Investigation of high performance multicolor quantum well infrared photodetectors and 4 x 4 three-color focal plane arra...

MISSING IMAGE

Material Information

Title:
Investigation of high performance multicolor quantum well infrared photodetectors and 4 x 4 three-color focal plane array for mid- and long-wavelength applications
Physical Description:
vi, 137 leaves : ill. ; 29 cm.
Language:
English
Creator:
Lee, Jung Hee, 1965-
Publication Date:

Subjects

Subjects / Keywords:
Quantum wells   ( lcsh )
Optoelectronic devices   ( lcsh )
Infrared detectors   ( lcsh )
Electrical and Computer Engineering thesis, Ph. D   ( lcsh )
Dissertations, Academic -- Electrical and Computer Engineering -- UF   ( lcsh )
Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 2000.
Bibliography:
Includes bibliographical references (leaves 130-136).
Statement of Responsibility:
by Jung Hee Lee.
General Note:
Printout.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 025854830
oclc - 46727534
System ID:
AA00013532:00001


This item is only available as the following downloads:


Full Text








INVESTIGATION OF HIGH PERFORMANCE MULTICOLOR
QUANTUM WELL INFRARED PHOTODETECTORS AND
4 X 4 THREE-COLOR FOCAL PLANE ARRAY FOR
N ID- AND LONG-WAVELENGTH APPLICATIONS












BY

JUNG HEE LEE


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


2000













ACKNOWLEDGMENTS


I wish to express my deepest gratitude to my advisor and the chairman of my

committee, Dr. Sheng S. Li, for his guidance, patience, encouragement, and support of

my research efforts. I would also like to thank Dr. Gijs Bosman, Dr. Arnost Neugroschel,

and Dr. Timothy J. Anderson for ser\ ing on my supervisory committee.

I am very grateful to Drs. Jung-Chi Chiang, Jerome Chu, and Xudong Jiang for their

assistance and valuable discussion in device fabrication and design. Special thanks are

extended to my laboratory colleagues, Chia-Hua Huang, Junhee Moon, Lin Jiang. .lion

Song, and Seung-Hwan Kim for their friendship and support in semiconductor processing

and measurements.

I am greatly indebted to my father, my parents-in-law, my wife, my dear son, and my

dear daughter for their emotional support, understanding, and unconditional love during

all the years of my life. Last, but not least, I cordially thank m\ late mother for her

devotion and endless love.

Finally, the financial support of ARPA, US Air Force Material Command. and

Advanced Device Technology, INC. is gratefully acknowledged.














TABLE OF CONTENTS


Page


A CKN O W LED G ENTS ............................................................................................ ii

TA BLE O F CON TEN TS................................................................... ............................. iii

A BSTRA CT................................................................................................................... v

1 IN TR O D U C TIO N ........................................................................................... 1

2 THEORETICAL STUDY ON QUANTUM WELL
INFRARED PHOTODETECTOR (QWIP) ...............................................10

2 .1 In tro d u ctio n ............................................................................................... 10
2.2 Calculation of Energy States in Quantum Well and Superlattice ............... 11
2.3 Transfer Matrix Method (TMM) for the Transmission Coefficient .............13
2.4 Strain Effect on the Band Structure of QWIP...............................................14
2.5 Absorption Coefficient on Intersubband Transition in QuantumWells........16
2.6 Q W IP Figures of M erit............................................................................... 19

3 A THREE-STACK INGAAS/ALGAAS/INGAAS BROADBAND
TRIPLE-COUPLED QUANTUM WELL INFRARED PHOTODETECTOR..26

3.1 Introduction.................................................... .............................. ......... 26
3.2 Device Design and Fabrication.....................................................................27
3.3 Results and Discussion ................................................ ......................... 28
3.4 C onclusions................................................................................... ........... 30

4 AN INGAAS/INALAS/INGAAS TRIPLE-COUPLED QUANTUM
WELL INFRARED PHOTODETECTOR FOR MWIR DETECTION.............37

4.1 Introduction ..................... ........................................................................ 37
4.2 Device Design and Fabrication ............................................... .............. 38
4.3 Results and D discussion ................................... ...................................39
4.4 C onclusions................................................................. ............ ......................40









5 AN ALAS/INGAAS/ALAS/INALAS DOUBLE-BARRIER
QUANTUM WELL INFRARED PHOTODETECTOR
OPERATING AT 205K AND 3.4 jum..............................................................47

5.1 Introduction ....................................................................................... ......... 47
5.2 Device Design and Fabrication ..................................... ......................... 48
5.3 Results and Discussion ........................................... .......................... 49
5.4 C onclusions................................................................ ...................... ............51

6 QUANTUM WELL INFRARED PHOTODETECTORS WITH
DIGITAL GRADED SUPERLATTICE BARRIER FOR LONG
WAVELENGTH AND BROADBAND DETECTION ...................................58

6.1 Introduction ................................................................................................... 58
6.2 Device Design and Fabrication .................................................................59
6.3 Results and Discussion ................................................ ...... ...................62
6.4 C onclusions............................ ................................................................ 65

7 HIGH SENSITIVITY QUANTUM WELL INFRARED
PHOTODETECTORS WITH LINEAR GRADED BARRIER.........................74

7.1 Introduction ........................ ........................................................................... 74
7.2 Device Design and Fabrication ............................ .................................75
7.3 R results and D discussion .................................................................................77
7.4 C onclusions.................................................................................................. 80

8 THREE-COLOR THREE-STACK QUANTUM WELL INFRARED
PHOTODETECTORS 4 x 4 FOCAL PLANE ARRAYS ..................................92

8.1 Introduction ............. ... ................................................................ ..............92
8.2 Device Design and Fabrication ............................. .................................93
8.3 Results and Discussion ........................................... .......................... 96
8.3.1 LW/LW/MW QWIP 4 x 4 FPA .......................................................97
8.3.2 LW/MW/SW QWIP 4 x 4 FPA ......................................................98
8.4 C conclusions ................................................ ............................. ............ 100

9 SUMMARY AND CONCLUSIONS............................................................... 123

R E FE R E N C E S ............................... ................................................... ..................... 130

BIOGRAPHICAL SKETCH ..................... ..........................1..37













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

INVESTIGATION OF HIGH PERFORMANCE MULTICOLOR
QUANTUM WELL INFRARED PHOTODETECTORS AND
4 X 4 THREE-COLOR FOCAL PLANE ARRAY FOR
MID- AND LONG-WAVELENGTH APPLICATIONS

By

Jung Hee Lee

December 2000

Chairman: Sheng S. Li
Major Department: Electrical and Computer Engineering

In this work, several novel high performance n-type quantum \ell infrared

photodetectors (QWIPs) and two three-color, three-stack QWIP 4 x 4 focal plane arrays

(FPAs) have been developed for 1-3 utm short-wavelength infrared (SWIR), 3-5 Vim mid-

wavelength infrared (MWIR), and 8-14 utm long-w\avelength infrared (LWIR) detection.

Multi-color, high-sensiti itN, and broadband detection have been achieved by using

different QWIP device structures and material systems.

The first QWIP for LWIR detection is a high performance broadband

InGaAs/AIGaAs/InGaAs triple-coupled (TC-) QWIP grown on semi-insulating (SI)

GaAs substrate. The full-width half-maximum (FWHM) of this QWIP was two times

larger than that of the normal QWIP structure. The second QWIP was a new MWIR TC-

QWIP using lattice-matched In,)53Gai47As/Inj i.AlIu48As/In ,.3GaAs; material systems

grown on InP substrate for 3-5 Itm detection with operating temperature up to 116K. Ihe







third QWIP structure was a new dual-mode (PV and PC mode) n-tN pe

A1As/InGaAs/AlAs/InA1As double-barrier (DB-) QWIP grown on InP substrate \\ith

peak detection \aa\elength at 3.4 uIm and operating temperature up to 205K for mid-

\wavelength infrared (MWIR) detection.

The fourth QWIP devices were two novel high performance InGaAs/AlGaAs/GaAs

QWIPs using digital graded superlattice barrier (DGSLB) grown on the GaAs substrates.

These new structures enable the broadband detection (54 62 %) and significantly!

improve the device performance under positive bias operation. The fifth QWIPs were a

high performance QWIP using InGaAs/AlGaAs linear graded barrier (LGB) QWIP

structure for the broadband (BB-) detection (6.5 16 4lm) and a high-sensiti ity

AlGaAs/InGaAs/AlGaAs double-barrier (DB-) LGB QWIP with peak responsivity of

4.38 A/W at 9.1 pmr and 35K.

Finally, two three-color, three-stack QWIP 4 x 4 focal plane arrays (FPAs) with a

pixel size of 100 x 100 pm2 were fabricated and characterized for the first time. Each

stack can detect unique peak so that three-color detection can be obtained through the

three-stack structure. These two FPAs were designed to detect the LW/LW/MW and

LW/MW/SW regimes, respectively. The excellent multi-color detection with three

identical peaks was successful achieved.













CHAPTER 1
INTRODUCTION


Since a physical object emits its energy into infrared (IR) spectrum radiation under

nature's normal environment, the IR radiation detectors are most useful devices to

observe the motions of the lives, the objects, and the surroundings without light. The IR

spectrum can be divided into short-wavelength infrared (SWIR, 1-3 plm), mid-

wavelength infrared (MWIR, 3-5 pm), long-\wa\elength infrared (LWIR, 8-14 lTm), and

very long-wavelength (VLWIR, >14 gm). The IR detectors can be grouped into two-t) pe

[1]: thermal-type and photon-type. The thermal-type detectors are called uncooled

thermal detectors because most thermal detectors do not require the cryogenic cooling

system while liquid helium, liquid nitrogen or thermo-electric cooler are needed to

operate the photon-type detectors. Three different thermal detectors have been studied.

which are the bolometer, the pyroelectric, and the thermopile. These thermal detectors

can observe changes in some phy sical property such as the temperature increase and the

electrical resistance by the heating effect due to the absorption of the incident IR

radiation, and usually respond equally to all wavelengths. However. the time constant of

a thermal detector is not short enough so that the applications are significantly limited.

For photon detectors, the spectral response results from the direct interaction between

photons and carriers in material. The response time of the photon detectors is very short

and sensitivity is one or two orders higher than thermal detectors.







A wide variety of detector materials such as HgCdTe. InSb. InGaAs, and PtSi ha\e

been used to detect desirable optical absorption, which have various band gap energies

(Eg). In general. IR detectors require very high quality material to improve the device

performance. Therefore, the detector materials should keep stability, uniformity and high

operability. The detection wavelength regions of IR detectors using interband transition

rely on the bandgaps of the detector materials so that the detection wavelength flexibility

will be largely compromised.

In recent years, quantum well infrared photodetectors (QWIPs) have been de eloped

for MWIR and LWIR detection. The quantum well (QW) can be formed due to the

conduction or the valence band offset when a smaller band gap material was grown

between two larger band gap materials. The quantized energy levels are formed inside the

quantum well so that the infrared radiation can be detected due to the optical transitions

from the ground state to the upper excited states. The quantum well must be doped with

donor (i.e., n-type QWIP) and the electrons, which occupied the ground state of the

quantum well by doping. can be photoexcited to an unoccupied excited state to generate

photocurrent for IR radiation detection. The separation of energy levels in the quantum

well can be varied by changing the well thickness and the mole fractions of the quantum

Well and barrier. Therefore, the detection wavelength of QWIP is comparatively not

dependent on the device material but rather the layer structure. The rapid progress of the

III-V material device growth technique such as molecular beam epitaxy (MIBE) [2] and

metalorganic chemical vapor deposition (MOCVD) has extensively encouraged the

development of a number of QWIP devices. The QWIP has a lot of advantages: high

uniformity, high yield, low cost. wavelength flexibility, multicolor capability, radiation







hardness. lo\\ 1/f noise, and large size staring arrays. Although the photovoltaic (PV)

mode IR detector under zero bias is attractive for some applications due to the low dark

current, low power dissipation. excellent noise property and fast integration time for

focal plane array (FPA), the photoconductive (PC) mode under biased operation is still

practical because the responsivity of the PC mode detector is much higher than the PV

mode operation.

Since West and Eglash [3] first observed the strong optical intersubband transition in

GaAs/AlGaAs multi quantum well (MQW) and Levine et al. [4] demonstrated the first

GaAs QWIP, various intersubband transition schemes and material systems have been

widely studied. For example, bound-to-bound (BTB) [4], bound-to-quasi-bound (BTQB)

[5], bound-to-miniband (BTNI) [6], and bound-to-continuum (BTC) [7] transition

schemes \were chosen to improve the device performance in the past ten )ears. Figures

1.1 and 1.2 show the schematic energy band diagram for n-type and p-type QWIPs.

respectively. The p-type QWIPs can allow the normal incident IR detection due to the

linear combination of p-like valence Bloch states without the grating coupler [8,9].

However, the n-type QWIP is more interesting than the p-type QWIP because the

performance of the n-type QWIP is much better and the grating coupler is no longer an

issue due to the advanced processing technology.

The multi-color detection for the MWIR and LWIR dual band detection is more

important to determine thermal feature on a target object because the absolute

temperature can be uniquely determined by the ratio of radiance measured at the twxo-

wavelength band. Therefore, the multiple wavelength IR detector can improve the device

performance including better discrimination and tracking. This multi-color detection can








be obtained by several structures. which are the multi-stack structure with one NIWIR

stack and one LWIR stack [10,11], asymmetrical QWIP structure for voltage tunabilit),

and QWIP structure with two occupied energy levels [12]. Figure 1.3 shows the different

structures for multi-color detection.

The excellent QWIP performance makes it possible to develop the large area, high

sensitive, high-speed, and low cost QWIP staring focal plane arrays (FPAs) which are

useful in some practical applications: the remote sensing of earth/atmosphere, infrared

astronomy, night vision, tracking, medical thermal imaging, temperature measurement,

and weather \watch [13]. Until recently, 128 x 128, 256 x 256. and 640 x 480 FPAs [14-

17] have been demonstrated with higher uniformity and higher yield and a 1024 x 1024

format FPA is possible at the present time. In addition, the multi-color QWIP FPA for

MWIR and LWIR dual band detection has been successfully demonstrated. Readout

circuitry is needed to transfer signals from the detector pixels to the output of the chip.

The Si-CMOS MUX is usually hybridized to the QWIP FPA for the imaging camera. The

readout integrated circuit (ROIC) with high-resolution and low-noise for all QWIP FPA

technologies is necessary to convert the photoresponse efficiently into digital form for

signal processing. Dex elopment of multi-color QWIP \with high performance is of prime

interest for FPAs and other image applications.

In chapter 2, we will describe the basic principles of the quantum well infrared

photodetectors, which include the calculation of the energy levels in the quantum well

and superlattice, the strain effect on bandstructure. the transfer matrix method (TMM) for

the calculation of the transmission coefficient, the absorption coefficient on the





5


intersubband transition in the quantum wells, and the QWIP figures of merit for de\ ice

performance characterization.

Chapter 3 will give a demonstration of an InGaAs/AlGaAs/InGaAs broadband (BB)

triple-coupled (TC-) QWIP for voltage tunable multicolor detection in the LWIR region.

This BB TC-QWIP was formed by using a three-stack structure without contact laNers

between the stacks in which each stack has a different period to appropriately distribute

the applied bias.

Chapter 4 will report a new MWIR triple-coupled quantum well infrared

photodetector (TC-QWIP) using an Ino.53Ga0.47As/In0.52Al.48As/Ino.3Gao.7As material

system grown on InP substrate for 3-5 utm detection. The peak wavelengths for this

device were ,pi = 4.6 gm and Xp2 = 3.7 p.m, which were due to the BTB and BTC

transitions, respectively.

Chapter 5 will show\ the characterization of a new dual-mode (i.e., PV and PC mIode)

operation n-type AIAs/InGaAs/AlAs/InAlAs double-barrier (DB-) QWIP with peak

detection wavelength at 3.4 p.m for mid-wav\elength infrared (MWIR) detection. This

device can be operated up to 205K with excellent performance.

In chapter 6, we first report two novel high performance InGaAs/AlGaAs/GaAs

QWIPs using a digital graded superlattice barrier (DGSLB) to achieve the staircase-like

graded barrier across the barrier region of the QWIP. The broadband detection w\as

achieved under positive bias and the normal photoresponse with \'er\ high response\ ity

\\wa observed under negative bias.

Next, the broadband (BB-) linear graded barrier (LGB) QWIP and the double-barrier

(DB-) LGB QWIP are presented in chapter 7. The spectral response% it) was measured at







T = 35, 60, and 77K. The broadband detection was obtained under positive bias

conditions in both LGB QWIPs in \which the detection w\avelength of the BB-LGB QWIP

was much broader than the DB-LGB QWIP. Under a negative bias condition, the peak

responsivity of the DB-LGB QWIP was found to be as high as 4.38 A/W.

Finally, two 4 x 4 focal plane array (FPA) three-color, three-stack QWIP structures

with a pixel size of 100 x 100 utm2 were grown on a semi-insulating (SI) GaAs substrate

in which each stack can cover LWIR, NIWIR, or SWIR regime, respectively. These 4 x 4

FPAs consist of three stacks with a buffer layer between the middle-stack and the bottom-

stack for multicolor detection. The first FPA (LW/LW/MW) has three distinct peaks of

12, 8.8, and 4.2 .tm \while the second FPA (LW/MW/SW) is sensitive to three bands \\ith

corresponding peaks of 7.9, 3.7, and 2.4 upm, respectively. These devices are

demonstrated in chapter 8.















(BTB N-QWIP)


(BTQB N-QWIP)


(BTM N-QWIP)
(BTM N-QWIP)


(BTC N-QWIP)


Figure 1.1. The schematic conduction band diagrams of n-type QWIPs.


I57


- %








LH1
HH1
HH2
r I~I


HH3


(TSL P-QWIP)


HH1
HH2


-, w MORMI-M3MIS R


HH3


(CSL P-QWIP)


HHI

inHInnnrIHnnr


UUUUU


HH4


(SBTM P-QWIP)


Figure 1.2. The schematic valence band diagrams of p-type QWIPs.


(*00


UUUUU


I~l~aei~0l00 S~


-ICI IL~I 1








SWIR


(three-color three-stack DB/BTC/BTQB QWIP)


MWIR


(two-stack two-color BTC/TC QWIP)


LWIR
TTTFi 111lli


LWIR


(BB-DGSLB QWIP)


(BB-LGB QWIP)


Figure 1.3. The schematic conduction band diagrams of n-type multicolor QWIPs.


m JE,













CHAPTER 2
THEORETICAL STUDY ON QUANTUM\ WELL INFRARED PHOTODETECTOR
(QWIP)

2.1 Introduction


The quantum well infrared photodetectors (QWIPs) based on intersubband transition

are composed of several periods of the doped quantum %wells (QWs) and the undoped

barriers. The maturity of the QWIP growth technology such as molecular beam epitaxy

(MBE) technique make it the rapid progress of QWIPs with high performance in the past

decades. Since the first GaAs/AlGaAs QWIP was developed, various QWIP structures

with different material systems and transition schemes ha\e been studied. For an

optimized QWIP device, the materials, the light coupling scheme, operating temperature,

and readout electronics must be significantly considered. The great success in the single

QWIPs has led the development of the focal plane array (FPA) and in recent years the

large (640 x 480) FPA with high uniformity has been successfully demonstrated [17]. In

addition. It is well known that it is important to understand the design and analysis of

infrared focal plane array readout circuitry with low-noise and high-resolution. The

QWIP (<30 %) has much lower quantum efficiency than HgCdTe (=60 %). However, the

QWIP has a number of advantages on wavelength flexibility, multi-color capability, high

uniformity, lower cost and so on. In order to improve the QWIP device performance, it is

motivated to understand the basic QWIP detector physics. In this chapter, the

fundamental principles on QWIP are described.







2.2 Calculation of Energy States in Quantum Well and Superlattice


Using the effective mass approximation, the motion of the electrons in quantum well

structure can be written by Schr6dinger wave equation as


[- V +V(z) T = ET, (2.1)
2m

where the m* is the effective mass, V(z) is the potential profile along z-direction. and z is

the grotIlh direction of the QW structure. When the wave function %as divided into t\o

components of the z-direction and x-y plane, the solution of Schridinger equation can be

simplified as [18]

Y(x,y,z)=e' e (z), (2.2)

By substituting (2.2) into (2.1), and we can get

h2 a2
S2 + V(z) y(z)= Ey(z), (2.3)
2m* az

Here. when we first consider the quantum well structure with infinite barriers, the

normalized wave function using the boundary conditions at -1/2 and 1/2 with quantum

well thickness of Ican be gi\en as


y,(z)= -cos( n nodd (2.4)


2 ( n&
= -ssmi- ne\en (2.5)
1/1

For the quantum well structure with finite barrier height (1 ), the condition of the allo\\ed

bound energy states can be obtained as


a tan(-1 = fl (2.6)
(2








a cot-I = -pl (2.7)


where

2m E
a (2.8)
h2


2m (V E)
2 (2.9)

These equations can be solved by numerical technique.

When the quantum wells and thin barriers are alternatively and periodically repeated,

it is called superlattice. In a quantum well structure with thick and finite height barriers,

the bound states are localized into the quantum well region, which can also be

exponentially deca ed into the barrier region. Therefore, there is no overlapping of bound

state wa\e functions between the adjacent wells. However, when the barrier thickness is

decreased, those bound state wave functions can be overlapped, which results in a

miniband. The Kronig-Penney model can be employed to calculate the dispersion relation

of the minibands. When the superlattice has a quantum well \\idth of a, barrier width of

b, barrier height of Vc, and period of d (= a+b), the solution of the Schridinger equation

for this superlattice can given as a Bloch theorem:

y(z + d) = e'"-"y(z), (2.10)

In a period,

y(z) = .4 e' + Be''-, in the barrier (2.11)

=Ce'" + De-'", in the well (2.12)


Then, in the following period,








y(z)= e':"[Ae''cl~'"- + Be ."''-/)], in the barrier (2.13)

= e''"Ce"' I-"0 + De-(('"") ', in the well (2.14)

where Sis if in which a and flare show n in (2.8) and (2.9), respectively. A, B, C, and D

constants can be obtained by the boundary conditions and the dispersion relations for the

bound and unbound states are given as [19]


cos(kd) = cos(aa)cosh(bfJ) -a sin(aa)sinh(bf,), for 0 < E < Vc (2.15)
2af

a2 + ,.
= cos(aa)cos(bS) sin(aa) sin(bd), for E > Vc (2.16)
2a6

Therefore, the minibands can be calculated from the above Eqs. (2.15) and (2.16).



2.3 Transfer Matrix Method (TMM) for the Transmission Coefficient


The energy levels and the corresponding vwa\e functions in a multi-quantum well

structure can be obtained by different methods [20-23] such as the standard analytical

method, Wentzel-Kramers-Brillouin (WKB) approximation. Kronig-Penney model, and

the variational principle, which are however not suitable for general purposes. Recently,

the transfer matrix method (TMM) [24], which involves the straightforward

multiplication of 2 x 2 matrices with no iterations, has been developed to analyze exactly

the multi-quantum well structures. When we consider an arbitrary multi-quantum \ell

structure with N+1 layers, the solution of the Schrodinger wave equation can be given as

/,, = A,,e -\ *e'- + B,,e' e (2.17)

where


A, =A2 =0,


(2.18)







A,, = k,,(d + d, + + d,_, ), n = 3, 4, 5, Y-, + 1 (2.19)

2m,2
k,,= (E- V) (2.20)


where A, and B, are the amplitudes of the wave function in +z and -z directions,

respectively, and d,, m,,, and V, are the thickness, the effective mass. and the potential of

the nth layer, respectively. Applying the boundary conditions at each interface gives as

follows

(A, (A.2 (A,) AA AN+ (
Bi B2 B3 BB

where N is the total number of the layers and

I ( e '"'''' r,,e-'" k. ,+ 2k,
IS t r e' k r,, = ^ ( 2.22
S t,, r,,e"- e'id k,,+k ,,+ k,, + k,,.

Here, we can set BN+i = 0 since there is no reflection wave in N+1'1 layer. Therefore, the

transmission coefficients can be calculated by A,,, / A,2 as a function of E and then each

energy states can be obtained.



2.4 Strain Effect on the Band Structure of QWIP


In order to grow the strain structure without misfit dislocations, the layer thickness

cannot be more than a certain threshold, which is called critical thickness. If the layer

thickness exceeds the critical thickness, misfit dislocations are created in the structure.

Matthrews and Blakeslee expressed the critical layer thickness as [25.26]

h I vYcos' [1i 2
c 8r(1 + v) cos \. b








where b= a/l2 and a is the lattice constant grown on the substrate. 0 is the angle

between the dislocation line and the Burges vector, A. is the angle between the slip

direction and the la\er plane direction, v is the Poisson ratio, d is the la\er thickness, c=

(a?-al)/al is the biaxial strain parameter in which al and a7 are the lattice constants of the

epilayer and the substrate. respectively. The value of the in-plane lattice constant (al) due

to straining is obtained by minimizing the total elastic strain energy of the barrier and the

well. The in-plane lattice constant (all) is intermediate between the unstrained lattice

constants of the layers, then [26]


aa, =a 1+ s/ 1+S, d (2.24)
1 S, d,

where S, are the shear moduli. di is the layer thickness, and E is the lattice mismatch

between layers. Figure 2.1 (a) shows the unstrained lattice layers having different

constants (a2 > a/) and the strained layers with in-plane lattice constant (a ). In an

unstrained condition, the heavy and light holes have the same potential and the first

heavy hole energy level is always the first valence band level because of the effective

mass difference between heavy and light holes. After straining, the splitting of the hea\y

hole and light hole at the valence band zone center (F) occurs in every band structure;

that is, under biaxial compression, the heavy hole band is higher than the light hole band.

and the interval between the first heavy and light holes is increased by splitting while the

light hole band is pushed to a higher le\el than that of the hea\\ hole under biaxial

tension. In addition, the conduction band is changed due to the strain in which the

conduction band can be moved upward and downward depending on biaxial compression

and biaxial tension, respectively. Figure 2.1 (b) shows the energy band shift bN strain in







detail. The energy band gaps due to strain for the heavy hole and light hole are given as

follows [27]

Eo = E, + AEOhh, (2.25)

Elh = Eo + AEolh, (2.26)

where Eo is the unstrained band gap and AEohh, and AEoih are the strained band gap shifts

for the heavy hole and light hole, respectively,


AEhh =L 2a C11 C12 C11 -12 C2 (2.27)



AE, = 2a c + bC + 2C12, b2 C11 12 /(2Ao) (2.28)
( C11 J I ) Cl I

where e can be a negative sign for compressive strain or a positive sign for tensile strain,

a is the hydrostatic deformation potential, b is the shear deformation, Ao is the spin-orbit

splitting energy, and Cy is the stiffness coefficient of the material.

When the quantum wells are highly doped, the electrons interact with each other so

that the energy level shift due to the many-body effects on quantum well structure can be

observed, which are direct Coulomb interaction, electron-electron interaction, electron-

phonon interaction and so on [28-33]. Therefore, those effects should be taken into

account in calculating the energy levels in the quantum well structure.



2.5 Absorption Coefficient on Intersubband Transition in Quantum Wells


The calculation of the energy states and the corresponding peak detection wavelength

in a QW structure was described in the previous subsections. In addition, the calculation







of the absorption coefficient on intersubband transition in a QW structure is needed to

determine the absorption strength and the absorption lineshape. The absorption

coefficient as a function of the photon energy, hw, can be defined as [5]

a(h) = h t nibher of transition per unit cell volimne and lime
a(thw) = co- (2.29)
inc'id'nt energy) flux

When %ec consider the intensity of the radiation field (/) and the optical intersubband

transition rate between the initial and final states (Wf,), the absorption coefficient can be

also given as

hno W, (f, f;)
a(hw) = (2.30)
QI

where ff and / are the Fermi-Dirac distribution function for the final and initial energy

states, respectively, Q is the volume of the QW, and WfT can expressed as [34]


W, = V,, '8(E, E,-hw), (2.31)

where Ef and E, are the final and initial energy states, respectively, and Vp is the

interaction potential for the photon absorption, which can be written as [34]

e e = e
V =- A.p Aojcos(k r-ot)^ p --- A e-e p, (2.32)
m m 2m

where is the polarization vector, m is the effective mass of the electron, and p is

momentum vector of the electron in the QW. In a simple one-band model, the wave

function for a quantum well structure is given as [35]


e gkp Y/,, (z) u(r). (2.33)


where Ad is the QW area, k = (k,,k,)is 2-D in-plane wave\'ector, p=(x,y) is 2-D







coordinate space vector, r = (p,z), ii(r) is the period part of the Bloch function, and

y,,(z) is an envelop function in the growth direction (z) for the n'" energy state.

Therefore, the transition rate can be rewritten by


W 2i = 2 f, f A ei. k-oe P, pi -(E E co),
h 2m I

2 / e 12
4--m A 10 -i i- ) { (E,-E-ho), (2.34)
h 4m 9z


where KYV,, T,)(FY Vj) for the intersubband case, -p=p =-ih-, and


e'r = 1 for the QW scale [5].

The monochromatic radiation field impinging on the QWIP structure can be

described by [36]

A(r,t) = A, lcos(k -r -(l) (2.35)

where A0 is the complex polarization vector and k is the wa\ evector. The electric field,

E(r,t), and the magnetic field, B(r,t), can be calculated by

E(r,t) = A( ) Asin(k-r- t), (2.36)
c at c

B(r,t)=VxA=-k x lAosin(k-r-o/), (2.37)

With these two E(r,t) and B(r,t), the intensity of the radiation field can be described by

averaging the Poynting vector (S) as follows

c nwB- A, 2
I =I S= ExB B -= (2.38)
where n is the refractive index and c is the speed of the light.
where n is the refractive index and c is the speed of the light.







Finally, by substituting (2.34) and (2.38) into (2.30), the absorption coefficient can be

expressed as [37]


a(ho) = 4' e 2 ) (E, E, -hco),


4r22 e 2dk M2 1 F/2
f;2 M f(-f ) (2.39)
cno) Bz(2)2 f' i (Ef -E, -hw +(r/12)

where M. = (- ih / m y, 1(a / 1z ,) and F is the broadening factor.



2.6 QWIP Figures of Merit


In order to design and characterize the QWIP device, the figures of merit such as dark

current, spectral responsivity, noise, and detectivity must be highly considered.

Moreover, these figures of merit can be controlled by the doping density in QW, the

number of QW, and the barrier profile for the optimization of QWIP performance.

It is well known that under a dark condition there are three electron transport

mechanisms, which are thermionic emission, thermally assisted tunneling, and direct or

trap-assisted tunneling. The thermionic emission due to the electrons transferred out of

the quantum wells is dominant at low biases and high temperatures while the thermally

generated carriers tunneling through the barriers are more important at high biases and

low temperatures. The dark current models for QWIP have been analyzed in the

published papers [38-47]. The dark current was calculated by assuming a constant

electric field across the QWIP active layer and thermodynamic carrier equilibrium in

which the dark current calculation was performed using WKB approximation including

image charge effect and it was compared with calculations using the Transfer Matrix







Method (TMM) [41]. The dark current calculation in Liu et al. [42] shows the good

agreement with the measured dark current at a range of temperature from 65 to 105K.

The dark current of QWIP due to the thermionic emission is given by [48]

I, = Aevn, (2.40)

where Ad is the detector area, e is the electronic charge. Vd is the average drift velocity,

and n is the density of mobile carriers, which can be expressed as

v, = pF/[l+(PF/v,)2]", (2.41)


n = (m /r 2 L,) (E)T(E, F)dE, (2.42)
E,

where u is the electron mobility, F is the average electric field. v, is the electron

saturation velocity, m* is the electron effective mass in the QW, Lp (=L,,+Lb) is the QWIP

length, and the f(E) is the Fermi-factor given by f(E) =[l+exp(E-E -E-E,))/kT]-,

respectively. T(E,F) is the bias-dependent tunneling current transmission coefficient for a

single barrier and can be obtained by WKB approximation, which is given by [42]

T(E, F) = 1, for E > Eb eFL,,, (2.43)


T(E,F) =exp 3'hF(m h) (E E-eFL1) ,


for E, -eFLp < E < Eb eFL,, (2.44)


T(E, F) = exp(- 3t4 (2me)') 2 E, E-eFL," (E,, E -eFL,, )2


for El < E < Eb eFL,, (2.45)

where L,, Lb, and mb are QW width, barrier thickness, and electron effective mass in the

barrier, respectively. Finally, the dark current can be expressed as follows,







S= Aem* jiF I.(E)T(E,F)dE. (2.46)
h L, 1 + (pF / v,)2 ,*

For the sake of simplicity, we can assume that T(E) = 0 for EEb.

Therefore, n and Id can be given by,

n = (m*kT /rh L,, )exp[-E,, / kT], (2.47)

mkT (F Er
I = A ,77kT j / + exp W (2.48)
d t L, L1+,/v, + kT

where Eac (=Eb-EI-EF) is the activation energy and Eb, El, and EF are the barrier energy,

the ground state energy, and Fermi energy, respectively. The Fermi energy (EF) can be

calculated from


Nd mrh2T lI + exp( )]. (2.49)
2 L,, I UkT

As a result, the dark current (Id) is exponentially related to the doping density and the

activation energy (E,,ac) as follows,


Id oc exp [- oc ex k" (2.50)
LkT kT

Therefore, the activation energy (Eac) can also be obtained from the slope of the

normalized dark current (IdT) versus the inverse temperature straight line on a semilog

scale.

The absolute spectral responsivity (Ri) can be given by quantum efficiency (q) and

photoconductive gain (g) as [49]

eArI e 2
R, == =g,. i- g, (2.51)
he hv 1.24








where e is the electronic charge, A is the wavelength of the incident photon, h is the Plank

constant, c is the speed of the light, v is the incident frequency, and qc is the collection

efficiency (rlg). The quantum efficiency (q) [49] is the average number of optical

transitions per each photon incident into the detector and described as

7 = (1/ 2)(1 R )(1 e-"'"i), where Rc is the reflection coefficient, m is the number of the

absorption pass, a is the absorption coefficient of the QW layers, and I is the total

thickness of the QWIP layer. The photoconductive gain (g = vd/L) can be expressed as

the ratio of the electron mean free path (Vdr) to the QWIP thickness (L) or the ratio of

photo-generated electron lifetime (r) to the transit time (L/vd) across the QWIP layer,

where Vd is the drift velocity of the photo-generated electrons. In general, the

photoconductive gain (g) is not required less than one, which can be more than one when

a photo-generated electron can circulate the circuit several times until it is recombined.

The spectral responsivity can also be calculated by

I
Ri = (2.52)
'i"

where Ip is the photocurrent output (A), and Pin is the input IR radiation power (W),

which can be expressed as

V
I = (2.53)
Rf

P, = AdH,,,. (2.54)

where Vp is the photovoltage of the photodetector, R, is the gain of the transimpedance

amplifier (TIA), Ad is the photodetector area (cm2), and Hi, is the input irradiance

(W/cm2), which is given by








V T OTH
Hi,, (2.55)


where Vr.,o is the photovoltage of the pyroelectric detector. T,, is the transmissivity of the

entrance \ window of the cr ogenic s\ stem. Td is the transmissivit) of the photodetector.

R, is the responsivity of the pyroelectric detector (V/W), and Ap is the active area of the

pyroelectric detector (cm2), respectively. The pyroelectric detector is used to calibrate the

input power of the infrared radiation from the blackbody IR source onto the

photodetector.

The noise in QWIP devices are due to the fluctuations in the velocityy of the carriers

and in carrier density, which are called Johnson noise and generation-recombination (g-r)

noise, respectively, and gi\ en as follows [50]

2 4kTAf'
Johnson noise : i= k (2.56)
R

g-r noise : i,2 = 4eglAf (2.57)

where Af is noise band\\ idth and R is the resistance of the detector.

Finally, the spectral detectivity (D*) is useful figure of merit in comparing different

detectors. The non-BLIP (background limited performance) detectivity, D*, can be

shown as [49]


D = R, A (2.58)


Under BLIP condition, which requires Ib (background photocurrent detected in

QWIP) > Id, it is well known that the BLIP detectivity (D*Nm.p) is independent of the

photoconductive gain and dark current, and it can be expressed as [51]








=D =* A A (2.59)
S- 2he 2, Q,,

where A is the wavelength; h is the Plank constant: c is the velocity of light; qr is the net

quantum efficiency, and Qb is the incident photon flux density from the background for a

given spectral bandwidth at peak \%a\ length, which can be calculated from


Qb= f-J2 c dA, (2.60)
4 Ae -1

where X, is the cutoff wavelength. k is the Boltzmann constant, and Tb is the background

temperature.













I I


N,


unstrainn)


unstrainn)


Ih
(compression)


Figure 2.1. (a) Unstrained and strained two-lai er schematic and (b) Energy band gap
shifts under strain.


a2



aJ


I- ^ ^


a1:


(strain)


(tension)


I


--












CHAPTER 3
A THREE-STACK INGAAS/ALGAAS/INGAAS BROADBAND
TRIPLE-COUPLED QUANTUM WELL INFRARED PHOTODETECTOR

3.1 Introduction


Multi-color quantum well infrared photodetectors (QWIPs) using different transition

schemes and structures have been widely investigated in recent Near [12]. Theoretical

studies of quantum-confined Stark effect in two- and three-coupled-quantum-w ell

(TCQW) structures have been reported for voltage tunable infrared detection [52,53], and

voltage-tunable multicolor triple-coupled QWIPs (TC-QWIPs) using

InGaAs/GaAs/AlGaAs and high-strain (HS) InGaAs/AlGaAsGaAsnaAs material systems

grown on GaAs for 8-12 am long-wavelength detection have been demonstrated recently

[54,55]. The high-strain InGaAs/AIGaAs/InGaAs TC-QWIP has shown excellent

performance and wavelength tunability in the LWIR spectral range [54]. Most QWIP

devices were developed \with sensitivity in the 3-5 ipm mid-vavelength infrared (MWIR)

or 8-12 upm long-wavelength infrared (LWIR) region. The multi-color QWIPs were also

investigated by using multi-stack quantum %well structure for both the mid-wavelength

and long-wavelength detection or voltage tuning of peak wavelength such as TC-QWIP,

%which usually has narrow bandwidth due to bound-to-bound state transition [54,55]. In

order to broaden the detection bandwidth, the bound-to-miniband (BTM) QWIP [56] and

asymmetrical quantum well structure with graded barrier [57] were proposed. Recently,

several broadband (BB-) QWIPs have been demonstrated using three- or four-well







structure in a unit cell that has different \\ell widths and barrier heights by \arl ing the

barrier layer composition [58].

In this chapter, we report a high performance InGaAs/AIGaAs/InGaAs broadband

triple-coupled quantum well infrared photodetector (BB TC-QWIP) for voltage tunable

multicolor detection in the 7-12 .lm long-wavelength band. This BB TC-QWIP was

formed by using three-stack structure without contact layers bet ween the stacks.



3.2 Device Design and Fabrication


This BB TC-QWIP device was formed by using a three-stack structure without

contact layers between the stacks. The device structure of each stack consists of an

asymmetrical triple coupled quantum \wells (TCQWs) with one deep In( .Ga,,75As

quantum well Si-doped to Nd = 7.0 x 1017 cm3 and two undoped Ino.12Ga, gsAs shallow

quantum wells separated by two 20A thin inner barriers (AlxGal.\As) between the 500A

thick barriers (AlxGal.-As). Each TC-QWIP stack has different quantum well widths.

periods, and barrier heights by varying Al composition so that each stack has different

peak detection wavelengths. The different periods were used to evenly distribute the bias

voltage drop across each stack, which are 3 periods for the bottom stack. 5 periods for the

middle stack, and 6 periods for the top stack, respectively. The quantum well \ idths and

barrier material of the TC-QWIP stacks are given as 55/40/35A and Alo0. IGa %,As for the

bottom stack, 55/40/40A and Alol, Gaij ,As for the middle stack, and 50.'40'40A and

Al0.06Ga1,94As for the top stack, respectively. Two 0.1 upm thick undoped GaAs layers

were grown on both sides of this TC-QWIP to reduce the dark current. Finally, the TC-

QWIP surrounded by the top and bottom ohmic contact layers (Si-doped to n = 2.0 x 1018

cm'3) was grown on the semi-insulating GaAs substrate. Table 3.1 shows the laser








structure of the stacked TC-QWIP. A mesa structure with 216 x 216 Ipm active area was

fabricated to characterize the device performance by using standard x\et chemical

etching. AuGe'Ni/Au was evaporated on the top and bottom of the mesa structure for

ohmic contacts. The 450 polished facet was processed for the back-illumination and IR

radiation coupling. The device was mounted at the edge of the hole on the 16-pin TO-8

socket. The silver-filled epoxy with low thermal resistance was used to bond the device

onto the package.



3.3 Results and Discussion


Figure 3.1 shows the schematic conduction band diagram and the calculated

intersubband transition energy levels of this stacked TC-QWIP at zero bias voltage. Each

TC-QWIP stack can detect specific peak wavelength and then the whole three-stack TC-

QWIP structure can cover the broad wavelength range by overlapping the detection

wavelength of each stack. The El -+ Ec bound-to-continuum (BTC) transition was mainly

detected at lower bias voltages while both El -+ Ec and El -+ E3 bound-to-bound (BTB)

transitions were observed at higher bias \oltages.

Figure 3.2 shows the dark current versus bias voltage (I-V) measured at T = 40, 52,

60, and 77K with a 1800 field of view (FOV) 300 K background photocurrent of the

device. The dark current curves show the asymmetric behavior due to the asymmetric

layer structure and the normal feature of n-type QWIPs. The device is under background

limited perfromance (BLIP) between -5.5V and 4.2V at T = 40K, and between -1.8V and

1.2V at T = 77K.

The photo response \was measured at T = 40 and 77K by using a 1/8 monochromator

and blackbody light source (T = 1273K) at a chopped frequency (200 Hz). Figure 3.3







shows the spectral responsivity of the TC-QWIP at T = 40K, (a) lower bias. and (b)

higher bias voltages. The applied bias voltage Nwas first distributed across the bottom TC-

QWIP because of the highest resistance by the highest barrier. When the applied bias

voltage was further increased the middle stack and the top stack can be biased one after

another. As clearly shown in Fig. 3.3, only El -+ Ec transitions of the bottom and middle

stacks were observed at lower bias voltages (Vb < -3.75V). The El -- Ec transition of the

top stack can also be observed with the increasing bias voltage (Vb > -4V). The peak

responsivity and full-width half-maximum (FWHM) at 1, = 8.7 jlm and Vb = -3.75V

were 0.48 A/W and AX/X, = 21 %, respectively When a higher bias voltage (Vb > -4V)

was applied to the device, the thick AlGaAs barrier was tilted to the thin triangle barrier

so that the EB -- E3 transitions can be detected by the tunneling through E3 bound state.

At the bias voltage of-4.5V, the spectral responsi\ities at ,pi = 8.4 tpm and Xp2 = 10.8 jim

were almost the same (0.82 A/W and 0.81 A/W, respectively). Therefore, a very broad

responsivity curve was achieved with AX/X,p = 21 % and AX/Xp2 = 20 % for this stacked

TC-QWIP. It is noted that this AX/Xp2 = 20 % is much broader than that of the one stack

high-strain TC-QWIP (AX/Ap = 10 %). The El -> E3 transition becomes the dominant

response peak at Vb = -4.75V. In particular, the peak spectral responsivity at Xp = 10.6

jtm and Vb = -5.2V, which was primarily due to the El -- E3 transition of the bottom

stack, was found to be 2.75 A/W. The tunable \wavelength for El -> E3 transition was

10.6-10.8 lim between -5.2V and -4.75V. Figure 3.4 shows the measured and the

calculated 300K background window current I-V curves with 1800 field of view (FOV),

which \\as in good agreement with each other. Figure 3.5 shows the spectral response ily

of the BB TC-QWIP measured at T = 77K and different bias voltages. The peak








responsivity at T = 77K and Vb = -4.25V was 0.62 A/W at .,, = 10.3 pm. The T = 40K

and 77K spectral responsivity curves were different at the same bias voltage. For

example, the El -- E3 transition at Vb = -4.25V and T = 77K was more dominant than

that at T = 40K, that is, the spectral responsivity by El -+ E3 transition was larger than

that of El -- Ec transition at T = 77K, but this was reversed at T = 40K, which h was

attributed to the more dominant thermionic-assisted tunneling (TAT) conduction through

E3 bound state at higher temperature.

We ha\e calculated the detectivity for the BB TC-QWIP at T = 40K and 77K by

using our noise model. The background limited performance (BLIP) detecti\ ity (D*BI iP)

at X = 10.6 pmr was found to be 1.98 x 11l0 cm-Hz/2/W at T = 40K and Vb = -5.2V. The

peak detectivity (D*) under non-BLIP condition was 5.54 x 109 cm-Hz"2/W at T = 77K,

-p = 10.3 p.m, and Vb = -4.25V.



3.4 Conclusions


We have fabricated and characterized a new InGaAs/AlGaAs/InGaAs broadband

triple-coupled quantum well infrared photodetector (BB TC-QWIP) for 8-14 pm long-

wavelength detection. In order to detect the broad wavelength range, the three-stack

structure that has three different peak wavelengths was created as a three-color QWIP.

The El -- Ec transition %\as obtained at lo\\er bias voltages (Vb < -3.75V) and both El --

E3 and El -+ Ec transitions were detected simultaneously at higher bias voltages (Vb > -

4V). The broader wavelength range can be detected by modifying the layer structure. The

positive temperature dependence of the responsivity was found from the measured results

at T = 40 and 77K.











Table 3.1. The layer structure of the BB TC-QWIP.


Layer Thickness (A) Dopant Concentration (cm3)
n GaAs (top contact) 5000 Si 2xl0'8
i GaAs 1000 none none
i Alo.06Gao.94As 500 none none
i Ino.12Gao.88As 40 none none
i Alo.o6Ga) 94As 20 none none
i Ino.12Gao.88As X 6 40 none none
i Alo.06Gao 94As 20 none none
n Ino.25Gao.75As 50 Si 7xl017
i Alo.o6Ga, 94As 500 none none
i Ino.12Ga,, sAs 40 none none
i Alo.osGao.92As 20 none none
i Ino.12Gao.88As X 5 40 none none
i Alo.osGao.92As 20 none none
n Ino.25Gao.75As 55 Si 7x1017
i Alo.osGao.92As 500 none none
i Ino 12Gao.88As 35 none none
i Alo.1 Gao.89As 20 none none
i Ino.12Gao.88As X 3 40 none none
i Alo.1 Gao.89As 20 none none
n In2sGa,, 75As 55 Si 7x1017
i Alo.01 Ga,, As 500 none none
i GaAs 1000 none none
n GaAs (bottom contact) 10000 Si 2x10'8
S.I. GaAs substrate 625 + 25 tam none none











Bottom stack


rtnrIzIzi


It IIII


E-
1-


I--X 3


++


X5


Bottom stack (Ino.25Gao ,As/Alo i1Ga ,89As/ Ino 2Ga,, ,As )
Middle stack ( Ino25GaU.75As/Al,,.,sGa0o92As/ In0 12Gao88As)
Top stack ( In0o25Ga0o75As/A li o,Gao094As/ In. 12Ga,,,As )

< intersubband transition energy at zero bias voltage >
Bottom stack: E, Ec (140 meV ), E, E3 ( 114 meV)
Middle stack : E, Ec (123 meV ), E, E3 ( 104 meV )
Top stack : EI Ec ( 106 meV ), E, E3 ( 98 meV)




Figure 3.1. The schematic conduction band diagram and the intersubband transition
energy of the BB TC-QWIP.


+


X6 %-


111 1 IT1


Middle stack


Top stack


- Ec


- Ec














100


10-8


10-12 1
-6 -4.5


-3 -1.5 0 1.5


3 4.5


BIAS VOLTAGE (V)










Figure 3.2. The dark current versus bias voltage (I-V) measured at T = 40, 52. 60 and
77K with the 180 field of view (FOV) 300K background x\indow current of the BB TC-
QWIP.




























7 8 9 10 11 12


WAVELENGTH (pm)
(a)


7 8 9 10 11 12


WAVELENGTH (prm)
(b)


Figure 3.3. The spectral responsivity of the BB TC-QWIP measured at T = 40K: (a)
lower bias voltage and (b) higher bias voltage.

















10-4


10-5




10-6




10-7


10-8 L
-5.5


-5 -4.5 -4 -3.5 -3 -2.5


BIAS VOLTAGE (V)













Figure 3.4. The 300K background window current with 1800 field of xie\ (FOV) of the
BB TC-QWIP: the solid line (calculated) and the dashed line (measured).


















0.7
T=77 K
-4.25V T=
0.6


S0.5


S0.4 -4V

z
0 0.3
a_ -3.75V
CO -3V
U)
Wl 0.2 -2.75V

-2.5V -3.5V
0.1 .25 -3.25


0
7 8 9 10 11 12 13

WAVELENGTH (pm)


Figure 3.5. The spectral responsivity of the BB TC-QWIP measured at T = 77K.













CHAPTER 4
AN INGAAS/INGAAS/INGAAS TRIPLE-COUPLED QUANTUM
WELL INFRARED PHOTODETECTOR FOR MWIR DETECTION


4.1 Introduction


The performance of the quantum well infrared photodetectors (QWIPs) have been

enhanced by various intersubband transition schemes and material systems with the

maturity of the molecular beam epitaxy (MBE) technology [12]. Different intersubband

transition schemes such as bound-to-bound (BTB), bound-to-quasi-bound (BTQB),

bound-to-miniband (BTM), and bound-to-continuum (BTC) can be chosen to improve the

device performance. For example, the BTC transition can contribute to more sensitive

QWIP by aligning the final state above the barrier and the BTQB transition mechanism

can reduce the dark current because the dark current due to thermionic emission is

exponentially decreased with the increasing barrier height. The as mmetrical QWIP

structures such as the step well, the linear graded barrier, and the coupled quantum well

can affect the device performance and detection x\a\elength due to the voltage tunable

wavelength shift for multicolor detection and the broadband detection on bias polarity.

Theoretical studies of the double- and the triple-coupled quantum well structures have

been reported [52,53]. The strong quantum Stark shift by the applied bias is predicted in

the triple- coupled quantum well (TCQW) structure for long wavelength infrared

detection. Recently, the triple-coupled quantum well infrared photodetectors (TC-QWIPs)








[54,55] using AlGaAs/InGaAs material systems grown on S.I. GaAs substrate have been

demonstrated for 7-14 p.m long-wavelength infrared (LWIR) detection in which a large

voltage tunable wavelength shift and high responsivity were obtained. In addition to

GaAs/AlGaAs, InGaAs/GaAs, and InGaAs/AlGaAs material systems grown on GaAs

substrate, Ino053Ga.~ As/Ino.52Al048As lattice-matched QWIP structures grown on InP

substrate [55,59,60] have been extensively studied, which allows the shorter wavelength

detection due to the large conduction band offset. The short- and mid-wavelength infrared

regions are of important for imaging and communication applications [61-67]. In this

chapter, we present a new mid-wa length infrared triple-coupled quantum well infrared

photodetector (MWIR TC-QWIP) using Ino.s5Gao.47As/In0.52Al, As/ Ino.3Ga0.As material

systems grown on InP substrate for 3-5 pm detection.



4.2 De\ ice Design and Fabrication


The basic device structure of the MWIR TC-QWIP consists of a 40A Ino04Gaf, ,As

quantum well Si-doped to Nd = 2.5 x 1018 cm-3 and t\wo undoped 25 and 20A shallow

Ino,3Gao7As quantum wells separated by two 12A thin Inuu,Alo.4As inner barriers. ,which

were sandwiched by the 300A thick Ino ,A,,,As barriers to form the unit cell. Ten

periods of the unit cell were grown to form the absorber layer. Finally, InGaAs contact

layers ( Si-doped to 2.0 x 1018 cm"3) on both the top and bottom of the 10- period QWIP

structure were grown on the semi-insulating InP substrate. Table 4.1 shows the layer

structure of this MWIR TC-QWIP. Figure 4.1 shows the schematic conduction band

diagram and the intersubband transition schemes (a) at zero and (b) negative bias







voltages. The first (E2) and second (E3) excited states 'were aligned due to the strong

asymmetrical coupling effect of the three InxGa,-xAs quantum wells and two thin

Ino.52A1048As inner barriers. As clearly shown in Figure 4.1(b), the thick Ino.52A1.48As

barrier turns into the thin triangle barrier at higher bias voltages so that the E,-E3

transition can be detected by tunneling mechanism of the photo-exciled electrons through

the E3 excited state.

The test mesa structure with an active area of 216 x 216 upm2 was created to

characterize the device performance through the top contact layer and device active

region down to the bottom contact layer by standard wet chemical etching. The

AuGe/Ni/Au was evaporated for electrical metalization. The test device was annealed at

T = 4500C for two minutes after E-beam evaporation. The 450 facet \\as polished on the

InP substrate for back-illumination IR radiation coupling. The silx er-filled epoxy cement

was used to attach the device to a package, which provides a low thermal resistance

between the device and the package. The device can be mounted on a round TO-type

package with 16 pins, which is one of the earliest IC packages. The ultrasonic wire-

bonding through a combination of pressure and rapid mechanical vibration was used with

gold wires for making electrical connection between the device and the package.



4.3 Results and Discussion


Figure 4.2 shows the dark current xersus bias \oltage (I-V) curves measured at T =

77, 97, and 116K with a 180" field of view (FOV) 300K background photocurrent of

MWIR TC-QWIP. The device is under background limited performance (BLIP) when the







applied bias voltage is between -4.8V and 3.3V at T = 77K while the device is

background limited between -2V and 0.8V at T = 116K.

The photocurrent was measured at T = 77 and 116K by using 1/8 monochromator and

blackbody light source (T = 1273K) at 200 Hz chopped frequency. Figure 4.3 and 4.4

show the spectral responsivity of MWIR TC-QWIP at different temperatures and bias

voltages. Only bound-to-continuum (BTC) transitions were observed at positive and

lower negative bias voltages (Vb<-3.5V at T = 77 and 116K). The bound-to-bound (BTB)

transitions were enabled at higher negative bias voltages (Vb>-4V at T = 77 and 116K).

Moreover, the BTB transition was dominant at biases higher than Vb = -4.5V. The peak

spectral responsivity for this MWIR TC-QWIP was 0.31 A/W at Xp = 4.6 Lim, V, = -5.5V,

and T = 77K. The BTB transitions at a fixed bias voltage and different temperatures for

the MWIR TC-QWIP were much less sensitive to the temperature increase because of the

very high barrier height as shown in Figure 4.3 and 4.4. The background limited

performance (BLIP) detectivity (D*BLI) at p = 4.6 plm was found to be 2.65 x 1010 cm-

Hz' /W at T = 77K and Vb = -4.5V. The peak detectivity (D*) under non-BLIP condition

was 1.44 x 1010 cm-Hz' -,W at T = 116K, p = 4.6 pm. and Vb = -4V.



4.4 Conclusions


We have demonstrated a new In0.45Ga,, cAs/In .,,Al 4 AAs/In0 .Ga, -As mid-\%wavelength

infrared triple-coupled quantum well infrared photodetector (MWIR TC-QWIP) grown

on the InP substrate. This device can be operated up to T = 116K and the spectral

responsivity was almost independent of the temperature. The BTC transition was





41


observed under both the negative and positive bias voltage conditions. %while the BTB

transition was detected at higher negative bias \oltages. The responsiit) due to BTB

transition was found to increase dramatically with increasing bias voltage. It is noted that

this device has better performance under negative bias voltages because both the BTB

and BTC transitions can be detected with higher responsivity and lower dark current.


















Table 4.1. The layer structure of the MWIR TC-QWIP grown on the InP substrate.


Layer Thickness (A) Dopant Concentration (cmn3)
n In 53GaU47As 10000 Si 2x1018
i Ino. 2Ao1048As 300 none none
i Ino.3Ga o.As 20 none none
i In0.52A0lo48As 12 none none
i Ino3GaoAs X 10 25 none none
i Ino,52Alo48As 12 none none
n Ino.4,Gao.55As 40 Si 2.5x 1018
i InO.52Alo, As 300 none none
n In o.3Ga0.47As 10000 Si 2x1018
S.I. InP substrate 625 25 pjm none none









(In,, 4Ga1, sAs/Ino.52Alo.4As/Ino.3Gao7As )


E3
E2




El


Ec





E3
E2


Intersubband transition energy at zero bias voltage: E,-E, (335 meV), E,-E, (270 meV)


Figure 4.1. The schematic conduction band diagram and the bound state energy of the
MWIR TC-QWIP grown on the InP substrate under a) zero bias voltage and (b) negative
bias conditions.


PL











10-2


10-6


10-8


10-10


10-121 1 I
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5


BIAS VOLTAGE (V)








Figure 4.2. The dark current versus bias voltage (I-V) curves measured at T = 77, 97, and
116K along with the 1800 field of view (FOV) 300K background photocurrent of the
MWIR TC-QWIP grown on the InP substrate.













0.4


0.3


0.2


0.1


3 3.5 4 4.5 5 5.5
WAVELENGTH (pm)


0.15



>- 0.1
(I--
z
00.05
LUJ


0


3.5 4 4.5 5 5.5
WAVELENGTH (pim)


Figure 4.3. The spectral responsivity of the MWIR TC-QWIP grown on the InP substrate
measured at T = 77K, under (a) negative bias and (b) positi\ c bias conditions.











0.15



S0.1

U)
z
O 0.05
a-



0
3




0.1


g
F-

> 0.05
z
CL


0
3
3


3.5


3.5


4 4.5 5
WAVELENGTH ([m)
(a)


4 4.5
WAVELENGTH
(b)


5
(rlm)


Figure 4.4. The spectral responsivity of the MWIR TC-QWIP grown on the InP substrate
measured at T = 116K, under (a) negative bias and (b) positive bias conditions.


5.5


5.5












CHAPTER 5
AN ALAS/INGAAS/ALAS/INALAS DOUBLE-BARRIER
QUANTUM WELL INFRARED PHOTODETECTOR
OPERATING AT 205K AND 3.4 Lpm

5.1 Introduction


Quantum well infrared photodetectors (QWIPs) by using a variety of transition

schemes and structures have been extremely investigated for long-wavelength (8-12 plm)

and mid-wavelength (3-5 p.m) regimes in recent years [12,68,69]. In particular, the

photovoltaic (PV) mode infrared detection is attractive for practical application due to the

low dark current, low power dissipation, excellent noise property, and fast integration

time for focal plane array (FPA) applications. However, in most of the QWIPs, the

photoresponse for the PV mode is usually much smaller than that of the photoconductive

(PC) mode. As a result, most of the studies for the PV mode detection have been focused

on improving the performance by using different QWIP structures such as double-barrier

quantum well (DBQW), asymmetric stepped well, and graded barrier [61,62,68-71]. The

mid-wavelength infrared (MWIR) detection using intersubband transition is of great

interest for imaging and target tracking applications [61-63]. The existing GaAs/AlGaAs

QWIP structure grown on GaAs substrate have a limitation due to the small conduction

band offset (AEc) for shorter wavelength detection. Therefore, InGaAs/AlAs quantum

well structures grown on GaAs and InP substrates have been studied for extending the

intersubband transition detection wavelengths to 1.9 pm and 1.55 utm [64-67].







In this paper, we report a new dual-mode (i.e., PV and PC mode) operation n-type

A1As/InGaAs/A1As/InAlAs double-barrier quantum well infrared photodetector (DB-

QWIP) with peak detection wavelength at 3.4 pmr for mid-wavelength infrared (IMWIR)

detection.



5.2 Device Design and Fabrication


The AIAs/InGaAs/A1As/InAlAs DB-QWIP was grown on the semi-insulating InP

substrate by the molecular beam epitaxy (MBE) technique. The basic device structure for

this DB-QWIP is composed of a doped In.53Gao.47As quantum well (44A, Si-doped to 2.0

x 1018 cm3) sandwiched between two ultra-thin undoped AlAs double-barriers (15A) and

separated by a thick (300A) undoped In0.52Al048As barrier, which are then repeated 10

times to form the active absorber layer. Two 0.5 p.m thick highly doped n' Ino 53Gao.47As

contact layers were grown on the top and bottom of the active DB-Q~hrP for ohmic

contacts. Although the thin AlAs double barrier was highly strained (3.7%), the

Inos53Ga047As quantum well and Ino.s5Al0.4As barrier layers were lattice-matched to InP

substrate. As a result, excellent surface morphology and high quality of quantum

well/barrier layers were obtained in this DB-QWIP structure. Table 5.1 shows the layer

structure of this DB-QWIP grown on InP substrate.

In order to characterize the device performance parameters, mesa structures with

active area of 216 x 216 .m2 were fabricated by using standard photolithography and wet

chemical etching through the 10-period detector layer onto the bottom contact layer.

AuGe/Ni/Au was deposited on a 50 x 50 p.m' area on the mesa structure for the top








contact and around the periphery of the mesa structure for the bottom contact. The IR

light coupling was achieved by using a 450 facet backside illumination.



5.3 Results and Discussion


Figure 5.1(a) and (b) show the schematic conduction band diagram and the transition

scheme, and the calculated transmission coefficient versus energy for the DB-QWIP,

respectively. The multi-layer transfer matrix method (TMM) was used to calculate the

transmission coefficient and the bound state energy levels in the Ino53Ga0.47As quantum

well [69]. The effect of electron-electron interaction was considered in this calculation.

The IR detection for this DB-QWIP is based on the intersubband transition from the

localized ground state (E,) to the quasi-bound state (E2) inside the double barrier well.

Using a lattice-matched In0.53Gao047As quantum well (44A) and In0.48A10.52As (300A thick)

barrier layer with InP substrate, an intersubband transition energy of 365 meV between

the E, and E2 bound states is obtained, which corresponds to a peak detection wavelength

at 3.4 pm. The ultra-thin (15A) wide band gap AlAs double-barrier was used to increase

the barrier height so that shorter wavelength detection can be achieved by the photo-

generated carrier conduction through the quasi-bound states (E2) above the In0.52Al0.48As

barrier. By adjusting the well width we can push up the quasi-bound state (E,) and hence

increase the energy spacing between the E, and E2 bound states for shorter wavelength

detection [64,65,67].

Figure 5.2 shows the dark current versus bias voltage (I-V) curves measured at T =

40, 77, 109, 137, 163, and 205K along with the 300K background window current with a







1800 field of view (FOV) for this DB-QWIP. The device is under background limited

performance (BLIP) at T = 77K, Vb < -2.5V and T = 109K, Vb <-2V. Due to the low

device dark current this DB-QWIP was able to operate at a temperature as high as 205K

while maintains good characteristics with slightly lower detectivity than the 77K

operation. Furthermore, the PV mode response was observed for temperatures up to 170

K under BLIP condition. Figure 5.3(a) and (b) show the spectral responsivity curves of

the DB-QWIP measured at different bias voltages and at T = 77 and 205K, respectively.

The peak responsivity for the PC mode at Vb = -3 V was found to be 0.159 A/W at T =

77K. This device showed a large peak responsivity for the PV mode, which was found to

be 19 mA/W and 9 mA/W at T = 77 and 205K, respectively. The peak detection

wavelength (,X) for this device was found to be independent of temperature and applied

bias voltage.

Figure 5.4(a) shows the peak responsivity versus temperature measured at different

biases (-3.5 V > Vb > 0). The results show that for small biases (i.e., Vb < -1.5 V) the

responsivity is nearly independent of temperature up to about 130K and then slowly rises

with increasing temperature (i.e.. positive temperature coefficient). It is interesting to note

that the positive temperature coefficient of the responsivity has been observed only in

QWIPs with short period (with number of periods less than 10) [68.72]. This is due to the

fact that responsivity depends on both the quantum efficiency and the photoconductive

gain (i.e., R = eArlg/hc; where q is the quantum efficiency, and g is the photoconductive

gain.). For the DB-QWIP device studied here. 10 periods of active absorber layer were

used, both the photoconductive gain and quantum efficiency are expected to vary \\ith







temperature due to the short period absorbed layer used in this structure. For typical

QWIPs with 50 periods of quantum wells, the responsivity is generally found to be

independent of temperature. The responsivity at higher bias voltages (Vb > -1.5 V) was

decreased with increasing temperature up to T = 163K. Figure 5.4(b) shows the peak

responsivity versus bias voltage measured at different temperatures for the DB-QWIP.

The responsivity was increased with increasing bias voltage and saturated at high biases.

The increase of responsivity with bias is due primarily to the increase in the

photoconductive gain with applied bias. This trend becomes more prominent at low

temperatures, as clearly shown in Figure 5.4(b). The responsivity for PV mode was

slightly decreased with increasing temperature. Finally, the peak detectivity (D*) and

collection efficiency, rig, were calculated from the measured responsivity and dark

current data. The peak detectivity and the collection efficiency for the PC mode under

BLIP condition were found to be D*BLUP= 7.28 x 1010 cm-Hz'2/W and rig = 4.5 % at =

3.4 tim, T = 77K, and Vb = -2.5 V, respectively. The peak detectivity for the PV mode

under BLIP was D*BLI = 8.9 x 1010 cm-Hzl2/W with a 0.7 % collection efficiency at Xp=

3.4 atm and T = 77K. And the PV mode detectivity under BLIP at T = 163K was D*B, =

3.42 x 1010 cm-Hz"2/W with a 0.5 % collection efficiency at X,= 3.4 pum.



5.4 Conclusions


In conclusion, we have demonstrated a new high performance dual mode operation

AlAs/InGaAs/AlAs/InAlAs DB-QWIP grown on the InP substrate for 3.4 pum MWIR

detection. The PC and PV dual-mode detection was observed in this device with excellent








performance characteristics up to 163K. Due to the very low device dark current.

excellent responsivity and BLIP detectivity were obtained for temperature below 163K,

while the PV mode detection under BLIP was achieved for temperatures below 170K.

The peak responsivities for PC and PV modes at T = 77K were 0.186 A/W at Vb = -3.5V

and 19 mA/W, respectively. The responsivity for PV mode detection was slightly

decreased with increasing temperature. The positive temperature coefficient of the

responsivity \\as observed at low biases and high temperatures. The performance of this

device was greatly enhanced by using the lattice-matched AIAs/InGaAs/AIAs/InAlAs

DB-QWIP structure grow n on InP substrate.













Table 5.1. The laser structure of the DB-QWIP gro\\n on the InP substrate.



Layer Thickness (A) Dopant Concentration (cm'3)
n Ino.53Gao.47As 5000 Si 2 x 10'
i Ino.52A10.48As 300 none none
i AlAs 15 none none
n Ino.53Gao.47As X 10 44 Si 2 x 1018
i AlAs 15 none none
i Ino.52Al1,48As 300 none none
n Ino.53Gao047As 5000 Si 2 x 10'8
S.I. InP substrate 625 + 25 pLm none none












AIAS / In.53Gao.47As / AIAS
El El


Ino.52Ao.48AS


- -


Ino.52Alo.48As


- M


0.2 0.4 0.6 0.8 1


Energy (eV)

(b)


Figure 5.1. (a) The schematic conduction band diagram and transition scheme of the DB-
QWIP and (b) the transmission coefficient versus energy calculated by the TMM.


0


-10


-30


-40


I I _




















10-2


300K BG




Li -6
W 10-8




205 K
1 163 K
10-10 137 K
109 K
77 K
40 K
10-12 L L I ___
-5 -4 -3 -2 -1 0 1 2 3 4 5

BIAS VOLTAGE (V)




Figure 5.2. The dark current versus bias voltage measured at T = 40. 77, 109, 137, 163,
and 205K, along with the 300K background photocurrent for the DB-QWIP.



























3 3.5 4 4.5 5


WAVELENGTH (jim)
(a)


3 35 4 4.5 5


WAVELENGTH (
(b)


I'm)


Figure 5.3. The spectral responsivity curves of the DB-QWIP measured at different
biases: (a) T = 77K and (b) T = 206K.


0.25

0.2

0.15

0.1

0.05


0 i.
2.5


0.14

0.12

0.1

0 08

0 06

0.04

0.02












0.3
0.25 = 3.4 .m
0.25

>. 0.2 -3.5V

> 0.15
z -2.5V



------ OV
0
40 70 100 130 160 190
TEMPERATURE (K)
(a)

0.3
0.25p = 3.4 pim
S 0.25
i 40K 109K
0.2 7K 137K

>0.15
z
o 0.1 205K
n 163K
LU
n, 0.05

0
-3.5 -3 -2.5 -2 -1.5 -1 -0.5 0
BIAS VOLTAGE (v)
(b)

Figure 5.4. (a) The peak responsivity versus temperature for the DB-QWIP measured at
different biases (-3.5V > Vb > 0) and (b) the peak responsivit\ versus bias voltage
measured at different temperatures.












CHAPTER 6
QUANTUM WELL INFRARED PHOTODETECTORS WITH DIGITAL GRADED
SUPERLATTICE BARRIER FOR LONG WAVELENGTH AND BROADBAND
DETECTION

6.1 Introduction


Quantum well infrared photodetectors (QWIPs) have been Nw widely investigated for the

mid-x\axelength infrared (MWIR) and long-wavelength infrared (LWIR) as well as for

multi-color or broadband infrared detection in recent years [12,73,74]. The multi-stack

structure is usually employed to obtain multi-color detection in the MWIR and LWIR

atmospheric spectral bands [73,74]. Voltage tunable QWIPs with asymmetrical triple-

coupled quantum well structures have also been reported for multi-color infrared

detection by using the quantum confined Stark effect [55]. The broadband infrared

detection has been achieved by using a wide variety of device structures with variable

well width and barrier height in the quantum \well [58,75-82]. Levine et al. have reported

a voltage tunable LWIR QWIP using graded barrier quantum wells to achieve large shifts

in the peak detection w\a\ length spectral line-width, and cutoff wavelength [82]. Duboz

et al. has studied the effect of asymmetrical barriers on the performance of GaAs/AlGaAs

QW'IPs [83]. In this chapter we first report two novel high performance

InGaAs/AlGaAs/GaAs QWIPs using digital graded superlattice barrier (DGSLB) to

achieve the staircase-like graded barrier across the barrier region of the QWIP, which in

turn gives the stepwise band gap (or composition) variation in the barrier region. This

band gap engineering approach has been widely used in III-V optoelectronic devices for







band gap variation and for enhancing device performance [84,85]. The new\ structures

enable the broadband detection and significantly improve the device performance under

positive bias operation.



6.2 Device Design and Fabrication


Two noel InGaAs/GaAs/AlGaAs QWIP structures using digital graded superlattice

barriers (DGSLB) were grown on the semi-insulating GaAs substrates by using

molecular beam epitaxy (MBE). The first DGSLB QWIP structure uses the InGaAs

quantum well and GaAs/AIGaAs digital graded superlattice barriers to form the DGSLB

QWIP device. The second DGSLB QWIP structure adds a thin undoped AlGaAs double

barrier on both sides of the InGaAs quantum well for electron wave function

confinement. The DGSLB structure was used to achieve the staircase-like graded barrier

in these devices. The standard MBE growth of the graded layer structure usually requires

pausing the growth to change and stabilize the source temperature for the desired

composition profile. As a result, it requires a longer growth time and may lead to more

oxygen to be incorporated into the graded layer during the growth interruption. The

compositionally digital graded superlattice barriers (DGSLB) of the QWIP structures

were grown using digital graded superlattices, which enable a step\vise composition

graded barrier to be formed without adjustment of the source temperature and the

AIGaAs composition (i.e., using a fixed (15%) Al composition). The DGSLB structure

can be obtained by using short-period superlattice structures with variable %%ellbarrier

thickness to change the Al mole fraction ratio and hence the energy band gap of the

graded barrier. Adjusting the duty cycle can change the well/barrier thickness for each







superlattice unit cell (5 periods. 20A thick). Therefore. using the DGSLB structure

without changing the source temperature setting greatly simplifies the growth procedure

and yields excellent after r quality. The DGSLB layers were formed by using five

superlattice unit cells in series in which the thin GaAs/Alo.15Gao.ssAs layers with a 20A

period were repeated 5 times for each superlattice unit cell. The device structure for the

broadband (BB-) DGSLB QWIP consists of a 50A Ino.2Gao.8As quantum well (Si doped

to 7 x 1017 cm-3) and a 500A GaAs/Alo.1iGao.ssAs DGSLB layer. Each superlattice unit

cell in the DGSLB-layer has a combination of five well/barrier thicknesses (17.6/2.4,

15.2/4.8, 12.8/7.2, 10.4/9.6, and 8/12A) to obtain the target Al mole fractions of x =

0.018, 0.036, 0.054, 0.072, and 0.09 from the substrate side for the staircase-like graded

barrier layer. In the double barrier (DB-) DGSLB QWIP, a thin (20A) undoped

Alo.lsGao.85As double barrier was grown between the DGSLB layers and the 88A

Ino.2Gao.sAs quantum well (Si doped to 7 x 1017 cm-3) to confine the electron wave

functions and to create a resonant state (E4) with the graded superlattice barrier. The

DGSLB layer is composed of five 100A thick superlattice layer each with 5 periods of

superlattices with (well/barrier) thicknesses of 18.4/1.6, 16.8/3.2, 15.2/4.8, 13.6/6.4, and

12/8A for the target Al mole fractions of x = 0.012, 0.024, 0.036, 0.048, and 0.06 in the

AlGa-xAs graded barrier layer. The 5000A contact layers (Si doped to 2 x 1018 cm-3)

were grown at a substrate temperature of 6000C, while the rest of the structure was gro% wn

at 5100C to avoid indium (In) desorption from the InGaAs layers. Table 6.1 and 6.2 show

the layer structures of the DGSLB for the BB- and DB-DGSLB QWIPs, respectively.

Finally, the complete layer structure of the BB- and DB-DGSLB QWIPs was shown in

Table 6.3.







Figure 6.1(a) and (b) show the schematic conduction band diagram and the calculated

transmission coefficient versus energy at zero bias using the multi-layer transfer matrix

method (TMM) [24] for the broadband (BB-) DGSLB QWIP, respectively. The dotted

lines denote the effective barrier height for each superlattice unit cell (5 periods,

20A/period), which can create a staircase-like barrier in the DGSLB layers. In this

calculation, the strain effect due to lattice-mismatch between the InGaAs QW and the

GaAs/AlGaAs barrier and the exchange energy\ due to the electron-electron interaction

were considered. The EI-E4, E1-Es, and E|-E6 transitions contribute to the broadband

detection under positive bias condition, while only transitions from the El to E6 states

were observed under negative bias because the photo-generated carriers need to surmount

the abrupt side of the barrier layers. Under positive bias condition, the effective barrier

will decrease gradually with increasing bias to the lowest superlattice barrier height and

then the bound states aligned by the DGSLB at zero bias will broaden with increasing

bias voltage to form a global miniband across the DGSLB layers and quantum wells.

Thus, the broadband response can be attributed by the bound-to-miniband (B-M) state

transitions under positive bias condition. On the contrary, the effective barrier height for

the photo-excited electron transport will be at its maximum under negative bias condition

and the slope of the DGSLB will be much steeper than under positive biases. Thus, the

capture probability of the photo-excited electrons due to the bound-to-bound (B-B)

transitions will be increased and normal spectral response is expected under negative bias

condition.

Figure 6.2(a) and (b) sho\\ the schematic conduction-band diagram and the calculated

transmission coefficient versus energI at zero bias for the double barrier (DB-) DGSLB







QWIP device. The broadband response was not observed in this device because the wave

function for the peak \wavelength detection is strongly confined by the thin Alo11sGau ssAs

double-barrier and is resonantly coupled to the wave functions of the E4-state in the

DGSLB region. Thus, normal spectral response with identical peak detection wavelength

(at 12 pim) due to the El to E4 state transitions "was obtained under both negative and

positive biases for this device.

In order to characterize the device performance, test mesa structures with active area

of 216x216 itm2 were fabricated by using standard photolithography and wet chemical

etching procedure. The AuGe/Ag/Au film was deposited by E-beam evaporation on the

top and periphery of the mesa structure for ohmic contacts and annealed at 4500C for 2

minutes. The test devices %were polished to 45" facet on the GaAs substrates for back

illumination.


6.3 Results and Discussion


We have performed the dark current-voltage (I-V) and spectral response

measurements on both DGSLB QWIPs under negative and positive bias conditions.

Excellent results were obtained in the photoresponse measurements on these devices.

Figure 6.3 (a)and (b) show the dark current density as a function of applied bias voltage

for the BB- and DB-DGSLB Q'WIPs measured at different temperatures (T = 35, 50, 60,

and 77K), respectively. The 300K background window currents with a field of view

(FOV) of 1800 were also given in Fig. 6.3 (a) and (b).

As expected in the asymmetrical quantum well structure, the dark currents and

photoresponse are also highly asymmetrical under positive and negative bias conditions.







which h is attributed to the different effective barrier profiles under negative and positive

biases, as explained previously. In both devices \%e have observed a much higher dark

current and photo-response under positive bias condition. This is due to the barrier

lowering of the DGSLB and the electron launching under positive bias condition. The

BB-DGSLB QWIP device is under background limited performance (BLIP) between -

IV and +0.75V at T = 35K and the BLIP temperature was 55K while the DB-DGSLB

QWIP is under BLIP between -2V and +0.35V at T =50K.

Figure 6.4 shows a comparison of the dark current density as a function of the electric

field at T = 35K for these two QWIP devices. Although the DB-DGSLB QWIP exhibits a

longer peak wavelength (12 utm peak) than the BB-DGSLB QWIP (11 pm peak) under

negative biases, the dark current density of the DB-DGSLB QWIP is slightly lower than

that of the BB-DGSLB QWIP due to the use of a thin undoped Alo.15Gac osAs double-

barrier around the InGaAs quantum well which tends to reduce the carrier transport

probability under dark condition.

The spectral response was measured at T = 35K for both the BB- and DB-DGSLB

QWIPs by using an 1/8 m grating monochromater, a calibrated blackbody IR source (T =

1273K), and a closed cycle liquid helium cryostat at 200 Hz chopped frequency. The

pyroelectric detector is used to calibrate the input poN\er of the infrared radiation from the

blackbody IR source onto the photodetector.

Figure 6.5(a) and (b) show the spectral responsivity of the BB-DGSLB QWIP at T=

35K under (a) negative and (b) positive bias conditions. The peak wavelength was blue-

shifted from 11 upm to 10.8 [pm between -0.75V and -1.25V under negative bias

condition. The absolute responsivity increases with the applied bias due to the increase in







photoconductive gain with increasing bias. The peak responsivities at ,p = 10.8 ram and

9.8 tpm were found to be 0.57 A/A and 1.07 A'W at Vb = -1.25V and +0.75V,

respectively. It is noted that a very broad spectral bandwidth was obtained under positive

bias condition in this device. The full-Ividth half-maximum (FWHM) spectral bandwidth

of this device at Vb = -1.25V was found to be AX/l = 130%' while FWHM' spectral

bandwidths at Vb = +0.75V and +0.5V were found to be AA/Ap = 62%' and 54%,,

respectively. This broadband detection feature was attributed to the formation of

miniband states by the overlapping of E4, Es, and E6 wave functions, which enables the

broadband detection from the El to the E4 Es, and Eo states transitions under positive bias

condition.

Figure 6.6(a) and (b) show the spectral responsivity of the DB-DGSLB QWIP at T

35K under (a) negative and (b) positive bias conditions. The maximum peak responsivity

at kp = 11.8 am was found to be 0.28 A/W at Vb = -1.5V and T = 35K. However, the

spectral responsivity was dramatically increased under positive bias condition due to the

graded barrier lowering and electron launching effect. The peak responsivity at Vb = +1V

was 3 A/W at ?p = 12 lm and T = 35K. The FWHM spectral bandwidth at Vb = -1.5V

and +1V were AA/kp = 11% and 17%, respectively. A slightly broader spectral bandwidth

detection was obtained under positive bias condition. The peak detection wavelength for

this device was attributed to the El to E4 state transitions.

The detectivity of both QWIPs was calculated from the results of the responsivity and

dark current measurements. Under background limited performance (BLIP) condition, it

is well known that the BLIP detectivity is independent of the photoconductive gain and

dark current. In the first DGSLB QWIP, the BLIP detectivity (D* .BP) at Vb = -0.75V and







0.75V were found to be 7.2 x 109 cm Hz12/W at hp = 11 Im and 1.3 x 1010cm HzI12/W at

Xp = 9.8 um, respectively. The BLIP detectivity (D*BLIP) at Vb = -1.5V and +1V for the

DB-DGSLB QWIP were found to be 5.8 x 109 cm HzI2/W at Xp = 11.8 pmr and 1.9 x 1010

cm Hz'//W at Xp = 12 [tm, respectively.



6.4 Conclusions


In this chapter, we have demonstrated a novel broadband (BB-)

InGaAs/AlGaAs/GaAs quantum well infrared photodetectors (QWIPs) using digital

graded superlattice barrier (BB-DGSLB QWIP) for broadband detection and a high

sensitivity double-barrier (DB-) InGaAs/AlGaAs/GaAs/AlGaAs DGSLB-QWIP for long-

wavelength infrared (LWIR) detection. For the BB-DGSLB QWIP, the peak responsivity

at p = 9.8 mrn was found to be 1.07 A/W at Vb = +0.75V and T = 35K, %with a

corresponding BLIP detectivity (D*BLIP) of 1.3 x 10l0 cm Hzl/2/W. A very broad spectral

response bandwidth (7 -16 pim) was obtained under positive bias condition in this device.

The full-width half-maximum (FWHM) spectral bandwidth at Vb = +0.75V and +0.5V

were found to be AAX/ = 62 % and 54%, respectively, for this device. As for the DB-

DGSLB QWIP device, a very large spectral responsivity and BLIP detectivity (D*BLIP)

(i.e., Ri = 3 A/W and D*BI IP = 1.9 x 10Io cm HZ/2/W) were obtained at Vb = +1V and Xp

= 12 jpm. The dark current density of the DB-DGSLB QWIP was found to be slightly

lower than that of the BB-DGSLB QWIP due to the use of a thin undoped Alo.0sGa, 85As

double-barrier around the InGaAs quantum well in this device. Excellent device

performance was obtained in both the BB- and DB-DGSLB QWIPs.









Table 6.1. The layer structure of the digital graded superlattice barrier (1) for the BB-
DGSLB QWIP.

Layer Thickness Average Dopant Concentration
(A) (x) (cm-3)
i GaAs X 5 8 0.09 none none
i Alo.i0Ga,, NsAs 12 none none
i GaAs X 5 10.4 0.072 none none
i Alo.15Ga.s85As 9.6 none none
i GaAs X 5 12.8 0.054 none none
i Alo.s1Gau, 8As 7.2 none none
i GaAs X 5 15.2 0.036 none none
i Alo.lsGao.ssAs 4.8 none none
i GaAs X 5 17.6 0.018 none none
i Alo.15Gao.85As 2.4 none none


Table 6.2. The layer structure
DGSLB QWIP.


of the digital graded superlattice barrier (2) for the DB-


Layer Thickness Average Dopant Concentration
(A) (x) (cm')
i GaAs X 5 12 0.06 none none
i Alo.s1Gao.ssAs 8 none none
i GaAs X 5 13.6 0.048 none none
i Alo.s1Gan 8iAs 6.4 none none
i GaAs X 5 15.2 0.036 none none
i Alo.lsGao.ssAs 4.8 none none
i GaAs X 5 16.8 0.024 none none
i Alo.isGau 8sAs 3.2 none none
i GaAs X 5 18.4 0.012 none none
i Alo.15Ga, siAs 1.6 none none















Table 6.3. The complete layer structure of the BB- and DB-DGSLB QWIP.


Layer Thickness Dopant Concentration
(A) _(cm3)
n GaAs (top contact) 5000 Si 2x1018
Digital Graded Barrier (1) X 5 500 none none
n In0.2Gao.sAs 50 Si 7x1017
Digital Graded Barrier (1) 500 none none
n GaAs (middle contact) 5000 Si 2x1018
Digital Graded Barrier (2) 500 none none
i Alo.15Gao.85As X 5 20 none none
n Ino.2Ga,, sAs 88 Si 7x1017
i Alo.isGao.s8As 20 none none
Digital Graded Barrier (2) 500 none none
n GaAs (bottom contact) 5000 Si 2x 018
S.I. GaAs substrate 625 + 25 pm none none



























In02G a, As


E3 E4 E5 E


0.05 0.1 0.15
Energy (eV)


(b)




Figure 6.1. (a) The schematic conduction band diagram and (b) the calculated
transmission coefficient versus at zero bias for the broadband (BB-) DGSLB QWIP.





69













E-3


GaAs /Alo. 1Gao 85As GaAs /Alo 15Ga ,,8As
EF ..... .......
EE
E,


In0.2Ga0oAs


(a)




E3 E4 E5





E
-10 -

-20



-30 I II
0 0.05 0.1 0.15 0.2
Energy (eV)



(b)




Figure 6.2. (a) The schematic conduction band diagram and (b) the calculated
transmission coefficient energy at zero bias for the double barrier (DB-) DGSLB QWIP.











c1-)
C\i





z
w



03
0
Ix
Ix


-2 -1.5 -1 -0.5 0 0.5 1
BIAS VOLTAGE (V)
(a)


100

10-2
10 -
-4
10-6


10-


10-
-2 -1.5
-2 -1.5


-1 -0.5 0 0.5 1
BIAS VOLTAGE (V)
(b)


Figure 6.3. The dark current density versus bias voltage for (a) the BB-DGSLB QWIP
and (b) the DB-DGSLB QWIP. The dashed line is the 300K background photocurrent.


100

10-2

10-

10-6

10-


10-1(


CMJ
E

>.-

z
w


111
a::

0
a:
C





















T= 35K


-60 -40 -20


0 20 40


ELECTRIC FIELD (kV/cm)












Figure 6.4. A comparison of the dark current density versus the electric field for the BB-
DGSLB QWIP (solid line) and the DB-DGSLB QWIP (dashed line).


100


10-2


10-6

10-8
io8


\ BB-DGSLB QWIP




DB-DGSLB QWIP \


E




O
>.-

CI.




Q:
0
0


-80


''''''''''''''''''' ''''
























9 10 11 12
WAVELENGTH (pnm)


1.I
T=35K

S1.2 +0.75V(9.8 m)

I- 0.9


03
0.6

S0.3 +0.5V (10 lm)


0
7 8 9 10 11 12 13 14 15 16 17 18
WAVELENGTH (gm)
(b)



Figure 6.5. The spectral responsivity of the BB-DGSLB QWIP de\ ice at T= 35K: (a) at
negative bias and (b) at positive bias condition.













0.4
-

0.3 -
>-

0.2 -

0-
O
0) 0.1
IJ.

0
10


11 12 13
WAVELENGTH (utm)


11 12 13


WAVELENGTH (pm)
(b)






Figure 6.6. The spectral responsivity of the DB-DGSLB QWIP at T = 35K: (a) at
negative and (b) at positive bias condition.













CHAPTER 7
HIGH SENSITIVITY
QUANTUM WELL INFRARED PHOTODETECTORS
WITH LINEAR GRADED BARRIER

7.1 Introduction


In recent years, quantum well infrared photodetectors (QWIPs) have been widely

developed for 3-5 mrn mid-wavelength infrared (MWIR) and 8-14 jtm long-\wavelength

infrared (LWIR) spectral regimes [12,86-96]. The matured III-V compound

semiconductor growth technology and the flexibility of the device structure have induced

the rapid development of various QWIP structures. The multi-color IR detectors for

practical imaging applications such as target discrimination, remote sensing system,

medical imaging. and tracking system have been extensively investigated by using the

multi-stack QWIP and voltage-tunable asymmetrical coupled quantum well structures. In

particular, the broadband QWIPs covering the entire LWIR region were demonstrated by

repeating the quantum wells with different well width and depth [58,78]. The asymmetric

structure was introduced to detect the broadband region [83,84]. This broadband

detection is required in the infrared (IR) spectrometers, which measure the interaction of

the IR radiation with experimental samples. To further optimize the QWIP performance,

considerable efforts have been devoted to the theoretical investigation of intersubband

optical absorption in various quantum well structures [97-100]. Theoretical study of the

intersubband and free-carrier absorption coefficients in the heavily doped QW structure

has also been reported [97]. The effective mass approximation which takes into account







the effects of strain and subband nonparabolicity has been used to calculate the electronic

states for the intersubband optical absorption coefficient in the strained double- barrier

quantum well structure [98]. The nonlinear spectral responsii itN \ersus applied bias has

been studied, in which the physical effects of QWIPs under localized excitation [99] and

carrier depletion or accumulation on the local electric field have been theoretically shown

in the QWIP devices [100]. In addition, the new light coupling schemes in the corrugated

(C-) QWIP [101] and the quantum-grid infrared photodetector (QGIP) [102] have been

developed, which showed efficient coupling of normal incidence light into these QWIP

devices.

In this chapter, we report two high performance QWIPs with linear graded barrier

which are a InGaAs/AlGaAs broadband (BB-) linear graded barrier (LGB) quantum well

infrared photodetector (QWIP) and a high-sensitivity AlGaAs/InGaAs/AlGaAs double-

barrier (DB-) LGB QWIP for broadband wavelength regime and long-waxelength

detection.



7.2 Device Design and Fabrication


These two linear grade barrier (LGB) QWIPs were grown on semi-insulating GaAs

substrate by using molecular beam epitax) (MBE). The top-stack for BB-LGB QWIP

consists of a 45A In0.26Ga0.74As quantum well (Si-doped to 7 x 1017 cm-3) and a 500A

linear graded barrier (AlxGal-xAs) as a unit cell in \which the Al mole fraction of linear

graded barrier was increased from 1.8% to 9% starting from the substrate side. This unit

cell was repeated 15 times for the whole BB-LGB QWIP structure. In the DB-LGB

QWIP, a thin 20A layer (Alo0.5Ga,,sAs) was grown between the 72A In0.26Ga,7 7As







quantum well (Si-doped to 7 x 1017 cm-3) quantum well and the LGB layer to obtain the

double-barrier structure which can contribute to confine the wave function. This structure

can induce the strong intersubband absorption from the ground state (El) to the excited

state (Es) which was resonantly created with the A1\Ga-.,As (x=1.8%~-9%) 500A linear

graded barrier (LGB). The quantum well and double-barrier with the linear graded barrier

were also repeated 15 times for the bottom-stack. Finally, the ohmic contact layers (Si-

doped to 2 x 1018 cm3) with different thicknesses (3000, 7000, and 5000A) were grown

on the top of the BB-LGB QWIP stack and on the bottom of the DB-LGB QWIP, and

between the two stacks, respectively. Table 7.1 shows the complete layer structure of the

BB- and DB-LGB QWIPs.

Figure 7.1 shows (a) the schematic conduction band diagram and the intersubband

transition scheme and (b) the calculated transmission coefficient by using the multi-layer

matrix method (TMM) for the BB-LGB QWIP. The photo-generated carrier transport

mechanism for the BB- and DB-LGB QWIP is definitely dependent on the applied bias

polarity as described in chapter 6. Under positive bias condition, the slope of the linear

graded barrier \\as gradually decreased with increasing bias so that the excited electron

for longer-wavelength detection can easily overcome the barrier. Therefore, the

broadband (BB) detection can be obtained by overlapping the E1-E3, E|-E4 EI-Es, and E1-

E6 transitions under positive bias condition. On the contrary, the linear graded barrier will

be much steeper and the barrier height will be at its maximum under negative bias

condition. Thus, the capture probability of the photo-generated electron carriers into the

quantum well was enhanced by the higher barrier height and then the longer-wavelength







detection due to the E1-E3 and Et-E4 transitions was not expected and normal spectral

response with two-peak wavelength was achieved under negative bias condition.

Figure 7.2 shows (a) the schematic conduction band diagram and the intersubband

transition scheme and (b) the calculated transmission coefficient at zero bias for the DB-

LGB QWIP. The high spectral response can be obtained by the thin Alo0.1Gao.gsAs barrier

which confine the %wa'e functions and then induce the strong intersubband absorption.

Normal spectral response was expected under both positive and negative bias condition.



7.3 Results and Discussion


The test mesa structures with an active area of 216 x 216 pm2 was processed to

characterize the device performance by using standard photo-lithography and wet

chemical etching procedure. The AuGe/Ag/Au (300A/1000A/1500A) was deposited on

the top and the periphery of the mesa structure for ohmic metalization. The device was

annealed at 4500C for 2 minutes. The GaAs substrate was polished for 450 facet back-

illumination.

Figure 7.3 (a) and (b) shows the dark I-V curves measured at T = 35, 50, 60, and 77K

for the BB- and DB-LGB QWIPs, respectively The 300K background window current

with 1800 FOV (field of view) was also given as the dashed lines in Fig. 7.3 (a) and (b).

The BB-DGB QWIP is under background limited performance (BLIP) between -4V and

+0.9V at T= 50K and TBLIP = 58K while the DB-LGB QWIP is under BLIP between -

1.6V and +0.7V at T = 60K and Ti| i, = 65K. As shown in Fig. 7.3 (a) and (b), the dark

current of the BB-LGB QWIP is several orders of magnitude higher than the DB-LGB

QWIP, which is due to the thin 20A double-barrier in the DB-LGB QWIP. In addition.







the asymmetrical dark currents can be attributed to the different carrier transport

properties depending on the bias polarity.

The spectral response measurement was performed with 450 facet back-illumination

by using an 1/8 m grating monochromater, a calibrated blackbody IR source (T =

1273K), and a closed cycle liquid helium cryostat at 200 Hz chopper frequency. The

operating temperatures of the devices for the spectral response measurement were T = 35,

60, and 77K for the BB- and DB-LGB QWIP.

Figure 7.4 shows the spectral responsivity of the BB-LGB QWIP measured at T =

35K under (a) negative and (b) positive bias condition. The maximum peak responsivity

at T = 35K and -3V was 1.61 A/W at a peak \wavelength with 9.9 pmr which is due to the

E1-Es transition. The Ei-E6 transition was also obtained in this measurement, which has

much lower responsivity than the EI-Es transition. The spectral responsivity was

increased with increasing bias voltage. The inset in Fig. 7.4 (a) shows the linear

dependence of the peak wavelength for the EI-E5 and Ei-E6 transitions under applied

negative bias voltage. The peak wavelength was blue-shifted with increasing bias voltage

ranging from 10.2 to 9.9 ulm and 8.5 to 7.5 upm for the EI-Ei and EI-E(, transitions,

respectively. The peak responsivity at +2.75V and T = 35K was found to be 1.75 A/W at

Xp = 11.9 prm. The three-peak wavelength (8.9, 10.2, and 11.9 Pm) was obtained at

+2.7V, which were due to the EI-E, Ei-Es. and EI-E4 transitions. The full-width half-

maximum (FWHM) spectral bandwidth of this device at +2.75V and T = 35K was Al/Xp

= 52%. This broadband detection covering from 6.5 to 16 p.m under positive bias

condition can be achieved by the overlapping of the EI-E6. EI-Es, E|-E4. and EI-E3

transitions. The Ei-E6 and EI-E5 transitions were first enabled at lower positive bias







voltage (+1V). The E|-E4 and E1-E3 transitions can be gradually detected with increasing

bias voltage. However, the contribution of the E1-E3 transition for the broadband

detection is much smaller than the other transitions at higher positive bias voltage. Figure

7.5 shows the spectral responsivit) measured at T = 60K under (a) negative and (b)

positive bias condition. The peak responsivity with 1.41 A/W at 10 mrn was obtained at -

3V and T = 60K. The wavelength tunability was ranging from 10.5 to 10 tm and 8.6 to

7.5 pim for the EI-E5 and E1-E6 transitions, respectively. The peak responsivity at +1.45V

was found to be 0.73 A/W at T = 60K and 1, = 11.1 p.m. The FWHM spectral bandwidth

at -3V was found to be AX/X, = 12 % while FWHM spectral bandwidths at +1V and

+1.45V were found to be AX/Xd = 42 % and 45 %, respectively. As shown in the results at

T =35K, the broadband wavelength detection was achieved at T = 60K under positive

bias voltage. Figure 7.6 shows the spectral responsivity measured at T = 77K under

negative bias condition. The blue-shift of the peak wavelength was also observed at T =

77K. The responsivity can not be measured at positive bias voltage because

Transimpedance Amplifier (TIA) was saturated due to the high dark current at T = 77K

under positive bias condition.

Figure 7.7 shows the spectral responsivity measured at T = 35K for the DB-LGB

QWIP under (a) negative and (b) positive bias condition. The peak responsivity due to

the EI-ES transition at T = 35K and -4.5V was found to be 1.23 A/W with a peak

wavelength at Xp = 9.2 ptm. The responsivity with 4.38 A/W at X, = 9.1 pm was obtained

at +3.5V and T =35K. The FWHM spectral bandwidths under positive bias voltage \ere

two times broader than under negative bias voltage because the Ei-E4 transition for the

longer-wav\elength detection can be obtained under positive bias condition. The FWHM







spectral bandwidth at +3.5V was AX/X = 27%, while FWHM spectral bandwidth at -4.5V

was Al/X, = 11%. The Ei-E4 transition was dominant at lower positive bias voltage

(<+2.5V) while the EI-Es transition was more dominant at higher positive bias voltage

(>+3V). The responsivity was dramatically increased with increasing bias voltage under

positive bias condition. Figure 7.8 shows the spectral responsivity measured at T = 60K

for the DB-LGB QWIP under (a) negative and (b) positive bias condition. The peak

responsivities at -4.5V and +2.5V were found to be 1.23 A/W at Xp = 9.2 p.m and

2.49A/W at X, = 9.8 pm, respectively. Finally, Figure 7.9 shows the spectral responsivity

measured at T = 77K for the DB-LGB QWIP under negative bias condition. The peak

responsivity at -4V and -3.5V were found to be 0.72 A/W and 0.43 A/W at Xp = 9.2 gm,

respectively.

The detectivity calculation for the BB- and DB-LGB QWIPs was carried out from the

measured dark current and the spectral responsivity under background limited

performance (BLIP) and non-BLIP. The BLIP detecti\it) for the BB-LGB QWIP was

found to be 1.3 x 1010 cm-HzV/W at +2.75V, 11.9 jim, and 35K while the non-BLIP

detectivity was 6.9 x 109 cm-Hz'/W at -2.45V, 10.1 jim, and 77K. For the DB-LGB

QWIP, the BLIP detectivity was 2.5 x 1010 cm-Hz ~/W at +3.5V, 9.1 upm, and 35K while

the non-BLIP detectivity was 6.4 x 109 cm-HzV/W at -4V, 9.2 ptm, and 77K.



7.4 Conclusions


In this chapter, we have developed an InGaAs/AIGaAs broadband (BB-) linear graded

barrier (LGB) quantum \vell infrared photodetector (QWIP) and a high-sensitivity







AIGaAs/InGaAs/AlGaAs double-barrier (DB-) LGB QWIP for broadband %waelength

regime and long-wvavelength detection. In the BB-LGB QWIP, the broadband detection

was achieved from 6.5 to 16 pim at +2.75V and T = 35K. The wavelength can be tunable

with blue-shift under negative bias condition. The maximum responsivity at 11.9 pim,

+2.75V, and 35K was 1.75 A/W with BLIP detectivity of 1.3 x 1010 cm-Hz'2/W and

FWHM bandwidth of AX/Xp = 52%. For the DB-LGB QWIP, the peak responsivity at Xp

= 9.1 lam \\as 4.38 A/W at +3.5V and T = 35K. The corresponding BLIP detectivity and

FWHM bandwidth were 2.5 x 1010 cm-Hz/V/W and AAX1, = 27%, respectively. The 20A

A1GaAs double-barrier contributed to the lo\ er dark current and the high sensitivity.














Table 7.1. The complete layer structure of the BB- and DB-LGB QWIP.


Layer Thickness Dopant Concentration
(A) _(cm3)
n GaAs (top contact) 3000 Si 2x10'1
i AlxGal.xAs (x=0.018~0.09) X 15 500 none none
n Ino.26Gao.74As 45 none 7x1017
i AlxGal.-As (x=0.018-0.09) 500 none none
n GaAs (middle contact) 7000 Si 2x10'8
i AlxGal.,As (x=0.018-0.09) 500 none none
i Alo.15Gao.85As X 15 20 none none
n In, ,Ga, 74As 75 Si 7x1017
i Alos15Ga,, yAs 20 none none
i Al,Ga.l,As (x=0.018-0.09) 500 none none
n GaAs (bottom contact) 5000 Si 2x1018
S.I. GaAs substrate 625 25 none none
pmin















V;iS. ,Ao.S .. ... *, i E6
-- E1


AlxGal-xAs
(=0.018-0.09)


In0.26Ga0o74As


E2 EE E E
2 3 4 5 6


0.05 0.1 0.15 0.2
Energy (eV)


0.25


(b)


Figure 7.1. (a) The schematic conduction band diagram and (b) the calculated
transmission coefficient versus energy) at zero bias for the BB-LGB QWIP.











Al0.5Gao X.As


AIlGal-,As


In026Gao.74As


0.05 0.1 0.15 0.2 0.25
Energy (eV)
(b)


Figure 7.2. (a) The schematic conduction band diagram and (b) the calculated
transmission coefficient versus energy at zero bias for the DB-LGB QWIP.






































-4 -3 -2 -1 0 1
BIAS VOLTAGE (V)
(a)


2 3 4


-5 -4 -3 -2


-1 0 1 2
BIAS VOLTAGE (V)
(b)


Figure 7.3. The dark current versus bias voltage for (a) the BB-LGB QWIP and (b) the

DB-LGB QWIP. The dashed lines is the 300K background window current.


10-4

J 1 0-6
0 io6

C1 10-8


0 10-10

10-12

10-14


1 0-4


10-6

10-8

10-10

10-12

10-14


3 4 5


























06.
6.5


2


S1.5
>-

> 1
O
z
0
a-
|) 0.5
LU

0 -
6.5


8 9.5 11 12.5
WAVELENGTH (pm)


8 9.5


11 12.5 14 15.5 17


WAVELENGTH (pm)

(b)



Figure 7.4. The spectral responsivity of the BB-LGB QWIP device at T = 35K: (a) at
negative and (b) positive bias condition.

























8 9.5 11 12.5
WAVELENGTH (pm)


8 9.5 11 12.5 14
WAVELENGTH (pm)


15.5 17


(b)





Figure 7.5. The spectral responsi\ity of the BB-LGB QWIP device at T = 60K: (a) at
negative and (b) positive bias condition.


Om
6.5


.5
6.5













1.2


0.9


0.6


0.3


6.5


8 9.5 11 125 1


V\ ,uar,,TH(pnm)








Figure 7.6. The spectral responsivity of the BB-LGB QWIP device at T = 77K under
negative bias condition.

























7.5 8.5 9.5 10.5 11.5
V\A\ELEGTH (p)
(a)


7.5 8.5 9.5


10.5 11.5 12.5


V\AVBENGTH (pm)
(b)



Figure 7.7. The spectral responsivity of the DB-LGB QWIP device at T = 35K: (a) at
negative and (b) positive bias condition.


0 '
6.5


0 '
6.5















1.5
T=60K
S1.2 -4.5V 7=9 2 pm

0.9 -4V





C(a)

-3.5.5V
Z 0.6










T60
3 ~ +2.75V
^2.5 /+2.5
fc 2 // +2V
C. -3V









wj 0.3
0 1

















6.5 7.5 8.5 9.5 10.5 11.5 12.5
WAVELENGTH (pm)
(b)









Figure 7.8. The spectral responsivity of the DB-LGB QWIP device at T = 60K: (a) at
negative and (b) positive bias condition.
negative and (b) positive bias condition.





















0.8

0.6

0.4

0.2


6.5 7.5 8.5 9.5 10.5


11.5


WuaTH (pm)









Figure 7.9. The spectral responsivity of the DB-LGB QWIP device at T = 77K under
negative bias condition.












CHAPTER 8
THREE-COLOR THREE-STACK QUANTUM WELL INFRARED
PHOTODETECTOR 4 x 4 FOCAL PLANE ARRAYS

8.1 Introduction

Recently, there have been great interests in developing the quantum well infrared

photodetectors (QWIPs) for the multi-color detection and the large focal plane array

(FPA). For multi-color detection in QWIP structure, there are mainly two approaches in

which one is to use the multi-stack structure [10,11,74] and the other is the asymmetrical

QWIP structure. The multi-color detection is highly desired in some practical

applications such as the discrimination and identification of the target. The mature of the

QWIP growth technique such as molecular beam epitaxy (MBE) make it possible. J. C.

Chiang et al. [54] has successfully demonstrated two-stack multi-color QWIP for MWIR

and LWIR dual-band detection, and voltage tunable multi-color triple-coupled (TC-)

QWIP for LWIR detection. Focal plane array (FPA) in IR imaging system has many

applications such as night vision, early warning system, monitoring the temperature

profiles, target detection, and discrimination. While single-color FPA is useful for distinct

targets, multicolor FPA with two or more spectral bands can improve the overall

performance when the targets are not clear [103]. Therefore, the advanced FPA requires a

simultaneous multicolor response in SWIR, MWIR, LWIR, or VLWIR bandwidths.

Recently, the large 640 x 486 GaAs/AlGaAs FPA with 9 pm cutoff wavelength and two-

color large 640 x 486 GaAs/AlGaAs FPA for LWIR and VLWIR regions have been







developed [17,104]. In addition, the 256 x 256 two-color MW/LW QWIP FPA with pixel

registration and simultaneous integration has been demonstrated [105]. Finally, the large

format and high uniformity QWIP focal plane array (FPA) cameras have been

successfully developed for IR imaging applications [106-107].

In this chapter, we have designed and demonstrated two three-color, three-stack

QWIP 4 x 4 focal plane arrays (FPAs) grown on semi-insulating (SI) GaAs substrate.



8.2 Device Design and Fabrication


The two (LW/LW/MW and LW/MW/SW) three-color, three-stack QWIP 4 x 4 FPAs

were grown on semi-insulating GaAs substrate. For the first QWIP 4 x 4 FPA of

LW/LW/MW three-color detection, the top-stack for LWIR detection (Xp 12 g.m)

consists of 62A In0.23Gao.77As quantum well (Si-doped to 7 x 1017 cm-3) separated by

500A thick undoped GaAs barrier as a period, which was repeated 5 times. The middle-

stack for LWIR detection (Xp 8.8 gpm) was formed by a 62A Ino.23Gao.77As quantum

well (Si-doped to 7 x 1017 cm-3) and a 300A Alo0.1Gao s9As barrier as a unit cell in which

a very thin GaAs (5A) was inserted between quantum well and barrier. This unit cell was

repeated 5 times. The fiN e-period MWIR QWIP structure for bottom-stack consists of a

25A Ino.35Gao.65As quantum well (Si-doped to 2 x 1018 cm-3) and a 300A Alo 38Gau 62As

quantum well as a unit cell in which a very thin GaAs (5A) was also grown between

quantum well and barrier. Each stack was surrounded by the undoped 500A GaAs spacer

layers, which can contribute to the lower dark current. The ohmic contact layers (4500A,

Si-doped to 2 x 1018 cm3) were grown on top of top-stack, bottom of bottom-stack, and

between top- and middle-stacks. Finally, the p-type Alo 38Gau 62As buffer layer (3000A,







Be-doped to 1 x 1016 cm-3) sandwiched by two ohmic contact layers was grown between

middle- and bottom-stacks. Table 8.1 shows the complete layer structure for this

LW/LW/MW three-stack, three-color QWIP 4 x 4 FPA. The second three-color, three-

stack QWIP 4 x 4 FPA can detect the three different bands of LW, MW, and SW

wavelength regimes. The top-stack for LWIR detection (Xp 7.9 4m) consists of 62A

InxGal-xAs (x=0.230.01) quantum well (Si-doped to 7 x 1017 cm3) separated by 300A

undoped AlxGa-_xAs (x=0.110.01) barrier as a unit cell in which a very thin layer was

inserted between quantum well and barrier and the unit cell was repeated 5 times. The

middle-stack for MWIR detection (Xp = 3.7 pim) was formed by a 25A InxGal.xAs

(x=0.350.01) quantum well (Si-doped to 2.5 x 1018 cm3) and a 300A undoped AlxGal.

xAs (x=0.380.01) barrier as a unit cell in which a very thin GaAs (5A) was also inserted

between quantum well and barrier as described in the top-stack. This unit cell was also

repeated 5 times. The unit cell of the bottom-stack with 10-period for SWIR detection

consists of 500A AlxGal.xAs (x=0.800.01) indirect-barrier (IB) and 24A InxGal-xAs

(x=0.370.01) quantum well (Si-doped to 2 x 1018 cm"3) with 5A GaAs thin layer

between quantum well and barrier. Each stack except for bottom-stack was surrounded

by the undoped 500A GaAs spacer layers. The ohmic contact layers (4500A) were grown

on top of top-stack (Si-doped to 2 x 1018 cm3), bottom of bottom-stack (Si-doped to 2 x

1018 cm3), and between top- and middle-stacks (Si-doped to 3 x 1018 cm3). Finally, the

AlxGal.xAs (x=0.380.01) buffer layer (3000A) sandwiched by two ohmic contact layers

(Si-doped to 3 x 1018 cm-3) was grown between middle- and bottom-stacks. Table 8.2

shows the complete layer structure for this LW/MW/SW three-stack, three-color QWIP 4

x 4 FPA. Figure 8.1 shows the schematic conduction band diagrams of the first




Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EXBCB3DG7_2Q2ES9 INGEST_TIME 2013-02-14T17:20:00Z PACKAGE AA00013532_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES