaxation pathways between nearby rare earths. The result is that as the rare earth concentration is decreased below 0.8 at% and increased above 1.3 at%, the total infrared and visible irradiance of the films decreases substantially. 00.0020.0040.0060.0080.010.01200.511.522.53Rare Earth Concentration (at%)Nd and Er Irradiance (mW/m2nm)00.00010.00020.00030.00040.00050.00060.0007Tm Irradiance (mW/m2nm) Nd Er Tm Figure 4-17 NIR irradiance as a function of rare earth concentration. Note that the maximum occurs near 1 at% for each rare earth.

PAGE 99

85 4.7 Comparison of Infrared to Visible Emission Shown in figures 4-18 to 4-20 are the peak intensities (in uW) of the infrared and visible peaks for each dopant over a range of rare earth concentrations, from ~0.5 at% to over 1.5 at%. Even though there is a difference in the infrared and visible intensities with changing deposition conditions, the concentration with the maximum peak intensity is very similar. The visible emission is affected similarly to the NIR emission by crystallinity, concentration quenching, charge flow, and phosphor field. The slower luminescence reduction with increasing concentration for Nd doped phosphors can be explained by the state distribution of Nd. As discussed in section 5.6.1, Nd seems to contribute shallower states than Tm or Er. The shallower state distribution means that lower fields are necessary to inject them. The lower fields produce lower energy electrons that are more suitable to exciting the lower energy levels that produce infrared luminescence.

PAGE 100

86 00.00010.00020.00030.00040.00050.00060.00070.000800.511.522.53Concentration (at%)Irradiance (mW/m2nm)00.000020.000040.000060.000080.00010.000120.000140.000160.000180.0002 800 nm 480 nm Figure 4-18 B40 (left ordinate) for the NIR emission and B40 (right ordinate) for the visible emission in ZnS:TmF3 for various Tm concentrations

PAGE 101

87 00.0020.0040.0060.0080.010.01200.511.522.53Concentration (at%)Irradiance (mW/m2nm)00.00010.00020.00030.00040.00050.00060.0007 900 nm 600 nm Figure 4-19 B40 (left ordinate) for the NIR emission and B40 (right ordinate) for the visible emission in ZnS:NdF3 for various Nd concentrations

PAGE 102

88 00.00050.0010.00150.0020.00250.0030.00350.0040.004500.511.522.53Concentration (at%)Irradiance (mW/m2nm) 980 nm 530 nm Figure 4-20 B40 (left ordinate) for the NIR emission and B40 (right ordinate) for the visible emission in ZnS:ErF3 for various Er concentrations

PAGE 103

CHAPTER 5 ELECTRICAL EFFECTS OF CHANGING TARGET DUTY CYCLES AND SUBSTRATE TEMPERATURE DURING RF MAGNETRON SPUTTER DEPOSITION 5.1 Introduction Presented in this chapter are data on the electrical characteristics of EL devices and their effects on the optical performance of ZnS:[RE]F3 electroluminescent phosphors. As described in section 2.7.5 the external charge and external voltage were used to calculate the current, internal charge and phosphor field of the device. The purpose of this study was to determine how well the devices operate and to what extent the electrical properties of the phosphors (e.g. electrical threshold and phosphor field) have on the infrared and visible emission from the devices. It will be shown that several electrical characteristics are the same for each of the rare earth dopants. In addition, it is hypothesized that an interface layer is formed during deposition and is responsible for several of the electrical properties observed. 5.2 Charge-Voltage (Q-V) Data The external charge versus voltage (Q-V) data for ZnS doped with each of the rare earths is shown in figures 5-1 to 5-3. Each pulse begins at the origin because the positive and negative pulses were recorded separately so no information about the leakage or polarization charges [114] could be obtained. The change in the falling edge slope (points D to E or I to J from figure 2-11) as voltage is increased is an artifact due to resolution changes of the oscilloscope during measurement and was not considered during analysis. By taking the difference in charge between points A and E or F and J the 89

PAGE 104

90 external charge transferred across the device can be determined. This uses the assumption that the slope from point A to B is the same as the slope from point D to E and also follows for the negative pulse. Based on each of the Q-V figures, this assumption is valid. Plotting the transferred charge for each voltage produces a plot similar to a brightness-voltage curve. This can be used to determine the electrical threshold of the device by tracing back the curve, similar to determining optical threshold from a B-V curve. This information can be used to relate the electrical threshold of the device with the optical threshold voltages. The threshold voltages for each phosphor are shown in figures 5-4 and 5-5. Figures 5-6 and 5-7 shows the Q-V traces at B40 for Tm and Nd doped samples as the deposition temperature is changed. Note that the transferred charge in the Tm doped devices decreases with increasing deposition temperature while the transferred charge remains constant for the Nd doped devices. Also notable is that as the deposition temperature increased the amount of external charge in the device decreased. Finally, there is no curving from points C to D and G to H indicating that there is no dynamic space charge in these devices [47].

PAGE 105

91 -3-2-10123-150-100-50050100150Voltage (volts)External Charge (c/cm2) 100 110 120 130 140 150 Figure 5-1 Typical Q-V data for ZnS:TmF3 at drive voltages between 100 and 150 volts

PAGE 106

92 -3-2-10123-200-150-100-50050100150200Voltage (volts)External Charge (c/cm2) 80 90 100 110 120 130 140 150 160 170 Figure 5-2 Typical Q-V data for ZnS:NdF3 at drive voltages between 80 and 170 volts

PAGE 107

93 -4-3-2-101234-150-100-50050100150Voltage (volts)External Charge (c/cm2) 80 90 100 110 120 130 140 150 Figure 5-3 Typical Q-V data for ZnS:ErF3 at drive voltages between 80 and 150 volts

PAGE 108

94 507090110130150170190210230203040506070Duty Cycle Ratio (doped/undoped)Threshold Voltage (volts) Tm Nd Er Figure 5-4 Electrical threshold voltages for each phosphor as a function of duty cycle

PAGE 109

95 60708090100110120130140130140150160170180190Deposition Temperature (deg. C)Threshold Voltage (volts) Tm Nd Er Figure 5-5 Electrical threshold voltages for each phosphor as a function of deposition temperature

PAGE 110

96 -2.5-2-1.5-1-0.500.511.522.5-150-100-50050100150Voltage (volts)External Charge (c/cm2) 140 150 160 180 Figure 5-6 Plot of Q-V of ZnS:TmF3 at B40 with increasing deposition temperature (140-180oC)

PAGE 111

97 -2.5-2-1.5-1-0.500.511.522.5-200-150-100-50050100150200Voltage (volts)External Charge (c/cm2) 140 160 180 170 Figure 5-7 Plot of Q-V of ZnS:NdF3 at B40 with increasing deposition temperature 5.3 C-V Data Charge versus voltage (C-V) data were obtained from the Q-V data as described in Chapter 2.7.6. Figures 5-8 to 5-10 show typical C-V plots for Tm, Nd, and Er doped ZnS. The total capacitance of the device before breakdown is typically near 10 nf/cm2 for each of the devices but this varies from 3 to 12 nf/cm2 depending on the phosphor thickness (varies from 200 to 1200 nm). For an ideal device the capacitance after phosphor breakdown should be solely the insulator capacitance and should be the same for each device. The capacitance of the insulator is 64 nf/cm2 (determined from an ATO thickness of 220 nm and a dielectric constant of 16). All of the Tm doped phosphors exhibited a difference of 10 nf/cm2 between the pre-breakdown capacitance and the post-breakdown capacitance. The Nd doped phosphors had a capacitance difference of 14

PAGE 112

98 nf/cm2 while the Er doped samples had a difference of 16 nf/cm2. The post-breakdown values were the same for each phosphor but different between the phosphors. This value was used as the insulator capacitance to generate the Qint-Fp plots described below instead of the physical insulator capacitance value of 64 nm/cm2. The low capacitance values imply that the phosphors have not broken down. This is not believed to be the case and will be discussed below. In addition to the lower than ideal capacitance values after breakdown, the C-V plot has a slope after breakdown implying that there is a resistive component. Finally, none of the phosphors exhibited C-V overshoot indicating that there is no dynamic space charge in these devices, as previously mentioned for the Q-V plots. 0510152025303540020406080100120140160Voltage (volts)Capacitance (nf/cm2) 150 140 130 120 110 100 Figure 5-8 Typical C-V data for ZnS:TmF3 at drive voltages between 100 and 150 volts

PAGE 113

99 0510152025303540020406080100120140160Voltage (volts)Capacitance (nf/cm2) 170 160 150 140 130 120 110 100 90 80 Figure 5-9 Typical C-V data for ZnS:NdF3 at drive voltages between 80 and 170 volts

PAGE 114

100 05101520253035404550020406080100120140Voltage (volts)Capacitance (nf/cm2) 150 140 130 120 110 100 90 80 Figure 5-10 Typical C-V data for ZnS:ErF3 at drive voltages between 80 and 150 volts 5.4 Qint-Fp Data Internal charge and phosphor field data (Qint-Fp) were also derived from the Q-V data as discussed in Chapter 2.7.7. The phosphor and insulator capacitance values are critical to the production of Qint-Fp plots. The values for the insulator capacitance and total device capacitance were taken from a C-V plot for each sample, as shown in the previous section. The phosphor capacitance was determined using tiitpCCCCC where Cp is the phosphor capacitance, Ct is the total device capacitance, and Ci is the insulator capacitance. Because the positive and negative pulse Q-V data is generated separately the Qint-Fp data is also calculated separately. The data for each pulse originally begins at the origin. First, the two separate plots are shifted horizontally by half of the

PAGE 115

101 distance from the y-axis to point E on the positive pulse graph. The two data sets are joined at the end point of the positive pulse data and the beginning point of the negative pulse data (points E and F). This join point can be seen in quadrant one of each plot and trends toward the upper right as the voltage is increased. The plots are finally centered vertically by setting the first and last points of the positive pulse equidistant from the x-axis. Because of this processing no information can be gained about the leakage charge. Figures 5-11 to 5-13 are Qint-Fp graphs for devices with each dopant as the voltage is increased. As the voltage increases in each device the phosphor field approaches a constant value, a phenomenon known as field clamping. Field clamping implies that the phosphor is completely broken down and is at odds with the capacitance data. A solution for this discrepancy is proposed below. As the deposition temperature was increased, the phosphor field and internal charge in the Tm doped samples decreased as shown in figure 5-14. Figure 5-15 shows that while the electric field in the Nd doped samples dropped as in the Tm doped samples the internal charge rose as the deposition temperature increased. Figure 5-16 shows that for the Er doped samples, the phosphor field decreased with increasing temperature except for at 180 oC. The differences in internal charge for the different dopants demonstrate the differences in the distribution of interface states for each material.

PAGE 116

102 -2-1.5-1-0.500.511.52-2-1012Phosphor Field (Mv/cm)Internal Charge (c/cm2) 100v 110v 120v 130v 140v 150v 160v Figure 5-11 Internal Charge vs. phosphor field for increasing voltage in ZnS:TmF3

PAGE 117

103 -1.5-1-0.500.511.5-2-1012Phosphor Field (Mv/cm)Internal Charge (c/cm2) 100v 110v 120v 130v 140v 150v 160v 170v Figure 5-12 Internal Charge vs. phosphor field for increasing voltage in ZnS:NdF3

PAGE 118

104 -2-1.5-1-0.500.511.52-2-1012Phosphor Field (Mv/cm)Internal Charge (c/cm2) 100v 110v 120v 130v 140v 150v Figure 5-13 Internal Charge vs. phosphor field for increasing voltage in ZnS:ErF3

PAGE 119

105 -1.5-1-0.500.511.5-2-1.5-1-0.500.511.52Phosphor Field (Mv/cm)Internal Charge (c/cm2) 140 150 160 180 Figure 5-14 Internal charge vs. phosphor field for ZnS:TmF3 as the deposition temperature is changed

PAGE 120

106 -1-0.8-0.6-0.4-0.200.20.40.60.81-2.5-2-1.5-1-0.500.511.522.5Phosphor Field (Mv/cm)Internal Charge (c/cm2) 140 160 180 170 Figure 5-15 Internal charge vs. phosphor field for ZnS:NdF3 as the deposition temperature is changed

PAGE 121

107 -2-1.5-1-0.500.511.5-3-2-10123Phosphor Field (Mv/cm)Internal Charge (c/cm2) 140 180 150 160 Figure 5-16 Internal charge vs. phosphor field for ZnS:ErF3 as the deposition temperature is changed 5.5 Time Resolved Electroluminescence Time resolved electroluminescence (TREL) data were taken as described in chapter 3.3.7. Figure 5-20 shows the luminescence of ZnS:TmF3 for the 480 nm and 800 nm emission during a 30s pulse. The pulses are normalized to the same maximum value. Notice that the NIR emission rises slower and peaks later than the visible. The large spikes at the beginning and end of the voltage pulse are due to electrical noise and are not pulses of luminescence. This noise is also seen in figures 5-21 and 5-22, TREL graphs for Nd and Er doped phosphors, respectively. The Nd phosphor graph shows the electroluminescence at 600 nm for voltage pulses of 5 s and 30 s while the figure 5-22 shows the electroluminescence of Er at 530 nm. As can be seen by the difference in decay rates in figure 5-21, excitation for radiative transitions occurs during the steady state portion of the voltage pulse in addition to during the rising edge. Figures 5-23 to 5

PAGE 122

108 27 show the data from the previous three graphs on a logarithmic scale to determine the decay constant for each material. The figures show that the each emission can be fitted with two decay constants. The 480 nm emission decays with 9 and 17 s time constants while the 800 nm emission had time constants of 11 and 19 s. The 600 nm emission of Nd had time constants of 3.3 and 15 s when excited by a 30 s voltage pulse and 2.5 and 13 s for excitation by a 5 s pulse. Finally, Er decays with 6 and 14 s time constants. 5.5.1 Discussion of TREL Data The luminescent decay of these devices highlights several aspects of their operation. The decay in all of the films is much faster than the decay from other ZnS phosphors with dopants in which luminescence originates from a forbidden transition, such as manganese and terbium [115]. Both manganese and terbium doped ZnS exhibit decay times on the order of a millisecond while the decay from all of the present dopants is on the order of 15 s. The decay rates for Tm are comparable to those reported by Sohn and Hamakawa [20], while the decay rate for Er is similar to primary decay rate reported by Wang et al. [116]. These fast decay rates imply that either the crystal field around the dopants is relaxing the selection rules, or the emission is dominated by non radiative recombination. Since the 4f transitions in rare earths are shielded by 6s shell it is unlikely that the selection rules would be broken enough to permit such fast decay. There is no evidence to this effect, i.e. no color changes in the emission that should accompany strong crystal field effects. While manganese is very similar in size to zinc the rare earth ions are much larger causing a larger lattice distortion. Also, terbium has been added as a phosphor dopant as an oxifluoride [117] and not a fluoride like the rare earths in this study. The ionic radius

PAGE 123

109 of oxygen (~140 pm) is much smaller than that of sulfur (~185 pm) for which it would substitute in the lattice. The incorporation of oxygen is likely to reduce the strain on the lattice generated by the large Tb ion, thereby reducing the effects of non radiative decay in this phosphor. If the fluorine rests in an interstitial site, there is no strain compensating ion incorporation in these phosphors, however, an interstitial fluorine ion, with an ionic radius of ~120 pm, will not add strain because it is smaller than the interstitial volume. Because relaxation of selection rules is unlikely, non-radiative decay is the most probable cause of the fast luminescent decay in these materials. The luminance of the blue emission from the Tm doped devices was ~0.02 cd/m2. The speculation of an increase in non-radiative decay is supported by the low irradiance of these devices compared to the brightness of annealed devices, as shown in table 2-3. As can be seen by the plots of the ZnS:TmF3 TREL, the decay of the infrared luminescence is slower than the visible. Tm has two possible excited states for luminescence near 800 nm, decay from the 1G4 (the same excited state as for the blue emission) to the 3H5 level or a transition from the 3F4 to the 3H6 ground state. The emission at 800 nm also peaks later in the voltage pulse than does the 480 nm emission. This is evidence that, since emission from the same level should decay at the same rate, the higher 1G4 level is pumping the lower 3F4 level. Figures 5-22 and 5-23 show that the luminescent decay in Nd doped devices is faster for a 5 s voltage pulse than for a 30 s pulse. The shorter pulse ends just as the luminescence is peaking while the longer pulse plateau endures for most of the luminescence decay period. This demonstrates that Nd continues to be excited during the entire voltage pulse and not just during the rising edge. A close look at the decay after

PAGE 124

110 the 30 s pulse for each material shows that each of the dopants experiences continuous excitation during the pulse. 02.557.510-5.00E-065.00E-061.50E-052.50E-053.50E-05Time (sec.)Intensity (arb. units) pulse 480 nm 800 nm Figure 5-17 Time resolved electroluminescence of the NIR and blue emission from ZnS:TmF3

PAGE 125

111 0246810120.00E+001.00E-052.00E-053.00E-054.00E-05Time (sec.)Intensity (arb. units) 5 ms pulse 5 ms lum. 30 ms lum. Figure 5-18 Time resolved electroluminescence of the visible emission from ZnS:NdF3 for voltage pulse durations of 5 and 30 s

PAGE 126

112 0501001502002503000.00E+001.00E-052.00E-053.00E-054.00E-05Time (sec.)Intensity (arb. units) Pulse 530 nm Figure 5-19 Time resolved electroluminescence of the visible emission from ZnS:ErF3

PAGE 127

113 y = 7.5537e-58876xR2 = 0.9795y = 15.779e-108815xR2 = 0.9950.010.1110-5.00E-065.00E-061.50E-052.50E-053.50E-05Time (sec.)Intensity (arb. units) 480 nm Figure 5-20 Log plot of TREL decay of the 480 nm emission from ZnS:TmF3

PAGE 128

114 y = 15.388e-93085xR2 = 0.9895y = 8.028e-52888xR2 = 0.91140.1110-5.00E-065.00E-061.50E-052.50E-053.50E-05Time (sec.)Intensity (arb. units) 800 nm Figure 5-21 Log plot of TREL decay of the 800 nm emission from ZnS:TmF3

PAGE 129

115 y = 95.078e-300249xR2 = 0.9976y = 2.6633e-67171xR2 = 0.92280.11101000.00E+001.00E-052.00E-053.00E-054.00E-05Time (sec.)Intensity (arb. units) Figure 5-22 Log plot of TREL decay of the 600 nm emission from ZnS:NdF3 for a 30 s voltage pulse

PAGE 130

116 y = 178.16e-398374xR2 = 0.9764y = 1.3493e-76030xR2 = 0.51770.11101000.00E+001.00E-052.00E-053.00E-054.00E-05Time (sec.)Intensity (arb. units) Figure 5-23 Log plot of TREL decay of the 600 nm emission from ZnS:NdF3 for a 5 s voltage pulse

PAGE 131

117 y = 880.96e-167196xR2 = 0.9956y = 169.8e-70353xR2 = 0.985111010010000.00E+001.00E-052.00E-053.00E-054.00E-05Time (sec.)Intensity (arb. units) Figure 5-24 Log plot of TREL decay of the 530 nm emission from ZnS:ErF3 5.6 Discussion 5.6.1 Q-V Analysis From the data presented in the previous sections it is clear that the electrical properties of rare earth doped ZnS change dependent upon the dopant. As shown in section 5.2 and figures 5-1 to 5-3, the critical voltage for charge injection (point B) decreases as the drive voltage is increased for each phosphor. This fact has often been reported [29,114] and can be explained by the amount of charge flowing at increasing voltages. As the amount of charge flowing through the device is increased more electron interface trapping states will be filled for each pulse. Electronic states with the deepest energy will fill first and continue filling to progressively shallower energies as the amount of charge increases. At higher voltages, shallower electron states are filled and

PAGE 132

118 charge from these states will tunnel inject at lower fields. Hence, charge injection begins at lower critical voltages as the drive voltage is increased. As the deposition temperature is increased the total external charge in the Tm doped devices dropped while the charge in the Nd doped phosphor remained constant (Figures 5-4 and 5-5). The amount of decreased charge appears to be concentration and dopant dependant. It is possible that the energy of electron trapping interface states induced by Tm have deeper energy distributions as compared to Nd. If this is the case there will be fewer shallow states filled in the Tm doped phosphors. The deeper trap states will require higher fields to inject charge through the device while a shallow distribution of electrons could still be injected at lower voltages (figure 5-25). Despite the decreased threshold voltage of the higher temperature films, the transferred charge remained the same for all of the Nd doped samples while it dropped for the Tm samples. Figure 5-25 Energy band diagram of an ACTFEL device showing how the distribution of interface states can affect the electric field necessary for tunnel injection

PAGE 133

119 Even though the critical voltage for charge injection drops with increasing drive voltage, the electrical threshold of the device is defined as the voltage obtained from the slope extrapolation of the Qe-Vmax plot as shown in figure 5-26. The electrical thresholds for phosphors with varying duty cycles and deposition temperatures were shown in figures 5-4 and 5-5. Data for the optical threshold for NIR emission were shown in chapter 4. It was expected that the optical threshold for NIR emission would be at a lower voltage than for visible emission because the NIR emission from each rare earth originates from a lower energy excited state than the visible emission. Shown in figures 5-27 to 5-29 are B-V curves comparing the optical threshold for visible and infrared emission from each material. The optical threshold voltages for visible and NIR emission are the same in all cases. In figures 5-30 and 5-31 the electrical and optical threshold voltages are compared for each phosphor. The optical and electrical thresholds are equal within experimental noise. It appears that when electrical threshold is reached, the electric field is sufficiently high that injected electrons have enough energy to excite both the visible and NIR emission excited states. While the NIR and visible optical thresholds are the same for ZnS, Kim et al. [118] have reported lower optical thresholds for NIR versus visible emission for rare earth doped GaN.

PAGE 134

120 00.20.40.60.811.21.41.61.8280100120140160180Voltage (volts)Transferred Charge (C/cm2) Figure 2-26 Transferred charge versus maximum applied voltage showing the electrical threshold for a typical ZnS:TmF3 device

PAGE 135

121 00.00010.00020.00030.00040.00050.00060.00070.00080.00090.001100120140160180Voltage (volts)Irradiance (mW/m2nm) 800 nm 650 nm 480 nm Figure 5-27 Irradiance from ZnS:Tm versus applied voltage showing the optical threshold is the same for NIR and visible emission

PAGE 136

122 00.0010.0020.0030.0040.0050.0060.0070.0080.0090.01100110120130140150160170180Voltage (volts)Irradiance (mW/m2nm) 892 nm 815 nm 602 nm Figure 5-28 Irradiance from ZnS:Nd versus applied voltage showing the optical threshold is the same for NIR and visible emission

PAGE 137

123 00.00050.0010.00150.0020.00250.003405060708090100110Voltage (volts)Irradiance (mW/m2nm) 1000 nm 660 nm 550 nm 530 nm Figure 5-29 Irradiance from ZnS:Er versus applied voltage showing the optical threshold is the same for NIR and visible emission

PAGE 138

124 608010012014016018020020304050607080Duty Cycle Ratio (doped/doped+undoped)Threshold Voltage (volts) Tm elec Tm opt Nd elec Nd opt Er elec Er opt Figure 5-30 Comparison of optical and electrical threshold voltages with changing duty cycle ratios for each dopant

PAGE 139

125 60708090100110120130140130140150160170180190Deposition Temperature (deg. C)Threshold Voltage (volts) Tm elec Tm opt Nd elec Nd opt Er elec Er opt Figure 5-31 Comparison of optical and electrical threshold voltages versus deposition temperature for each dopant 5.6.2 C-V Analysis As mentioned in the previous section, the critical voltage for charge injection shifts to lower voltages as the applied voltage is increased. This is easy to see in the C-V plots for each dopant (Figures 5-8 to 5-10). The ATO dielectric has a dielectric constant of 16 resulting in a capacitance of ~64 nf/cm2 for the 220nm thick layer. In each case the capacitance after the critical voltage for charge injection is less than expected, i.e. is ~20-30 nf/cm2 for every sample. The lower capacitance value implies that the phosphors do not completely break down above the critical voltage. As will be discussed below, the phosphor does appear to be completely broken down. This implies that the unexpectedly small insulator capacitance is not a bulk effect. In addition, there is a resistive component after the phosphor has broken down as evident from the positive slope of the

PAGE 140

126 C-V curve after electrical breakdown, this behavior has been observed in ZnS:Mn thin films [48]. In the ideal model of an ACTFEL device, after break down the resistivity should be close to zero. To explain these data, formation of an interface layer with high electrical breakdown strength is postulated. This layer could be formed at the ATO/phosphor interface during deposition, by oxidation of the phosphor surface, and/or reaction at the phosphor/aluminum interface. It has been shown that sputter deposited ZnS grows as columnar grains, but that there is an equiaxed polycrystalline layer (~100 nm thick in ZnS:Mn doped with KCl) at the dielectric interface [119]. Also, the films are exposed to air before the deposition of the final contact making an oxide layer probable. If the interface layer or layers change thickness with deposition temperature and are more resistive than the phosphor, then there would be a constant change between the capacitance before conduction onset and the capacitance after conduction onset. An explanation of the possibility of an interface layer is discussed in section 5.6.4. 5.6.3 Qint-Fp Analysis The contribution of the rare earth dopants to the energy distribution of electron trapping states in the phosphors was discussed above in section 5.6.1 is consistent with data in figures 5-14 to 5-16. As the rare earth concentration increases with deposition temperature, the internal charge in the Tm doped films decreases, while internal charge increases for Nd doped phosphors even with the decrease in phosphor field. If the higher concentration of Tm leads to a higher concentration of deeper energy states, a lower internal charge would be expected because the field will not be strong enough to tunnel inject the charge from these deeper states. If Nd were to contribute shallower states to

PAGE 141

127 the distribution, then even with a lower phosphor field, there will be more charge injected at a lower field. The internal charge versus Er concentration was constant within experimental noise, implying that Er did not significantly change the energy distribution of trapping states. The relation between NIR emission, phosphor field and internal charge in Tm doped films is shown in figure 5-32, while figures 5-33 to 5-36 show the same for Nd or Er doped devices. For Tm and Er, increased NIR peak intensity correlates with an increased phosphor field but not increased internal charge. This implies that increased brightness results from a hotter electron distribution (i.e. increased phosphor field), not more electrons (increased internal charge). This is consistent with the conclusion above that both Tm and Er have deeper energy trapping state distributions. The charge trapped in the deeper states needs a higher field for injection and results in hotter ballistic electrons. In the case of Nd, there is no clear relation between the phosphor field and infrared intensity. In addition, increased internal charge is observed as the Nd concentration increases, while the infrared brightness decreases. These observations are consistent with a shallow energy trap distribution for Nd since shallow traps would require lower fields to inject charge and lead to cooler ballistic electrons, plus electrons in shallow traps are more likely to be excited by other electrons, increasing the internal charge. Without an increase in brightness with an increase in field or charge, brightness would be expected to increase with increasing concentrations. While this expectation is realized at low concentrations, the decrease in brightness above 1 at% Nd can be attributed to non-electrical effects such as concentration quenching and decreased crystallinity at higher Nd concentrations, as discussed in the previous chapter.

PAGE 142

128 00.10.20.30.40.50.60.70.80.9100.511.522.53Tm Concentration (at%)Normalized Value B40 Qint pos Qint neg Fp pos Fp neg Figure 5-32 Normalized internal charge, phosphor field and NIR brightness versus Tm concentrations in ZnS:TmF3. Note that while the average of internal charge is nearly constant, the trend for both B40 and Fp is down as the temperature increases. This correlation is discussed in the text

PAGE 143

129 00.10.20.30.40.50.60.70.80.91130140150160170180190Deposition Temperature (deg. C)Normalized Value B40 Qint pos Fp pos Fp neg Qint neg Figure 5-33 Relation of internal charge and phosphor field with NIR brightness for ZnS:NdF3 with changing deposition temperature

PAGE 144

130 00.10.20.30.40.50.60.70.80.9120304050607080Duty Cycle Ratio (doped/doped+undoped)Normalized Value B40 Qint pos Qint neg Fp pos Fp neg Figure 5-34 Relation of internal charge and phosphor field with NIR brightness for ZnS:NdF3 with changing target duty cycle

PAGE 145

131 00.10.20.30.40.50.60.70.80.9100.511.522.5Concentration (at%)Normalized Value B40 Qint Figure 5-35 Relation of internal charge with NIR brightness for various Nd concentrations in ZnS:NdF3

PAGE 146

132 00.10.20.30.40.50.60.70.80.91130140150160170180190Deposition Temperature (deg. C)Normalized Value B40 Qint pos Fp pos Fp neg Qint neg Figure 5-36 Relation of internal charge and phosphor field with NIR brightness for ZnS:ErF3 with changing deposition temperature. Note that the brightness correlates with Fp and not with the internal charge 5.6.4 Interface Layer Discussion The Qint-Fp plots (Figure 5-11 to 5-13) show that each of the phosphors exhibits pseudo field clamping, as evidenced by the near constant phosphor field at higher voltages. Field clamping occurs when the charge is free to flow through the phosphor so that any increase in the phosphor field due to increased voltage is canceled by increased charge at the interface resulting in a counter field which results in a constant phosphor field with increasing voltages. When the phosphor field is not constant but increases slightly in proportion to increasing voltage, due to incomplete cancellation by charge accumulation, then pseudo field clamping is observed. Field clamping is normally observed only during complete breakdown of the phosphor because of the need for rapid

PAGE 147

133 charge movement. Based on the C-V data in figures 5-8 to 5-10, it appears the phosphors in this study did not completely break down, because the capacitance above the voltage required for charge injection was lower than that of the insulator alone. However, if there is an interface layer that does not breakdown above the voltage necessary for charge injection, the bulk phosphor could breakdown enough to produce field clamping while appearing to remain capacitive. If there is an interface layer below the aluminum contact then there could be charge build up on both sides of the phosphor. A single amorphous ZnS interface layer created during deposition is probable. Thin film ZnS grown by spray pyrolysis can be amorphous [120] and the successive ionic layer adsorption and reaction technique (SILAR) has produced films that are amorphous for the first 250 nm and then become polycrystalline [121]. Sputtered and electron beam evaporated ZnS:Mn films have exhibited a 100 nm to 200 nm thick layer of small equiaxed grains before the columnar growth typical of ZnS films [118, 122]. Because the rare earths are much larger than Zn or Mn (~100 pm ionic radii for rare earths compared to ~70 pm for Zn and Mn) an amorphous layer instead of a small grained layer is not unlikely. It is hypothesized that a layer of amorphous ZnS between 80 and 170 nm thick is grown during deposition and that the thickness of the amorphous layer depends on the rare earth dopant and the deposition temperature. Using the C-V data collected and knowing the capacitance of the ATO layer (64 nf/cm2, as discussed above) the required capacitances of the bulk ZnS film and the interface layer can be calculated. The capacitance of a material is given by dCr 0

PAGE 148

134 where o is the permittivity of free space, r is the relative dielectric constant, and d is the thickness. If there is no interface layer, based on the C-V data, the dielectric constant of the ZnS film needs to be ~4.5. The typically cited dielectric constant of ZnS is between 8.0 and 8.5 [123, 124]. A dielectric constant of <5 for a polycrystalline layer of ZnS seems low, however, the American Institute of Physics Handbook warns that Discrepancies in the dielectric constant of the order of 10% are frequently found in the literature. [125]. In addition, the dielectric constant is dependant on the temperature and crystallinity of the material as well as the measurement frequency [122, 125, 126]. The dielectric constant of thin film BaTiO3 has changed from ~20 to over 100 when the film is changed from amorphous to polycrystalline [122]. On the basis of poor crystallinity for the bulk ZnS and a drive frequency of 2.5 kHz of ~6.4 (20% lower than typically reported) does not seem unreasonable. Because of its amorphous nature the dielectric constant assigned to the interface layer is 4.0 (50% lower than typically reported). Calculations of layer thicknesses were done using the previous equation and ilftbCCCC1111 where Ctb is the total capacitance below the critical voltage, Cf is the bulk ZnS film capacitance, Cl is the interface layer capacitance, and Ci is the insulator capacitance for the capacitance below the injection voltage. The layer thicknesses above the injection voltage were calculated using iltaCCC111 where Cta is the total capacitance above the injection voltage. From the dielectric constants listed above and the capacitances from the C-V data, interface layers of ~35%,

PAGE 149

135 ~22%, and ~27% the total film thickness for Tm, Nd, and Er doped phosphors respectively provide capacitances within 10% of those measured when the deposition temperature was changed. Figures 5-37 and 5-38 show the interface layer thicknesses of ZnS:TmF3 and ZnS:NdF3 calculated for various deposition conditions. Samples deposited at the same temperature with varying sputter target duty cycles obtained results within 10% of the measured capacitance for a constant interface layer thickness. Studies have shown that the capacitance above the injection voltage increases to ~64 nf/cm2 with annealing [127]. This is in agreement with the decrease in interface layer thickness with increasing deposition temperature. The decreasing interface layer thickness is attributed to increased atomic mobility resulting in faster crystallite formation during deposition and the conversion from amorphous to polycrystalline when annealed. Studies of ZnS:Mn support this description having shown that the fine grained layer in those devices exhibits strong crystal growth with annealing [122].

PAGE 150

136 80100120140160180200130140150160170180190Deposition Temperature (Deg. C)Interface Layer Thickness (nm) Temp Duty Figure 5-37 Calculated interface layer thicknesses for ZnS:TmF3 as a function of deposition temperature and duty cycle (Duty cycle data plotted at the duty cycle + 100 i.e. a duty cycle ratio of 50 is plotted at 150)

PAGE 151

137 60708090100110120130140150160130140150160170180190Deposition ConditionInterface Layer Thickness (nm) Temp Duty Figure 5-38 Calculated interface layer thicknesses for ZnS:NdF3 as a function of deposition temperature and duty cycle (Duty cycle data plotted at the duty cycle + 100 i.e. a duty cycle of 50 is plotted at 150)

PAGE 152

CHAPTER 6 CONCLUSIONS 6.1 Deposition Effects on the Physical Properties and Optical Properties of ZnS:RE Phosphors The effects of deposition conditions during simultaneous R.F. magnetron sputtering of undoped and doped ZnS targets on the electroluminescent emission of near infrared and visible light from ZnS ACTFEL devices doped with TmF3, NdF3, or ErF3 have been studied without annealing. It was shown that changing the target duty cycle (50% to 100%) in the dual target deposition system or changing the substrate temperature (130-190C) can dramatically change the properties and performance of these devices. EDS and EPMA showed that the rare earth (Tm, Er, or Nd) concentrations increased with increasing deposition temperature. This increase was attributed to increased thermal desorption of the host species as the temperature was raised, consistent with a decrease in ZnS deposition rate as the deposition temperature was raised. It was also shown that as the concentration of the rare earth was increased by either an increased deposition temperature or increase in the doped sputter target duty cycles, the crystallinity of the phosphor film decreased. The RE ions substitute for Zn ions in the lattice and the large rare earth ions create strain in the crystal lattice leading to decreased crystallinity. The concentration of rare earth is one of the most influential factors for controlling the electroluminescent power from the device. The brightest NIR and visible emission 138

PAGE 153

139 was produced by phosphors with a rare earth concentration of 0.8 to 1.1 at% for each dopant. The near infrared emission from these ZnS:RE phosphors was increased form 300% to 700% by decreasing the rare earth concentration from 2 at% to 0.9 at%. The decrease in brightness at rare earth concentrations >1 at% was attributed to concentration quenching and reduced crystallinity. Time resolved data from the visible emission of each phosphor allowed calculation of the luminescent decay times which were 100 times faster than expected from these materials after annealing. This fast decay was attributed to a large fraction of the excited electrons decaying non-radiatively due to poor crystallinity of the as deposited samples. In addition, it was found that excitation of the luminescent centers occurs during the plateau of the driving waveform in addition to the rising edge of the pulse increasing the irradiance when driven by longer pulses. The NIR emission from Tm has a slower rise time and a slower decay time than the visible emission suggesting the possibility that the 1G4 level that produces the visible emission is feeding the 3F4 level that produces some of the NIR emission. 6.2 Electrical Properties of ZnS:RE Phosphors The optical threshold voltages for visible and near infrared emission were expected to be different because of the differing energies in the excited states responsible for the luminescent transitions, however the experimental data showed that the threshold voltages were the same for the NIR and visible emission for each phosphor. The optical threshold voltage was equal to the electrical threshold voltage in each case. It was speculated that the field necessary to create electrical breakdown was sufficient to accelerate injected electrons to high enough energy to excite both the NIR and visible

PAGE 154

140 transitions. While the electric field in the films for each dopant decreased for the thinner films deposited at higher deposition temperatures, the internal charge through phosphors doped with Nd increased while the internal charge in the Tm phosphors decreased. This difference was attributed to the depth of the interface states as modified by dopant. It is proposed that Nd creates a shallower energy state distribution than Tm or Er, so Nd doped phosphors will have more electrons injected from shallow state traps even at the lower field. Finally, all of the devices exhibited capacitances that were lower than expected after electrical breakdown and they also exhibited pseudo-field clamping. The low capacitance implies that the phosphor is not fully breaking down, but this fact is not consistent with the observation of field clamping, which requires fast charge transport. It was postulated that an amorphous interface layer with low dielectric constant, high electrical break down strength was formed during deposition. While the majority of the phosphor layer fully breaks down at the threshold voltage, allowing charge to flow fast enough for field clamping to occur, the interface layer does not breakdown and continues to contribute to the capacitance, and lowers the value from that of the insulator alone. Calculations show that the interface layer is consistently 35%% as thick as the total deposited film for ZnS:Tm and 22%% of the thickness for ZnS:Nd, indicating that the layer is formed during deposition of the films and not after removal from the deposition chamber. The interface layer will significantly decrease the brightness of the device because of its higher breakdown strength and amorphous nature.

PAGE 155

LIST OF REFERENCES 1. P.D. Rack, P.H. Holloway, Mat. Sci. Eng. R21, 171 (1998) 2. P.H. Holloway, M. Davidson, N. Shepard, A. Kale, W. Glass, B.S. Harrison, T.J. Foley, J.R. Renolds, K.S. Schanze, J.M. Bonecella, S. Sinnott, D. Norton. Proc. Conf. 5080 Cockpit Disp. X, (2003) 3. V. Zeninari, B. Parvitte, D. Courtois, V.A. Kapitanov, Yu.N. Ponomarev, Infrared Phys. and Tech. 44, 253 (2003) 4. S. D. Smith, A. Vass, P. Bramley, J. G. Crowder, C. H. Wang, IEE Proc. Optoelectron. 144, 266 (1997) 5. V. Weldon, J. OGorman, J.J. Perez-Camacho, D. McDonald, J. Hegarty, J.C. Connoly, N.A. Morris, R.U. Martinelli, J.H. Abeles, Infrared Phys and Tech. 38, 325 (1997) 6. Autospeed.com, Performance News 134 (2001) 7. N. Barnes, Auto. Eng. (London) 27, 44 (2002) 8. T. Gibson, Amer. Heritage of Invent. and Tech. 14, 47 (1998) 9. Lord Bowden of Chesterfield, IEE Proceedings 32, 436 (1985) 10. B.E.A. Saleh, M.C. Teich, Fundamentals of Photonics (John Wiley and Sons Inc., New York 1991) 11. R. Sakai, T. Katsumata, S. Komuro, T. Morikawa, J. Luminescence 85, 149 (1999) 12. D-J. Kim, H-M. Kim, M-G. Han, Y-T. Moon, S. Lee, S-J. Park, J. of Vac. Sci. Technol. B 21, 641 (2003) 13. M.G. Tomilin, J. Optical Tech. (A Translation of Opticheskii Zhurnal) 70, 454 (2003) 14. J. C. Hitt, J. P. Bender, J. F. Wager, Crit. Rev. Sol. State & Mat. Sci. 25, 29 (2000) 15. M. Leskel, J. Alloys and Compounds 275, 702 (1998) 141

PAGE 156

142 16. M. Shikada, Telecommunications, 26, 4p (1992) 17. S.M. Goodnick, M. Dur, R. Redm er, M. Reigrotzki, N. Fitzer, Proceed. of SPIE 3940, 217 (2000) 18. P. D. Keir, C. Maddix, B. A. Baukol, J. F. Wager, B. L. Clark, D. A. Keszler, J. Appl. Phys. 86, 6810 (1999) 19. S. H. Sohn, D. G. Hyun, A. Yamada, Y. Hamakawa, Appl. Phys. Lett. 62, 991 (1993) 20. S. H. Sohn, Y. Hamakawa, Appl. Phys. Lett. 62, 2242 (1993) 21. C. R. Ronda, T. Jstel, H. Nikol, J. Alloys and Compounds 275, 669 (1998) 22. J. Carter, D. Snyder, J. Reichenbaugh, IEEE Semiconductor Thermal Meas. and Manag. Symp. 357 (2003) 23. J.P. Silveira, F. Grasdepot, Infrared Phys. and Tech. 37, 143 (1996) 24. F.L. Perdrotti, S.J. Leno, S. Perdrotti, Introduction to Optics, 2nd ed. (Prentice-Hall, Englewood Cliffs, New Jersey, 1987) 25. A. N. Krasnov, Prog. Cryst. Growth & Charact. Mat. 37, 123 (1998) 26. G. Hrknen, M. Leppnen, E. Soininen, R. Trnquist, J. Viljanen, J. Alloys and Compounds 225, 552 (1995) 27. S.Ochiai, T. Kato, T. Ogawa, K. Kojima, Y. Uchida, A. Ohashi, M. Ieda, T. Mizutani, Conf. on Elec. Ins. and Dielec. Phenom. 681 (1994) 28. D.H. Smith, J. Luminescence 23, 209 (1981) 29. J.F. Wager, P.D. Keir, Ann. Rev. Mat. Sci. 27, 223 (1997) 30. W.D. Callister, Materials Science and Engineering an Introduction (John Wiley and Sons Inc., New York 1997) 31. K.A. Neyts, P. DeVisschere, J. App. Phys. 68, 4163 (1990) 32. S.M. Sze, Physics of Semiconductor Devices (Wiley-Interscience, New York, 1981) 33. C. Kittel, Introduction to Solid State Physics (John Wiley and Sons Inc., New York, 1986)

PAGE 157

143 34. YH. Lee, YS. Kim, BK. Ju, IEEE Transactions on Electron Devices 46, 892 (1999) 35. XY. Jiang, ZL. Zhang, WM, Zhao, ZG. Lui, SH. Xu J. Phys. Condens. Matter 6, 3279 (1994) 36. H. Zhao, Z. Xu, Y. Wang, Y. Hou, X. Xu, Displays 21, 143 (2000) 37. S. Okamoto, E. Nakazawa, Y. Tsuchiya, Jap. J. Appl. Phys. 28, 406 (1989) 38. J.W. Mayer, Electronic Materials Science: For Integrated Circuits in Si and GaAs, (Macmillan, New York, 1990) 39. G. Blasse, B.C. Grabmaier, Lumineacent Materials (Springer-Verlag, Berlin, 1994) 40. N. Miura, T. Sasaki, H. Matsumoto, R. Nakano, Jpn. J. Appl. Phys. 31, 295 (1992) 41. E. Bringuier, J. Appl. Phys. 75, 4291 (1994) 42. T.D. Thompson, J.W. Allen, J. Crystal Growth 101, 981 (1990) 43. E. Bringuier, J. Appl. Phys. Lett. 60, 1256 (1992) 44. W.E. Howard, O. Sahni, P.M. Alt, J. Appl. Phys. 53, 639 (1982) 45. I. Tanaka, Y. Izumi, K. Tanaka, Y. Inoue, S. Okamoto, J. of Luminescence 87, 1189 (2000) 46. S. Shih, P. D. Keir, J. F. Wager, J. Appl. Phys. 78, 5775 (1995) 47. P.D. Keir, Dissertation, Electrical and Computer Engineering, Oregon State University (1999) 48. J.S. Lewis, Dissertation, Materials Science and Engineering, University of Florida (2000) 49. Y.A. Ono, Electroluminescent Displays, (World Scientific Publishing Co., Singapore, 1995) 50. S-S. Sun, R. Khormaei, SPIE High-Resolution Disp. and Projection Sys. 1664, 48 (1992) 51. I. Khormaei, Dissertation, Electrical and Computer Engineering, Oregon State University (1995)

PAGE 158

144 52. P.D. Keir, H. Le, R.L. Thuemler, H. Hitt, J.F. Wager, Appl. Phys. Lett. 2421 (1996) 53. H. Jiang, Y. Zhou, B.S. Ooi, Y. Chen, T. Wee, Y.L. Lam, J. Huang, S. Liu, Thin Solid Films 363 25 (2000) 54. M. Suzuki, Y. Maeda, M. Muraoka, S. Higuchi, Y. Sawada, Mater. Sci. Eng. B 54, 43 (1998) 55. L-J. Meng, M.P. dos Santos, Thin Solid Films 289, 65 (1996) 56. J.W. Bae, H.J. Kim, J.S. Kim, N.E. Lee, G.Y. Yeom, Vacuum 56, 77 (2000) 57. T.M. Miller, H. Fang, R.H. Magruder III, R.A. Weller, Sensors and Actuators, A: Phys. 104, 162 (2003) 58. H. Kim, C.M. Gilmore, J. Appl. Phys. 86, 6451 (1999) 59. F. Aduodija, H. Izumi, T. Ishihara, H. Yoshioka, Appl. Phys. Lett. 74 3059 (1999) 60. H. U. Gdel M. Pollnau, J. Alloys and Compounds 303, 307 (2000) 61. D.J. Robbins, J. Appl. Phys. 54, 4553 (1983) 62. P. D. Rack, A. Naman, P. H. Holloway, S. S. Sun, R. T. Tuenge, MRS Bulletin 21, 49 (1996) 63. S. Nakamura, Y. Yamada, T. Taguchi, J. Cryst. Grow. 214, 1091 (2000) 64. S. Okamoto, E. Nakazawa, Y. Tsuchiya, Jap. J. Appl. Phys. 28, 406 (1989) 65. A.C. Cefalas, S. Kobe, Z. Kollia, E. Sarantopoulou, Crystal Eng. 5, 203 (2002) 66. L.E. Tannas, Flat-Panel Displays and CRTs (Van Nostrand Reinhold, New York, 1985) 67. J.J. Brehm, W.J. Mullin, Introduction to the Structure of Matter (John Wiley and Sons Inc., New York, 1989) 68. J. Ballato, J.S. Lewis, P.H. Holloway, MRS Bulletin 24, 51 (1999) 69. T. Hirate, T. Ono, T. Satoh, SID International Disp. Res. Conf., 693 (2002) 70. K.A.Dunn, K. Dovidenko, A.W. Topol, G.S. Shekhawat, R.E. Geer, A.E. Kaloyeros, MRS Symp. Proc. 695, 41 (2002)

PAGE 159

145 71. Y-J. Wang, C-X. Wu, M-Z. Chen, M-C. Huang, J. Appl. Phys. 93 9625 (2003) 72. D.R. Lide, H.P.R. Frederikse, eds. CRC Handbook of Chemistry and Physics 78th ed. (CRC Press, Boca Raton, 1997) 73. D.B. Hollis, F.R. Cruickshank, M.J.P. Payne, J. Non-Cryst. Solids 293-295, 422 (2001) 74. B. T. Collins, J. Kane, M. Ling, R. T. Tuenge, S. S. Sun, J. Electrochem. Soc. 138, 3515 (1991) 75. S. Tanaka, S. Morimoto, K. Yamada, H. Kobayashi, Z. Zhang, X. Jiang, J. Cryst. Grow. 117, 997 (1992) 76. CT. Hsu, J. Cryst. Growth 208, 259 (2000) 77. A. B. Stambouli, S. Hamzaoui, M. Bouderbala, O. K. Omar, Appl. Energy 64, 207 (1999) 78. XY. Jiang, ZL. Zhang, WM, Zhao, ZG. Lui, SH. Xu J. Phys. Condens. Matter 6, 3279 (1994) 79. M. Potenza, IEEE Communications Magazine 34, 96 (1996) 80. L. Meng, C. Li, Phys Status Solidi A 119,677 (1990) 81. M. Wachtler, A. Speghini, K. Gatterer, H. P. Fritzer, D. Ajo, M. Bettinelli, J. Am. Ceram. Soc. 81, 2045 (1998) 82. E. W. Chase, R. T. Hepplewhite, D. C. Krupka, D. Kahng, J. Appl. Phys. 40, 2512 (1969) 83. F. J. Bryant, A. Krier, Phys. Stat. Sol. 81, 681 (1984) 84. A. K. Alshawa, H. J. Lozykowski, J. Electrochem. Soc. 141, 1070 (1994) 85. A. Ryer, Light Measurement Handbook (International Light Inc., Newburyport, Massachusetts, 1998) 86. P. R. Ehrmann, J. H. Campbell, J. Am. Ceram. Soc. 85, 1061 (2002) 87. Y.A. Ono, H. Kawakami, M. Fuyama, K. Onisawa, Jpn. J. Appl. Phys. 26, 1482 (1987) 88. A. Abu-Dayah, S. Kobayashi, J.F. Wager, Appl. Phys. Lett. 62, 744 (1993)

PAGE 160

146 89. E. Bringuier, J. Appl. Phys. 66, 1314 (1989) 90. R.L. Thuemler, MS Thesis, Electrical and Computer Engineering, Oregon State University (1997) 91. A. Abu-Dayah, MS Thesis, Electrical and Computer Engineering, Oregon State University (1993) 92. W.M. Ang, S. Pennathur, L. Pham, J.F. Wager, S.M. Goodnich, J. Appl. Phys. 77, 2719 (1995) 93. M.E. Abu-Zeid,. A.E. Rakhshani, A.A. Al-Jassar, Y.A. Youssef, Physica Status Solidi (A) Appl. Res. 93, 613 (1986) 94. M. Fatemi, Review of Progress in Quantitative Nondestructive Evaluation 5B (Plenum Press, New York, 1986) 95. K. Nakamura, T. Fuyuki, H. Matsunami, Jap. J. Appl. Phys. 37,4231 (1998) 96. S.O. Pillai, Solid State Physics, (New Age Inc.1998) 97. J.N. Spencer, G.M. Bodner, L.H. Rickard, Chemistry: Structure and Dynamics (John Wiley and Sons Inc., New York, 2003 98. P.K. Petrov, Z.G. Ivanov, S.S Gevorgyan, Mat. Sci. Eng. A 288, 231 (2000) 99. M. Warashina, M. Tajima, Japanese J. Appl. Phys. Part 1, 35 120 (1996) 100. V. D. Scott, G. Love, Quantitative Electron-Probe Microanalysis (Wiley & Sons, New York, 1983) 101. D.E. Newbury, D.C. Joy, P. Cechlin, C. Fiori, J.I. Goldstein, Advanced Scanning Electron Microscopy and X-ray Microanalysis (Plenum Press, New York, 1986) 102. Y. Kawakami, K. Omae,. A. Kaneta, K. Okamoto, T. Izumi, S. Saijou, K. Inoue, Y. Narukawa, T. Mukai, S. Fujita, Physica Status Solidi A Appl. Res. 183, 41 (2001) 103. J.H. Zhang, X.W. Fan, J. Cryst. Growth 121, 769 (1992) 104. A. Ashour, H.H. Afifi, S.A. Mahmoud, Thin Solid Films 248, 253 (1994) 105. T.W. Hickmott, G. Ehrlich, J. Phys. Chem. Solids 9, 47 (1958) 106. H.S. Hsu, Chemicool.com (H.S. Hsu, Boston, 1997)

PAGE 161

147 107. M. Winter, Webelements.com (WebElements Ltd, UK, 2003) 108. S. Lombardo, S.U. Campisano, G.N. Hoven, A. Polman, MRS Symp. Proc. 422, 333 (1996) 109. L-J. Meng, C-H. Li, G-Z. Zhong, J. Luminescence 39, 11 (1987) 110. M. Aozasa, H, Chen, K, Ando, Thin Solid Films 199 129 (1991) 111. N.P. Barnes, E.D. Filer, C.A. Morrison, Conf. Proc. Lasers Electro-Optics Soc., 286 (1996) 112. T. Honma, K. Toda, Z. Ye, M. Sato, J. Phys. Chem. Solids 59, 1187 (1998) 113. S. Guy, M. Malinowski, Z. Frukacz, M.F. Joubert,. B. Jacquier, J. Luminescence 68, 115 (1996) 114. R, Young, A.H. Kitai, J. Electrochem. Soc. 139, 2673 (1992) 115. J.P. Kim, Dissertation (Materials Science and Engineering, University of Florida, 2001) 116. Y-J. Wang, C-X. Wu, M-Z. Chen, M-C. Huang, J. Appl. Phys. 93, 9625 (2003) 117. J.P. Kim, M.R. Davidson, B. Speck,. D.J Moorehead, Q. Zhai, P.H. Holloway, J.Vac. Sci. Tech., Part A 19, 2490 (2001) 118. J.H. Kim, N. Shepard, M. Davidson, P.H. Holloway, Appl. Phys. Lett. 83, 641 (2003) 119. Q.Zhai, J.S. Lewis, K.A. Waldrip, K. Jones, P.H. Holloway, M. Davidson, N. Evans, Thin Solid Films 414, 105 (2002) 120. H.H. Afifi, S.A. Mahmoud, A. Ashour, Thin Solid Films 263, 248 (1995) 121. S. Lindroos, T. Kanniainen, M, Leskela, MRS Bulletin 32, 1631 (1997) 122. H. Venghaus, D. Theis, H. Oppolzer, S. Schnid, J. Appl. Phys. 53, 4146 (1982) 123. T. Yamamoto, Jap. J. Appl. Phys. 42, L514 (2003) 124. T. Tsuchiya, S. Ozaki, S. Adachi, J. Phys. Condens. Matter 15, 3717 (2003) 125. J.S. Browder, M.E. McCurry, S.S. Ballard, Appl. Opt. 23, 537 (1984)

PAGE 162

148 126. Y. Drezner, S. Berger, M. Hefetz, Mat. Sci. and Eng. B87, 59 (2001) 127. A. Kale, Dissertation (Materials Science and Engineering, University of Florida, 2001)

PAGE 163

BIOGRAPHICAL SKETCH William Robert Glass III (Bill) was born in Bridgeport, Connecticut, on December 18, 1973. He lived in Monroe, Connecticut, until 1980 when his family moved to Fairfield, Connecticut. He graduated from Fairfield High School in 1991. He attended Rensselaer Polytechnic Institute (RPI) from 1991 until 1995 when he graduated with a bachelors degree in physics with a minor in astronomy. He attended Lehigh University from September 1995 until December 1996 when he earned his masters degree in physics. He stayed at Lehigh University as a visiting scientist until his enrollment at the University of Florida in August of 1997. He received his Ph.D. from the Department of Materials Science and Engineering under the advisement of Dr. Paul H. Holloway. His areas of research have included development of infrared emitting electroluminescent flat panel displays as well as the structure and phonon properties of chalcogenide glasses. Mr. Glass interned at the Air Force Research Laboratories Space Vehicles Directorate during the summer of 2002 and was presented with a special service award for his research performed there. Mr. Glass has also won several awards from the American Vacuum Society for his presentations during the annual meeting of their Florida chapter. 149


Permanent Link: http://ufdc.ufl.edu/UFE0002403/00001

Material Information

Title: Sputter Deposition of Rare Earth Doped Zinc Sulfide for Near Infrared Electroluminescence
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0002403:00001

Permanent Link: http://ufdc.ufl.edu/UFE0002403/00001

Material Information

Title: Sputter Deposition of Rare Earth Doped Zinc Sulfide for Near Infrared Electroluminescence
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0002403:00001


This item has the following downloads:


Full Text












SPUTTER DEPOSITION OF RARE EARTH DOPED ZINC SULFIDE FOR NEAR
INFRARED ELECTROLUMINESCENCE















By

WILLIAM ROBERT GLASS III


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


2003

































Copyright 2003

by

William Robert Glass III















ACKNOWLEDGMENTS

First of all I would like to thank my advisor, Dr. Holloway. He has been the best

advisor I have known. It was an honor to work with him. I would also like to thank Dr.

Mark Davidson. Without his help, both mentally and physically, I would not have been

able to reach my goals. It was also a pleasure to work with all of the people out at

Microfabritech including Barbara, Diane, Scott, Chuck, Andreas, and Maggie.

Ludie, of course, deserves a huge thank you. Ludie is the best secretary a group

could ever have. Ludie was never without a smile and made things go smoother than I

could ever have imagined. I appreciate all of the members of Dr. Holloway's group

including Ajay, Nigel, Jie, Dave, etc for all of their help and support.

I, of course, want to thank my parents for their support and love. Without them I

would never have been able to make it to where I am today.

Finally, I want to thank my wife Jackie. Without her I would have been lost. She

is the best thing that has ever happened to me. Words are not enough to express my love

to her.
















TABLE OF CONTENTS
Page

A C K N O W L E D G M E N T S ......... ................................................................................... iii

L IST O F TA B L E S ......... ....... ........ ....... .. ......... .. .... ............ .. vii

LIST OF FIGURES .......................... ....... .................. viii

A B S T R A C T .......................................... ..................................................x iii

CHAPTER

1 IN TRODU CTION ................................................. ...... .................

2 B A C K G R O U N D .................... .... .............................. .......... ........ .......... .. ....

2 .1 In tro d u ctio n ............................................................................... 3
2.2 Infrared Em hitters ................................................ .................... .... 6
2.3 Electrolum inescent D vice Structure ........................................... ............... 8
2 .4 D ev ic e P h y sic s ................................................................................................ 1 1
2.5 A C TFEL D M materials .................................................. .............................. 17
2.5.1 Substrates .................... ........ ..............17
2 .5 .2 In su lato rs ..............................................................19
2.5.3 Conductors ................ ... .............. ........22
2.6 Phosphor Luminescence ................................... .. ...............24
2.6.1 H ost M materials ................................................ ............... 24
2.6.2 Lum inescent Centers ................................ ........................................... 27
2.6.3 R are Earth D oped ZnS.................................... .................. 29
2 .6 .3 .1 Z n S :T m ........... ........................................................ ......... .... 30
2.6.3.2 Z nS:E r ...................... .................. .................................... 31
2 .6 .3 .3 Z n S :N d ..................................................................... ................. 3 2
2.7 Electrical and Optical Characterization................................... ............. .........32
2.7.1 B rightness versus V oltage ........................................ ...................... 33
2 .7 .2 T hresh old V oltag e ........................................................... .....................3 5
2.7.3 Efficiency versus V oltage...................................... ......................... 35
2.7.4 Electrical Testing ...... ............................ ........ .... .. ............ 37
2.7.5 Charge versus Voltage (Q-V)................................ ............... 38
2.7.6 Capacitance versus V oltage..................................... ....................... 42
2.7.7 Internal Charge versus Phosphor Field....................................................43
2.7.8 Maximum Charge versus Maximum Voltage ........................................46









3 EXPERIMENTAL PROCEDURE................................... .................................... 48

3.1 Substrate and Target Preparation................................... .................................... 48
3.2 Sulfide Sputter Deposition System .................................................................... 48
3.3 T op C contact D position .......................................................................... .... ... 52
3.4 Sam ple H handling and Storage ........................................ ......................... 53
3.5 Sputtered Film Characterization................................ ......................... ....... 53
3.5.1 Thickness Measurements......................................... 54
3.5.2 X -ray D iffraction (X RD ) ............................................................ .......... 54
3.5.3 E lectrolum inescence........................ .... ............ ... .......................... .............56
3.5.4 Photoluminescence and Photoluminescent Excitation.............................59
3.5.5 Electron M icroprobe.................... ................. .................. 60
3.5.6 Energy Dispersive X-ray Spectroscopy (EDS) on a Scanning Electron
M icroscope (SEM ) ......................................... ........ .. .. .. .. ..............61
3.5.7 Time Resolved Electroluminescence ............................... ............... 62
3.5.8 Electrical M easurem ents ........................................ ........................ 63

4 PHYSICAL EFFECTS OF CHANGING TARGET DUTY CYCLES AND
SUBSTRATE TEMPERATURE DURING RF MAGNETRON SPUTTER
D E P O SIT IO N ............. .. ........... ...................................................................... 6 5

4 .1 Introdu action ........ ........................ ....... .................................. .... 6 5
4.2 Spectra ........................................................................65
4.3 Target D uty Cycle A lteration ........................................ .......... ............... 69
4 .3 .1 C on centration .............................. ........................ .. ........ .... ............69
4 .3 .2 C ry stallinity ............................................................69
4 .3.3 T sickness .......................................................................................... .. ....... 7 1
4 .4 .4 T hresh old V oltag e ........................................................... .....................72
4.4.5 Infrared Em mission ....................................... .................................73
4.5 Deposition Temperature Effects................................ ......................... ....... 74
4 .5 .1 C on centration .............................. ........................ .. ........ .... ............74
4 .5.2 C ry stallinity ................................................................... 75
4 .5.3 T sickness ......................................................... ....... ........... 76
4 .5 .4 T hresh old V oltag e ........................................................... .....................77
4.5.5 Infrared Em mission ....................................... .................................78
4.6 D discussion ..................................................................... ......79
4.7 Comparison of Infrared to Visible Emission.................................................. 85

5 ELECTRICAL EFFECTS OF CHANGING TARGET DUTY CYCLES AND
SUBSTRATE TEMPERATURE DURING RF MAGNETRON SPUTTER
D E P O SIT IO N ............. .. ........... ...................................................................... 89

5.1 Introduction ......................................... ... ...... .. ... ... ........... 89
5.2 Charge-V oltage (Q -V ) D ata ........................................ ........................... 89
5.3 C-V D ata ..................................... .................. ............. ........... 97
5 .4 Q int-F p D ata ......... .............. ............................. ................................ 100
5.5 Time Resolved Electroluminescence..................... ........... ... ............. 107


v









5.5.1 Discussion of TREL Data............................ ....................... 108
5.6 Discussion ......... ...... .. ........................ ............... 117
5.6.1 Q -V A analysis .................................... ................. ................ 117
5.6.2 C -V A naly sis ........................ .... ................ ... ....... .... ...........125
5.6.4 Interface Layer Discussion .......................... .......................... 132

6 CONCLUSIONS ....................................... ...... ... .. .............138

6.1 Deposition Effects on the Physical Properties and Optical Properties of ZnS:RE
Phosphors .....................................138
6.2 Electrical Properties of ZnS:RE Phosphors............. ...... .................. 139

LIST OF REFERENCES ......... ...................................... ........ .. ............... 141

B IO G R A PH ICA L SK ETCH ............ .................................................... .....................149
















LIST OF TABLES

Table pge

2-1 List of insulators used in ACTFEL devices and their properties of interest ...............22

2-2 Properties of ZnS and SrS ......... .... ....... ................. .............................. 25

2-3 Optical properties of common sulfide based EL materials..............................28

2-4 Physical properties of ZnS............................................. ................. ............... 30
















LIST OF FIGURES


Figure pge

2-1 Sketch of phosphor-LE P lam p ............... .................................... .............................

2-2 Cross-sectional view of (a) normal and (b) inverted ACTFELD structure ...................9

2-3 Equivalent circuit for an A CTFEL device ..................................................................11

2-4 Energy band diagram illustrating the five primary physical processes responsible for
ACTFEL device operation ................................... .......... .................. 12

2-5 Energy band diagram of an ACTFEL device with and without space charge in the
phosphor layer ................. ......... .................... 18

2-6 Energy level diagrams and radiant transitions of Tm3+, Nd3+, and Er3......................26

2-7 Impact cross sections of the 3F4 and 1G4 levels in Tm3+ [78] ..........................31

2-8 Brightness vs. voltage curve showing the threshold voltage.............. ............ 34

2-9 ACTFELD efficiency versus drive voltage..... ...................................36

2-10 Schematic of a Sawyer-Tower test setup........................................ ............... 37

2-11 Trapezoidal waveform with important points marked for reference.........................38

2-12 Typical Q -V plot............................... .... .......... ...... ........ .. 40

2-13 Typical C-V plot...................... ........... .. ........... .............. .. 43

2-14 Typical Q int-Fp plot ......................................................... ........ ... ...... ...... 45

2-15 Typical Q max-V max plot ........................................ .................. ............... 47

2-16 Typical Q emax-V max plot ................................................. ................................ 47

3-1 Schematic of the sputter system used for RF magnetron sputtering ...........................50

3-2 View of sample platter showing substrate positions and spaces for additional
substrates ............................................................... .... ..... ........ 51









3-3 Schematic of the heating system in the sputtering system .......................................52

3-4 Back view of the sample on the test stage.................................... .............. 57

3-5 Spectral sensitivity of the Ocean Optics #13 grating ...............................................58

3-6 Side view of the sample stage and fiber optic detection system .............. ...............59

3-7 System to measure time resolved luminescence and electrical data .........................63

4-1 Electroluminescent spectrum of ZnS:TmF3 ..................................... .................66

4-2 Electroluminescent spectrum of ZnS:NdF3 ............. ........... ..... ........ ........... 67

4-3 Electroluminescent spectrum of ZnS:ErF3 ............................... ............ ............ 67

4-4 Energy levels of rare earth ions and transitions luminescence producing transitions
observed in Figs. 4-1, 4-2 and 4-3................ ......... ........................ ............... 68

4-5 Effect of target duty cycle on the Tm, Nd, and Er concentrations in the ZnS films
m measured by ED S and EPM A ....................................................................... ..... 70

4-6 Effect of duty cycle ratio on the full width at half maximum of the 28.50 x-ray
diffraction peak of ZnS ........... ............. ........... ... ............ ................ 71

4-7 Normalized thickness of the rare earth doped ZnS films. Deposition times were
changed to attempt to achieve the same thickness for each rare earth film. ............72

4-8 NIR threshold voltages of the doped ZnS films with varying deposition duty cycles 73

4-9 Effect of target duty cycle on the near infrared emission of each rare earth ..............74

4-10 Concentration of each rare earth in the ZnS films as a function of substrate
temperature during deposition measured by EDS..............................................75

4-11 Increasing FWHM of the ZnS 28.5 diffraction peak as the deposition temperature is
increased ................................... .................. ............... ........... 76

4-12 Decreasing phosphor thickness with increasing deposition temperature ................77

4-13 Optical turn on voltage variation with increasing deposition temperature for each
m material .............. ... ......................................................... 78

4-14 Decrease of near infrared irradiance with increasing deposition temperature ..........79

4-15 Comparison of NIR turn on voltage and phosphor thickness as deposition
tem perature is varied ............................. ................ ...... ... .... ........ 81









4-16 Comparison of NIR turn on voltage and phosphor thickness as duty cycle and
deposition tim e is varied ................................................ .............................. 82

4-17 NIR irradiance as a function of rare earth concentration. Note that the maximum
occurs near 1 at% for each rare earth. ........................................... ............... 84

4-18 B40 (left ordinate) for the NIR emission and B40 (right ordinate) for the visible
emission in ZnS:TmF3 for various Tm concentrations ........................................86

4-19 B40 (left ordinate) for the NIR emission and B40 (right ordinate) for the visible
emission in ZnS:NdF3 for various Nd concentrations............................................87

4-20 B40 (left ordinate) for the NIR emission and B40 (right ordinate) for the visible
emission in ZnS:ErF3 for various Er concentrations...................... .............88

5-1 Typical Q-V data for ZnS:TmF3 at drive voltages between 100 and 150 volts ..........91

5-2 Typical Q-V data for ZnS:NdF3 at drive voltages between 80 and 170 volts.............92

5-3 Typical Q-V data for ZnS:ErF3 at drive voltages between 80 and 150 volts..............93

5-4 Electrical threshold voltages for each phosphor as a function of duty cycle ..............94

5-5 Electrical threshold voltages for each phosphor as a function of deposition
temperature ............... ......... .... ............ ............. ............ 95

5-6 Plot of Q-V of ZnS:TmF3 at B40 with increasing deposition temperature (140-1800C)96

5-7 Plot of Q-V of ZnS:NdF3 at B40 with increasing deposition temperature ................97

5-8 Typical C-V data for ZnS:TmF3 at drive voltages between 100 and 150 volts...........98

5-9 Typical C-V data for ZnS:NdF3 at drive voltages between 80 and 170 volts ............99

5-10 Typical C-V data for ZnS:ErF3 at drive voltages between 80 and 150 volts ..........100

5-11 Internal Charge vs. phosphor field for increasing voltage in ZnS:TmF3...............02

5-12 Internal Charge vs. phosphor field for increasing voltage in ZnS:NdF3 ...............03

5-13 Internal Charge vs. phosphor field for increasing voltage in ZnS:ErF3 ................104

5-14 Internal charge vs. phosphor field for ZnS:TmF3 as the deposition temperature is
c h a n g e d ...................................................................... 1 0 5

5-15 Internal charge vs. phosphor field for ZnS:NdF3 as the deposition temperature is
c h a n g e d ....................................................................... 1 0 6









5-16 Internal charge vs. phosphor field for ZnS:ErF3 as the deposition temperature is
c h a n g e d ...................................................................... 1 0 7

5-17 Time resolved electroluminescence of the NIR and blue emission from ZnS:TmF31 10

5-18 Time resolved electroluminescence of the visible emission from ZnS:NdF3 for
voltage pulse durations of 5 and 30 s.............. ................... ..................... 111

5-19 Time resolved electroluminescence of the visible emission from ZnS:ErF3........... 112

5-20 Log plot of TREL decay of the 480 nm emission from ZnS:TmF3.........................113

5-21 Log plot of TREL decay of the 800 nm emission from ZnS:TmF3.........................114

5-22 Log plot of TREL decay of the 600 nm emission from ZnS:NdF3 for a 30 gs voltage
p u lse .............................................................................................1 1 5

5-23 Log plot of TREL decay of the 600 nm emission from ZnS:NdF3 for a 5 gs voltage
p u lse .............................................................................................1 16

5-24 Log plot of TREL decay of the 530 nm emission from ZnS:ErF3 ..........................117

5-25 Energy band diagram of an ACTFEL device showing how the distribution of
interface states can affect the electric field necessary for tunnel injection ............118

2-26 Transferred charge versus maximum applied voltage showing the electrical
threshold for a typical ZnS:TmF3 device .................................... ............... 120

5-27 Irradiance from ZnS:Tm versus applied voltage showing the optical threshold is the
sam e for N IR and visible em mission ............................................. ............... 121

5-28 Irradiance from ZnS:Nd versus applied voltage showing the optical threshold is the
sam e for N IR and visible em mission ............................................. ............... 122

5-29 Irradiance from ZnS:Er versus applied voltage showing the optical threshold is the
sam e for N IR and visible em mission ............................................. ............... 123

5-30 Comparison of optical and electrical threshold voltages with changing duty cycle
ratios for each dopant ................................................ .. ...... .. ............ 124

5-31 Comparison of optical and electrical threshold voltages versus deposition
tem perature for each dopant ........................................................ ............. 125

5-32 Normalized internal charge, phosphor field and NIR brightness versus Tm
concentrations in ZnS:TmF3. Note that while the average of internal charge is
nearly constant, the trend for both B40 and Fp is down as the temperature
increases. This correlation is discussed in the text...... .... ................................. 128









5-33 Relation of internal charge and phosphor field with NIR brightness for ZnS:NdF3
with changing deposition temperature ....................................... ............... 129

5-34 Relation of internal charge and phosphor field with NIR brightness for ZnS:NdF3
w ith changing target duty cycle ........................................ ......................... 130

5-35 Relation of internal charge with NIR brightness for various Nd concentrations in
Z n S :N dF 3 ....... .... ....... .. ........................................................... ........... .. 13 1

5-36 Relation of internal charge and phosphor field with NIR brightness for ZnS:ErF3
with changing deposition temperature. Note that the brightness correlates with Fp
and not with the internal charge ........................................132

5-37 Calculated interface layer thicknesses for ZnS:TmF3 as a function of deposition
temperature and duty cycle (Duty cycle data plotted at the duty cycle + 100 i.e. a
duty cycle ratio of 50 is plotted at 150)....................................... ............... 136

5-38 Calculated interface layer thicknesses for ZnS:NdF3 as a function of deposition
temperature and duty cycle (Duty cycle data plotted at the duty cycle + 100 i.e. a
duty cycle of 50 is plotted at 150) ............... .................... ........... ............ ...137















Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

SPUTTER DEPOSITION OF RARE EARTH DOPED ZINC SULFIDE FOR NEAR
INFRARED ELECTROLUMINESCENCE

By

William Robert Glass III

December, 2003

Chair: Dr. Paul H. Holloway
Major Department: Materials Science and Engineering


Near infrared emitting alternating current thin film electroluminescent (ACTFEL)

phosphors were fabricated by simultaneous R.F. magnetron sputtering from both a target

of doped ZnS and an undoped ZnS target. The intensities of both near infrared (NIR) and

visible emission from ZnS doped with thulium (Tm), neodymium (Nd), or erbium (Er)

fluorides were dependent on deposition parameters such as target duty cycle (varied from

25 to 100% independently for the two targets) and substrate temperature (140-180C),

with lower temperatures giving 400% better NIR brightness. By optimizing the rare earth

concentration between 0.8 and 1.1 at%, the near infrared irradiance was improved by

400% for each dopant. The increase in brightness and optimal concentrations are

attributed to decreased crystallinity and increased dopant interaction at higher rare earth

concentrations. The brightness increase with decreasing deposition temperature was

attributed to a reduction of thermal desorption of the ZnS during deposition, and









consequently thicker films and optimized rare earth concentration. Luminescent decay

lifetimes were short (20-40 [tsec) because of a high concentration of non-radiative

pathways due to defects from the strain of the large rare earth ions on the ZnS lattice.

The threshold voltage for visible and near infrared emission was identical despite

emission of NIR and visible light resulting from electrons relaxing from low and high

energy excited levels, respectively. The optical threshold voltages were identical to the

electrical threshold voltages, and it was concluded that at the voltages necessary for

electrical breakdown, the accelerated electrons had enough energy to excite either the

visible or NIR emitting levels. Phosphors doped with Nd exhibited increased internal

charge at higher dopant concentrations despite a reduction in phosphor field (i.e. reduced

applied voltage) In contrast; the charge did not change appreciably for Er and decreased

for Tm doped films at reduced fields. The charge differences were attributed to dopant

effects on the distribution of states near the interfaces. It was postulated that Nd doped

devices have a shallower state distribution, while the majority of states in Tm doped

devices are deeper and require higher fields for tunnel injection. The electrical behavior

of all of the devices also demonstrated that field clamping occurred despite non-ideal

phosphor breakdown during device operation. It is postulated that a high breakdown

strength, low dielectric constant, interface layer is formed during deposition, and reduces

capacitance before and after phosphor breakdown and results in field clamping. The

thickness calculated for the interface layer decreases with increasing deposition

temperature implying that the layer is formed during deposition, and this decreasing

thickness results from increased atomic mobility at higher temperatures.














CHAPTER 1
INTRODUCTION

Recently, much interest has been given to technologies for emitting visible light for

use in flat panel displays. One of these technologies is the alternating current thin film

electroluminescent device (ACTFELD) [1]. While visible emitting ACTFEL devices

have garnered much attention, little attention has been given to infrared emitting devices.

Near infrared emitting ACTFEL devices are suitable for applications that require

mechanically robust, thermally stable devices than have lower power consumption than

infrared emitting resistive devices.

ZnS doped with various rare earths ions are promising materials for the

development of infrared emitting ACTFEL phosphors [2]. While phosphors such as ZnS

doped with Tm, Nd or Er emit blue, orange, and green visible light, they also emit

strongly in the near infrared region (0.7-2 um). However, infrared emission from these

phosphors is undesirable when used for their visible output. In this study the relationship

between visible and infrared emission and the determination of the deposition conditions

necessary for maximizing the infrared output of these devices has been performed.

Chapter 2 will present background information on infrared emitting devices as well

as a review of ACTFELD structures and operation. In chapter 3 the experimental and

characterization methods and equipment used in this study will be presented. In chapter

4, results will be presented that show a dramatic increase in the infrared output of the rare

earth doped ZnS by alteration of deposition conditions during R.F. magnetron sputtering.

It will be shown that the rare earth concentration is a critical parameter determining the






2


intensity of infrared and visible emission. Chapter 5 will present electrical

characterization data and a discussion of the factors limiting the output of these materials

and devices. Finally, conclusions will be presented in chapter 6.














CHAPTER 2
BACKGROUND

2.1 Introduction

Much work has recently been done on the development of visible thin film

phosphors for use in flat panel displays. Thin film phosphors which emit in the infrared

have often been overlooked. While infrared phosphors do not have the same markets as

their visible counterparts, there are applications in which infrared alternating current thin

film electroluminescent devices (ACTFELDs) are desirable. Industry can use infrared

emitting devices for absorption based gas sensors or production of thermal bandages and

the auto industry has investigated infrared systems for improving safety during night

driving [3]. Military applications include night vision, friend/foe identification, scene

projectors for night mission training, and infrared portable computer screens for night

operations.

Industrial gas sensors operate by light absorption of a gas through the vibration-

rotation bands of polar molecules. When these bands are centered at wavelengths

characteristic of the bending and stretching of the molecules, the absorption depends on

the number of molecules in the light path [4]. For example, devices with emission at 761

nm can be used to detect oxygen [5]. Currently thermal sources, such as tungsten lamps,

are the light sources for most gas sensors. However, advances in semiconductor

technology and a decrease in component costs can lead to the replacement of filtered

thermal sources in gas sensors.









Automobile companies such as Daimler Chrysler are testing infrared illumination

systems to make night driving safer. Daimler Chrysler has fitted active night vision

systems onto its Jeep Grand Cherokee and had tested the system on a bus. The bus's

night vision system allows the driver to "see" further than with conventional headlights

without blinding oncoming drivers [6]. Other auto companies currently investigating

night vision include Acura, Cadillac, and Volvo [7].

The United States military has wanted to engage its enemies under cover of night

since the revolutionary war. Such attacks proved to be extremely dangerous until

effective night vision equipment was developed. The first true night vision systems were

developed during World War II in the form of infrared sniper scopes. These scopes

emitted an infrared light that the scope could detect and turn into a visible picture. While

current military practice focuses on passive night vision (the amplification of existing

light), active night vision may be a more effective tool. During desert storm military

helicopters had infrared aiming lights installed on their landing skids to avoid sand dunes.

The helicopters were in no danger of being seen because the Iraqi army did not have near

infrared detection devices [8]. IFF (identification friend or foe) has concerned the

military since World War II. IFF was developed in England to avoid shooting down their

own planes when they returned home. IFF is a concern whenever aircraft are in the sky

[9].

Infrared emitting phosphors can be used in each of these applications. A

phosphor is a material that emits light when excited by an energy source. Emission that

ceases within 10 nanoseconds of the excitation is known as fluorescence [10]. Longer

lasting luminescence, known as phosphorescence, can last hours [11]. The exciting









energy can be photonic, electronic, ionic, or thermal. Thin film phosphor devices usually

operate in one of several ways. Photoluminescent devices are excited by higher energy

photons from sources such as ultraviolet lamps or lasers [12]. Cathodoluminescent

devices, such as televisions, operate by the emission of electrons from a tip or electron

gun that strike the film [13]. Electroluminescent devices use an applied electric field

across the phosphor to induce luminescence [14].

Research into rare earth doped zinc sulfide has been concentrated on the search

for efficient red, green, and blue phosphors; infrared emission from these materials was

overlooked or actively discouraged to improve the efficiency of visible emission. Zinc

sulfide doped with thulium is a blue emitting phosphor whose emission is generally too

weak for use as a display phosphor however; it exhibits significant near infrared (NIR)

emission [15]. Neodymium and erbium doped zinc sulfide also emit in both the visible

and infrared regions. Neodymium emits in the orange and erbium emission is stronger in

the green regions, with weaker emission in the red. Unlike thulium and neodymium, the

infrared emission from erbium has been of interest, mainly for telecommunications [16].

Strontium sulfide also has been studied as a host for rare earth phosphors. While SrS is a

better host for blue devices due to its superior electron high-field transport properties

[17], ZnS is better for infrared. Hot electrons (the excitation source in electroluminescent

ZnS doped with TmF3 or other rare earths, as shown below) in ZnS do not appear to have

enough energy to excite shorter wavelength luminescent centers [18]. This leads to

decreased blue emission compared to SrS, but these electrons can stimulate infrared

emission. As discussed below, the ratio of infrared to visible emission is dependent on

deposition conditions.









Modification of the phosphors, including codoping with alkalis such as lithium,

has been tested to improve the visible brightness of ZnS:RE films by lowering the

symmetry around the rare earth [19]. These alterations succeeded in decreasing the

infrared to visible ratio. In addition, others have introduced oxygen into the phosphor

films in an effort to increase the visible luminescence. While this was effective in

increasing the blue emission in ZnS:TmF3, it also increased the infrared output. These

increases are thought to result from reduced non-radiative transitions at sulfur vacancies

[20]. The non-radiative transitions are caused by the defects produced at the sulfur

vacancies. It is possible to improve the crystallinity of the ZnS by annealing etc. without

needing to add oxygen. Finally, because of the decrease in infrared emission, doping

with alkalis should be avoided if an infrared emitter is desired. For these reasons rare

earth doped ZnS phosphors used for infrared emission are often deposited simply as

fluorides.

2.2 Infrared Emitters

There are several sources of infrared light other than thin film devices. The most

common are light emitting diodes (LEDs), lasers, and thermal emitters. Infrared LEDs

are the analog of the common visible light LEDs. One of the possible drawbacks with

LEDs is that they are limited to a fairly large size compared to the possible pixel size of

an electroluminescent thin film. This makes LEDs undesirable for screen applications

such as scene projectors or more flexible applications such as thermal bandages.

However, rare earth ACTFLED phosphors can be used in LEDs for other applications by

depositing the phosphor on a blue emitting GaN chip and using the blue light to photo

excite the phosphor (Figure 2-1). A major drawback of such a design is a loss of

efficiency [21].










Combination of light
emitted by the LED
and the phosphor



\ Phosphor grains



LED chip









Figure 2-1 Sketch of phosphor-LEP lamp

Infrared lasers can be much more intense than infrared ACTFEL devices but they

are usually limited to applications that an ACTFELD would not be suited for. Infrared

lasers are useful for directional applications such as target identification but fail when a

more omni-directional device is needed. In addition a lasers emission wavelength is

unstable with temperature, drifting several nanometers as the temperature changes [22].

Applications such as gas sensors need stable light sources to function properly. Infrared

lasers can also have problems with long-term stability due to amplitude variations when

wavelength modulated [23].

Thermal emitters are similar to the filament of an incandescent light bulb. The

main differences are the material used and the temperature of the glower. A common









type of thermal emitter is the Globar. Globars are silicon carbide rods that are heated a

desired temperature. The emission of the globar approximates that of a blackbody source

at the same temperature [24]. Two of the drawbacks of thermal emitters are that they

need to be heated to elevated temperatures to emit strongly in the near infrared and

because of their blackbody character they do not emit at distinct wavelengths but instead

over a wide spectrum.

2.3 Electroluminescent Device Structure

Electroluminescent devices are flat electrically driven light emitters that use an

electric field to produce luminescence without heat generation. The structure of an

ACTFELD is essentially that of a dielectric-phosphor-dielectric sandwich. A complete

device consists of a conductor, insulator, phosphor, insulator, conductor stack deposited

on a substrate [25, 26]. Thin film electroluminescent devices have two basic designs

based on the same structure. Typically, a 'normal' device is deposited on a transparent

substrate with a transparent conductor and insulator between the phosphor and the

substrate. The top dielectric may be transparent or opaque and the top conductor is often

reflective. A so-called 'inverted' structure is the same layer sequence deposited on an

opaque substrate with a transparent top insulator and conductor. An inverted structure is

viewed through the top electrode while a regular device is viewed through the substrate

(Figure 2-2)[15].

























































Figure 2-2 Cross-sectional view of (a) normal and (b) inverted ACTFELD structure


Reflecting Electrode

Top Insulator


Phosphor


Bottom Insulator

Transparent Electrode


Glass Substrate


Transparent Electrode
Top Insulator


Phosphor



Bottom Insulator

Reflecting Electrode


Opaque Substrate









Both standard and inverted ACTFEL devices are commercially used. The choice

of which structure to use depends on the application and processing requirements. The

typical transparent substrate structure has several advantages over the inverted structure.

One advantage is that if a suitable top conductor, such as aluminum, is used then the

device experiences self-healing breakdown [27]. Self-healing causes the top electrode to

pull back from short circuit paths such as pinholes and defects preventing catastrophic

device failure. The electrode maintains effective contact to the rest of the device while

isolating the short. Another advantage of this structure is its inherent durability. Since

this device is viewed through the substrate, the films are protected. An advantage of the

inverted structure is higher processing temperatures. At about 600C the glass substrate

commonly used for visible emitting normal structures begins to buckle and melt. Using

an inverted structure, a higher melting temperature material, such as silicon, can be used.

A disadvantage of the inverted structure is that self-healing top electrodes are not

possible with transparent conductors. This means that the phosphor must have a very low

defect density for the device to be reliable.

Another ACTFEL device structure, commonly used for testing, is the one-insulator

or "half stack" structure. As the name implies, a half stack device is the same as either a

standard or inverted device except that one of the insulators is missing, while a "full

stack" device has both insulators. The removal of this insulating layer from the device

reduces the time needed to produce a device by eliminating one of the processing steps.

Another advantage of half stack devices is that they tend to run at lower voltages than a

comparable full stack device. However, half stack devices leave the phosphor layer more

exposed than full stack devices and therefore exhibit poor long term reliability.







2.4 Device Physics
Understanding the basic physics of ACTFEL devices give insight into how they
may be improved. An ACTFEL device can be modeled as circuit in which the phosphor
is represented as a capacitor shunted by back-to-back Zener diodes with the insulators
represented as capacitors [28] (Figure 2-3). Operation of an ACTFEL device follows five
basic steps. They are (1) electron injection from interface states, (2) electron transport
across the phosphor, (3) excitation of luminescent centers, (4) photon emission from
radiative recombination, and (5) electron trapping [29]. These steps are shown in figure
2-4.




Cil




Cp


Ci2

Tb


Figure 2-3 Equivalent circuit for an ACTFEL device
















1 2


Figure 2-4 Energy band diagram illustrating the five primary physical processes
responsible for ACTFEL device operation


When the applied voltage is below the threshold voltage, the electrical circuit

characteristics are that the Zener diodes are below their breakdown voltage. Hence, an

ACTFELD below electrical threshold can be modeled simply as three capacitors. The

capacitance for each of the layers can be modeled as parallel plate capacitors using the

following equation.











t

where C is the capacitance of the layer, Er is the relative permittivity, So is permittivity of

free space, A is the area, and t is the thickness of the layer [30]. The equation for the

whole device is simply that of three (or two in the case of a half stack) capacitors in

series,

SC p C, top C bottom
pC, top p CpC bottom C, topC, bottom

where Cp is the capacitance of the phosphor and Citop and Cibottom are the capacitances of

the top insulator and bottom insulator respectively. For the half stack device this

equation simplifies to

C cpC,'
Cp +C,

When the applied voltage becomes high enough, the phosphor reaches its threshold

voltage; the circuit behaves as though the Zener diodes have reached their breakdown

voltage; and the capacitance of the device is now just that of the insulating layers.

Therefore, during device operation, injection of electrons from the insulator-phosphor

interface into the phosphor occurs when a voltage large enough to breakdown the

phosphor is applied to the device. When threshold is reached, the electrons trapped in

interface states can tunnel into the conduction band of the phosphor [31]. The large

electric field in the phosphor layer accelerates the electrons to ballistic energies and they

travel across the phosphor. Sufficiently hot electrons may excite the host or non-

luminescent centers which then transfer energy to the luminescent dopant, or the

electrons may directly strike the luminescent centers causing impact excitation or impact









ionization. After this collision, the electrons are again accelerated and the process

continues. Once an electron travels across the phosphor from either the interface or from

impact ionization, it will be captured by interface states on the other side of the phosphor.

It is possible that electrons can be trapped in bulk states creating a space charge on the

other side of or throughout the phosphor. Once the next voltage pulse arrives, the

polarities of the electrodes are switched and the process begins again in the opposite

direction.

The interface between the insulator and the phosphor can be modeled after a

Schottky barrier. The tunnel emission for a Schottky barrier is given by [32]




J & E2 exp- 8s- 2~m (qB)2
3qhE


where E is the electric field, m* is the electron effective mass, q is the charge of an

electron, (B is the barrier height, and h is Planck's constant. For interface state

emission, the equation must be modified by replacing the barrier height with the interface

trap depth. While the tunneling is temperature independent, the device current is

temperature dependent. Thermionic emission has been suggested as an additional

mechanism for charge injection. The Richard-Dushman equation for thermionic

emission is [33]



=r 2mq e -w/


where Je is the electric charge flux, c is the temperature multiplied by Boltzmann's

constant, m is the mass of an electron, q is the charge of an electron, < is Planck's









constant divided by 2B, and W is the work function. This equation is for the metal-

vacuum interface, so in the ACTFELD case the equation must be modified to take the

phosphor's electron affinity into account. Roughening of the insulator-phosphor

interface creates a wider interface region resulting in a broader distribution of interface

trap energy. This can lower the field necessary to turn on the EL device [34].

Once the electrons have been injected into the phosphors conduction band, they

must be accelerated to high enough energies to induce luminescence (typically >2eV).

The electric field in the phosphor can be calculated by rearranging Maxwell's equations

for a series of capacitors yielding


E=
P E Kdp +Epd, to


where Ep is the phosphor electric field, ; is the dielectric constant, d is the thickness of

the layer, and the subscripts i and p are for the insulator and phosphor, respectively.

Inserting typical values for the dielectric constants and the thickness yield electric fields

of about 2 to 2.5 MV/cm. Electrons accelerate very quickly in this high field. Their

energies are limited by scattering, which can occur by several mechanisms, including

low-energy quantum states [35]. Interface roughening, as mentioned earlier, broadens the

energy distribution of traps at the interface. A broad energy distribution will allow

tunneling of electrons in higher energy states to occur at lower electric fields. The

acceleration due to the weaker field will result in lower energy ballistic electrons. The

lower fields will not accelerate the electrons to as high an energy as would a large field.

The energy levels necessary for infrared radiative transitions lie lower than those for

visible emission, so it would appear that the lower energy electrons would result in









increased infrared emission at lower voltages. However, this has not been tested, so it is

unknown how the relative emission from visible and infrared emitting transitions will be

affected.

Energetic electrons may cause excitation of the host material or directly excite the

luminescent centers in the phosphor. As the excited host ions return to a lower energy,

the excited electrons may transfer energy through exciton states to the luminescent

centers in the device or lose the energy to phonons, plasmons or Auger transitions [36].

With high enough energies the hot electrons can interact with the luminescent centers

promoting ground state electrons to higher energy levels. As previously mentioned, the

electrons can either be promoted to the conduction band of the host or to a higher level

within the atom through impact ionization and impact excitation respectively [37]. The

probability of an interaction is related to the impact cross section which will be discussed

in the phosphor luminescence section. An electron that is impact excited to a higher

energy level can then de-excite radiatively or non-radiatively. Non-radiative de-

excitation usually occurs through phonon generation. Phonon energies are small

compared to photon energies, usually about 20 meV [38]. Radiative de-excitation occurs

through photon generation with the photon energy matching the energy level transition of

the electron [39]. When the electron promoted into the conduction band of the host

material is carried away by the electric field, it will either impact an ion in the phosphor

or be carried to the interface. A luminescent center can only emit light when it captures

another electron through a non-radiative transition from the conduction band into one of

the atoms excited states. If the band gap of the host is a lower energy than the excited

state of the luminescent center, visible or near IR emission is greatly reduced [40].









The previous description does not take into account space charge, a very common

occurrence in ACTFEL devices [41]. Some of the electrons or holes in the phosphor may

be trapped in bulk trap states and create a space charge. The space charge will produce

bending of the bands near the interface causing the field across the phosphor to be non-

uniform. If holes are concentrated near the cathode then the field will have an increased

strength near the cathodic interface and lower strength as it approaches the anode (Figure

2-5). Space charge is presumed to result from ionization of deep traps at the interface,

field emission from bulk traps, or band to band impact ionization and subsequent hole

trapping [42,43,44]. Space charge in SrS phosphors has been photo-induced [45]. Space

charge generation in ZnS:MnCl has been attributed to the impact ionization of zinc

vacancies that are part of chlorine-zinc complexes [46]. Zinc-fluorine complexes formed

when using fluorides instead of chlorides as the starting compounds could lead to similar

states.

2.5 ACTFELD Materials

2.5.1 Substrates

The substrate for a standard ACTFELD needs to be transparent, smooth, robust,

and preferably inexpensive. The substrate of a visible ACTFELD is often Corning 7059

soda-lime glass. Corning 7059 glass has a softening temperature of about 600C so rapid

thermal annealing below 650C is possible but anything higher will deform the substrate

[47]. Smaller samples, up to 2 inches square, may be annealed up to 8500C for short

times. In addition, Corning 7059 glass is free of alkalis; so alkali diffusion into the

device is avoided [48]. For phosphors requiring higher temperature anneals or for mid-
















No Space Charge


Figure 2-5 Energy band diagram of an ACTFEL device with and without space charge in
the phosphor layer










infrared applications, Coming 7059 glass is an unsuitable choice. High

temperature glass is often too expensive to be a viable option, but silicon is a suitable

choice for use with inverted structures or mid-infrared applications. Silicon is readily

available and inexpensive and, with proper doping, can be used as the bottom contact for

the inverted structure. A silicon substrate will withstand annealing up to 14000C before

melting, so high temperature processing is limited by the robustness of the deposited

layers. Silicon has already been used for active matrix displays where each pixel was

controlled using a circuit array on the wafer [49].

2.5.2 Insulators

In a full stack device the phosphor is sandwiched between two dielectric layers and

in a half stack device the phosphor is deposited onto a dielectric layer. The insulator

affects the phosphor-insulator interface that determines the interface states that play a

large role in the production of the current necessary for light generation [50]. More

importantly, these layers contribute to the stability of the device by preventing large

currents from flowing through the phosphor when the device is driven at the large

voltages, typically 2 Mv/cm, needed for electrical breakdown. Because of the high

electric fields present during device operation, the insulator needs high dielectric

breakdown and needs to be as defect free as possible. The insulator should also prevent

charge leakage into the phosphor layer. In addition, the dielectric layers need high

thermal stability to withstand heat treatments and the insulators also need to adhere well

to the phosphor and the contacts. Also, in order to prevent the diffusion of foreign

species into the phosphor layer, the insulator should be chemically stable. Finally, as

with the bottom contact in a standard structure, the dielectric layer should be as









transparent as possible to the emission wavelengths of the device. So, the essential

insulator requirements for use in ACTFEL devices are as follows [51]:

1. Sufficient dielectric breakdown electric field, FBD

2. High relative dielectric constant, Er

3. Small number of defects and pinholes

4. Good adhesion to phosphor and contacts

5. Transparency

6. Good thermal and chemical stability

7. Small dielectric loss factor, tan6

In order to have efficient device operation, as much of the applied voltage as

possible should be dropped across the phosphor layer. The proportions of the voltage

dropped across the phosphor and insulators are determined by the capacitance of the

phosphor, Cp, and the capacitance of the insulator, Ci. As discussed in section 2.4, the

capacitances of the layers are determined using


C EoEr
t

where So is the permittivity of free space, Er is the relative permittivity, and t is the

thickness of the layer.

In order to maximize the voltage drop across the phosphor, the capacitance of the

insulator should be much larger than the capacitance of the phosphor. Using the above

equation, either the insulator should be very thin or the relative dielectric constant of the

insulator should be large. Unfortunately, charge leakage has been shown to occur in

insulators that are thinner than 50 nm [52]. As noted above, high dielectric breakdown

strength is necessary for insulators because if the phosphor becomes a virtual short after









breakdown then the additional voltage will be dropped across the insulators increasing

the electric field they experience. The thinner the insulator the larger the field; however,

most insulators with high dielectric constants have low breakdown strengths. In addition,

insulators with high dielectric constants often exhibit propagation breakdown, which

occurs when a small portion of the insulator breaks down forming a short that heats up

the insulator leading to catastrophic failure. On the other hand, many insulators with

lower dielectric constants experience self-healing breakdown in which the breakdown

areas become an open instead of a short circuit so they do not exhibit catastrophic

breakdown. See Table 2-1 for a list of insulators and their properties [49].

Pinholes and defects in the insulator should be minimized to prevent device failure.

If the insulator experiences propagating breakdown, a pinhole or defect can lead to failure

of the entire device. Stability of the device also requires that the insulating layers adhere

well to the contacts and the phosphor. Insulators with poor adhesion will cause the

device lifetime to be short. Obviously, the bottom insulator of a standard ACTFEL

device has the same requirement as the bottom contact in that it needs to be transparent to

the emitted light. Again, like the bottom conductor, the insulators need to be able to

withstand the thermal processing of the device. The bottom insulator must also be

chemically stable so that it does not affect the conductivity of contacts such as ITO, or

modify the composition of the phosphor layer. Finally, the insulator must be able to

maintain the charge balance in the device. The insulator can cause charge loss or leakage

disrupting the proper function of the ACTFELD. Because of this it is believed that

leakage charge, as can occur with thin layers, negatively affects device operation [53].

For this reason, the loss factor of the insulator should be kept small.










Table 2-1 List of insulators used in ACTFEL devices and their properties of interest

Insulator Deposition r FBD EOrEFBD Breakdown
Method* (MV/cm) (pC/cm2) Mode**

SiO2 Sputtering 4 6 2 SHB
SiOxNy Sputtering 6 7 4 SHB
SiOxNy PVCD 6 7 4 SHB
Si3N4 Sputtering 8 8 to 9 4 to 6 SHB
A1203 Sputtering 8 5 3.5 SHB
A1203 ALE 8 8 6 SHB
SiAION Sputtering 8 8 to 9 4 to 6 SHB
Y203 Sputtering 12 3 to 5 3 to 5 SHB
Y203 EBE 12 3 to 5 3 to 5 SHB
BaTiO3 Sputtering 14 3.3 4 SHB
Sm03 EBE 15 2 to 4 3 to 5 SHB
HfO2 Sputtering 16 0.17 to 4 0.3 to 6 SHB
Ta205-TiO2 ALE 20 7 12 SHB
BaTa206 Sputtering 22 3.5 7 SHB
Ta205 Sputtering 23-25 1.5 to 3 3 to 7 SHB
PbNb206 Sputtering 41 1.5 5 SHB
TiO2 ALE 60 0.2 1 PB
Sr(Zr,Ti)03 Sputtering 100 3 26 PB
SrTiO3 Sputtering 140 1.5 to 2 19 to 25 PB
PbTiO3 Sputtering 150 0.5 7 PB
BaTiO3**** Press/sinter 5000 ? ? ?
Westaim proprietary*** Press/sinter 1700 ? ? ?



2.5.3 Conductors

In a normal ACTFEL structure, the emitted light must be able to pass through the

bottom insulator and the bottom contact. This means that the most important property of

the bottom contact is that it is transparent at the desired wavelengths. In addition, the

bottom contact must have an electrical resistivity low enough not to affect the

capacitance of the device or cause resistive heating. Finally, the contact must also be

able to withstand the thermal processing of the device. For wavelengths longer than 1150

nm it is possible to use doped silicon as a bottom conductor in addition to being the

substrate. This has the benefits of reducing the number of deposition steps needed as

well as the advantage of silicon's tolerance for higher temperature processing.









For visible applications and those in the near IR the most common material for a

bottom conductor is ITO, indium tin oxide, an alloy of -90 wt% In203 and 10 wt% SnO3.

ITO can be deposited in several ways including RF magnetron sputtering, plasma ion-

assisted deposition, focused ion beam, and pulsed laser deposition [54-58]. The ITO

layer is typically 200 nm thick with a resistivity of 1x10-4 Q-cm. This provides a sheet

resistance of 5-10 Q/o. ITO is transparent over the visible and near infrared range

because the bandgaps of In203 and SnO3 are both about 3.5eV. ITO is conductive due to

oxygen vacancies and Sn4+ ions occupying In3+ sites creating shallow donors a few meV

below the conduction band [59]. So the conductivity of the layer can be reduced during

annealing if not suitable protected. In addition, transparent conductors may cause

reliability problems for inverted structures because they do not exhibit self healing like

some opaque conductors.

The final layer of a normal ACTFEL device is typically an opaque top contact.

Like the bottom contact, the top contact must be highly conductive but does not need to

survive high temperature processing since annealing can be completed before the top

contact is deposited. In addition, this layer must be able to adhere well to the insulating

layer in the case of a full stack device or the phosphor in the case of a half stack device.

Aluminum is almost always the material of choice for the top contact. Aluminum has

many advantages including low cost and low resistivity. In addition, aluminum adheres

well to most insulators which makes it good at self-healing around short circuits. Finally,

aluminum is easy to deposit using either evaporation or sputtering. Aluminum's low

melting temperature of 660C is both an advantage and a disadvantage. The low melting

temperature makes aluminum easy to thermally evaporate, but is undesirable for inverted









structures when high temperature processing is needed. For inverted structures other

metals with higher temperature tolerances, such as tantalum, molybdenum, or tungsten,

can be used [49].

2.6 Phosphor Luminescence

2.6.1 Host Materials

Determining the host material is an important first step in designing an

electroluminescent device. Thulium, erbium, neodymium, and dysprosium all have

excited state energy levels between 6666 and 14000 cm-1 above the ground state at zero

(corresponding to photons with wavelengths of 1500 nm to 715 nm in the NIR) [60]

(Figure 2-6). ZnS provides a suitable host for these infrared emitters ions for several

reasons. The typical phosphor fields during ACTFEL device operation exceed 1 MV/cm

and depend on the thickness and dielectric constant of the phosphor [61]. The dielectric

breakdown strength of ZnS is -1.5 MV/cm [62]. This dielectric strength is sufficient for

ZnS to act as an insulator below threshold and act as a conductor at high fields. For near

infrared devices the bandgap of the host must be at least 1.6eV while visible phosphors

need a bandgap of 3.1eV. ZnS has a bandgap of ~3.7eV at room temperature and is

transparent from below 400 nm to past 10 [im [63]. The glass substrates typically used

for this type of device are only transparent to about 4 [im; however, a silicon or

chalcogenide substrate could be used for longer wavelengths [10]. The ITO often used as

a transparent conductor has a plasma edge, the long wavelength cutoff for transparency

dependent on carrier concentration, at -1.8 im. As mentioned above, silicon can be used

as substrate and bottom conductor for longer wavelength applications solving this

problem.









In addition, hot electron distribution in ZnS is less energetic than in SrS, another

common ACTFELD host material [18]. Low energy (cooler) electrons can only excite

ground state electrons to the lower energy states on the luminescent centers. This lower

energy electron distribution shifts the strong luminescent emission for dopants such as

Tm and Nd from the visible to the near infrared. An example of this is that SrS:Nd

produces an orange-white light while ZnS:Nd produces only orange. This is because the

neodymium doped ZnS phosphor has no emission shorter than 530 nm. The emission at

600 nm due to relaxation from the 2H11/2 to the 419/2 ground state is active in both hosts,

but the higher 4G7/2 and 4G9/2 levels that produce shorter wavelength emission are more

active in the SrS phosphor [64]. The difference is unlikely to be an effect of the host

lattice symmetry due to the shielded nature of the 4f transitions in rare earths but the lack

of high energy electrons would produce weaker visible emission compared to infrared

emission [65]. Table 2-2 shows a comparison of the properties of ZnS and SrS [49].

Table 2-2 Properties of ZnS and SrS
Item llb-Vlb compound Ila-Vlb compound
Material ZnS SrS
Melting point (C) 1800-1900 >2000
Band Gap (eV) 3.6 4.3
Transition type Direct Indirect
Crystal Structure Cubic zinc blende or Rock Salt
Hexagonal wurtzite (NaCI type)
Dielectric constant 8.3 9.4
Lattice constant (A) 5.409 6.019
Ionic Radius (A) 0.74 1.13
lonicity 0.623 >0.785











20-


15--


25--



20--



15--



10--



5--


0


E
eC
00
M
cc


E
C


E
E
s
La
Iv


41 3Q

41,IQ
_~~~~~ S_ ____ 2____ 4


Tm3*


20-


> 10-


5-
5-


0 1


Nd3.


ZH-


=1= I4s


111i
32


' --iii
--- -- ---____4____


E
0
In
if


411


41132






41


Er3+


Figure 2-6 Energy level diagrams and radiant transitions of Tm3 Nd3+, and Er3


41 -
E E F410-
C 3HC
n C2


3H4 5-




3H, 0-


.2H112
-1112


4G712


-I- __


'15tr


A4F3z









The host material must also be insulating below its threshold voltage. As discussed

in the insulator section, this is to ensure a sufficient voltage drop across the phosphor. In

order to maximize the electric field across the phosphor layer, the capacitance of the host

should be low. Finally, electroluminescent phosphors are often annealed at temperatures

in excess of 500C so the phosphor host needs to have a melting temperature well above

this. In summary, ZnS is an excellent choice for the host material for rare-earth doped

NIR emitting EL phosphors.

2.6.2 Luminescent Centers

When choosing a host and a luminescent dopant combination care must be taken

to ensure that the two are compatible. The size and charge of the dopant will affect its

performance in each host. The luminescent dopant should be incorporated into the lattice

without creating too many defects, since defects can act as non-radiant relaxation sites

and reduce the luminance from the device. Also, the charge of an incorporated ion must

be accounted for. If the charge of an ion in a substitutional site is different than the

displaced ion, the charge difference must be compensated to maintain charge neutrality

for the solid. For this reason charge compensators, such as interstitial F1- and C11

compensating the 3+ rare earth ions substituting for Zn2+ ions, are introduced. Table 2-3

shows a list of common host and luminescent dopant ions for a sulfide based system [49].

The emitted light in ACTFEL devices comes from the luminescent dopant not host

material. These centers in phosphors luminesce by one of two mechanisms. The first is

through the recombination of electrons trapped in deep donor states and holes trapped in

deep acceptor states. The recombination energy depends on the trap depths of the donors

and acceptors the as well as band gap of the host because of its effect on the










Table 2-3 Optical properties of common sulfide based EL materials
Phosphor Emission Luminance Luminous
layer color L (cd/m2) efficiency
material 1 kHz 60 Hz q(lm/W)(1 kHz)

ZnS:Mn Yellow 5000 300 2-4

ZnS:Sm,F Reddish-orange 120 8 0.05
ZnS:Sm,CI Red 200 12 0.08
CaS:Eu Red 200 12 0.05
ZnS:Mn/Filter Red 1250 75 0.8

ZnS:Tb,F Green 2100 125 0.5-1
ZnS:Mn/Filter Yellow-green 1300 80 ---
CaS:Ce Green 150 10 0.1

ZnS:Tm,F Blue 2 <1 <0.01
SrS:Ce Blue-green 900 65 0.44
ZnS/SrS:Ce Bluish-green 1500 96 1.3
ZnS/SrS:Ce/Filter Greenish-blue 220 14 0.2
CaGa2S4:Ce Blue 210 13

SrS:Ce,Eu Eggshell-white 540 32 0.4
SrS:Ce/CaS:Eu Paper-white 280 17
ZnS:Mn/SrS:Ce Yellowish-white 2450 225 1.3

trap depth. Examples of this type of phosphor are ZnS doped with Al or Cl as donors and

Ag, Au, or Cu as acceptors. This type of phosphor is not used in ACTFEL devices

because the high electric fields in the phosphor destabilize the traps and sweep the

electrons and holes toward opposite sides of the film [66].

The second type of radiative relaxation operates through the electronic transitions

of the luminescent ions. This type of phosphor depends on the energies of the ground

state and excited states of the individual ions and not the host. The quantum mechanical

selection rules governing electronic transitions are important to emission from this type

of relaxation. The spin and parity selection rules, governing transitions between states

depending on the electronic spin or the symmetry of the stationary state wave function

respectively, determine if a particular transition is allowed or disallowed. Essentially,

these rules state that transitions are allowed only if they are between states with the same









spin and disallowed for electronic shells with the same reflection property of the

waveform, also called parity. In other words transitions between thep and s shells or the

fand d shells are permitted, but transitions within a particular shell or transitions between

the d and s orf andp shells are forbidden [67]. Luminescent ions in this category include

the 3+ ionized rare earths and transition metals ions such as Mn2+, Ag1+, and Cu+.

Trivalent rare earth ions have filled 6s shells, incompletely filled 4fshells, and

empty 5d shells. This leads to the two types of excited state to ground state

recombination that are observed in rare earths. The first occurs when there are electrons

excited from the 4finto the 5d shell, as in Ce3+ and Eu2+. Because the transition is

between the d andforbitals the parity rule is not broken. However, the transition in Eu2

is spin forbidden while the transition in Ce3+ is not. This means that the decay time of

Ce3+ is much faster (several ns) than the decay time of Eu2+ (several [is) [39,68]. Since

the 5d orbital has a higher energy than the 6s shell, 5d-4f transitions can be strongly

affected by the crystal field around the luminescent center and can shift in wavelength

depending on the host. The other type of transition is the intra-shell 4f-4ftransitions.

This type of transition occurs in rare earths such as Tm3+, Nd3+, and Er3 [39,68].

Because these are intra-shell transitions they are forbidden by the parity selection rule

and thus have longer decay times. Since electrons in the 4f shell are shielded by those in

the 6s shell, these transitions are relatively well shielded from crystal field effects and

characterized by sharper transition lines.

2.6.3 Rare Earth Doped ZnS

A variety of techniques have been used to deposit rare earth doped zinc sulfide

films, including CVD (chemical vapor deposition), MOCVD (metal oxide chemical

vapor deposition), and thermal evaporation [69-71]. At room temperature ZnS exhibits









two crystal structures, a zincblende cubic structure called sphalerite and the hexagonal

wurtzite phase. The properties of these phases are given in Table 2-4 [72]. The effect of

phase on the luminescent centers is minimal because they have similar properties and

symmetry. As mentioned above; Tm, Nd, and Er emit by parity forbidden 4fintrashell

transitions. Because of this the spectra from each of these dopants exhibits little change

due to crystal field effects and each one typically has luminescent decay times in the ms

range [73].

Table 2-4 Physical properties of ZnS
Parameter Value
Zincblende Wurtzite
Lattice Constant (A) 5.409 a=3.814; c=6.258
Mass Density (g/cm3) 4.08 4.1
Melting Point (K) 2100 2100
Heat of Formation (kJ/mol(300K)) 477 -206
Specific Heat (J/kg-K(300K)) 472
Debye Temperature (K) 530

2.6.3.1 ZnS:Tm

ZnS doped with Tm exhibits both visible and infrared emission. The NIR peak is

near 800 nm while the visible emission is largely in the blue at 480 nm with weaker

emission possible in the red region at 650 nm [74]. Photoluminescent excitation of

thulium doped zinc sulfide takes place through efficient energy transfer from the ZnS

host to the luminescent centers [75]. Sputter deposited films have exhibited the highest

photoluminescent infrared to blue intensity ratios [76]. Infrared electroluminescence is

often achieved by direct impact excitation of the rare earth by hot electrons [77]. The

low hot electron energy in ZnS, as mentioned above, makes it less likely that direct

impact excitation will produce higher energy blue light resulting in weaker visible

emission. In addition, in a ZnS host, the impact cross section of the 1G4 level, from

which blue light is produced, is much smaller than that of the 3F4 energy level, from








which the 800 nm infrared light is produced [78] (Figure 2-7). Due to this difference in

excitation mechanisms, ZnS:Tm excited by PL tends to have increased blue luminescence

while EL favors the infrared emission.


P1 -._9 ..T. .. ... r.r .


0 I r
.5. 7-s i
o











I 2? a 4 6 '7 0 9 10

ELECTRON ENERY (sv)



Figure 2-7 Impact cross sections of the 3F4 and 1G4 levels in Tm3+ [78]

2.6.3.2 ZnS:Er

Erbium is the rare earth most people think of when considering infrared. This is

because of the proliferation of erbium-based infrared telecommunications equipment.

Erbium is doped into fiber optic cables for transmission in the fibers absorption minimum

at 1550 nm [79]. In addition to the 1350 and 1550 nm lines used for telecommunications,

ZnS doped with erbium also emits in the NIR at 990 nm, weakly in the red (660 nm) and

strongly in the green (530 nm). Unlike ZnS:Tm in which the infrared emission originates









from a lower energy state than the visible, ZnS:ErF3 green emission at 530nm originates

from 4S3/2 (18900 cm-') with decay to the 4115/2 ground state while the near infrared at

990nm starts at from the higher 2F7/2 (20100 cm-1) and decays to the 4111/2 (10100 cm-1)

excited state [80]. The emission of NIR light from the higher energy state implies that

ZnS is a poor host choice for infrared emitting Er however emission from the 4111/2 state

to the ground level also emits at -1000nm.

2.6.3.3 ZnS:Nd

Neodymium is a rare earth element with four transitions in the near infrared.

These transitions have wavelengths of 900 nm, 1060 nm, 1365 nm, and 1800 nm [81].

The transitions responsible for the 900 and 1060 nm emission are from the 4F3/2 to the

419/2 and 4111/2 levels respectively, and the excited state is lower in energy than the 2F11/2

and 4G7/2 states from which the visible emission originate. (Figure 2.6) In previous

studies ZnS:Nd was found to have the highest near infrared electroluminescence of any

ZnS:RE phosphors [82]. Direct current electroluminescence of ZnS:Nd films has also

been found to be more efficient than ZnS:Tm films under similar conditions [83].

2.7 Electrical and Optical Characterization

Optical characterization is useful because the ultimate goal of most ACTFELD

research is light output of a specific color at the lowest possible input power. The

characterization properties of most interest are brightness versus voltage, power

efficiency, and the emission spectrum of these devices. Because the electrical properties

of an ACTFEL device are critical to its EL performance, electrical characterization is

useful for understanding the fundamental materials properties. Four types of electrical

data will be discussed: charge versus voltage (Q-V), capacitance versus voltage (C-V),









internal charge versus phosphor field (Qint-F), and maximum charge versus maximum

applied voltage (Qmax-Vmax).

2.7.1 Brightness versus Voltage

The luminance of an ACTFEL device is very sensitive to the voltage waveform as

well as the drive frequency and amplitude. The most common types of waveforms are

sinusoidal or trapezoidal. If a trapezoidal waveform is used the luminescence is

dependent on the amplitude of the pulse and the pulse width as well as the rise and fall

times of the pulses. Drive frequency has a large effect on the radiant output of an

ACTFELD. The emission intensity increases markedly (often nearly linearly) as the

drive frequency increases. As the frequency increases there are more pulses per given

time yielding more excitation of the luminescent centers and higher brightness.

However, the luminescent centers need time to de-excite. If the drive frequency is faster

than the luminescent relaxation time, EL output will be saturated and further increases in

frequency do not produce increased luminance. Also, the device may heat up at these

higher frequencies leading to decreased luminescence due to thermal quenching, i.e.

increasing probability of non-radiative recombination [84]. Frequencies of up to 7 kHz

have been used but typical drive frequencies are 60 Hz to 2.5 kHz.

Visible emitting phosphor brightness is discussed in terms of luminous flux, i.e. the

photon flux convoluted with the wavelength response function of the human eye, with

units of lumens, nits or candela/m2 [85]. However, infrared phosphor brightness must be

specified using the irradiance (W/cm2) of the emitted light, since the human eye response

is zero. Brightness versus voltage (B-V) is a typical measure of ACTFEL devices, i.e.

the brightness at the wavelength of interest measured at steadily increasing applied

voltages. A typical B-V curve is shown in Figure 2-8.










0.016

0.014

0.012

E 0.01

0.008

a
0.006

0.004

0.002


0 50 100 150 200 250
Voltage (volts)


Figure 2-8 Brightness vs. voltage curve showing the threshold voltage

Of course, the physical properties of the phosphor, and not just the voltage pulse,

affect the luminescence of the device. Three properties that have a large effect on the

ACTFELD operation are the thickness of the phosphor film, the insulator capacitance

(see above), and the concentration of the luminescent dopant. As the phosphor becomes

thicker there are more luminescent centers for an accelerated electron to impact. In

addition, a thicker phosphor will have a lower capacitance yielding a larger voltage drop

across the phosphor layer. Both of these effects cause thicker phosphor devices to be

brighter than thinner ones. However, as the thickness increases the operating voltage also

increases, which is undesirable because of the size and expense of high voltage supplies.

The choice of insulator affects how the voltage is dropped across the device as

discussed above. Recall that as the insulator capacitance increases with respect to that of

the phosphor, the threshold voltage of the device will decrease because more of the









voltage will be dropped across the phosphor instead of the insulator. Therefore a thin

dielectric layer with a high dielectric constant and high breakdown strength is desired.

Finally, the doping concentration of the luminescent impurity has a large effect on

the emission characteristics of an ACTFEL device. At low concentrations, typically <

lmol% for most materials, the luminance increases steadily with increasing dopant

concentration. Once a maximum emission is reached at an optimum concentration of

activator, the luminance will decrease with further concentration increases. This decrease

has generally been attributed to concentration quenching, which results from interactions

between neighboring centers that lead to non-radiative relaxation through self quenching

or contact with killer centers such as defects or impurities [86].

2.7.2 Threshold Voltage

The optical threshold voltage, Vth, is the voltage at which the device begins to

emit light, and is dependent on several physical properties including the capacitance of

the insulator and the thickness of the phosphor. There are several definitions of threshold

voltage, but the most common is the voltage axis intercept of the extrapolation of the

maximum slope portion of a B-V curve (Figure 2-8). Another common definition of EL

threshold is that voltage at which a certain brightness value is achieved, such as Icd/m2

for visible emitters. For this study the optical threshold voltage will be determined using

the first method. A second type of threshold voltage is that where current is transferred

across the phosphor in charge versus voltage (Q-V) tests. This is a measure of electrical

threshold, and is commonly different from the optical threshold voltage.

2.7.3 Efficiency versus Voltage

In addition to the brightness, the power efficiency of a device is an important

quantity. Power consumption is a critical concern for any device that needs to use









batteries, e.g. portable displays or sensors. Hence it is desirable to know the light

produced per unit power input. For visible emission this is termed the luminous

efficiency and is described in lumens per watt. As discussed above, for infrared emitters

lumens are inapplicable so power efficiency is expressed as watts of optical output per

watt of electrical input. The input power can be determined using

1 t+1
P = v(t')i(t')d',
Art

where A is the area of the device, T is the period of the driving waveform, v(t) is the

applied voltage, and i(t) is the current. The output power can be determined by knowing

the sensitivity of the detector (determined by using a calibrating light source with a

known power output). This output power efficiency can be plotted versus the input

voltage of the device yielding an efficiency versus voltage plot (Figure 2-9).






0.3-



0.2-



0.1




150 160 170 180 190 200 210
Applied Voltage (V)


Figure 2-9 ACTFELD efficiency versus drive voltage









2.7.4 Electrical Testing

The typical circuit used for electrical testing (Figure 2-10) employs a Sawyer-

Tower arrangement with either a sense capacitor or a sense resistor [29]. An arbitrary

waveform generator is used to produce a voltage pulse that is then amplified and used to

drive the circuit consisting of a series resistor, the ACTFEL device and a sense element in

series. The series resistor is used to limit the current to the ACTFELD in the case of

catastrophic failure. If the sense element is a capacitor then the external charge of the

device is measured, while external current can be measured by using a resistor as the

sense element. If the capacitor is used its capacitance value must be much larger than

that of the device. If a resistor is used as the sense element, its resistance is typically near

100Q. In the case of a resistor, if the resistance is too large, the dynamic response of the

device will be delayed as the RC time constant of the circuit increases.


V2(t


Vt(t)


V3(t)


Figure 2-10 Schematic of a Sawyer-Tower test setup









The typical waveform used for testing is a bipolar trapezoidal waveform as

discussed above and shown in figure 2-11. The pulses have a rise and fall time of -Ss, a

30s plateau, and a frequency between 60 Hz and 2.5 kHz. The labels A-J are used to

designate important points during the cycle. Most of the points are self explanatory

except for points B and G. These points are to designate the electrical threshold voltages,

i.e. the voltages at which the phosphor begins to conduct charge. This labeling scheme is

common in the literature and will be used for throughout this document in discussions of

electrical properties and the matching of points on Q-V, Q-F data curves with points on

the driving waveform [29].



C D



Vmax



A E




H I



Figure 2-11 Trapezoidal waveform with important points marked for reference

2.7.5 Charge versus Voltage (Q-V)

The most basic measure of the electrical characteristics of ACTFEL devices is the

charge versus voltage (Q-V). A Q-V plot displays the charge stored between the external

terminals of the capacitive ACTFEL device versus the voltage across the terminals.









When the device is driven below the electrical and optical threshold voltages, the plot is

simply a straight line with a slope equal to the total capacitance of the device (assuming

leakage current is negligible). When the device is driven above electrical and/or optical

threshold, the Q-V plot becomes a hysteresis loop due to the dissipative charge

conduction through the device. Hence, the electrical threshold can be determined as the

voltage at which the plot becomes hysteretic. The voltage drop across the ACTFEL

device can be found from V2 and V3 by using

VEL(0 = V2(0t- V3(0.

When using a sense capacitor, the external charge can be determined using

qexr (t)= C V3 (t)

where Cs is the sense capacitance. When using a sense resistor, it is first necessary to

calculate the current passing through the device. Since the current through each element

will be the same, the current through the ACTFEL device can be determined using

VIQ -V2=Qt)
sees

The external charge can then be found by integrating this current over time, such that

t
qe,(t) = i(t)dt.
0

An example Q-V plot using the labeling scheme from above is shown in figure 2-

12. The voltage labeled Vto in figure 2-12 is the turn on voltage of the device, which is

different from the electrical threshold voltage due to the polarization charge, Qpo,. The

plus and minus superscripts are to signify which occurred during the positive waveform

pulse and which occurred during the negative pulse. Positive is defined as when the

voltage pulse is applied to the Al electrode and negative is when the pulse is applied to









the transparent electrode. The polarization charge is the result of charge buildup at the

phosphor-insulator interface creating a charge imbalance in the phosphor at the

conclusion of a pulse. The polarization charge may help the next pulse because the built-

up charge at the phosphor-insulator interface creates an electrical field that is the same

polarity as that of the pulse. The threshold voltage may be defined as

Vth = lim V,o(Qpo).
Qpo->0

The polarization charge may be reduced by the leakage charge, Qleak, over the time

between voltage pulses (during segments EF and JA). Leakage charge results from

electrons escaping relatively shallow states due to the polarization field.







o
-2 tI Q
E B









-200 -100 0 100 200

Applied Voltage (V)


Figure 2-12 Typical Q-V plot
Qcond
H "relax


-200 -100 0 100 200
Applied Voltage (V)


Figure 2-12 Typical Q-V plot

The other types of charge occur during the voltage pulses. The conduction charge,

Qcond, is the charge conducted through the ACTFEL device from device turn-on until the









end of the voltage plateau (segments BD and GI). The conduction charge is essentially

the charge flow responsible for impact excitation or ionization of the luminescent centers,

assuming the charge conduction between electrical and optical thresholds is small. Qrelax

is the relaxation charge, the charge that flows during the plateau portion of the voltage

pulses (segments CD and HI). The term relaxation charge is used because the flow of

this charge, at a constant voltage, sets up an electric field opposed to the total field across

the phosphor. Finally, the maximum charge, Qmax, is the charge at the maximum voltage

measured across the sense element for a given applied voltage. Qmax differs from the

other charges discussed in that it is taken in reference to the zero point of the charge axis

instead of being referenced to a specific point of the driving voltage waveform. Qmax is a

useful term for evaluating transferred charge, as will be shown later.

The Q-V curve is useful for determining several parameters of an ACTFEL device.

First, as discussed above, below the turn-on voltage the slope of the Q-V curve is a

measure of the total device capacitance. Second, above the turn-on voltage the phosphor

is assumed to be a conductor, which means that the Q-V slope can be used as a measure

of the capacitance of the devices insulator layer(s). Care must be taken when

determining this because several factors may skew this measurement in a real device. If

the phosphor is not completely shorted then the slope will be less than the insulator

capacitance due to some remaining phosphor capacitance and/or resistance. If there is a

build-up of space charge in the phosphor, the slope may be larger than that of the

insulator capacitance [28]. Third, the area inside the Q-V curve is proportional to the

input electrical power density delivered per pulse [87].









2.7.6 Capacitance versus Voltage

A capacitance versus voltage (C-V) plot allows the measurement of the dynamic

capacitance of an ACTFEL device against the voltage across the terminal during the

rising edge (segments AC and FH) of the voltage pulse. C-V plots are derived from the

data obtained during measurements for Q-V; the slope of the Q-V plot during the rising

edges of both pulses is plotted. The capacitance is calculated using


C(v) =i(t)
dv(t)
dt

where i(t) is the current derived using the inverse of the previous equation for dqext(t),

dqex, (t)
i(t) =
dt

These reduce to the following formula,

dqex, (t)
C(v) =
dv(t)

An example of a C-V plot is shown in figure 2-13. The total physical capacitance,

Ctcv, is the capacitance of the whole device below the turn-on voltage and is usually in

good agreement with the capacitance calculated from dielectric constant and film

thickness measurements, Ctphy. Ciphys, the capacitance above turn on is simply that of the

insulators, since the phosphor is shorted (broken down) at higher voltages. Cicv, is the

measurement of the capacitance above turn-on and, in the ideal case, should equal the

calculated insulator capacitance. Usually, however, the value of Cicv is either above or

below the value of the ideal case. Cicv less than the calculated value occurs when the

phosphor does not completely short, resulting in some remnant capacitance. Cicv can be

larger than Ciphys if there is dynamic space charge built up in the phosphor that decreases









the total phosphor field, as discussed previously. Space charge can also lead to CV

overshoot, a sudden increase followed by a decrease at higher voltages in the capacitance

when the device turns on.



25-
; I
I

20 /
.


Cphs Vohys
1 15 1



Vtol
"10:------vt-------



0 50 100 150
Voltage (V)


Figure 2-13 Typical C-V plot

2.7.7 Internal Charge versus Phosphor Field

Internal charge versus phosphor field (Qint-Fp) is a technique used to provide

information field clamping or charge relaxation, which are difficult properties to

determine from a Q-V plot [88]. In a Q-V plot the total charge capacitivee plus phosphor

charge) and total device voltage are studied, not just the charge and voltage across the

phosphor. The Qint-Fp data are the charge transported in the phosphor layer and the

electric field in the phosphor layer only.

The internal charge of the phosphor can be determined from








(C, + C )
q(t) = q,Q) Cp [v2 (t)-3(t)]
C,

where q(t) is the internal charge, Ci and Cp are the insulator and phosphor capacitances,

respectively, qext is the external charge, and v2 and v3 are the voltages measured on each

side of the device. The phosphor field is obtained using

1 qqt (t) ) v3(t)]}
f p = d C,- [v2 3(

where dp is the phosphor thickness. These equations basically use the raw data from a Q-

V curve, remove the capacitive displacement charge, and remove the voltage drop across

the insulators to calculate the field across the phosphor. The equations are developed

from the equations used to describe an ACTFELD with a phosphor layer free from space

charge [89]. A typical graph of Qint-Fp is shown in figure 2-14. Unlike a Q-V plot the

Qint-Fp loop goes in a clockwise direction. A Qint-Fp plot shows several of the same

quantities as in a Q-V plot, but these are only charges and fields in the phosphor. The

charge information shown is Qcond, the conduction charge transported across the

phosphor, Qpo,, the polarization charge stored at the phosphor/insulator interface, Qleak,

the leakage charge between the voltage pulses, Qrelax, the relaxation charge flowing

during the voltage pulse plateau, and Qmax, the maximum charge across the phosphor

(Figure 2-14). The other information available is the steady state field, Fss. Field

clamping, when the charge flow through the device is sufficient to counteract the

increasing field generated by increasing the applied voltage, can be determined by

comparing Fss at different voltages above the threshold voltage. If there is field

clamping, then Fss will be independent of voltage above threshold. Some devices









demonstrate field overshoot because of dynamic space charge effects that are usually

manifested around points B or G in the Qint-Fp plot [90].



21


li: \1 1---:--------IE
--i:Q = ^ -5
-1 0 1
SPhos Q dF d



Figure 2-14 Typical Q plot
-B

.2 I l l I 1 I l l l l |. l .. I |

-2 -1 0 1 2
Phosphor Field (MV/cm)


Figure 2-14 Typical Qint-Fp plot

Reduction of Q-V data to Qint-Fp data depends on knowing the capacitance of the

phosphor and insulator(s) as well as the thickness of the films. Uncertainty in the

capacitances results in distortion in the Q-Fp plots. The capacitance values can be refined

using their effects on the shape of these plots [91]. Often the capacitance values are

adjusted to obtain a vertical slope for the BC and GH portions and a horizontal slope for

the ED and IJ segments of the plot (Figure 2-14). If there is a large amount of space

charge then the capacitance values will be larger than the physically measured values. If

the phosphor thickness value is incorrect, there is inaccuracy in the phosphor field. So,

great care must be taken when using a Qint-Fp plot versus Q-V or C-V plots to make sure

that the data are meaningful.









2.7.8 Maximum Charge versus Maximum Voltage

The final electrical characterization technique, maximum charge-maximum

voltage (Qmax-Vmax), measures the transferred charge at the maximum pulse amplitude for

several applied voltages [92]. Typically there are two types of charge measured, the

internal charge, Qmax, and the external charge, Qemax. The internal charge values can be

taken from the maximum charge points (points D and I) in the Q-Fp plot, while the

external charge values, known as a Qemax -Vmax or AQ-V plot, can be taken from the

maximum charge values of the Q-V plot [87].

Figure 2-15 shows a typical Qmax-Vmax plot for an ACTFEL device. Often charge

values from both the positive and negative pulses are plotted simultaneously to determine

if the charge transfer in the device is symmetric. Care must be taken because the

accuracy of the Qmax values is affected by the accuracy of the phosphor and insulator

capacitances. The Qmax-Vmax plot looks similar to a B-V curve and is a measure of the

internal charge needed for a desired brightness. Qemax-Vmax data are more reliable than

Qmax-Vmax data because it is directly measured so capacitance inaccuracies are

unimportant. Above threshold, the slope of the Qemax-Vmax plot is proportional to the

insulator capacitance. If the slope is too small then there is insufficient transferred

charge. If the slope is too large then there is more than the expected transferred charge,

possibly caused by dynamic space charge. The voltage derivative of Qemax-Vmax (Figure

2-16) is a direct measure of the capacitance and can give information about how the

capacitance changes with voltage. Dynamic space charge can lead to C-V overshoot

which is easily seen in this type of plot.





47



2.2

2 2









120 140
Vm


Figure 2-15 Typical Qmax-Vmax plot



-25


g 20


S15-


10
'yh /\
I ;

J "0
^"[

? :: -
10- -^* ^ --


Figure 2-16 Typical Qemax-Vmax plot


160 180
VaxV)


120 140 160 180
Vmax (V)














CHAPTER 3
EXPERIMENTAL PROCEDURE

3.1 Substrate and Target Preparation

Zinc Sulfide doped with rare earth fluoride thin films were deposited onto 2.5 x 5

cm 7059 glass coated with 360 nm of a polycrystalline indium tin oxide (ITO) transparent

conducting electrode and 160 nm of amorphous aluminum titanium oxide (ATO)

transparent dielectric layer obtained from Planar Systems. The substrates were cleaned

in a UVOCS Inc. ultraviolet light ozone cleaner for six minutes in air to remove organic

contaminants. They were then blown clean with dry nitrogen to remove any particles. In

addition, substrates (2.5 x 5 cm) of bare 7059 glass were prepared by the same methods

for simultaneous coating, the used for film thickness measurements and destructive

testing. Several targets were used for deposition of the doped ZnS films. All of the

doped targets were pressed powder, were 5 cm in diameter, 0.65 cm thick, and were

manufactured by Target Materials Incorporated. The targets included ZnS doped with

1.5mol% of either 99.9% pure TmF3, NdF3, or ErF3. In addition, a CVD grown plate of

pure, undoped, dense ZnS from Morton Thiokol was cut with a diamond saw and used as

an undoped target. All targets were conditioned for one hour after any break in the

vacuum of the deposition system.

3.2 Sulfide Sputter Deposition System

Films were deposited by RF planar magnetron sputtering in a high vacuum

chamber using a Leybold Trivac rotary vane pump for backing and roughing, and a

Leybold 1600W magnetic levitation turbomolecular pump with a Leybold Mag.DriveL









controller. The ultimate pressure of the system varied between 6 x 10-7 and 2 x 10-6 Torr

as measured by a hot filament ionization gauge. The system is designed to run as many

as three sputter sources simultaneously. An Angstrom Science Onyx 2 magnetron

sputtering gun was used for the undoped target and an AJA A300 magnetron sputtering

gun was used for all of the doped targets. During dual rare earth depositions the

Angstrom Science gun held the thulium doped target. The target face of the Angstrom

Science gun was 10 cm from the substrates while the target face of the AJA gun was 5

cm away. Power to each gun was supplied by an RFPP RF5S radio frequency controller

with an RFPP matching network. Duty cycles of 25, 50, 75, and 100% were used with a

consistent pulse width of 40 milliseconds and varying delays between each pulse.. RF

power was set to 120 watts in all cases. By independently varying the duty cycles used

for two targets (e.g. undoped ZnS and ZnS:RE), the concentration of the rare earth (RE)

fluoride in the thin films can be varied. A schematic of the deposition chamber is shown

in figure 3-1. The substrates were held on a multi-sample platter consisting of four 2 x 2"

sample mounting positions that was rotated at 11 seconds per cycle to ensure that the film

deposited at each sample position was identical. A schematic of the sample holder and

sample positions is shown in figure 3-2. Deposition rates varied from 4.0 to 12.5 nm/min

depending on the duty cycles of the targets and the substrate temperature. Deposition

times were varied to maintain film thicknesses between 0.2 and 0.6 itm, depending on the

experimental run. These parameters resulted in deposition times from 50 minutes to 220

minutes, depending on the target materials and duty cycles.





































Figure 3-1 Schematic of the sputter system used for RF magnetron sputtering





































rotation


Figure 3-2 View of sample platter showing substrate positions and spaces for additional
substrates


Ultra high purity argon was used as the sputter deposition gas. The gas was

introduced into the chamber using Unit UFC 1100A 20, 50, and 100 seem mass flow

controllers for three inlet lines. The argon pressure was regulated using the flow

controller and a throttle valve before the turbo pump. Using this method, the argon

pressure was maintained at 2x10-2 Torr measured by a baritron capacitance gauge.

The substrates were radiatively heated by an array of resistive carbon cloth

filaments. A graphite plate was situated above the heater clothes and used as the seat for









the sample platter. The platter holding the samples rested on four 1.25 cm high ceramic

feet to reduce conductive heating. The sample positions were square holes in the platter

with small ledges for the samples to rest on two sides. See figure 3-3 for a schematic of

the heating system. Deposition temperatures were measured by a thermocouple

positioned just above, but not contacting, the platter surface.


Figure 3-3 Schematic of the heating system in the sputtering system

3.3 Top Contact Deposition

All of the devices tested were of the half stack configuration, i.e. no top dielectric

layer was deposited. A stainless steel shadow mask was used to create an array of

aluminum contacts directly on the phosphor surface. The aluminum was thermally









evaporated onto the phosphor using an Edwards Coating System E306 thermal

evaporator (base pressure of 1x10-5 Torr). The aluminum contacts were 0.3 cm diameter

circles between 190 and 250 nm thick. The bottom conductor was the ITO layer. The

ITO was buried under the ATO and phosphor layers so these layers needed to be

removed before the bottom contact can be connected. To achieve contact, the phosphor

layer and the ATO layer were removed by scratching with a diamond scribe. Once the

layers were removed, a multimeter was used to test conductivity in the scratched area to

ensure that the ITO layer was exposed. Once ITO contact was confirmed, an indium wire

was melted with a soldering iron into the scratched area to create a contact to the ITO.

3.4 Sample Handling and Storage

Substrates and incomplete devices (devices with no top contact) were stored in a

nitrogen cabinet under a steady dry nitrogen flow. Completed devices were stored either

in the nitrogen cabinet or in a standard cabinet under normal room humidity. Time

delays from a day to over a year occurred between deposition of the phosphor and

deposition of the top contact. Varying time delays also occurred between sample

completion and device testing. All samples were handled with latex or nitrile gloves

and/or with tweezers. Note that storage in a humid environment after device completion

and the various time lags during device construction did not appreciably affect device

performance.

3.5 Sputtered Film Characterization

The sputtered films were characterized using a variety of techniques including

optical interferometry, x-ray diffraction (XRD), electron microprobe (EMP), energy

dispersive x-ray spectroscopy (EDS), photoluminescence (PL), photoluminescent









excitation (PLE), electroluminescence (EL), time resolved electroluminescence, and

electrical measurements. The details are provided below.

3.5.1 Thickness Measurements

Optical interferometry [93] was used to measure the thickness of each deposited

film. The films deposited on the bare 7059 glass substrates were used to avoid

interference from the ITO/ATO layers. The index of refraction of the film (2.5) and the

substrate (1.5) is known. Upon shining a beam of light onto the sample, interference

patterns will be created from reflection at the air-film and film-substrate interfaces. The

frequency of the interference fringes is dependent on the thickness of the film and the

optical index. Using an in-house developed Excel macro, the film thickness can be

determined by curve matching a calculated pattern to the experimental pattern.

3.5.2 X-ray Diffraction (XRD)

X-ray diffraction [94] was used to evaluate the ZnS crystallinity. The

diffractometer was a Phillips model APD 3720 operated at 40 kV and 20 mA. The

wavelengths used were from Cu K, lines at 0.15406 and 0.15444 nm. The Cu Kp was

blocked using a nickel filter. The diffractometer was scanned over the range of 26.50 to

31.5 to encompass the primary emission peak of both cubic and hexagonal ZnS at 28.50.

The goniometer scanned 0.01 per second with a step size of 0.01.

X-ray diffraction is used primarily to determine phase of a material but it may also

be to determine crystal size, strain of the lattice, film thickness, and semi-quantitative

composition analysis [95]. These parameters can be extracted from the diffraction peak

intensity, width and position.

Atoms can scatter x-rays, other photons, and electrons. Diffraction consists of the

constructive and destructive interference of the scattered wave. Constructive interference









results in a diffraction signal causing an intensity peak while destructive interference

results in no signal. Constructive and destructive interference is the result of the

periodically arranged atoms in a crystalline solid. The atomic alignment necessary to

cause constructive interference is defined by Bragg's law

nk = 2dhk sinO

where n is the order of the diffraction (typically 1), h is the wavelength of the incident

radiation, dkl is the spacing between the atomic layers with Miller indices of (hkl), and 0

is the angle between the beam of the incoming radiation and the normal of the plane of

atoms [96].

ZnS has two crystal structures, a cubic structure commonly called sphalerite and a

hexagonal structure called wurtzite. The crystal planes that can produce constructive

interference vary with each crystal structure. For example, face centered cubic lattices,

such as sphalerite, can only produce reflections if the indices are all even or all odd [97].

Sphalerite has an intense diffraction signal from the (111) plane at 28.580. Wurtzite has

an intense diffraction signal from the (100) plane at 26.940 and another intense peak at

28.53 from the (002) plane. If the films are thinner than the penetration depth of the x-

rays (typically a few microns for ZnS) the peak heights will be artificially adjusted if the

films are not all the same thickness. Due to the thinness of the deposited films in this

study (<1 jtm) and the penetration depth of the x-rays, diffraction scans of films

deposited on ATO/ITO substrates also exhibit diffraction peaks from ITO. Since there is

variation in the film thickness from sample to sample the full width at half maximum

(FWHM) of the peaks is used to compare the crystallinity of the films. As crystallinity

decreases the FWHM of the peaks increases until, in the case of an amorphous material,









the XRD pattern appears as a series of low broad undulations. In addition, the peak

position can be used to determine if the film is strained because strain will cause an

increase or decrease in the interatomic distance which, using Braggs law, will affect the

value of 0 [98].

3.5.3 Electroluminescence

Electroluminescent brightness was measured using various detectors depending on

the wavelength range. The excitation source was a custom built driver based on a design

by Planar Inc. The EL driver produced trapezoidal voltage pulses that had a rise time of

5 microseconds, a plateau width of 5, 30, or 800 microseconds (typically used at 30

microseconds), <5 microsecond fall time, and a frequency of 2.5 kilohertz. The high

voltage for the driver was supplied by a Sorensen DCS 600-1.7 high voltage power

supply. The current from the supply was limited to 0.025 amps and the voltage to 300

volts. The input pulses traveled through a 125+5 ohm resistor positioned before each

terminal of the device. The sample to be measured was placed on the sample holder as

shown in figure 3-4. The sample was placed on a glass slide attached to a mounting card

and held in position by pogo pins that also acted as leads to the device. The pogo pins

were connected to terminals on the card that was then placed into a card holder attached

to an x-z translation stage for alignment with the detector.

The detector for 350 to 1200 nm was an Ocean Optics S2000 silicon CCD with

Ocean Optics spectroscopic grating #13 installed. (See Figure 3-5 for the response of

grating #13.) The data were processed by computer using OOIBase32. OOIbase32 is a

program written by Ocean Optics Inc. to gather and process data received by the Ocean

Optics detectors. OOIbase32 collects and displays spectral data in real time over a range

from 200 nm to 1600 nm with integration times as short a 5 ms. Other detectors, used









mainly for time resolved electroluminescence and described below, included an Oriel

77341 photomultiplier for visible emission and an Oriel 71654 germanium detector for

near infrared emission. Calibration of the silicon CCD and photomultiplier tube was

done using an Oriel 63358 45W tungsten halogen calibrated lamp. Calibration of the

germanium detector was done using a 99.9+% efficient blackbody source.


sample -



pogo pll-


-mouiting card



-glass slide


- pogo pil


SLUUUUUU i


Figure 3-4 Back view of the sample on the test stage

The light path from the sample to spectrometer was an Ocean Optics VIS-IR

optical fiber with an attached 74-VIS collimating lens. The card and translation stage

assembly were installed in a test housing designed to minimize stray light. For a










schematic of the test stage assembly see figure 3-6. Other detectors are described in the

following section.


Grating #13


20 -
200 39 400 590 600 700 00 90
Wauelength (rm)


Figure 3-5 Spectral sensitivity of the Ocean Optics #13 grating


300 linesomm
Blazed at 500 nm


10D0 1100 120












[1I 1



I

S--- sample

"- pogo
IiL









Figure 3-6 Side view of the sample stage and fiber optic detection system

3.5.4 Photoluminescence and Photoluminescent Excitation

Photoluminescent brightness was measured using the same detectors used for

electroluminescence [99]. The excitation source was an Oriel model 66902 lamp with a

300W xenon bulb. Broadband light from the xenon lamp was monochromatized by an

Oriel Cornerstone 74100 spectrometer with 3 mm slits. Emitted light was focused on the

entrance slits of an Oriel MS257 monochromater. An Oriel 77265 photomultiplier tube

was used for detecting visible and near ultraviolet emission from 300 to 800 nm. The

detector used from 800 nm to 2ptm was a germanium detector. The detector for 2 to 5 um

was a thermoelectrically cooled lead selenide detector. Signal detection and chopping for

noise reduction was controlled by an Oriel Merlin control unit. Traq32, a program









created by Oriel Inc., controlled the MS257 and Cornerstone spectrometers. Traq32 was

written specifically to control Oriel spectrometers and to collect and process data. Using

Traq32, all spectrometer functions and data acquisition parameters can be specified.

Unlike like silicon detector discussed above, data are collected by Traq32 by scanning the

wavelength range, not at all wavelengths simultaneously. Data from Traq32 and

OOIbase32 can be easily imported into Microsoft Excel for data processing and analysis.

3.5.5 Electron Microprobe

The electron microprobe [100] was one method used to determine film

composition. A JEOL Superprobe 733 was used. Primary electrons were generated by

thermionic emission from a tungsten filament. The operating voltage was 8 kV. Since

the samples were on nonconductive bare glass substrates or on ITO/ATO substrates.

Because high beam currents are used during microprobe analysis (-20 nA) all of the

samples including the samples with ITO were evaporation coated with carbon to prevent

charging. For electron microprobe analysis (EMPA), characteristic x-rays generated

from the inelastic ionizing collisions of electrons in the sample are used to quantitatively

determine elemental concentrations. The X-rays may be energy analyzed using

dispersion by wavelength (wavelength dispersive spectrometry-WDA) or energy

dispersion (EDS). For this study energy dispersive analysis was used but at higher

currents, as discussed above, than the EDS analysis discussed below. The microprobe

data are quantitated based upon materials standards for the desired elements. Film

compositions were also determined with a x-ray spectrometer on an SEM as detailed

below.









3.5.6 Energy Dispersive X-ray Spectroscopy (EDS) on a Scanning Electron
Microscope (SEM)

EDS was used to verify film composition. A Hitachi S450 SEM with a Princeton

Gamma-Tech Prism digital spectrometer as the x-ray energy analyzer was used. Primary

electrons were generated by thermionic emission from a tungsten filament. The

operating voltage was 20 KV. The minimum usable voltage (10 KV was set by the fact

that the L line emission from the rare earths require this energy to be excited. 20 KV was

used, even with a greater penetration and excitation depth, because of reduced analysis

errors as compared to those found when using the lower accelerating voltage with rare

earths. The samples measured were on ATO/ITO substrates and the ITO was sufficiently

conductive to not require surface coating but the samples were daubed along their edge

with carbon paint to make electrical contact with the sample holder and reduce charging

from the sample. Collection time was twenty minutes to ensure high enough signal to

noise. Rare earth and rare earth fluoride standards were used as references for

determining the rare earth and fluorine concentration in each sample. For the other

elements, standardless quantification was used.

The high current electron microprobe analysis and the low current EDS use the x-

rays produced from atomic ionization induced by high energy electron bombardment.

Inelastic scattering of the energetic electron causes an inner shell electron to be ejected

from the atom. When an outer shell electron de-excited to fill the inner shell hole either

an Auger electron or a characteristic x-ray will be emitted. For EDS, the emitted x-rays

are collected by a silicon diode producing a charge pulse proportional to the energy of the

incident x-ray. These pulses are then amplified and processed to produce an energy

spectrum of the incoming x-rays [101].









3.5.7 Time Resolved Electroluminescence

Time resolved electroluminescence [102] was performed with an experimental

setup similar to that of photoluminescence measurements. The sample was placed in the

same position used for photoluminescence; however the sample was excited using the EL

driver and sample holder described in the electroluminescence section. A Tektronics

2024 digital oscilloscope or Tektronics TDS 3014 B digital oscilloscope was added to the

setup in the following manner. Channel one, called V1, of the oscilloscope was

connected before the resistor to the positive input terminal of the sample holder. Channel

two, called V2, was connected to the positive side of the holder after the resistor.

Channel three, called V3, was connected to the negative side of the device after the

resistor. Channel four of the scope was connected to the detector that was required using

a splitter and BNC cable to connect the detector to the oscilloscope and Merlin detection

system simultaneously (Figure 3-7).













sample
PMIT
[ r mono clu'omater


oscilloscope
II It Computer






VI, V2, V3




Figure 3-7 System to measure time resolved luminescence and electrical data

3.5.8 Electrical Measurements

Electrical data were taken with the samples in position for electroluminescence

measurements. Leads from the oscilloscopes were connected in the same manner as for

time resolved electroluminescence measurements. Using V=IR and the known

resistance, the current through the device can be determined by subtracting the value of

V2 from V1. Using the setup shown in figures 3-7 and 2-10, V3 corresponds to the

current through the sample when divided by the value of the sense resistor and this was

verified by subtracting the signal of V2 from that of V1. The sense resistor was a 125+5%

ohms. The PMT was connected directly to channel four of the oscilloscope when time

resolved measurements were made. The horizontal resolution of the scopes was set to

either 40 or 50 microseconds per division. This resolution provided information on either






64


a positive or negative pulse. The trigger value was 20 volts on the positive pulse edge.

The vertical resolution was dependent on the voltage of the pulses or the signal from the

PMT. The data was either sent to a computer via a GPIB cable or saved directly to disk

in the oscilloscope. The data was processed using Excel. The processing included

determining the external charge of the device during operation. The charge was

determined by integrating the current through the device over time. In addition, the

capacitance and electric field in the device were determined by further processing of the

data as detailed in section 2.7.














CHAPTER 4
PHYSICAL EFFECTS OF CHANGING TARGET DUTY CYCLES AND
SUBSTRATE TEMPERATURE DURING RF MAGNETRON SPUTTER
DEPOSITION

4.1 Introduction

In this chapter, the data on the effects of deposition conditions of ZnS:[RE]F3,

where RE is Tm, Nd, and Er, are presented. The objective of this study was to determine

the effects of sputter deposition parameter changes on infrared electroluminescent

intensity and to compare results from various rare earth dopants to draw trends to apply

to other lanthanides. It was found that changing the substrate temperature and the

sputtering target duty cycles modified several structural properties of the phosphors that

affect the infrared and visible emission. Duty cycle changes are listed as 100 multiplied

by the ratio between the duty cycle of the target doped with 1.5% rare earth fluoride to

the total duty cycles of the doped target and the undoped target. So a ratio of 50 means

that each of the targets was sputtering 100% of the time (100/(100+100) = .5 x 100 = 50)

while a ratio of 33 means that the doped target was sputtered 50% of the time while the

undoped target was sputtered 100% of the time (50/(100+50) = .33 x 100 = 33). The

substrates were heated so that the thermocouple described in chapter 3 measured

temperatures ranging from 130 C to 190 oC.

4.2 Spectra

None of the as-deposited phosphors exhibited photoluminescence. The xenon lamp

used as an excitation source was not intense enough to produce luminescence from this

condition. However, typical electroluminescence spectra obtained for as-deposited ZnS







doped with Tm, Nd, or Er are shown in figures 4-1 to 4-3. The spectrum from ZnS:TmF3
has two major peaks at 480 nm and 800 nm and one minor peak at 650 nm. These
correspond to the 1G4 -* 3H6, 3F4 3H6, and 3F3 3H6 transitions, respectively. The
ZnS:NdF3 spectrum exhibits one major visible peak at 600 nm and two major NIR peaks
890 nm and 1080 nm as well as several minor peaks. The major peaks are from the 2H11/2
419/2 for the visible emission and the 4F3/2 4 9/2 and 4F3/2 411 /2 transitions for the
NIR emission. The ZnS:ErF3 phosphor has several major peaks. The emission at 530,
550, 660, and 1000 nm correspond to the 2H11/2 -4115 2, 43/2 4115/2, 4F9/2 4115/2, and
4111/2 4115/2 transitions respectively. The energy levels and transitions are shown in
figure 4-4


0.003

0.0025
E
N 0.002
E
0.0015
.m
0.001

0.0005


0


300


400 500 600 700 800
Wavelength (nm)


900 1000 1100 1200


Figure 4-1 Electroluminescent spectrum of ZnS:TmF3


r,












0.012



0.01



E 0.008
E

0.006
U
C

S0.004



0.002



0
300 400 500 600 700 800 900 1000 1100 1200
Wavelength (nm)



Figure 4-2 Electroluminescent spectrum of ZnS:NdF3


0.0016


0.0014


0.0012
E
E 0.001-

E
S0.0008
U
2 0.0006


0. 0004 I -


0.0002


0
300 400 500 600 700 800 900 1000 1100 1200
Wavelength (nm)


Figure 4-3 Electroluminescent spectrum of ZnS:ErF3










20 -


15-*


25--


20--


15--


10--


5-


0-


10-


-


0-


I I I I 4FU2


I 41 3|

41,r
_ __ ',____________4|,


Tm3+


20-



15-






uJ
510-


5-


0o 1


Nd3"


IHjji


4Fr
--- -- -- ------- '4 I.


41
*li


Er3+

Figure 4-4 Energy levels of rare earth ions and transitions luminescence producing
transitions observed in Figs. 4-1, 4-2 and 4-3.


J-3
E E 3F
E E T
C C
w 3H5



-- --- --- 'H.
3H4


I f f 3 H 6


4G7/2


' '


.2H,11
-1112









4.3 Target Duty Cycle Alteration

Changes in the duty cycles of the sputtering targets affect the infrared emission

intensity of the ACTFEL devices. The possible duty cycles for both the undoped target

and the rare earth doped target were 100%, 75%, 50%, or 25%. If one target was set to a

duty cycle below 100% then the other was set to be on 100% of the time. The duty

cycles are listed as the ratio of doped target on time divided by on times of the doped and

undoped targets. The concentration of rare earth corresponding to each of the duty cycle

ratios is different for each rare earth and is discussed in the following section.

4.3.1 Concentration

The effect of duty cycle on the concentration of the individual rare earths, as tested

by EDS on the SEM and EPMA, is shown in figure 4-5. The trend was for the

concentration of each of the rare earths to increase as the relative duty cycle on the doped

target increased. As the duty cycle was changed the concentration of thulium in the

phosphor increased from 0.6 at% to 1.4 at%. As with the thulium doped samples,

increasing the duty cycle increased the neodymium concentration in the phosphor. The

Nd concentration rose from 0.55 at% to over 2.0 at%, while the concentration of Er in the

ZnS film exhibited the least change with changing duty cycle.

4.3.2 Crystallinity

The full width at half maximum (FWHM) of the 28.50 x-ray diffraction peak of

ZnS, which is observed from both the sphalerite (from the 111 plane) and wurtzite (from

the 002 plane) phases of ZnS, was used to characterize the crystallinity of the ZnS:[RE]F3

films. The FWHM increased for all of the films as the rare earth doped targets duty cycle

increased indicating that the host became less crystalline with increasing rare earth

concentration. The Tm and Er doped films experienced an increase in the FWHM of the







70


ZnS peak of over 30% while the data for the Nd doped films are too sparse to detect a


trend (Figure 4-6).


2.5
*Tm
a Nd
A Er
a 2-



1.5

U

c 1



0.5



0
20 30 40 50 60 70 80
Duty Cycle Ratio (dopedltotal)


Figure 4-5 Effect of target duty cycle on the Tm, Nd, and Er concentrations in the ZnS
films measured by EDS and EPMA










1 -
Nd
0.9 Tm

0.8

S0.7T



0 0.5


0.4

0.3

0.2
20 30 40 50 60 70 80
Duty Cycle ratio (doped/total)


Figure 4-6 Effect of duty cycle ratio on the full width at half maximum of the 28.50 x-ray
diffraction peak of ZnS

4.3.3 Thickness

The undoped target was further away from the substrate (8 cm) than the doped

targets (6 cm) yield resulting in a slower deposition rate for the pure material. In

addition, the sputter process changes the surface morphology of the targets as material is

sputtered causing the deposition rate to change slightly (-10%) from one deposition to

the next. For each film, the deposition time was changed in an effort to maintain a

uniform thickness between the samples of the same material. This effort was successful

for the Tm and Er doped films, however there was a large difference in thickness for the

Nd doped phosphors. Figure 4-7 shows the film thicknesses normalized to the thickest

film for each material and shows that the film thicknesses were usually within 5% of the










average for ZnS:Tm and ZnS:Er however, there was a large discrepancy in ZnS:Nd

thicknesses.


I-
",
o
--' 0 .6 ---------------------------


z 0.5


0.4


0.3
20 30 40 50 60 70 8
Duty Cycle Ratio (doped/total)


Figure 4-7 Normalized thickness of the rare earth doped ZnS films. Deposition times
were changed to attempt to achieve the same thickness for each rare earth
film.


4.4.4 Threshold Voltage

The NIR optical threshold voltage of each of the materials is shown in figure 4-8.

As will be shown in Chapter 5, the turn on voltage for infrared and visible emission is

identical. The turn on voltage for the Tm doped samples rose slightly as the Tm target

duty cycle increased but the majority of samples maintained a turn on voltage of

approximately 100 volts. The Nd doped films exhibited a turn on voltage near 200 volts

for the lower duty cycle ratios, but decreased to 130 volts for the higher duty cycle ratios.










The turn on voltage for the Er doped devices was consistently 110 volts except for the

lowest duty cycle ratio.


230
Tm
210 -- Nd T T

190

0 170

150
0
> 130


70
110

90 ---

70

50
20 30 40 50 60 70 80
Duty Cycle Ratio (doped/total)


Figure 4-8 NIR threshold voltages of the doped ZnS films with varying deposition duty
cycles

4.4.5 Infrared Emission

Alteration of the target duty cycles had a large effect on the emission intensity of

the near infrared emission. The effect of duty cycle on the different materials is shown in

figure 4-9. The brightness of the near infrared peak was highest for each of the rare

earths near the 50 ratio. The Tm emission maximum was at a duty cycle ratio of 57 and

the intensity decreased as the duty cycle ratio decreased. In contrast, the maximum Nd

and Er doped phosphor brightness were at lower duty cycle ratios and exhibited rapid

declines in infrared emission as the duty cycle ratio increased. There were similar trends

for the visible emission from each phosphor.












*Tm
0.9 _HNd
AEr

0.8

0.7

= 0.6

S0.5 T

E 0.4 T
o
z
0.3

0.2

0.1

0
0 ------------------------------------
20 30 40 50 60 70 80
Doped to total ratio (x100)


Figure 4-9 Effect of target duty cycle on the near infrared emission of each rare earth

4.5 Deposition Temperature Effects

The substrates were radiatively heated by resistive carbon cloth heaters located

below the sample stage to temperatures between 130C and 190C. The duty cycle ratio

that produced the brightest infrared emission at a substrate temperature of 160 C was

used for each of the rare earth dopants to study the effects of varying the substrate

temperature. In addition, the deposition time for each material was the same (50 min for

Tm and Er and 120 min for Nd) at each of the deposition temperatures.

4.5.1 Concentration

The effect of deposition temperature on the concentration of the different rare earth


dopants is shown in figure 4-10. As the temperature of the substrate was increased the

concentration of thulium, as tested by EDS and EPMA, in the deposited phosphor film

increased from below 0.5 at% to over 2 at%. The concentration of Tm rose steadily










between 130C and 170C with a sharp increase at 1800C. As with the thulium doped

samples, increasing the deposition temperature increased the neodymium and erbium

concentrations in the phosphors. The Nd concentration rose from below 1 at% to 1.5

at%. The concentrations of Er rose from 0.5 at% to 1.5 at% between 140C and 190 C.

The Nd and Er concentrations experienced sharp rises at the higher tested temperature,

similar to the thulium doped films.


o 2


."-
"-

C 1.5
0

, 1
LU
Si
Cu


0 4-
130
130


140 150 160 170 180
Deposition Temperature (Deg. C)


Figure 4-10 Concentration of each rare earth in the ZnS films as a function of substrate
temperature during deposition measured by EDS


4.5.2 Crystallinity

The full width at half maximum (FWHM) of the 28.50 x-ray diffraction peak of

ZnS, observed from both the sphalerite and wurtzite structures, increased for all of the

films as the deposition temperature increased indicating that the host became less










crystalline at higher temperatures. The Tm and Nd doped phosphors experienced an

increase in the FWHM of the ZnS peak of 30% while the FWHM of the Er doped

phosphor increased by 50% (Figure 4-11).


1.2 -

1.1

1

0.9

0.8

0.7

0.6

0.5

0.4

0.3
130


140 150 160 170 180
Deposition Temperature (Deg. C)


Figure 4-11 Increasing FWHM of the ZnS 28.5 diffraction peak as the deposition
temperature is increased


4.5.3 Thickness

As the deposition temperature was increased the thicknesses of each of the

phosphor layers decreased as shown in figure 4-12. The reduction in the thickness of the

films ranged from 55 to 30% of the maximum thicknesses obtained between 140C and

150C.














*+Er
0.95 -- Tm
A Nd


-.8 I 85

0.8

0.75
0 I
z 0.7


0.65

0.6
130 140 150 160 170 180 190
Deposition Temperature (Deg. C)


Figure 4-12 Decreasing phosphor thickness with increasing deposition temperature

4.5.4 Threshold Voltage

The turn on voltage also decreased as the deposition temperatures increased,

presumably due to the reduced film thickness (Figure 4-13). For the Tm doped films the

turn on voltage decreased from the maximum voltage of 130 volts at the lowest tested

temperatures (140 C) to 90 volts at the 180C deposition temperature. The effects of

deposition temperature on the turn on voltages of the Nd based phosphor were similar to

those of the thulium doped sample. The turn on voltage was at a maximum at the lowest

deposition temperatures and then fell with increasing temperature. For ZnS:ErF3 the turn

on voltage dependence on deposition temperature was smaller than for the other

materials, but higher deposition temperatures produced the lowest turn on voltages.











140
Tm
130 m Nd
A Er
120

0
> 110

100
0o 90


0- -
80

70

60
130 140 150 160 170 180 190
Deposition Temperature (deg. C)


Figure 4-13 Optical turn on voltage variation with increasing deposition temperature for
each material


4.5.5 Infrared Emission

Deposition temperature had a distinct effect on the emission intensity of the near

infrared and visible emission as shown in figure 4-14. The near infrared brightness was

highest at the 140 C deposition temperature for each of the rare earth dopants. Increasing

deposition temperature steadily reduced the infrared emission in each case. The overall

intensity loss was close to 80% in all cases.







79




Tm
0.9 Nd
0.8 Er
0.8

0.7

S0.6

0.5

=. 0.4

c 0.3

0.2

0.1

0
130 140 150 160 170 180 190
Deposition Temperature (Deg. C)


Figure 4-14 Decrease of near infrared irradiance with increasing deposition temperature

4.6 Discussion

It is clear that changing the RE concentration and substrate temperature critically

affected the properties of the phosphors. The reason behind most, if not all, of the

deposition temperature effects is because of Zn and S thermal desorption during

deposition. As the deposition temperature was raised the rare earth concentrations for

each of the phosphors increased. This is attributed to faster thermal desorption of the

host species than the rare earth dopants. This desorption is based on a lower sticking

coefficient for Zn and S at elevated temperatures. Thermal desorption has been used

previously to affect zinc and sulfur concentrations in materials such as ZnSxSex-l [103]

and decreasing thickness with increasing deposition temperature in ZnS films deposited









by spray pyrolysis has been attributed to re-evaporation [104]. The rate of desorption is

given by the Arrhenius type equation [105]


Rde k des n n exp OEn
s = k = V. exp QRT )


where Rdes is the rate of desorption, kdes is a desorption rate constant, 0 is the coverage,

Edes is the desorption activation energy, and Vn is the frequency factor of desorption.

The changes in concentration due to duty cycle variations are simply explained by the

increase in the amount of time the doped target was sputtered compared to the undoped

target. The variations from the expected trend for each material are the result of changing

sputtering target morphologies affecting the sputtering rates.

In addition to and because of the changing the rare earth concentrations, the

higher desorption rates at higher deposition temperatures modified the thickness and

crystallinity of the films. Since the deposition times for the temperature series films were

the same, the increased desorption of the host material as the temperature was increased

resulted in thinner films, as was shown in figure 4-12. Because the thickness was

decreased, a lower electric field was necessary to breakdown the phosphors resulting in

lower threshold voltages. The decrease in threshold voltage with increasing substrate

temperature correlates with the decrease in thickness, observed by the normalized values

for each shown in figure 4-15. The correlation between film thickness and turn on

voltage is supported by the duty cycle series (Figure 4-16).







81





0.95

0.9

0.85

S0.8

S0.75

E0.7 o
z -*-Tm turn on
0.65
.65 Nd turn on

0.6 -- Er turn on ,
o- Tm thickness
0.55 E- Nd thickness
A- Er thickness
0.5
120 130 140 150 160 170 180 190
Deposition Temperature (Deg. C)


Figure 4-15 Comparison of NIR turn on voltage and phosphor thickness as deposition
temperature is varied















09

I I \
S0.8


S0.7


0
Z 0.6 -- Tm turn on
Nd turn on
Er turn on
0.5 -- Tm thickness
[- Nd thickness ,
A- Er thickness .. .. -.
0.4 ..
20 30 40 50 60 70 80
Duty Cycle Ratio (doped/total)


Figure 4-16 Comparison of NIR turn on voltage and phosphor thickness as duty cycle and
deposition time is varied

Because the ZnS phosphors are crystalline as deposited, the change in crystallinity

with changing deposition conditions could be measured using XRD. It could be expected

that the crystallinity of the films would increase with increasing deposition temperature

due to the increased mobility of the sputtered species caused by increased thermal energy.

However, the crystallinity of the phosphors decreased as the deposition temperature

increased, as indicated by the increasing FWHM of the ZnS 28.5 diffraction peak. The

decrease in crystallinity is the result of increasing amounts of rare earths being

incorporated into the ZnS matrix. The rare earths incorporate substitutionally on the zinc

sites. The ionic radius of Zn2+ is 88 pm while the ionic radii of Nd3+, Er3+, and Tm3+ are

112, 103, and 102 pm respectively [106,107]. The rare earths have an average radius that

is 20% larger than the Zn ion resulting in the dopants creating more strain in the crystal









lattice. The duty cycle data support this interpretation since samples were deposited at

the same temperature and there is a general increase in rare earth concentrations

correlating with a decrease in crystallinity. As the rare earth concentration increases

defect formation results in an increasingly poorer crystallinity and an increasing number

of defects.

The infrared emission was affected by each of the changes in the properties of the

devices. Decreasing the thickness of the film yields a smaller volume of phosphor to

produce photons. Because of this, the changing thicknesses of the phosphor films can

mask how the concentration affects the NIR irradiance of the luminescent centers.

Normalizing the thicknesses of the films gives a more accurate view of how the

crystallinity and concentration affect the infrared output. This procedure is supported by

the nearly linear correlation between brightness and film thickness shown in Figure 4-16.

The data in Figure 4-17 show that the optimal concentration for all of the rare earths is

near 1 at%. The number of data points in Figure 4-17 is large because the data is from

both the duty cycle series and the deposition temperature series. This number compares

well with reports in the literature for the maximum visible luminescence from these

materials [74, 108,109]. Changes in irradiance are controlled by two competing

processes as the rare earth concentration increases. Higher rare earth concentrations

mean that there are more luminescent centers available to radiate. This would result in a

steadily increasing brightness as the luminescent center concentration was increased.

However, higher rare earth concentrations result in poorer crystallinity and possible non

radiative interaction between neighboring luminescent centers. It has been shown that

ZnS doped with Eu has poorer crystallinity than ZnS doped with an element closer in size







84


to Zn, such as Mn [110]. As the crystallinity of the phosphor decreases there are more

defects in the film. The defects can act as non radiative relaxation sites decreasing the

radiative efficiency of the devices. Finally, dopant to dopant interaction is another

increasingly prominent method of non radiative relaxation. The increasing concentration

of the rare earths can lead to concentration quenching [111-113, 86] caused by dipole-

dipole interactions or other non-radiative relaxation pathways between nearby rare earths.

The result is that as the rare earth concentration is decreased below 0.8 at%/ and increased

above 1.3 at%, the total infrared and visible irradiance of the films decreases

substantially.


0.012


0.01
E

3 0.008
E

. 0.006
I-

W 0.004
C-

0.002



0


0.0007


0.0006


0.0005 E


0.0004 E


0.0003 .2
T-

0.0002 E
I-

0.0001


0


0 0.5 1 1.5 2 2.5 3
Rare Earth Concentration (at%)


Figure 4-17 NIR irradiance as a function of rare earth concentration. Note that the
maximum occurs near 1 at% for each rare earth.









4.7 Comparison of Infrared to Visible Emission

Shown in figures 4-18 to 4-20 are the peak intensities (in uW) of the infrared and

visible peaks for each dopant over a range of rare earth concentrations, from -0.5 at% to

over 1.5 at%. Even though there is a difference in the infrared and visible intensities with

changing deposition conditions, the concentration with the maximum peak intensity is

very similar. The visible emission is affected similarly to the NIR emission by

crystallinity, concentration quenching, charge flow, and phosphor field. The slower

luminescence reduction with increasing concentration for Nd doped phosphors can be

explained by the state distribution of Nd. As discussed in section 5.6.1, Nd seems to

contribute shallower states than Tm or Er. The shallower state distribution means that

lower fields are necessary to inject them. The lower fields produce lower energy

electrons that are more suitable to exciting the lower energy levels that produce infrared

luminescence.







86



0.0008 0.0002
800 nm
0.0007 0480 nm 0.00018

0.00016
0.0006
E 0.00014
N 0-0005 0.00012

0.0004 0.0001

0.00000008
0 0.0003
0.00006
0.0002 -
S0 0.00004
0.0001
00001 0.00002

0 0
0 0.5 1 1.5 2 2.5 3
Concentration (at%)


Figure 4-18 B40 (left ordinate) for the NIR emission and B40 (right ordinate) for the
visible emission in ZnS:TmF3 for various Tm concentrations