|Table of Contents|
Table of Contents
Chapter 1. Introduction to spectral imaging
Chapter 2. Introduction to resonance ionization image detectors
Chapter 3. CS photoionization and lasers
Chapter 4. CS RIID characterization
Chapter 5. Time-resolved experiments in RIID
Chapter 6. Feasibility of sealed-cell CS RIID construction
Chapter 7. Buffer gas resonance ionization imaging detector with avalanche amplification of signal
Chapter 8. Conclusions and future work
A CS ATOMIC VAPOR
RESONANCE IONIZATION IMAGING DETECTOR
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
Dedicated to Diyora, and all the Temirovs.
Now, when I am on the eve of a very important point of my career, looking back at
the last few years spent in the graduate school, I want to say that they have been certainly
very important period. They have been a period of learning science, discovering new
philosophy and cultures, appreciating virtues related not to the work environment, but
also valued in simple, everyday relationships among people and friends. For all of that I
owe the debt of gratitude to a number of people, first of all to my supervisor, Dr. James
Winefordner, great scientist, yet modest person, understanding and guiding advisor.
I would like to thank Dr. Nicolo Omenetto for his inspiring, contagious and
philosophical enthusiasm about science and life. I would also like to thank Dr. Benjamin
Smith, whose office door has always been wide open with him ready to help, support, and
advise on any occasion. I deeply express my gratitude to Dr. Oleg Matveev and his
family, a brilliant scientist and a person who is actually guilty for me being in this school.
He and his family have helped me a lot to settle down and adjust to both school and life
in Gainesville. I would like to say many thanks to Dr. Igor Gomrushkin and his family. He
has been a great friend since my first days in the University of Florida and has given
helpful advice and support many, many times.
I would like to thank the entire Winefordner group of students, past and present,
with whom I have had a chance to boil in the same big "graduate school pot."
Particularly, I would like to thank Dr. Michael Shepard for being a truly big friend. I
would especially thank Dr. Janina Gutierrez. The friendship with her has certainly been
and will always be treasured by me and my family. I would like to thank the Harrison
group students for contributing to the friendly and pleasant atmosphere around our joint
I want to thank Ms. Jeanne Karably for her kindness and help with paperwork. I
greatly appreciate the prompt and high quality job done by the Machine and Electronic
Shops of the Department of Chemistry.
I thank my parents, brothers and their families. Despite being physically far, I
believe they have always been with me in their hearts and thoughts. I also would like to
thank all my relatives and friends in Uzbekistan, who have wished me luck.
I want to thank my wife, Firuza, for her support and care. And finally, I would like
to thank the Creator for blessing me and my wife with the birth of our little angel
TABLE OF CONTENTS
ACKNOW LED GM ENTS ................................................................................................. iii
ABSTRA CT...................................................................................................................... vii
1 INTRODUCTION TO SPECTRA L IM AGIN G ............................................................. 1
2 INTRODUCTION TO RESONANCE IONIZATION IMAGE DETECTORS .............9
RIID Principle of Operation ......................................................................................... 9
Sealed-cell M ercury RIID ........................................................................................... 11
Active Elem ent for RIID ............................................................................................ 12
Conclusion .................................................................................................................. 16
3 CS PHOTOION IZATION AND LA SERS ................................................................... 17
Cs Photoionization: Theoretical Considerations ........................................................ 18
Lasers: Availability..................................................................................................... 20
Cs Photoionization: Experim ental Studies ................................................................. 22
Two-step Ionization............................................................................................. 22
Three-step Ionization........................................................................................... 24
Lasers: Practicality...................................................................................................... 26
Conclusion .................................................................................................................. 27
4 CS RIID CHARACTERIZATION ................................................................................ 28
Experim ental Setup..................................................................................................... 28
Experim ental Results and Discussion......................................................................... 31
Im age Acquisition and Processing ...................................................................... 31
Signal vs. H igh Voltage....................................................................................... 33
Signal vs. Num ber D ensity of Cs Atom s............................................................. 35
Signal and Resolution vs. Laser Pulse-energy .................................................... 37
Detector Efficiency and Spatial Resolution ........................................................ 39
Conclusion .................................................................................................................. 40
5 TIM E-RESOLVED EXPERIM ENTS IN RIID ............................................................. 41
Tim e-Resolved M easurem ents in M ercury RIID ....................................................... 42
Experim ental Setup ............................................................................................. 42
Experim ental Results and Discussion......................................................................... 43
Theoretical Considerations.................................................................................. 43
Tim e-Resolved M easurem ent Results................................................................. 46
Low-m ass Ion Desorption ................................................................................... 49
Absolute Tim e-Resolution of Hg RIID ............................................................... 53
Tim e-Resolved M easurem ents in Cesium RIID ......................................................... 54
Dual-W avelength Tim e-Resolved Im aging with Cs and Hg...................................... 56
Conclusion .................................................................................................................. 59
6 FEASIBILITY OF SEALED-CELL CS RIID CONTRUCTION................................. 61
Cs RIID : First Attempt ............................................................................................... 61
Cs Compatibility: Theoretical Considerations............................................................ 63
Cs Compatibility: Experim ental Verification............................................................. 66
Conclusion .................................................................................................................. 71
7 BUFFER GAS RESONANCE IONZATION IMAGING DETECTOR WITH
AVALANCHE AM PLIFICATION OF SIGNAL...................................................... 72
Buffer Gas RIID : Design............................................................................................ 73
Buffer Gas RIID : Experim ental Results ..................................................................... 78
Avalanche Amplification of Cs Ionization Signal in Flam e....................................... 80
Experim ental Setup ............................................................................................. 81
Results and Discussion........................................................................................ 83
Conclusion .................................................................................................................. 90
8 CONCLUSIONS AND FUTURE W ORK .................................................................... 92
Future W ork................................................................................................................ 94
REFERENCES .................................................................................................................. 97
BIOGRAPHICAL SKETCH ...........................................................................................101
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
A CS ATOMIC VAPOR
RESONANCE IONIZATION IMAGING DETECTOR
Chair: James D. Winefordner
Major Department: Chemistry
Many modem analytical applications demand spectrally selective imaging devices
with high sensitivity and resolution. Spectral selectivity of an image detector would be
especially useful for imaging static or moving objects in highly scattering or high
background environments. For such applications, the detector must be able to detect
photons with a very small spectral or frequency distribution, typically less than 30 GHz.
The resonance ionization imaging Detector (RIID) proposed in this research is based on
an atomic vapor. The superior figures of merit of the RIID are determined mainly by
three factors: (1) extremely high spectral resolution, which comes from the narrowband
absorption of light in atomic vapor; (2) stepwise laser ionization of the excited atoms
with up to 100% efficiency; (3) and a large, almost hemispherical field-of-view.
To accomplish the goal of the project, a novel cesium RIID, the second existing
RIID (after the previously developed mercury RIID), has been developed and evaluated.
Cs is one of the few elements that are suitable for the construction of the RIID and can
offer certain advantages, such as high sensitivity due to strong atomic transitions of Cs
atoms and high spectral resolution, typically determined by the Doppler broadened
linewidth at room temperature (0.38 GHz). Also, the infrared detection wavelength of Cs
RIID at 852 nm allows important biological applications of the detector, because
radiation in this spectral range is safe for biological tissues and has good penetration
depth. The imaging capabilities of the Cs RIID have been evaluated, varying the RIID
parameters such as applied high voltage, the number density of Cs atoms and laser pulse
energies. The materials inert to Cs reactivity (sapphire, titanium, YAG) have been
experimentally evaluated; the results have proven the feasibility of the construction of a
compact, sealed-cell Cs RIID.
The flexibility of the resonance ionization imaging detectors has been demonstrated
with a non-imaging time-resolved mode of operation. Time-resolved ionization signal
measurements in addition to imaging information can enhance the performance of the
RIID by providing the basis for determination of the frequency shift from the resonance
for the photons to be detected. A dual-wavelength, time-resolved resonance ionization
detector with Cs and Hg vapors has been demonstrated. Such detector operating in time-
resolved mode could be very useful in imaging of two closely located targets.
INTRODUCTION TO SPECTRAL IMAGING
The ultimate goal of this research was to develop a two-dimensional spectral
imaging detector of photons with superior figures of merit compared to available spectral
imaging techniques. Generally, two categories of photon detectors are used in analytical
instruments: single-channel and multi-channel. The detector, developed in this work, falls
into the second category, being two-dimensional. By a brief comparison of the detector
with conventional two-dimensional detectors used in analytical instruments and
applications, the motivation and basis for this research will be shown.
There are many modem applications demanding spectrally selective imaging
devices with high sensitivity and resolution (spectral and spatial). Spectral selectivity of
an image detector would be especially useful for imaging static or moving objects in
highly scattering or high background environments. Examples of applications that would
benefit from high spectral resolution detectors are chemical and biological Raman
imaging [1, 2], laser-Doppler velocimetry , optical communications and satellite
tracking . For these applications, the detector must be able to distinguish very small
frequency differences in signal, typically less than 1 cm-1 (30 GHz) .
In most conventional imaging techniques, the multi-channel detector is a charge-
coupled device (CCD) with several thousand pixels in a rectangular arrangement.
Typically, CCDs have efficient spectral response over a broad range of wavelength,
which is favorable in a number of applications. However, in applications where high
spectral discrimination is needed, the CCDs must be coupled with some type of spectral
filter. Commonly available spectral filters are acousto-optic tunable filters (AOTFs) ,
liquid-crystal tunable filters (LCTFs) , liquid-crystal Fabry-Perot interferometers
(LCFPIs) , dual-grating filters (DGFs) , scanning line imaging monochromators
(SLIMs) , fiber-optic bundle arrays (FOBAs) , and dielectric interference filters
(DIFs) . All of these filters have some critical drawbacks. Major drawbacks are broad
bandwidth and/or limited transmission. AOTFs, for example, have relatively high
transmission efficiency (40 50%), but large spectral bandpass on the order of 50 cm-';
thus, they can not provide high spectral resolution. LCTFs and LCFPIs have smaller
bandpasses, on the order of 10 cm'1, and are commonly used as filters in commercial
Raman imaging instruments [1, 13, 14]. However, the transmission efficiency is low (15 -
25%). DGFs, which use two parallel, holographic transmission gratings to isolate images
according to wavelength, have a high transmission efficiency (75%) but large bandpasses
(200 cm'1). SLIMs have bandpasses limited by the spectrometer used (typically
1-10cm'), and generate line-scanned images which require a long analysis time to
produce a 2-D image of the sample. FOBAs have similar spectral characteristics;
however, the spatial resolution is limited by the diameter of the fiber, and current fiber
imaging quality is typically worse than other imaging filters. All these imaging filters
suffer from a trade-off between spectral resolution and throughput, as well as image
degradation. A narrowband atomic vapor resonance ionization imaging detector (RIID),
described in this work, is capable of true 2-D imaging with high sensitivity and spectral
resolution. These essential figures of merit of the system are determined by the physical
characteristics and spectroscopy of the active element of the system. The principle of
operation of the detector is based on absorption of radiation by atoms at the detection
wavelength, which typically is a resonance transition from the ground state, and
subsequent photoionization of the excited atoms. The narrowband atomic absorption
feature of the RIID, as will be shown in detail below, is the basis for high spectral
resolution of the detector. The laser photoionization method employed in the RIID makes
it a very sensitive photon detector.
As an example of RIID application, the principle of Raman imaging with the RIID
will be shown, because in the last few years, Raman spectroscopy has developed into a
2-D (wide-field) chemical and biological imaging technique and commercial instruments
are available for a variety of applications [15, 16]. The wide-field advantages are best
realized when high-fidelity images at a limited number of wavelengths (in the simplest
case, one wavelength) provide sufficient chemical and spatial information. Figure 1-1
shows a general schematic of RIID operation as a Raman imaging device. Let us assume
that the analyte has two spectral features with different Raman shifts, AARi and AIIR2,
indicated in Figure 1-1 as a red pentagon and a blue triangle. For example, these regions
could be healthy and diseased regions of a biological tissue in real-case applications. The
sample is globally illuminated with an expanded excitation laser detuned from the
resonance wavelength by exactly one of the Raman shifts (e.g., A2Ri). Thus, by detuning
the incident wavelength by an appropriate Raman shift, the Raman scatter of interest can
be made to coincide with the detection wavelength of the RIID. The atoms in the RIID,
excited with the radiation Raman scattered from the sample, will be photoionized with
the assistance of additional lasers. The ions formed will then be accelerated to the MCP
incorporated in the RIID, and electrons due to ions impinging on the MCP will be
accelerated to the phosphor screen of the RIID and form an image. The image produced
in the RIID will ultimately be captured with a CCD camera, as is done in conventional
SRaman scattered CCD
Figure 1-1. Schematic diagram of Raman detection with the RIID. The red pentagon and
blue triangle represent two regions of sample with Raman shifts of AZI) and
A2R2, respectively. The excitation laser is tuned to 2A-A2R1 such that upon
interaction with the Raman active sample, the scattered light will be shifted
into resonance with the signal transition of the atomic vapor.
For this example of the detector application, the RIID and a conventional Raman
imager can be compared. In fact, in Figure 1-1, replacing the RIID with an LCTF tuned
to /1, would represent a schematic of a conventional Raman imaging device (with
focusing and collection optics omitted for simplicity). However, there remains a major
difference between the RIID and conventional Raman technology. Conventionally, the
collected Raman signal captured by the CCD camera is reduced in intensity due to the
limited transmission efficiency of the LCTF (typically 25%). In this way, only a portion
of the inherently weak Raman signal reaches the detector. In the case of the RIID, an
amplified Raman signal will be detected by the CCD, because Raman scatter from the
sample first will be converted into ion/electron pairs by stepwise photoionization after
absorption in the atomic vapor. This can be done with efficiency close to 100%. The ions
(or electrons) representing the Raman signal are then amplified in the MCP before
forming an image on the phosphor screen and detection by a CCD. As a result, the RIID
simultaneously acts not only as a filter but also as an efficient amplifier of the Raman
signal, eliminating the need for an expensive image-intensified CCD camera.
The narrow absorption linewidth of the atoms makes the system inherently
selective to the photons with specific wavelength. The system offers very high spectral
selectivity and sensitivity. The spectral resolution can potentially be in the sub-MHz
region. It has been shown that atomic vapor narrowband imaging detectors and filters can
have a spectral efficiency higher than any existing imaging system  (Figure 1-2).
Spectral efficiency or luminosity-resolving power (LR) product, the most significant and
universal figure of merit of any spectroscopic system, is the product of the resolving
power (R) and luminosity (L). Resolving power R is equal to 2/A2, where X is the
average wavelength of two spectral lines separated by the minimum resolvable interval
A2. Luminosity or throughput is defined as the product of the solid angle of acceptance,
.Q (sr) and the projected (acceptance) area, A (cm2). Figure 1-2 shows the calculations of
LR product for several selected spectroscopic techniques with typical parameters . In
principle, all known types of spectrometers can provide a resolving power (R) as high as
105-106. However, for most conventional spectrometers there is an inherent problem of
decreasing throughput as resolving power increases. In other words, the product of these
two parameters is a constant. It has been shown that the LR product for atomic vapor
narrowband imaging detectors, in particular resonance ionization image detectors, can be
higher than 109 cm2 sr. Another important feature of atomic vapor detectors is that under
certain conditions (laser-cooled atoms), there is no trade-off between L and R; i.e., the LR
product is not constant, which means that an improvement in luminosity does not come at
the expense of decreased resolution (Figure 1-2). This trend is observed for heterodyne
detection technique as well. However, heterodyne systems have an inherently low field-
of-view, hence low luminosity in addition to complexity of the systems. Large
interference filters (Fabry-Perot or Michelson) can offer theoretically unlimited small
bandpass, but are extremely angle-sensitive and require additional pre-filters for efficient
background rejection. The RIIDs provide extremely narrow bandwidth (typically less
than a GHz) and simultaneously a hemispherical field-of-view, and therefore exhibit an
exceedingly high signal-to-noise ratio, which is influenced directly by the LR product.
102 Hg RIID (laser cooled)
10 Room temperature
C4 -2 -
"gi'- ----: u^s
Diffraction grating (ideal)
1o Conventional 4 min
0-& -p*& 111l 1 Infrared lItm
Heterodyne detection 1W 0.25 lin
1 10 . .*.... . . .. . 1.. 1. ......
105 106 107 10s 10 100
Figure 1-2. Calculated LR product for selected spectroscopic systems .
Several imaging devices based on resonance absorption of radiation in an atomic
vapor have been reported. The first resonance ionization imaging detector based on
mercury atoms was developed by Matveev et al. . A three-step ionization scheme
was used with the 253.7 nm resonance transition of Hg in the first step (the detection
wavelength). The electrons created in the ionization process were accelerated toward a
luminescent screen by a voltage applied between the screen and the quartz input window.
A sealed-cell compact version of the Hg RIID with microchannel plate was reported .
A resonance fluorescence imaging monochromator (RFIM) based on mercury was
described by Finkelstein et al.,  and Matveev et al.,  with one and two-step
excitations, correspondingly. In the RFIM, a fluorescence of the excited atoms is
detected. The spectral line profile of naturally occurring mercury atoms is 25 GHz. The
best reported spatial resolution has been 130 tm .
A cesium-based RFIM has been described and evaluated for space communication
satellite tracking by Korevaar et al. . The most recent cesium RFIM was reported by
Pappas et al., with less than 200 tm spatial resolution. The spectral resolution of 400
MHz was limited by the Doppler broadened line profile of Cs atoms at room temperature
. The spectral resolution of the Cs RFIMN was further improved (270 MHz) using a
Doppler-free excitation scheme . A detector of far IR radiation using Rydberg atoms
has been reported by Drabbels and Noordam with 300 (Lm resolution . They produced
a sheet of Cs Rydberg atoms in a vacuum chamber (10-6 mbar) by a UV light and directed
an IR radiation, carrying the image of the mask, perpendicular to the sheet. The detector
was sensitive to all IR photons with energies sufficient to ionize weakly bonded Rydberg
atoms, and therefore showed poor spectral selectivity.
A RFIM is similar in principle of operation to a RIID in that both detectors are
based on the narrowband absorption of radiation by an atomic vapor, and both detectors
have similar spectral resolving power. However, signal detection in RFIM is based on
fluorescence and in the RIID on photoionization. Typically, the cross section of
photoionization is 3-4 orders of magnitude larger than the fluorescence cross section. The
superiority of an RIID over RFIM in terms of sensitivity is obvious.
With this brief review of the modem spectrally selective 2-D imaging techniques,
we proceed to the next chapter, which will introduce the RIID design and principle of
operation in detail. Once again, the ultimate goal of this work was to develop a
methodology and device for a two-dimensional photon detection and imaging based on a
narrowband absorption in atomic vapors. To accomplish the goal, a novel cesium RIID
has been developed and extensive temporal studies of the mercury RIID have been
INTRODUCTION TO RESONANCE IONIZATION IMAGE DETECTORS
This chapter describes the general and detailed principle of operation of an RIID.
The performance of the sealed-cell Hg RIID, the only existing RIID until this research,
will be discussed along with some inherent and design problems discovered during the
extensive previous studies [18, 19, 21, 25-27]. Also, chemical elements, that could
potentially be used as an active element in RIIDs, will be listed with the comparison of
RIID Principle of Operation
Figure 2-1 shows a general detailed schematic of the sealed-cell RIID using a three-
step ionization scheme. The principle of operation of the RIID is based on the
narrowband absorption of radiation by an atomic vapor in a cell and subsequent laser
ionization of the atomic vapor filling the cell. The cell consists of an input window and a
phosphor screen, which are coated with a thin layer of metal in order to apply a high
accelerating voltage. Also incorporated in the cell is a microchannel plate (MCP) for
amplification of the photoionization signal. An "image carrier" beam at 2A (detection
wavelength) enters the cell through the input window and excites the atoms from the
ground state to the first excited state. This excitation occurs only in the region of the
atomic vapor which is a 2-D projection of the object being imaged. Depending on the
application, the image carrier beam, reflected or scattered from the object, can be either
of the same wavelength as a probe laser (from a static object) or its wavelength can be
Doppler- or Raman-shifted with respect to the probe laser beam. In the latter case, this
shift should be compensated by detuning the probe laser beam in order to bring the
scattered radiation to the detection wavelength at 2A (see for example Figure 1-1). Two
other laser beams at the wavelengths of 22 and 2A3, coming in through the side windows as
a thin sheet of light, form a planar region where atoms are excited to higher states and
photoionized. This ionization region should be as close to the input window as possible in
order to obtain the highest possible signal.
Ion k2 Ionization Region ja lm
W Atom /
Electron _"_" _
l to CCD
... .... .
Window 4 MCP Screen
Figure 2-1. Schematic diagram of a sealed-cell compact RIID.
By switching the polarity of the voltage applied between the input window and the
MCP, either formed ions or electrons can be detected. Figure 2-1 shows the case where
the ions are accelerated to the MCP. Electrons produced by charged particles striking the
MCP are accelerated by an electric field to a luminescent phosphorr) screen, where the
image of the object is captured by the CCD camera. In terms of design, the RIID is
similar to a standard image intensifier. Image intensifiers use a photocathode to transduce
an optical signal into an electrical signal; the electrons, amplified by the MCP
incorporated in the image intensifier, are then accelerated toward a luminescent screen. In
the RIID, the photocathode is replaced by a high-density atomic vapor. The optical image
is transduced into an electrical signal by the laser stepwise ionization of the atomic vapor.
Sealed-cell Mercury RIID
Previously a sealed-cell Hg RIID was designed and evaluated in our laboratory.
This compact cell was 5 cm in diameter and 4 cm in width, with a front input widow
diameter of 2.5 cm and side window widths of 0.7 cm [19, 27]. The cell was constructed
according to the diagram shown in Figure 2-1 (NPP Radian, Moscow).
In laser resonance ionization photon detection, which is the basic principle of the
RIID, it is desirable to have an efficient and simple photoionization scheme using a
minimal number of lasers and minimal energy of laser radiation. A stepwise
photoionization scheme of Hg has been the subject of many studies [18, 21, 25, 28, 29].
In all schemes, the first step was a resonance transition from the ground state at 2A =
253.65 nm. Figure 2-2 shows two successfully realized and studied schemes for the Hg
RIID routes of photoionization, one of which will was used in this work and will be
discussed in following chapters.
It has been shown that the sealed-cell Hg RIID was a very sensitive detector of
light; as low as 1000 photons have been detected by image summation . The detector
showed a 130 jim spatial resolution when the ionic component of the imaging signal was
detected . However, a few inherent and design problems have been encountered with
the detector. The detection wavelength of the detector (253.65 nm) was in the deep UV
region. First of all, this limits the application of the detector to important in vivo medical
applications such as imaging of blood circulation and tumors, because of the harmful
effects of UV radiation to human tissue. In addition, studies have shown an unwanted
photoelectric effect from the metal coated input window and the surface of the MCP due
to the UV transition of Hg [19, 30]. The energy of A,1 = 253.65 (4.89 eV) is higher than
the work-function of the metal coatings of the input window and MCP. These non-
selective electrons are a source of additional noise in the measurements. Recent studies
have shown that the detector suffers from temporal image distortions due to surface
charges accumulated on the input window .
'S' 0S, 'P -D
ns ns np nd
/rzizz//// LLL/J/////, 10.44 eV
S 8.85eV -8
7.73 eV 7- 7
4.89 eV 6
Figure 2-2. Partial energy level diagram and selected transitions of mercury.
Active Element for RIID
The performance and figures of the merit of an RID are determined by the physical
characteristics and spectroscopy of the active element of the system. As mentioned
earlier, the narrowband atomic vapor detectors with Hg and Cs have been reported in the
literature [18-24]. The choice of the active element is limited to a few atoms such as Rb,
Cs, Hg and inert gases which provide sufficient atomic vapor density at room temperature
to absorb a large number of resonance photons at the detection wavelength. Depending
on the application, one may be advantageous over another . Table 2-1 lists potentially
suitable elements for the construction of the RIID, along with some important related
Table 2-1. A list of elements suitable for RIID construction.
Element Hg Cs Rb He Ne Ar Kr
IP (eV) 10.4 3.9 4.2 24.6 21.6 15.8 14.0
MP(C) -39 29 39
,nsa(cm-3) 6x1013 4x10
Table 2-1 shows the first ionization potential, melting point and saturated vapor
density of potentially candidate active elements for RIID. The listed three metals, Hg, Cs,
and Rb, are preferred over noble gases, because the noble gases are difficult to ionize and
their electronic transitions lie in the deep UV spectral region, which may be difficult to
reach with commercially available lasers. Among the three metals, Rb may need
additional heating in order to obtain a high enough vapor density, due to its high melting
point. That leaves only Hg and Cs, two elements based on which resonance ionization
imaging detector's have been developed to date and those are the subject of this research.
It is pertinent here to compare closely essential physical and spectroscopic
characteristics of these two elements (Table 2-2), since the active element determines the
detector's operation and performance.
Table 2-2. Comparison of Hg and Cs properties.
Properties\Element Hg Cs
Detection wavelength (nm) 253.65 852.12
Oscillator strength of the detection transition,f 0.025 0.72
table isotopes 7 1
Atomic linewidth at room temperature, AX (GHz) 25 0.38
Chemical reactivity Low High
Photoelectric effect(PE) Yes No
Biological applications Limited Yes
The detection wavelength for Hg is 253.65 nm and for Cs is 852.12 nm. The
oscillator strength of the transition at the detection wavelength is much greater for Cs.
Mercury has 7 stable isotopes, whereas Cs has only one. The atomic line broadening at
room temperature is 25 GHz for natural Hg, which is a Doppler broadening with all
mercury isotopes taken into account. Doppler broadened atomic linewidth of
monoisotopic Hg atoms is approximately 1 GHz. Doppler broadened atomic linewidth is
only 0.38 GHz for Cs, making Cs more spectrally selective active element for the RIID.
The calculated cross sections of the transitions at the detection wavelength for Cs and
monoisitopic Hg atoms are 5.6x 10" cm2 and 5.7x 1013 cm2, respectively, showing the
better sensitivity of the detectors based on Cs atoms, provided that the number density of
atoms is sufficient to absorb the large number of signal photons. The saturated number
density of Hg atoms at room temperature is 3 orders of magnitude greater than that of Cs.
The UV detection wavelength of mercury may cause an unwanted photoelectric effect
(PE) from the metallic surfaces of the detector, which has been observed for sealed-cell
Hg RIID. For Cs, this effect is not expected and will be shown experimentally to not
occur. The infrared detection wavelength of Cs, which is safe for human tissue and can
penetrate deeply through a tissue, will allow biological applications of the detector. The
UV radiation at 253.65 nm may be harmful for human tissue. Despite the apparent
disadvantages, Hg RIID can provide better spatial resolution and signal-to-noise ratio
because it operates at a shorter wavelength. The LR products of Hg RIID and selected
high spectral resolution devices were compared in Figure 1-2. Figure 2-3 compares the
LR product for two Cs and Hg RIIDs with 100 cm2 acceptance area. It should be
mentioned that there is no practical limitation for construction of the detector with
considerable larger size.
-Hg, 1.00 GHz
.....Cs, 0.38 GHz
I 1 ' ".. ..." ...........
105 106 107 108 109 101
Figure 2-3. Calculated LR product for Cs and Hg RIIDs for Doppler-broadened
absorption profiles at room temperature.
In addition to better spatial resolution, mercury is not particularly reactive with the
cell materials such as quartz, metal coatings on the windows, electrodes, and thus a
sealed-cell Hg RIID was easily constructed. However, some difficulties must be
overcome in order to construct a sealed-cell Cs RIID due to the reactivity of Cs. The
feasibility of sealed-cell Cs construction from compatible with Cs materials will be
discussed in Chapter 6.
In this chapter, the general principles of operation of a resonance ionization
imaging detector have been discussed. The Hg RIID developed prior to this work proved
that the RIID could provide efficient photon detection with comparable imaging
performance to conventional techniques. Also, it has been shown that Cs offers certain
advantages over Hg as an active element in RIID. In the following chapter, ionization
schemes of Cs and laser systems appropriate for the RIID are experimentally
CS PHOTOIONIZATION AND LASERS
In order to reach the best performance of the Cs RIID, an ionization scheme had to
be carefully chosen, for it is the major factor determining the quantum efficiency
(sensitivity) of the detector. If high-energy pulsed radiation is used for photoionization,
theoretically there is no fundamental limitation to obtaining a near-unity ionization
efficiency. A single-photon detection capability of the RIID has already been
demonstrated . A pulsed RIID based on the photoionization with powerful lasers is
therefore advantageous over the cw diode laser-based RFIM's mentioned in Chapter 1
with regard to quantum efficiency. Currently, commercially available diode lasers cannot
provide sufficient energy for effective photoionization.
Despite the advent and achievements in the field of lasers in general and especially
in the field of tunable lasers, still there is no single laser that would satisfy the end user in
terms of availability, practicality, cost, tunability, power characteristics. For example,
tunable dye lasers are a valuable research tool, covering a wide spectral range, from deep
UV to mid-IR. However, they are difficult to use and maintain, in addition to being
unable to provide sufficient power for some applications. Tunable and powerful sources
of radiation would broaden the range of applications of lasers and/or improve the
performance of existing applications.
This chapter describes the spectroscopy of Cs as well as the laser system used in
these studies. An efficient ionization scheme will be evaluated based on the spectroscopic
parameters of Cs from the viewpoint of the availability and practicality of the commercial
lasers. Results of the non-imaging Cs resonance ionization studies will be shown.
Cs Photoionization: Theoretical Considerations
Cs is an unique metal, an element with the lowest ionization potential of all
elements (3.89 eV). Due to its unique properties, Cs atoms have been a subject of many
studies [32-39]. Cs atoms have been used as narrowband atomic filters in a number of
studies [40, 41, 42]. When choosing an excitation scheme for resonance ionization
imaging detectors, two factors are most important, the cross section of the transitions and
the availability of radiation sources for chosen transition. The energy required to saturate
the transition is directly proportional to the cross-section of the transition, which can be
approximated with the following formula (in cm2):
S= 8.19x1020 fk "J2ik (3-1)
where, fk is the oscillator strength of the given transition dimensionlesss), Aik is the
detection transition and AAD is Doppler broadened atomic linewidth at the given
conditions (both in nm). For Cs, the natural linewidth is 5 MHz and the Doppler-
broadened linewidth at room temperature is 380 MHz. The width of the absorption
profile is determined by external factors, such as temperature and nature of the
environment, whereas the oscillator strength is an intrinsic property directly determining
the cross-section of the transition.
Figure 3-1 shows two ionization routes for Cs. The transitions of Cs
6s2S/2->6p2P3/2 (852.12 nm) 6p2P312->6d2D3/2 (917.23 nm), shown in Figure 3-1, have
an oscillator strength of 0.72 and 0.33, respectively, and are the two strongest transition
of Cs. Logically, an efficient ionization scheme should include these transitions.
2 11/ 2 -F.V 2D-512
ns np nd
/////////// /////////'//// 3.89 eV
I 7- 2.81eV
Figure 3-1. Partial energy level diagram and selected transition of Cs atoms.
The excited 6d2D32 state lies 1.08 eV below the ionization limit of Cs (3.89 eV). Ideally
the photoionizing laser energy h v should match the energy difference between the excited
state and the ionization limit. The photoionization cross-section is frequency-dependent
and drops drastically when photoionization energy considerable exceeds the ionization
limit . The simplest laser available for photoionization of Cs from the excited 6dD312
state is the fundamental harmonic wavelength of Nd:YAG laser at 1064 nm with photon
energy of 1.16 eV. In order to obtain the highest possible efficiency (-100%), all steps
in the stepwise ionization must be saturated. Calculated cross-sections of the
6s2S12-->6pP32 (852.12 nm) 6p2P3/2->6d2D3/2 (917.23 nm) transitions according to
formula (3-1) are 5.6xl0"11 cm2 and 2.5xl011 cm2, respectively, for the Doppler-
broadened absorption profile of Cs at room temperature (380 MHz). Therefore, saturation
of these transitions in the experiments can be easily reached by commercially available
compact, low-cost diode lasers. The photon saturation irradiance of the transition is given
by (in photons.cm-.s -):
I o- (3-2)
where, r is the lifetime of the excited state (s), assuming the laser beam profile matches
the atomic absorption profile at the given conditions and the laser pulse duration is
shorter than the lifetime of the excited state. For the Cs transition from the ground state
(852.12 nm), the saturation irradiance is calculated to be -3x1017 photons.*cm-2.*s-1, which
can be reached with typical diode lasers with a power of a few mW.
However, the situation is dramatically different for the final photoionization step
due to its non-resonance nature. The calculated or experimental value of the
photoionization cross-section from the excited 6d2D3/2 state have not been found in
literature, but can be expected to be on the order of 10-16 1017 cm2, which is 5 6 orders
of magnitude smaller than the strong transitions of Cs mentioned above. Therefore, it is
critical to have a very high-power pulsed laser in the third step for obtaining the highest
possible ionization efficiency.
Once the question with the spectroscopy of the atoms clear and the transition are
chosen, the next important issue is the selection of light sources for excitation and
photoionization of the chosen transitions. One of the most often used and flexible light
sources in atomic spectroscopy is the tunable dye laser and it is useful for a wide
wavelength range of the electromagnetic spectrum. However, the dyes tunable in the IR
region are very inefficient and unstable, and require frequent change of the toxic dye
solution. Radiation at 852.12 nm can be obtained with a dye laser operating with a
solution of Styryl-9 in DMSO (Lambda Physik, Acton, MA). The maximum energy
conversion efficiency given by the manufacturer is only 9% when pumped by a XeCl
Excimer laser at 308 nm. For laboratory conditions, this efficiency is expected to be even
less. Despite this, the dye laser with Styryl-9 can operate with relatively stability and
acceptable efficiency at 852.12 nm for short period of time. The only available dye for
917 nm, HITCI, has 4% efficiency and is much less stable compared to Styryl-9. The
more efficient dye Styryl-14 has recently been discontinued, reflecting the interest in all
solid-state and semiconductor lasers.
A number of solid-state lasers have been developed as the result of the research in
the area of tunable solid-state lasers and non-linear crystals, providing an efficient
tunable source of radiation in the IR spectral region [44, 45]. Advantages of solid-state
lasers over dye lasers are compactness, efficiency and ease of operation. Lasers based on
LiF:.-F2+ color centers are very attractive as highly stable and powerful sources of tunable
radiation, exhibiting high gain and a low threshold of laser generation. A stable, efficient
LiF:F2+ laser, tunable in the 800-1200 nm region, has been evaluated and described in the
literature [46, 47]. A unique and practical source for excitation of Cs atoms would be a
LiF:F2+ laser capable of producing the 6s2S,/2--6p2P3/2 (852.12 nm) 6p2P3/2->6d2D3/2
(917.23 nm) transitions simultaneously since it would simplify the spatial alignment,
beam geometry, and the timing of the excitation pulses. This feature has already been
realized with tunable dye lasers and described in the literature .
Cs Photoionization: Experimental Studies
Both ionization schemes shown in the Figure 3-1 have been experimentally
investigated using the lasers available in the laboratory, including tunable dye lasers
pumped by an excimer laser at 308 nm, and LiF.F2+ color center pumped with various
sources. The ionization schemes were evaluated for a Cs resonance ionization detector
(RID), by measuring the ionization signal from a cell shown in the Figure 3-2. The cell is
an evacuated Pyrex cylinder (60 mm long, 25 mm diameter) containing solid Cs (Opthos
Instruments). The cell is furnished with a pair of nickel electrodes with feedthroughs for
application of voltage and detection of ionization signal. The design of the cell allows the
light can go in only through two side windows and can not be used for imaging.
Figure 3-2. A schematic diagram of the cell used for Cs RID evaluation.
Figure 3-3 shows the schematic diagram of the experimental setup of the two-step
ionization studies. Two dyes lasers (Molectron DL-II, Portland, OR) pumped by an
excimer laser (LPX-240i, Lambda Physik, Acton, MA) operated with XeCl (A = 308 nm)
were used in the experiments. The first dye laser at 852.12 nm entered the cell from one
end of the cell and excited the Cs atoms from the ground state into the first excited state
(62P3/2). The energy difference between the excited state and ionization limit is 2.44
eV. This corresponds to laser energy with a wavelength of- 508 nm.
Beam splitter Mirror
o| 5308 nm
S. ...Cs cell
E) Oscilloscope __
c,,- ______ \,852 nm ~^-508 nm
Figure 3-3. A schematic diagram for two-step ionization setup.
The counter-propagating second laser was frequency scanned in the tunability
range of the dye from -490 nm to 530 nm and had a maximum in the vicinity of 508
nm. The ionization signal was observed with an oscilloscope. Figure 3-4 shows the
ionization signal as a function of ionization laser tunability.
S100- 490 nm
t0 75 -
.o 530 nm
-25 -- -- -- -- -- -- -- --
-15 -10 -5 0 5
Figure 3-4. Two-step ionization signal of Cs as a function of wavelength.
For three-step Cs ionization, a system based on Nd:YAG laser and LiF:.F2+ crystal
color center laser was used. The schematic diagram is shown in the Figure 3-5.
Glass Raman 1 'nm KTP 'nm
Plate shifter r
% ~ M5
Oscilloscope 9 nm
;.--, / 852 nmvSJ LiF:F \
G, rating 1- !? | '
Figure 3-5. A schematic diagram for three-step ionization setup.
The laser beam with the diameter of -6 mm at 1064 nm from Q-switched Nd:YAG
(Laser Photonics) laser was used to pump the Raman wavelength shifter with Ba(N03)2
crystal (Solar, Belarus). The Energy of the Nd:YAG laser was -80 mJ, with 10 ns pulse
duration in multi-mode regime with repetition rate of 1-10 Hz. The telescope was used to
reduce the diameter of the laser beam to -2 mm. The shifted radiation at 1198 nm (-12
mJ) was focused by the cylindrical lens L1 into the KTP crystal for the second harmonic
generation. The radiation at 599 nm (-2 mJ) was separated from the fundamental 1198
nm by M1 cold mirror (Edmund scientific). Radiations A, = 852.12 nm and A.2 = 917.23
nm were obtained simultaneously by pumping the LiF:-F2+ crystal with 599 nm radiation.
The size of the crystal was 40x20x6 mm3 and the input windows were cut at Brewster
angle to reduce the reflection. The 599 nm laser beam was directed by an aluminum
mirror Mz and focused by a lens L2 onto the crystal. A resonator was formed by a pair of
aluminum mirrors (M4, M5) and M3 hot mirror (Edmund Scinetific). The hot mirror had a
high reflectance for 750-1000 nm and high transmission for the pump radiation (599 nm).
The diffraction grating with 1200 gr/mm was used as the wavelength selection element in
the resonator. The distance between M3 and the diffraction grating was -10 cm. A blaze
angle of the diffraction grating was chosen to be small (5) in order to expand the
tunability of the laser. Each of the two resonator mirrors M4 and M5 reflected a portion of
the zero-order diffracted beam. The mirrors could be adjusted separately to provide two
independently tunable output radiations at 2A, = 852.12 nm and 22 = 917.23 nm. A portion
of the fundamental radiation of Nd:YAG laser at 1064 nm was directed and focused to
the cell by a glass plate, M6 mirror and L3 lens. The ionization signal of Cs was
monitored in an oscilloscope by applying a 12 V bias voltage to the electrodes as in the
case of two-step ionization. Energy of the lasers at A, = 852.12 nm and 22 = 917.23 nm
were -50 [J each. The energy of the photoionization laser at 1064 nm (the third step) was
The results of the experiment are shown in the Figure 3-6. The curve 1 represents a
resonance three-step (852-*917->-1064) ionization signal of Cs atoms. The signal,
although very small, was observed in the absence of 1064 nm radiation due to Cs
photoionization with the same 917 nm radiation in the third step (curve 3). In addition,
there could be some contribution of the collisional excitation/ionization in the signal
since the cell was under saturated vapor pressure. Also, a small signal was observed from
1064 nm radiation only (curve 2). In this case, the signal was believed to be due to
collisional excitation of Cs atoms and photoionization with 1064 nm radiation. The ratio
of the signal when all three beams were present to the signal when only 852 and 917 nm
radiations were present was more than 350. The ionization signal with three-step scheme
was approximately 200 times larger compared to the signal with 852->917 two-step
scheme (Figure 3-4). This means that with the powerful Nd:YAG laser for the third
photoionization step, an ionization efficiency improved more than 2 orders of magnitude
for the given experimental conditions.
0 20 40 60
Figure 3-6. Three-step ionization signal of Cs; curve 1 represents a resonance three-step
(852-*917--> 1064) ionization signal, curve 2 signal from 1064 nm radiation
only, curve 3 signal with two wavelength present (852->917). Note that
signal values in curves 2 and 3 are multiplied by 100 for illustration purposes.
Despite the dramatic increase in an ionization efficiency of the three-step
ionization, the configuration shown in Figure 3-5 was too complicated and far from
practical, especially for applications to imaging. Obtaining an appropriate wavelength for
pumping the LiF.F2+ crystal was done in multiple steps involving two non-linear
phenomena, first shifting the fundamental output of Nd:YAG laser and then second using
the harmonic generation from the wavelength shifted radiation. This could be done only
with very low efficiency (approximately 2% in this case).
As an alternative source of radiation to pump the LiF.F2+ crystal, a single-
wavelength (-610 nm) dye laser with Rhodamine B was constructed. A dye solution
circulating cuvette was placed between a pair of mirrors, one with 100% reflectance, the
other one with 20% reflectance, thus forming a resonator without any dispersive element.
The dye laser was pumped with an XeCl excimer laser (308 nm). The stronger laser beam
to pump the LiF:F2+ crystal, compared to the case depicted in Figure 3-5, was obtained
with a pulse energy of approximately 5 mJ. Despite this improvement, when pumped
with this radiation, the output of the LiF:.F2+ crystal did not considerably improve,
probably due to beam quality at 610 nm. In addition, coupling this system with an
Nd:YAG laser for efficient three-step ionization would require a complicated
In this chapter the efficiency of three-step ionization scheme for Cs has been
theoretically discussed and experimentally validated. However, the laser system for the
three-step ionization scheme was not practical and was not applied to present imaging
studies. The only practical and reproducible laser system on hand suitable for Cs
photoionization was two-step ionization with dye lasers pumped by an excimer laser.
Using this two-step ionization scheme, the imaging capabilities of Cs RIID was evaluated
and will be presented in the next chapter.
CS RIID CHARACTERIZATION
A novel Cs RID reported in this work, inherited many design features from the
sealed-cell compact Hg RIID, developed and evaluated in our laboratory . In this
chapter, the first images obtained with the Cs RIID will be shown. All experiments
reported here have been performed using the two-step photoionization scheme
(852->508) of Cs, described in the previous chapter. The dependence of the images on
the parameters of the system, such as high voltage applied to the system, Cs vapor
density, and exciting/ionizing laser pulse energies, will be evaluated.
Figure 4-1 illustrates the schematic of the experimental setup used for evaluation of
the detector in imaging mode of signal detection. The detector consisted of a metal-
coated input window, a 25 mm long linear Cs metal dispenser, and a microchannel plate
coupled with a phosphor screen in a vacuum chamber as shown in Figure 4-1. A six-way
cross chamber was constantly evacuated with a mechanical and turbo-molecular pumps.
Vapor of Cs was produced using the linear dispenser (SAES Getters USA Inc., Colorado
Springs, CO). The dispenser contained a mix of alkali metal chromate (Cs2CrO4) with a
reducing agent to provide a controlled amount of alkali metal. The Cesium chromate salt
and reducing agent mixture was held within a nichrome metal container having a
trapezoidal cross section with a slit to allow evaporation of the alkali metal (Figure 4-2).
When heated under vacuum by flowing a current through, the dispenser released atomic
Cs as the result of an oxidation-reduction reaction.
Figure 4-1. Schematic diagram of Cs RIID setup.
- FT type-
Figure 4-2. Schematic diagram of the Cs metal dispenser [SAES Getters USA Inc.
A two-step ionization scheme described in the previous chapter was used for the
imaging experiments. The first dye laser beam, at the detection wavelength of
A/= 852.12 nm, entered the cell through the input window carrying an image of the
cross-shaped transmission mask and excited the Cs atoms from the ground state into the
6 P3/2 state. This excitation occurred only in the region of the atomic vapor which was
illuminated by A1 laser beam passing through a mask and the perpendicular A2 laser beam.
The second photoionizing dye laser beam, at 22 508 nrim, was directed from the side,
perpendicular to A,, as a thin vertical sheet of radiation and ionized the Cs atoms. The
distance between the input window and MCP was 30 mm and could be adjusted. The
distance between the MCP and the phosphor screen was fixed at 10 mm. Figure 4-3
shows the photograph of the test image with dimensions, which is a four-wing cross-like
shape cut through thin metal sheet. Wings do not meet at the center, so the center of the
cross is opaque.
Figure 4-3. Photograph of the test mask with dimensions.
Figure 4-4 shows the voltage division between electrodes (input window, MCP,
phosphor screen) of the RIID. A negative voltage between -9 kV and -11 kV was applied
to the MCP and the input window was grounded to accelerate Cs ions to the MCP. The
phosphor screen was held positive relative to the MCP in order to accelerate the electrons
created by ions striking the MCP.
r / \ _80 Mn
120 MQ 30 M
Figure 4-4. Voltage division between electrodes of the Cs RIID.
Experimental Results and Discussion
Image Acquisition and Processing
The images of the test mask formed on the phosphor screen were captured by a
monochrome CCD camera (PC23, Supercircuits, USA). The CCD camera was interfaced
with a computer via an image acquisition board (PCI 141, National Instruments, USA).
The image frames were synchronized by pulses from the excimer laser. The images were
acquired in 8-bit gray-scale mode (256 shades of gray, 0 for black and 255 for white)
with the IMAQ Vision Builder 6.0 software package (National Instruments, USA).
Figure 4-5 shows examples of images that could be obtained and processed with the
Figure 4-5. Two and three dimensional representations of: a) a single image; b) a filtered
single image (see text for filter description); c) average of 20 images.
given hardware and software for images acquisition. Three dimensional views of the
images were produced in IMAQ Vision Builder and presented for illustration purposes.
In Figure 4-5, a single image, a single filtered image and the average of 20 images are
compared (a, b and c). The IMAQ Vision Builder convolution filter highlights regions in
the image where there are sharp changes in pixel values. These regions correspond to the
boundaries of the mask and other noisy pixels that may be present in the image. Table 4-1
shows the improvements in the image signal-to-ratio. In this work, a signal-to-noise ratio
of the images was defined as the ratio of a maximum pixel value of the image to the
standard deviation of the pixel values along the vertical dashed line as shown in Figure
4-5a. All the images reported in this chapter are averages of 20 individual images.
Table 4-1. Signal-to-noise improvement with image averaging and filtering.
Signal Noise Signal-to-noise
a) 143 7.8 18
b) 252 10.7 24
c) 164 6.4 26
Signal vs. High Voltage
Figure 4-6 shows the dependence of signal on the high-voltage applied to the
system. In order to correlate the image intensity with the photoionization of Cs, signal
was calculated as the total intensity of the images, i.e. sum of all the pixel intensities of
the images of the test mask. Calculated in such a way from the images, signal or image
intensity would be proportional to the resonance photoionization signal of the Cs atoms.
An almost exponential signal increase with high voltage is supposed to be determined by
the non-linear gain of the MCP with the applied voltage. The S/N ratio as defined in the
previous section was monitored and reached a maximum at about -10.1 kV; and any
further increase of the negative high voltage did not improve the signal-to-noise ratio. At
40-, I I -,
9.6 9.8 10.0 10.2 10.4 10.6
Figure 4-6. Signal vs. high voltage applied to the system.
-10.5 kV, the pixel intensity saturated, reaching the maximum gray level of 255,
indicating that the images obtained at potential more negative than -10.5 kV would not
truly represent the photoionization of the atoms. Further correlation of the image intensity
and ionization signal would require attenuation of the luminescence of the phosphor
screen. A conclusion from these measurements was that the optimal operating voltage
was -10.1 kV for the given configuration of the detector and experimental conditions.
Other system parameters were studied at the optimal voltage.
Signal vs. Number Density of Cs Atoms
Absorption of resonance photons is directly proportional to the number of
absorbers n, i.e. number density of Cs atoms which was varied by changing the current
flowing through the linear dispenser. The studied range of number densities between 107
and 108 cm-3 was significantly small compared to saturated number densities of Cs atoms
at room temperature (n,, 4.1010 cm3). The number density of Cs atoms was estimated
by measuring the absorbance with a hollow cathode Cs lamp. Figure 4-7 shows the series
of images taken at different number densities of Cs atoms. Even at such low number
density of Cs atoms, images with high contrast ratio were obtained, indicating large
photoionization signal. Such high signals were due to the strong cesium transition at the
detection wavelength, hence large cross-section of the absorption.
Figure 4-7. Series of images at different Cs number densities (cm3). From left to right,
densities are: 1.3x 107, 2.4x107, 6.1x107; bottom row: 1.2x 108, 1.9x 108.
Figure 4-8 is a log-log plot of signal (total intensity of images) vs. Cs number
density. Theoretically, the signal should depend on the number density of atoms via the
expression of the absorption factor given by :
a = exp[-cn/] (4-1)
where I is absorption path length (in cm), i.e. the penetration depth of 2A into Cs vapors, a
is the absorption cross-section (in cm2) defined in equation (3-1) and n is the Cs number
density (in cm3). Generally, both 1 and a depend on the detection wavelength [49, 50]. In
order to achieve a high signal level the following condition must be fulfilled:
onl t 1 (4-2)
For optically thin conditions (nl -* 0), which was valid for our case, the absorption factor
is linearly related to the number density of atoms n for small values of n. At high number
density of atoms, a reaches its limiting or saturation value. It is seen from the Figure 4-8
that at the given experimental conditions, the signal vs. number density of Cs atoms
shows similar to typical atomic absorption experiments behavior. Saturation level of the
signal was not reached in the studied range of Cs number densities.
0.1 1 10
Log Cs number density (x108 cm3)
Figure 4-8. Signal vs. number density of Cs atoms.
The experimental values of the absorption coefficient, number density of Cs atoms and
absorption path length were 5.6x10-" cm2, 1.9x108 cm3 and 5 cm respectively. The
product of these quantities was equal to 0.05. In the experiments the number density of
Cs atoms were estimated by measuring the absorbance with a hollow cathode Cs lamp.
At n = 1.9x 108 cm3, saturation of the pixel intensity of the CCD camera was observed.
When dealing with atomic population of atoms with strong resonance transition,
such as Cs atoms, one must consider the effect of radiation trapping (diffusion). The
excited Cs atoms may be de-excited by emitting resonance radiation at 852.12 nm before
being ionized. This process may occur repeatedly and result in ionization at points far
away from the point where the original incident photon was absorbed. This phenomenon
is more pronounced for dense atomic populations. Despite the low Cs number densities
studied in this work, with an increase in the number density of atoms, a small degradation
(<10%) of spatial resolution of the images was observed (Figure 4-7), which was
attributed to the resonance radiation trapping (diffusion) effect. The effect of resonance
radiation trapping on the spatial resolution of the detector is expected to be minimal or
even negligible if one uses an efficient ionization scheme and lasers. Because of the
broad linewidth of the laser used in our experiments, the efficiency of ionization was low.
Therefore, for the best detector performance, it will be critical to have a laser system
capable of providing 100% efficient ionization, which is theoretically possible for a
method based on laser photoionization of atoms.
Signal and Resolution vs. Laser Pulse-energy
Figure 4-9 shows dependence of the signal on the irradiance of the excitation and
ionization lasers. The linewidths of the dye lasers used in the experiments were estimated
to be 1-1.5 cm1 (30 45 GHz), very large compared to the Doppler broadened profile of
Cs atoms at room temperature (380 MHz). Again, because of to this broad linewidth of
the lasers, the pulse energies of the lasers were not enough to saturate the transitions,
even the strong transition from the ground state at 852.12 nm, having oscillator strength
of 0.72. This can be seen from the plots of signal vs. laser irradiances in Figure 4-9. The
laser irradiances were varied using a set of calibrated filters.
80 . .. .
S 10 20 30 40 50 60
Irradiance of X, (,JIcm -2)
0 2 4 6 8
Irradiance of 42 (mJ/cm')
Figure 4-9. Signal vs. laser pulse energy: left 2A = 852.12 nm; right A2 = 508 nm.
Figure 4-10. Spatial resolution vs. laser pulse energy at X2. Irradiance: left 0.7 mJ/cm;
right 40 mJ/cm2.
Both signal and spatial resolution depend on the power characteristics of the lasers
used in the excitation scheme. It was especially important for the final ionizing step
because of the low cross-section of photoionization due to non-resonance nature of the
transition. Figure 4-10 illustrates the improvement in the spatial resolution with increase
in the irradiance of the second photoionizing laser beam at 508 nm. Two images were
taken with laser beams unfocused and tightly focused with a cylindrical lens. With
focusing, the irradiance of the laser beam increased from 0.7 to 40 mJ/cm2 (see
improvement in Figure 4-10).
Detector Efficiency and Spatial Resolution
The quantum efficiency (QE) and the ionization efficiency (IE) of the detector were
estimated as the following:
QE = # of laser induced charges (43)
QE : ------ (4-3)
# photons absorbed
JE # of laser induced charges (-
/E = -----------(4-4)
# of Cs atoms in the illumination volume
These two quantities are related by:
QE = axlE (4-5)
where, a is a fraction dimensionlesss) of photons absorbed or absorption factor (see
equation 4-1). The efficiencies were calculated from the measurement of the signal as a
current from the phosphor screen by 50 Q termination. Figure 4-11 shows an oscilloscope
trace of the measurement (average of 128 shots). The number of electrons calculated
from Figure 4-11 was 5.5x108. Assuming the MCP gain as 104, it can be concluded that,
on average, 5.5x104 ions were created per laser pulse. The number of Cs atoms in the
ionization volume was estimated to be -107. The ionization efficiency according to the
equation 4-4 is 0.005. Equation 4-5 then gives the value of the quantum efficiency of the
detector as QE = 0.05x0.005 = 2.5.10-4. Using the equation (4-3) the number of absorbed
photons was calculated to be 2.2-108.
0.5 1.0 15 2.0 2.5
Figure 4-11. Signal measured as a current from phosphor screen of the detector.
Absolute spatial resolution of the system with the current configuration and parameters
was determined by using a 1951 USAF test chart and was found to be 280 JIm.
Ultimately, an optimized RIID should have a spatial resolution limited by the resolution
of the MCP and/or image capturing device (CCD) on the order of 1 .m.
In this chapter, the Cs RIID was evaluated in the imaging mode of signal detection.
The conclusion of the chapter is that, for comprehensive evaluation of the detector it is
important to have an efficient ionization scheme with powerful lasers to perform the
ionization. In the next chapter, important and interesting fundamental temporal studies
performed in the non-imaging signal detection mode with both the Cs and Hg RIIDs
separately and in combination will be shown.
TIME-RESOLVED EXPERIMENTS IN RIID
As predicted by Matveev , there are four main noise sources that degrade the
figures of merit of a resonance ionization imaging detector: (1) thermally or collisionally
excited atoms in the state, which is to be excited only by the radiation to be detected; (2)
multiphoton ionization (MPI) of atoms by radiation in the second and/or third steps of
stepwise ionization; (3) thermal ionization (TI) of atoms; and (4) photoelectric effect (PE)
due to background radiation and radiations of the second and/or third steps. The
photoelectric effect can occur from the thin metallic film of the input window, side
windows and from the MCP. This chapter is dedicated to studies of some of these noises
by monitoring the time-resolved ionization signal in the RIID. Of these noises, PE in
particular, was observed in previously evaluated compact, sealed-cell Hg RIID [19, 27].
Understanding the sources of these noises is essential for both the Cs and Hg RIIDs. The
time-resolved signal detection mode of the RIID operation is based on a current
measurement from the phosphor screen using an oscilloscope. Time-resolved ionization
measurements in addition to imaging information can enhance the performance of an
RIID by providing the basis for determination of the frequency shift from the resonance
for the photons to be detected as was done in . Such possibilities will be discussed in
this chapter, along with a unique combination of Cs and Hg in a single RIID with dual
Time-Resolved Measurements in Mercury RIID
Three-step ionization of Hg (see Figure 2-2) in the experiments was achieved by
using two Molectron DL-II dye lasers pumped by a XeCl Excimer laser at 308 nm.
Radiation at 2/ = 253.65 nm (6'S0 -->'63pO) was generated by frequency doubling the
507.3 nm radiation from a dye laser with Coumarin 500. Radiation at 22 = 23 = 435.8 nm
for the transitions 63J0 -->7SI and 73S, -> ionization continuum were generated from the
second laser with Coumarin 120 dye solution. Pulse energies of 10 pJ and 450 gJ were
measured for radiation k, and X2, correspondingly, operating at 10 Hz. The linewidths of
the lasers were between 1 cm1 to 1.5 cm1.
a.... ... _
I I '
SII to CCD
...~~ ~..... ............... ......., ...o,..........
n I I
Ring MCP phosphor screen
S1 ^ . I----
Figure 5-1. Schematic diagram of the vacuum chamber Hg RIID.
Two versions of the mercury RIID were used in these experiments. The first one
evaluated in references [19, 27] and mentioned in Chapter 2 was the sealed-cell Hg RIID.
This compact cell was 5 cm in diameter and 4 cm in width, with a front input widow
diameter of 2.5 cm and side window widths of 0.7 cm. The second one was a semi-sealed
cell vacuum chamber RID as shown in Figure 5-1. The distance between the input
window and the MCP in this case was 40 mm. The end of the sealed quartz tube served
as an input window in this case. Mercury vapor pressure in the cell could be varied from
10-5 to 2 10-3 Torr. In this study, depending on the task, radiations at X2 and X3 were
focused with a cylindrical lens into a rectangular sheet approximately 0.1 0.7 mm in
width or with a spherical lens into a spot with predetermined diameters. The laser beam
with a wavelength of 253.7 nm, which carried the imaging information in these
experiments, could be expanded by a telescope to a 1.5 cm2 spot size.
Experimental Results and Discussion
In our case, the spectral line width of Al dye laser (A V/,ser 30 GHz) was much
broader than the width of the Doppler-broadened Hg atomic lines at room temperature
(A to = 1 GHz) and much of the incident radiation was not capable of being absorbed.
Therefore, the Al beam impinged on the channels of the MCP and created an unwanted
photoelectric effect (PE). In the imaging signal detection mode, the response of the
detector to PE electrons was identical to that of the mercury ions. If a collimated beam
was directed towards the RIID, the image formed on the phosphor screen due to PE
electrons would be identical with the image formed due to selectively ionized mercury
atoms. This PE image was easily observed experimentally when laser beams A2 and 23
were blocked from the RIID. Under normal operating conditions, the image formed
contained both PE and ionization components. Generally speaking, these two components
of the signal could be discerned in the imaging mode only if a rather short pulsed signal
was detected and time-resolved information could be acquired from every pixel of the
RID. The mercury RIID was not designed to acquire time-resolved imaging information
from every pixel in the nanosecond range. In this experimental study, the time-resolved
signal was measured with digital oscilloscope from the whole phosphor screen. Shown in
Figure 5-2 is a time-resolved measurement, typical for non-imaging signal detection
mode of the sealed-cell Hg RIID, where a second peak between the photoelectric and the
resonance ionization peaks was observed . The exact reason for the appearance of the
second peak is still not clear; however, after the experiments it is believed that the most
probable reason is desorption of low-mass ions (Na, Ca, or K) from the surface of a side
window [52, 53]. The flight time of the Hg ion can be varied by changing the distance
between the ionization region and the MCP, as well as by changing the high voltage
applied to the system.
Resonance Ionization signal
Figure 5-2. Typical time-resolved measurement in a sealed-cell Hg RIID .
The duration and temporal behavior of the ionization signal depended on the
penetration depth I of resonance radiation into the volume of RID. As described in
reference , 1 depended on the Al wavelength. The width of the ionization area is
narrowest (or the penetration depth is shortest) when the radiation is tuned exactly into
resonance. The absorption factor (a) of the atomic vapor is described by equation (4-1).
The time-resolved signal in the RIID cell can be described by the first derivative da/dl,
when 1 (distance) is substituted in terms of time t. For mercury vapor at room temperature
(T = 300 K), the number density of atoms, n, is approximately equal to 4.3x 10"3 cm-3, the
cross section a is -6x 10'13 cm2 for the resonance transition, and the Doppler-broadened
linewidth is 1.05 GHz. Approximately 60% of resonance radiation tuned to the center of
the absorption line will be absorbed for path length of 0.05 cm under this condition. If
there is some detuning from resonance, then this path length will be longer. The time
response of the ionization signal will be longer as well in the case of detuning far from
the center of the absorption line. In Figure 5-3, the results of time-resolved signal
calculations based on equation (4-1) are shown . As can be seen, the shape of the
signal depends on the frequency shift and the spectral composition of the detected light:
narrowband or broadband. The signal for broadband radiation (quasi-continuum source)
was normalized with the signal when very narrowband radiation was tuned to the center
of the line.
Here, narrowband radiation is considered as the laser beam with a spectral
linewidth, much smaller than the atomic linewidth at the given experimental condition
(Avlaser << Avatom). And a broadband radiation means that the spectral linewidth of the
laser beam is much broader than the atomic linewidth (Aviaser>> Avatom). Narrowband
radiation 200 MHz shifted from the center of the line can not be resolved as well as for
the case when it is shifted by 500 600 MHz. This is understandable since the first
derivative versus frequency has zero at the center of the line and has a maximum between
500 600 MHz. In Figure 5-3, the ionization signal was plotted vs. distance (in arbitrary
units) from the input window, which represented the absorption pathlength. As was stated
earlier, temporal behavior of the signal would be the same.
Distance from input window (a.u.)
Figure 5-3. Signal profile vs. absorption path length for Hg atoms at room temperature;
1 for narrowband laser (Aviaser s AVatom), 2 for broadband laser (Avia.ser >>
Ava.tom) and four cases when the narrowband laser is frequency-shifted from
Time-Resolved Measurement Results
The linewidth of the dye laser (-30 GHz) was approximately 30 times greater than
the Doppler-broadened mercury line. For such a line, the shape of the time-resolved
signal should not be very dependent on laser frequency shift. To observe and simulate the
frequency-dependent ionization signal by manipulating the width of the ionization region,
an experiment with different thicknesses of the 22 beam in the sealed-cell RIID was
performed (Figure 5-4). In the first case, the thickness of the beam sheet was -0.1 mm
(solid line) and, as predicted, the time response of the ionization signal was determined
by the duration and the shape of the dye laser pulse, which has two partially overlapping
peaks with time of ~20 ns between them. In the second case, the thickness of the beam
sheet was -5 mm. The duration of the ionization signal was longer than in the first
example and supposedly could be controlled by the width of the ionization region
0 100 200
300 400 500
Figure 5-4. Signal dynamics in case of narrow 0.1 mm (solid line) and 5 mm (dashed
line) 22 beam.
In Figure 5-5, the results of another experiment are shown. The 22 laser beam was
focused into a sheet and directed to three locations of the RIID, close to the MCP (1), to
the center of the cell (2) and close to the input window (3). The ion peak close to the
MCP has smallest amplitude, apparently due to absorption of radiation within the cell.
Consistent with theoretical calculation, shown in Figure 5-3, for broadband radiation,
peak 3 must have larger amplitude than peak 2 in the middle, and peak 1 must have the
smallest amplitude. In a first approximation, these data show that the time-resolved
signal can carry useful information about spectral composition of the detected radiation.
It must be emphasized that if the detected radiation has very narrow linewidth, it is not
necessary to have the whole time-resolved signal from the RIID cell. From Figure 5-3, it
can be derived that the information about the frequency of detected radiation can be
obtained by taking only two measurements. For example, from a point close to the input
0 100 200 300 400
5I I I 80
500 600 700 800
Figure 5-5. Spatial scan of 22 beam showing three location of the beam inside the RIID
cell; 1- close to the MCP, 2 in the center of the cell, 3 close to the input
window and from a point further than the region where all curves are crossed. If the
RIID is calibrated in terms of frequency shift, then the ratios of signal amplitude at these
two points will unambiguously determine the frequency shift from resonance for the
detected radiation. The reason for taking two measurement points on opposite sides of the
region where all curves cross is seen from Figure 5-3.
Low-mass Ion Desorption
As we hypothesized above, the second peak shown in Figure 5-2 was due to
desorption of low-mass ions from the surface of side window. To verify this hypothesis
and to study possible ways to eliminate noises related to the low-mass ions, another set of
experiments was carried out in a semi-sealed version of the RIID as depicted in Figure 5-
1. The chamber was under constant evacuation by a turbo-molecular vacuum pump and a
constant supply of mercury atoms was produced from a drop of mercury within the
chamber. The mercury vapor pressure could be varied from 10-6 Torr to the level of
saturated vapor 2x 10-3 Torr by changing the active aperture of a throttle-valve between
the chamber and the turbo pump. To eliminate the effect of absorption of A, radiation by
mercury atoms, a sealed 12 cm long quartz tube with a diameter of 25 mm was placed in
the chamber as shown in Figure 5-1. A Chevron type MCP with P-20 phosphor screen
was used to detect ions. The distance between input window (i.e., the end of the quartz
tube) and the MCP was set to 40 mm.
It was noticed (Figure 5-6) that when A2 was aligned precisely in parallel with input
window, no second peak was observed. The distance from the side windows of the
sealed-cell RIID to the MCP was approximately 1 cm. For the vacuum chamber, this
distance was 16 cm, which was sufficient to eliminate any noise-producing desorbed
ions. Focused and unfocused A2 beams showed clearly distinguishable difference in pulse
duration. As can be seen from Figure 5-6 when the A2 beam was tightly focused with a 60
cm focal length spherical lens, the ionization signal duration was 20 ns. When the 2
beam was directed into the chamber unfocused with a diameter of 6 mm, the pulse
duration was 130 ns.
Figure 5-6. Effect of focusing of 22 beam on the ionization signal.
When the A2 beam was directed to the surface of the input window at small (-5)
angle, a signal from low-mass ions was seen on the oscilloscope. In order to form clearly
distinguishable and repeatable desorbed ions from the surface of the input window, the 22
beam was focused; the results are shown in Figure 5-7. As can be seen, the second peak
in this case is split, showing three discrete peaks compared to the sealed-cell RIID case
(Figure 5-2). By blocking the 2A laser beam, the photoelectric (PE) and the resonance
ionization (RI) peaks were eliminated, which corresponds to the solid-line trace on the
oi* *' .':
S -0.010 .
U) -0.015 RI
.o -0.025 -
0 1000 2000 3000
Figure 5-7. Low-mass ion desorption from the surface of input window with the 22 beam.
Dashed line shows peaks due to photoelectric effect from MCP (PE),
resonance ionization of Hg atoms (RI) and low-mass ion desorption. Solid line
is the case when 2A beam was blocked, showing only peaks due to low-mass
By appropriate alignment of the lasers, it was possible to image all three peaks. In
Figure 5-8a, spatially separated images of three peaks related to three different physical
processes are shown. The positions of all the peaks were precisely where they were
expected to be geometrically. When A, was blocked, the image of only one spot
(Figure 5-8b) due to low-mass ion desorption was seen on the phosphor screen. In order
to further investigate the origin of the peaks that, according to our conclusion, were due
to low-mass ion desorption from the surface of the input window an experiment with a
XeCl excimer laser (308 nm) was performed (Figure 5-9).
Figure 5-8. Imaging the processes in the RIID.
Figure 5-9. Low-mass ion desorption from the surface of input window excimer laser
In this experiment, 6 mm diameter excimer laser beam with 1 mJ a pulse energy was
hitting the surface of the input window at -10 angle. This experiment proved the non-
selective nature of the peaks, indicating that the peaks were not related to resonance
excitation and/or ionization of Hg atoms.
It should be emphasized that we use the term ion desorption. However, the exact
physical mechanism of ion creation [52, 53], when a surface of a solid material is
illuminated by comparatively soft laser radiation is not known. In fact, we never noticed
any traces of a physical damage on the surface of the input window in our case, unlike the
techniques of MALDI and/or LIBS, where the surface is ablated by the laser.
Absolute Time-Resolution of Hg RIID
The next experiment was performed to evaluate the absolute time resolving ability
of the RIID, i.e., to verify if a small disturbance in the distribution of the ion cloud,
created as a result of resonance ionization of mercury atoms, could be time-resolved. The
laser beam of the second step (A2) was expanded into 1 cm wide horizontal sheet by
focusing with a cylindrical lens and directed to the Al laser beam path. Paper strips with
different widths were inserted into the 22 beam partially blocking its radiation. A gap in
the pulse, proportional to the width of the strip was observed. In Figure 5-9, the shape of
the pulse is shown when instead of strip a 0.54 mm diameter pin was inserted into the
beam. The half-width of the gap was 30 ns. This experiment demonstrated not only the
potential time resolution of RIID but also one of the possible applications to detect two
closely placed targets when they are not distinguishable by the imaging device. Ideally, if
time resolution was very high even the shape of the target could be determined. For such
tasks, the Resonance Ionization Imaging Detectors have tremendous advantage since it
practically does not detect any ambient light, responding only to the target illuminated by
1000 2000 3000
Figure 5-9. Absolute time resolution of the RIID; 30 ns gap in the resonance ionization
peak of Hg atoms corresponds to a pin with 0.54 mm diameter.
Time-Resolved Measurements in Cesium RIID
A time-resolved ionization signal in Cs RIID was shown in Figure 4-10. As
expected no photoelectric effect was observed due to radiation at the detection
wavelength of the Cs RIID (852.12 nm). Figure 4-10 was taken when a laser beam
852.12 nm laser beam entered the RIID through the input window, with the 508 nm laser
beam coming into the RIID from the side as a thin sheet of light (so called standard
imaging geometry). In order to further investigate the photoelectric effect, the geometry
of the beams was reversed. The ionization laser beam at 508 nm impinged on the MCP
carrying the image of the test mask and the 852.12 nm laser beam entered into the cell
from the side as a thin sheet of light. Spatially separated photoelectric and resonance
ionization images of the test mask from the phosphor screen are shown in Figure 5-10.
This was achieved by directing the 508 laser beam into the RIID at a small angle.
Figure 5-10. Photoelectric effect from the MCP in the Cs RIID when the laser
wavelengths are 2A = 508 nm and A2 = 852.12 nm.
It should be noted that the spatial resolution of the mage due to PE electrons was
better than the image die to Cs ions. This is because the PE image was formed by
electrons traveling the distance only between the MCP and the phosphor screen. As for
the RI image, in this case the image was due to Cs ions that had to travel to the MCP first.
For better spatial resolution the ionization region should be closer to the MCP. Spatially
separated photoelectric and resonance ionization peaks were time-resolved and the result
is shown in Figure 5-11.
The low-mass ion desorption process discussed for the Hg RIID was not observed
for 508 nm radiation in the Cs RIID, at least at the available laser pulse energies.
However, the presence of PE due to 508 nm radiation indicated that the work function of
the electrons from the MCP corresponded to the laser energy with a wavelength greater
than 508 nm but definitely smaller than 852.12 nm. Therefore, PE noise will not be
present in the Cs RIID with the three-step ionization scheme described in Chapter 3.
500 1000 1500
Figure 5-11. Time-resolved measurement in Cs RIID. PE is a peak due to photoelectric
electrons from the MCP and RI is a peak due to Cs ions.
Dual-Wavelength Time-Resolved Imaging with Cs and Hg
Coincidentally, the same dye laser with Coumarin 500 solution was used for both
Cs and Hg ionization. The first harmonic of the laser at -508 nm was used for ionization
of Cs and the second harmonic of the laser at 253.65 nm was used for Hg excitation. This
provided a ground for a very interesting and intriguing idea of combining the two vapors
in a single RIID. Simultaneous ionization of Cs and Hg atoms was investigated using a
two-step ionization scheme for Cs (852 nm -> 508 nm) and a three-step ionization
scheme for Hg (254 nm -+ 436 nm -+ 436 nm); the wavelengths are rounded off for
simplicity. This unique simultaneous ionization of Cs and Hg atoms was accomplished
using only two dye lasers and a LiF.-F2+ color center laser. The Cs atoms were produced
using the described above alkali metal dispenser. Hg atoms were produced from a small
drop of a mercury metal placed in the vacuum chamber. The experimental setup is shown
in Figure 5-12.
---- ~610 y
CT / 852
<--' 0 c Mask 254
00 --- o 508 '254
50 f 436 Rl 508
Figure 5-12. Setup for dual-wavelength time-resolved imaging with Cs and Hg vapors.
Two Molectron DL II dye lasers were pumped by the excimer laser to generate
radiations with a wavelength of 254, 508 and 436 nm. A third, homemade dye laser,
based on a amplified spontaneous emission at -610 nm was used to pump the LiF.'F2+
color center crystal in order to obtain the detection wavelength of Cs at 852 nm, which
illuminated the cell from the front side, carrying the image of the test mask. The UV
beams for Hg ionization (254 and 436 nm) were counter propagating, parallel to the MCP
to avoid the formation of the photoelectric effect from the MCP. The time-resolved
ionization signal could be measured simultaneously with image acquisition. The images
due to both Cs and Hg atoms are shown in Figure 5-13, along with the diagrams of the
beams forming the images.
Figure 5-13. Dual-wavelength imaging with Cs and Hg vapors.
The image on the top right was taken with all four beams present. The image of the
test mask was due to Cs ionization, and the stripe was formed by resonantly ionized Hg
atoms. By blocking the appropriate laser beams, separate images due to only cesium or
only mercury could be taken (middle and bottom images). The separate images
confirmed that the Cs and Hg vapors were ionized simultaneously, but independently.
This was confirmed by turning off the heating current of the Cs metal dispenser. With no
Cs atoms present, the only image observed was a stripe. The simultaneous ionization of
Cs Hg was time-resolved by monitoring the signal in an oscilloscope and the result is
436 nm 8 rm
a i 254 rnm
.. .... ..............
.I .I....; .
436 nm 2
shown in Figure 5-14. Despite the precautions, a small PE peak due to scattered from the
parts of the RIID UV radiations was observed at t = 0. The ratio of the times of flight of
the Hg and Cs ions could be calculated as follows:
tg M F200.59 1.22
tcs VM V132.905
where, M is a molar mass of the atoms. The experimental value of this ratio was -1.21,
showing a good agreement with the calculations.
0.5 1.0 1.5 2.0 2.5
Figure 5-14. Time-resolved ionization of Cs and Hg atoms.
The flexibility of the resonance ionization imaging detectors has been demonstrated
with non-imaging, time-resolved mode of operation of Hg RIID. This mode of signal
detection has provided improved figures of merit and could be performed simultaneously
with the imaging mode of signal detection. Furthermore, the non-imaging mode of RIID
operation provided a correction method for photoelectric noise during the imaging mode
of operation. This was achieved by temporal resolution of the ionization signal from the
photoelectric signal. A new source of noise, most likely due to low-mass ion desorption
from the walls of the RIID cell, has been discovered. A method to eliminate this noise has
been demonstrated. The feasibility of using time-resolved information to determine the
frequency shift of detected radiation has been discussed.
Also, an enhanced resonance ionization imaging detector with dual-wavelength
imaging capabilities has been demonstrated using Cs and Hg vapors. This version of
RIID with simultaneous imaging and time-resolved modes of signal detection could
broaden and improve the applications of the RIID. For example, it adds valuable imaging
information to the detection of two closely located targets.
FEASIBILITY OF SEALED-CELL CS RIID CONSTRUCTION
In this chapter, the issues related to Cs reactivity are discussed. Cesium is a strong
reducer that generates an explosive reaction with water and is not compatible with many
materials such as aluminum, copper and zinc. Even the alloys of noble metals such as
gold and silver are not compatible with Cs and other alkali metals as will be shown.
However, through a literature search and experimental work, materials that are inert to Cs
have been found. The possibility of sealed-cell Cs RIID construction will be shown in
Cs RIID: First Attempt
Our very preliminary experiments with Cs was performed in a semi-sealed cell
which consisted of a sapphire cylinder confined between Al coated input window and
phosphor screen. The schematic of the experimental setup is shown in Figure 6-1. The 20
mm long sapphire cylinder had a diameter of 50 mm. The sapphire cylinder was
sandwiched between 3 mm thick sapphire input window and phosphor screen. The
phosphor screen was a 6 mm thick lead glass substrate coated with P22G phosphor with
7.5 mg/cm2 surface density. The inside surface of the input window was coated with a 20
nm layer of Al. A 5 nm thick Al layer was deposited on the phosphor surface. A 25 mm
long linear dispenser of Cs, described in Chapter 4 (Figure 4-2), was installed on the
bottom part of the cell; through two 3x0.3 mm2 slits, dispenser terminals were connected
to a current supply. The cell was evacuated through the same slits. The dispenser was
heated by flowing current through it to release vapor of atomic Cs. The sapphire cylinder
between the input window and phosphor screen was heated by a wire coil wrapped
around it to minimize any Cs atom adsorption on the inner surface of the cylinder. The
cell was installed inside the six-way cross chamber depicted in Chapter 5 (Figure 5-1),
which was evacuated down to 105-10.6 torr pressure. A negative high voltage in the
range of 5-8 kV was applied to the input window and phosphor screen was grounded.
1 Screen UR
10 -10 Torr vacuum
Figure 6-1. The schematic of the semi-sealed Cs RIID.
The two-step ionization scheme (2A = 852.12 nm; 22 = 508 nm) mentioned in
Chapter 3 was used for Cs ionization. The laser beams were counter-propagating. The
reason for counter-propagating geometry of the beams was the illumination of the
phosphor screen by both lasers when the beams were directed to the cell from the front
side. The very thin Al coating on the phosphor screen along with the phosphor layer was
not capable of blocking light from reaching the CCD. With lasers counter-propagating
parallel to the phosphor screen, the resonance ionization image was observed. Electrons
created by selective photoionization of Cs were accelerated to the phosphor screen and a
bright stripe was observed. However, the detector was operable for only a few hours. The
aluminum layer on the surface of phosphor screen and input window reacted with Cs
forming some sort of amalgam. As a result, an electric field between the screen and
window could no longer be maintained. Figure 6-2 shows a photograph of the cell taken
from the input window side of the cell. The first thing to note was that the input window
was completely transparent, almost nothing was left from the 20 nm layer of aluminum
coating of the window. The phosphor layer of the screen was also completely exposed.
Furthermore, the phosphor layer of the bottom half part of the screen, which was close to
the dispenser, was destroyed too, indicating the incompatibility of the P20 phosphor with
Cs at high temperatures.
Figure 6-2. Photograph, showing the damage of the phosphor screen due Cs reactivity.
Cs Compatibility: Theoretical Considerations
For the initial construction of a sealed-cell Cs RIID, it was experimentally observed
that there was incompatibility of Al with cesium metal due to the solubility of the metals
in each other. Al was chosen because of its availability and ease to work with. Search in
literature for materials inert to Cs reactivity resulted in Table 6-1, which lists the
compatibility of materials with Cs . Even though the cell temperature in the
experiments was much lower than 450C (AL alloys), Al still reacted with Cs. Therefore,
the temperature, given in the table and up to which the metals are compatible, was
considered as an indication of the degree of reactivity of the alloys with cesium. As can
be seen, noble metals such as gold and silver, which would first come to mind in terms of
inertness, were not compatible at all with alkali metals.
Table 6-1. Compatibility of materials with alkali metals .
Compatible with alkali metal up to ('C)
Material Li Na K Rb and Cs compatibility
Mg alloys n.c." n.c,. 300 300 Metal solubility, oxygen exchange
Al alloys n.c, .350 400 450 Metal solubility
Cu alloys 300 400 400 400 Metal solubility
Ag and its alloys n.c. n.c. n.c. n.c. High metal solubility
Au and its alloys n.c. n.c. n.c. nc. High metal solubility
Zn coatings n.c. n.c. n.c, n.c, High metal solubility
Pb and its alloys n.c. nc. n.c. n.c. Very high metal solubility
Sn and its alloys n.c. n.0. n.c. n.c. Very high metal solubility
Fe 500 700 700 700 Nonmetallic impurities
Low-alloy steels 500 700 700 700 Nonmetallic impurities
Ferritic steels 500 700 700 700 Nonmetallic impurities
High-Cr steels 500 700 700 700 Nonmetallic impurities
Austcnitic steels 450 750 750 750 Nonmetallic impurities
Ni alloys 400 600 600 600 Flow velocity
Mo alloys 1000 1000 1000 1000 Nonmetallic impurities
W alloys 1000 1000 1000 1000 Nonmetallic impurities
Ti alloys 700 700 700 700 Nonmetallic impurities
Zr alloys 700 700 700 700 Nonmetallic impurities
Y alloys 700 700 700 700 Nonmetallic impurities
Nb alloys 700 700 700 700 Nonmetallic impurities
Ta alloys 700 700 700 700 Nonmetallic impurities
Sintered AlO3 350 500 500 500 Thermomechanical action
stab. ZrO2/CaO 350 350 350 350 Intergranular corrosion
stab. ThO2/Y20, 400 550 550 550 Intergranular corrosion
Glass n.c. 250 250 250 Chemical reaction
UO2 750 Excess of oxygen
UC 750 Nonmetallic impurities
Sn.c. = not compatible.
According to Table 1-1, the most compatible metals would be W and Mo, followed
by Ti, Zr, etc. The more readily materials Ni and Cr also seemed to be good candidates.
Based on this table, a few elements were selected as a potential metals for coating the
input window and the phosphor screen. Penetration of these materials by high energy
electrons was simulated using Casino V2.42 (monte Carlo Simulation of electron
Trajectory IN sOlids) software package (http://www.gel.usherb.ca/casino/). Calculations
for different materials and different electron energies were performed. Figure 6-3 shows
the simulation results for different materials bombarded with 10 keV electrons.
100 100- *
J o \:J80- U Ni0 5keV
C Ti i i 10 keV
0o- 60. .* 15keV
tB 0 \ 60 *
w 40- 40 -
a a |
S20- 20 .
Z 0- z 0 *
0 200 400 600 800 1000 0 200 400 6o00 800oo 1000 1200
Penetration depth (nm) Penetration depth (nm)
Figure 6-3. Penetration depth of electrons for different materials and electron energies.
Penetration depth decreased with the increase of the atomic number of elements.
From Figure 6-3a, it can be seen that the Ti is easily penetrated by electrons. Therefore Ti
can be used to coat the phosphor screen with a thick layer to make it optically opaque, but
transparent to electrons which will reach the phosphor layer and cause a luminescence.
The coating of the input window, on the contrary, should be as thin as possible to
transmit the most of the photons to be detected. Again, Ti is well-suited for the purpose;
the bombardment of Ti metal by electrons with different energies was simulated (Figure
Cs Compatibility: Experimental Verification
Two custom-made (Del Mar Venture, CA) luminescent screens coated with a layer
of Ti metal were experimentally evaluated. One of the screens was a conventional P20
phosphor coated with 200 nm Ti. The second screen was a novel luminescent YAG
crystal which was coated with 80 nm of Ti. YAG crystals have proven to be highly
resistant to alkali metals at high temperatures . Figure 6-4 show the experimental
setup used for evaluation of the luminescent screens.
Polished E Luminescent
Al disk Screen Oscilloscope
308 nm _
Figure 6-4. Schematic diagram of the setup for luminescent screen test.
A cell was formed with 3 mm thick well polished Al disk electrode and one of the
luminescent screens. Both screens were tested in exactly same way. Negative voltage up
to -18 kV was applied to the Al electrode. The luminescent screen was grounded through
a 50 Q resistor and the current, i.e. the number of electrons per second striking the screen
was monitored by an oscilloscope. Simultaneously, the image from the screen was
captured with a CCD camera. Photoelectric electrons were produced by illuminating the
Al electrode with the excimer laser beam (A = 308 nm). The excimer laser beam with a
pulse energy of 300 .tJ and a diameter of 3 mm impinged the center of the Al electrode at
30 angle. Measurements were performed at 7. 10-6 Torr pressure. Figure 6-5 shows signal
measured from the screens with the oscilloscope at -12 kV applied to the Al electrode.
0.5. YAG screen tested with excimer P20 screen tested with excimer
E5 -15 E -20- FWHM =25-30 ns
2 FWHM = 25-30 ns
c -2.0- *SI
U i -30-
-2.5- Ma m Max =4.71 mV
Max = 3.75 mV
-0.5 0,0 05 1.0 1.5 20 -0.5 0.0 05 10 1 5 20
Time (ps) Time (1ps)
Figure 6-5. Current due to electrons striking the luminescent screens.
The calculated number of electrons for both screens was approximately 1.5x 107. At
approximately -1 kV applied to the Al electrode, the signal amplitude reached the level
shown in Figure 6-5 and further increase of the voltage did not noticeably change either
the time-resolved signal intensity, or the time of flight of the electrons. More informative
results were obtained from the measurements of the image intensities. Figure 6-6 shows
series of images obtained with the P20 and YAG screens at different voltages. Each
image is an average of twelve individual image acquisitions. With increase in the voltage
the images became brighter, indicating an increase in the number of electrons penetrating
the metal coating. The experimental results were in good correlation with the simulation
results. At 6 kV a very faint image was obtained; according to Figure 6-3b, a very small
percent of electrons can penetrate 200 nm Ti layer with 5 keV energy. At energies greater
than 10 keV, almost all of the electrons can be expected to penetrate the 200 nm metal
layer according to Figure 6-3, which could be concluded from the plot of image
intensities vs. accelerating voltage in Figure 6-7. In general, P20 phosphor was a more
efficient luminescent substance.
6kV 10 kV
14 kV 18 kV
Figure 6-6. Series of images at different accelerating voltages. Above row P20 phosphor,
bottom row monocrystalline YAG.
* P20 coated with 200 nm Ti
* YAG coated with 80 nm Ti
S 0 *
14 16 18
Figure 6-7. Image intensity vs. accelerating voltage.
As for the optical parts of the cell, sapphire was chosen due to its known resistance
to alkali metals. In the work by Bouchiat et al., electrical conductivity of glass and
sapphire cells exposed to cesium vapors are compared . The walls of cells made of
different glasses became electrically conductive when exposed to Cs vapors (typical
resistances approximately 103 Q) whereas monocrystalline sapphire cells maintained a
resistance on the order of 109 Q in similar conditions. The measurements were done by
capacitively coupling the outer surface of the cells to the inner surface . They
concluded that in the case of glass cells, the conductance was due to Cs atoms adsorbed
on the inner walls of the cells. For the sapphire cell, they could see no indication of wall
surface conductivity, attributing the conductance to collisional processes occurring
between the Cs atoms, especially at number densities of more than 1014 cm-3. The
conclusion was made based on the cubic temperature dependence of surface conductance
of the inner cell walls. The adsorption energy for alkali atom glass wall was estimated
to be 0.66 eV . In the work by De Freitas et al., the adsorption-desorption of cesium
in a fused quartz cell was estimated 0.2 eV and was found to weakly depend on
Figure 6-8. A photograph of the sapphire Cs cell with Nb electrodes.
An experiment with a sapphire cell with niobium electrodes inside was performed.
A photograph of the cell is shown in Figure 6-8. The cell was approximately 8 cm long
cylinder with a diameter of 15 mm. The distance between electrodes was -15 mm. The
cell was evacuated and sealed with solid Cs inside. Niobium is a metal, compatible with
Cs according to Table 6-1. A high voltage up to 20 kV was applied to the electrodes to
the electrodes and VA characteristics of the cell was obtained. Figure 6-9 shows the
experimental results, the dependence of cell conductivity and current flowing through the
cell vs. applied high voltage.
I04- % 24-
0.3- Ill 2.2-.
0.2 B 2,
021 20 *. S 5
2 4 6 8 10 12 14 16 18 20 22 2 4 6 18 0 12 14 16 18 20 22
HV (kV) HV (kV)
Figure 6-9. VA characteristics of the sapphire Cs cell with Nb electrodes; a) conductance,
b) current vs. high voltage.
At room temperature, metallic Cs adsorbed on the inner walls of the cell which can
be seen by eye (inner walls were coated with a thin, golden-color film). The cell was
heated as a result of the applied voltage and the current flowing through the cell. If the
applied voltage is considered as an indication of the temperature of the cell, then Figure
6-9a shows a dependence very similar to the temperature-dependence of the conductance
discussed in reference , where the conductance in the sapphire was attributed to
collisions between Cs atoms in gas phase. However, as shown in Figure 6-9b, the drastic
decrease of the current with the voltage could be attributed to desorption of Cs atoms
from the walls, which could lead to the conclusion that the atoms adsorbed on the inner
surface of the cell was responsible for the conductance of the cell. As the results reported
in references [56, 57] were most likely interpretation of the results, the decrease of the
current with the voltage in our experiments was not clearly understood.
The current flowing through system dropped and stayed constant at a level of about
2 mA (see Figure 9-6b). The resistance of the cell increased from a few tens of kQ up to
-9 MQ due to the heat from the current flowing through the cell. At 5 kV applied
voltage, the temperature of the cell was measured to be -200 C. Two important
conclusions from the experiment are: (1) Cs metal adsorbed on the walls Cs metal could
not be completely evaporated even from sapphire; (2) however, intense electric field (>13
kV/cm) was realized in a sapphire cell filled with Cs atoms without producing a
discharge. The reasons that the wall conductivity could not be eliminated from sapphire
even at high temperatures could be due to surface and composition quality of the sapphire
used for the construction of the cell.
In this chapter, the possibility of constructing a sealed-cell Cs RIID construction
using sapphire and metals that are compatible with Cs has been shown. The chemical
inertness of the sapphire to Cs was confirmed in experiments with the sapphire cell and
Cs metal inside the cell. Sapphire, heated up to 200 C, did now show any signs of
reaction with Cs. In glasses heated up to such high temperatures, an irreversible reaction
between silicon oxide and alkali metal will occur and glasses become discolored and
brittle. The high electric field up to 13 kV/cm, produced in the sapphire cell with Cs
metal, did not cause an electric discharge and breakdown. For the metallic parts and
coatings of the windows of the potential sealed-cell Cs RIID, Ti was chosen among a few
other suitable metals based on a literature search. The electron penetration of Ti was
theoretically and experimentally studied by evaluating luminescent screens with P20
phosphor and YAG crystals.
BUFFER GAS RESONANCE IONZATION IMAGING DETECTOR WITH
AVALANCHE AMPLIFICATION OF SIGNAL
It has been shown that the high spectral resolution, which comes from the
narrowband absorption in atomic vapor combined with the efficiency of laser ionization,
furnishes the detector with superior figures of merit compared to conventional spectral
imaging techniques. However, there are certain applications where detectors with broad
bandpass are required. For example, Raman bands of molecules are typically -30 GHz or
broader. Narrowband RIID responds to only a small fraction of the Raman signal.
Despite this intrinsic disadvantage of the RIID for the detection of Raman scatter, based
on the sensitivity and high luminosity, the RIID still can be more efficient compared to
conventional Raman imagers. In addition there are ways of broadening the spectral
response of the RIID, while keeping the flexibility and efficiency. One way involves
collisional broadening of the atomic lines by addition of a buffer gas.
In reference , sub-Doppler absorption spectroscopy of Cs atoms was studied in
a sub-micron thin cell. Thin cells are very attractive idea for the design of RIID, because
they not only allow sub-Doppler high spectral resolution, but also provide improved
spatial resolution. In fact, the ultimate imaging with atomic vapors, both in terms of
spectral and spatial resolution, would be in atomic sheets.
In this chapter is discussed a completely different concept of a detector based on
atomic vapors. The novel detector is based on avalanche amplification and electrodeless
optical detection of resonance ionization signal and could address the question of line
broadening in a buffer gas and a very thin cell could be designed. The cell contains no
MCP or phosphor screen. The image is formed by optical emission of the buffer gas
excited by collisions with high energetic electrons accelerated in high frequency electric
field. Electrons are created as a result of resonance ionization of the atoms and form an
avalanche effect when accelerated in high electric field. This detector takes advantage of
the selective nature of the resonance ionization method and the sensitivity of particle
detectors, which are well developed and applied in the field of elementary particle
physics. This chapter discusses the design and the fundamental idea behind the buffer gas
RIID, along with the preliminary work done in this direction.
Buffer Gas RIID: Design
The photoionization detection of atoms in a buffer gases has been studied in many
experiments. The first successful experiment on the detection of single atoms of Cs by
means of two-step photoionization in buffer gases was shown by Hurst et al. . Cs
atoms formed by the molecular fission of 252Cf were photoionized and the resulting
photoelectrons drifted into a proportional counter and were detected. The first optical
emission detection of resonance ionization signal of Hg atoms in a buffer gas (Ne) was
reported by Matveev et al. .
As atoms are detected in a buffer gas, the time of interaction of atoms with the laser
beam, i.e. the residence time of atoms in an exposed region is defined by their diffusion
in the gas. The time of diffusion of atoms from a spherical region with a diameter a can
be expressed :
a2 *6 (7-1)
where D (cm 2/s) is the diffusion coefficient at a pressure of 760 Torr and P is the buffer
gas pressure (Torr). In order that all the atoms in this region are probed, the laser pulse
repetition rate must satisfy the condition:
f =-- (7-2)
for atmospheric pressures time of diffusion changes approximately from 2.5 ms to
250 ms as the diameter increases from 0.1 to 1 cm.
Electrons and ions produced by the ionization of Cs atoms diffuse by multiple
collisions with the buffer gas atoms. In the absence of an electric field, the diffusion
distribution is Gaussian and the mean free path of diffusion, I is :
-(7 ~ (7-3)
where, n is the number of atoms per unit volume (cm3) and oc() is the energy dependent
collision cross section (cm2). It is desirable to have rather high pressures in the cell (of
the order of 1 atm or higher), which leads to more efficient collisional excitation of buffer
gas atoms and improvement in the spatial resolution of the cell by limiting the free escape
of ions from the ionization region.
If the charge carriers (ions and electrons) are exposed to an electric field, an
ordered drift along the field will be superimposed over chaotic diffusion. The drift
velocity is directly proportional to the applied electric field and inversely proportional to
the pressure. Therefore, in terms of charge acceleration, high pressure may seem to have
a negative influence. Obviously, there is an optimal electric field strength for a given
Front view Pon
0/O OS 0 00 0 \
0 0 OSI @Supply
k3 I Pin electrodes
Figure 7-1. The schematic diagram of the buffer gas RIID.
The design of the buffer gas RIID and the principle of operation with three-step
ionization are shown in the Figure 7-1. Again, as described before, the image carrier
beam at the detection wavelength (A,) enters through the input window and excites the
atoms in the cell. Laser beams at A12 and A3? further excite and ionize the atoms. The
electrons formed are accelerated towards the very thin wire (pin) electrodes, which are
installed outside of the cell. There is no direct contact between the atoms and the
electrodes. However, the thickness of the cell wall separating the electrodes should be as
small as possible. Upon acceleration, the electrons form an avalanche effect, as shown in
Figure 7-2. The avalanche electrons will collisionally excite the buffer gas atoms and as a
result, a gas discharge localized to the point of ionization in the vicinity of the electrode
is formed (Figure 7-1).
In order to keep the signal proportional to the primary ionization, the avalanche
amplification factor should match the linear dynamic range of the detector. The lower
limit of this range corresponds to the field strength at which electron multiplication starts.
\ gas ions
Figure 7-2. Formation of an avalanche effect .
The upper limit is the starting point of new secondary and tertiary avalanches, which will
merge with the primary avalanche and no longer depend on the primary ionization. The
secondary and tertiary avalanches are due to photoelectrons produced by energetic UV
photons. At very high electric field strengths, the accelerated electrons also liberate
electrons from inner shells of gas atoms. This leads to de-excitation of excited gas atoms
by UV photon emission. These energetic UV photons will produce further electrons by
ionizing the gas atoms. This problem can be solved by addition of a quenching gas which
will absorb energetic UV photons, usually an organic gas. Most organic compounds in
the hydrocarbon and alcohol families are efficient in absorbing photons in the relevant
ranges. The molecules dissipate the excess energy by elastic collisions.
One should keep in mind, that not only the strength of the applied electric field is
important, but also the form of the field is critical. When the applied electric field is
continuous in time (Figure 7-3a), the signal will have a threshold character. In other
words, as the number of produced charged particles increase, they begin to have an effect
on the external field and saturation occurs. Signal amplitude is then no longer
proportional on the primary ionization (corresponds to the Isa, Figure 7-4). However, if
the field is applied in a form of short pulses (Figure 7-3b) with a duration of r, the signal
amplitude will then be different for different numbers of primary ionization products
(Figure 7-4: I,2,3).
ANWMkkAA PAMI MJVi
t r t
Figure 7-3. Continuous and pulsed high frequency electric field.
Figure 7-4. Signal behavior vs pulse duration.
Other advantages of a pulsed electric field are that it allows one to apply a stronger
electric field without entering the saturation stage, and it gives time for ions and electrons
to recombine. In a continuous electric field, the ion/electron pair does not have much of a
chance to recombine. This is exaggerated at high field strengths, which can lead to
accumulation of charges on the cell walls. Thus, electric field investigations include
optimization of the pulse duration to produce a discharge whose emission intensity is
proportional to the initial number of electrons created by resonance ionization process of
the atoms. The repetition rate of the electric field must match with the repetition rate of
Figure 7-5 illustrates the principle of image acquisition is shown in the buffer gas
RIID. The dichroic mirror, transparent for the image carrier beam at the detection
wavelength, is placed at 45. The gas discharge emission in is reflected from the mirror
and collected by a CCD camera.
X Dichroic mirror
Figure 7-5. The schematic of the image acquisition in buffer gas cell.
Buffer Gas RIID: Experimental Results
The experiments were performed in a cell, the cross-sectional diagram of which is
shown in Figure 7-6. The cell was constructed with a stainless steel body and a quartz
front window. Electrodes were 0.3 mm copper wires inserted into small capillary tubes.
Approximately 200 capillaries were bundled and tightly packed into a glass tubing with a
diameter of 15 mm. The glass tubing with electrodes then was inserted into the opening
in the stainless steel body of the cell. To prevent the direct contact of the electrodes with
the gas molecules and atoms, a thin layer of transparent vacuum UV glue was spread onto
the surface of the bundle of the electrodes and cured with the excimer laser. The cell
contained Ne gas and Hg atoms from a drop of mercury inside the cell.
Quartz ........... Stainless
... .. ...., ,
0.3 mm wire
Figure 7-6. Buffer gas RIID cell and the Ne gas discharge.
A continuous oscillating electric field with a frequency of 30 kHz and an amplitude
of 12 kV was applied to the electrodes. The pressure inside the chamber was varied and
images of the Ne gas discharge were manually taken with a digital camera. Figure 7-7
shows a series of images taken at different pressures of the buffer gas.
Figure 7-7. Ne gas discharge at pressures (left to right, in Torr): 15, 30, 120, 240.
The Ne gas discharge was initiated randomly and did not respond to the resonance
ionization of Hg atoms. As stated above, the applied field parameters are critical and a
pulsed field with certain frequency and amplitude must be applied to produce a controlled
discharge proportional to the selective ionization of atoms. Unfortunately, it was not
possible to construct such a power supply due to technical difficultness. However, the
formation of a gas discharge, localized to the tip of the electrode and the response of the
detector to the pressure change observed in the experiments are promising. The concept
of the buffer gas RIID with avalanche amplification and optical detection of resonance
photoionization of the atoms could be very interesting continuation of RIID development.
Avalanche Amplification of Cs Ionization Signal in Flame
The avalanche effect was studied in flame with two major goals: (i) to better
understand the conditions for avalanche with possible application of the results to the
design of the RIID; (ii) to apply the effect of avalanche amplification of signal to the
fundamental laser-enhanced ionization (LEI) technique. LEI is well known as one of the
most sensitive elemental analysis techniques, which takes advantage of the spectrally
selective nature of laser excitation of analyte atoms and of thermal environment at which
atoms are produced, leading to their effective collisional ionization . Ionization of Cs
atoms, enhanced by two-step excitation, was detected in hydrogen and propane flames.
This work not only proved the feasibility of the idea of avalanche amplification in flame
environment, but also by employing the effect of avalanche amplification of electrons, a
limit of detection (30 fg/ml) for laser-enhanced ionization (LEI) technique was achieved
. The most often used atomic reservoir for LEI is a flame, and the best reported limits
of detections (LOD) are in the range of tens of pg/ml . There are three main factors
limiting the LOD of atoms in a flame when using the LEI method, namely: (i)
fluctuations of the background current between the electrodes which detect the signal [63,
64], (ii) non-selective ionization of atoms and molecules in the flame and, (iii) Johnson
noise of the input resistance of the signal preamplifier . As shown by Matveev and
Omenetto [63, 64], there is a several orders magnitude gap between experimental limits
of detection and the theoretical values which can be calculated by taking into account
these main sources of noise. As suggested by Matveev , one of the most promising
ways to bridge this gap is to take advantage of the avalanche amplification of the signal
as is done in the field of nuclear physics when high-energy nuclear particles are detected
with gas-filled counters . Previously, the avalanche effect has been obtained for
mercury atoms and an inert gas contained in a quartz cell at atmospheric pressure with
argon and P-10 (10% methane in argon) gases as a buffer . The resulting detection
limit was approximately 15 Hg atoms in the volume illuminated by the laser radiation. In
this work the possibility of the effect of proportional avalanche amplification of an LEI
signal in a flame reservoir has been experimentally demonstrated.
The experimental setup (Figure 7-8) consists of two Molectron DL-II dye lasers,
pumped by the XeCl excimer laser (LPX-1 lOi, Lambda Physik, Acton, MA), a laboratory
constructed burner supporting propane and hydrogen fueled flames, and electronics for
signal measurement. Cesium atoms were excited into the 8d 2Ds/2 state by tuning the dye
lasers to the 6s2S/2 -- 6p2P3/2 (// = 852.12 nm), and 6p2p32 -> 8d2Ds/2 =2 = 621.3 nm)
transitions, respectively. Excited Cs atoms were collisionally ionized in the flame, and
the ionization signal was detected with a 0.4 mm diameter Pt wire (12 mm long)
electrode set 4-5 cm above the burner head. A negative voltage of up to -5 kV was
applied to a planar (screen) electrode consisting of a stainless steel wire mesh (1 x3 cm2).
The planar electrode was positioned just outside the flame (barely touching the flame) at
the same height as the probe. Introduction of this planar electrode was necessary in order
to prevent electrons originating in the primary combustion zone of the flame from
reaching the probe and contributing to the background noise. The counter-propagating
laser beams were coaxial with the Pt probe and had a diameter of 3 mm. The dimension
of the probe volume, defined by the dimensions of the exciting laser beams was 3x3x12
mm3. The distance between the 0.4 mm Pt probe and the mesh electrode was -10 mm.
/ --] ,/ ~]
7.5 MQ 15OpF -
0-5 kV Preamplifier Oscilloscope
Figure 7-8. Schematic diagram of the experimental setup
A high-voltage power supply (Series 205, Bertan Associates, Inc.) was connected
to the planar electrode and the Pt probe was connected to the signal detection system,
consisting of a laboratory-built, charge-sensitive, low-noise preamplifier and a digital
oscilloscope (Tektronix, TDS 620A). By changing the polarity of the applied voltage,
either the ions or electrons could be collected with the Pt probe. Radio frequency
interference noise was minimized by placing the burner and probe inside a shielded metal
box. The internal 20 MHz lowpass filter of the oscilloscope was used to further reduce
Results and Discussion
A preliminary set of experiments was performed in a hydrogen/air flame with a
detection scheme using an electric field formed between the Pt wire probe and the burner
head. A positive high voltage was applied to the probe while the burner head was
grounded. The distance between the probe and burner head was 4 cm. The two-electrode
scheme used was characterized by poor S/N due to the presence of high background
noise. The source of this noise was identified as the electrons produced in the primary
combustion zone of the flame, which is located close to the burner head. In the strong
electric field between the probe and the burner head, these non-selective electrons can
easily reach the probe, causing a high level of background noise. Due to this high
background, it was not possible to produce a stable avalanche effect because all electrons,
no matter how they originated, experienced the same avalanche effect under these
conditions. At rather moderate voltages (-3 kV), electrical breakdown between the probe
and burner head occurred. As mentioned above, the third planar electrode was
introduced to overcome this problem. Both the probe and the burner head were kept
positive with respect to the planar electrode, which was at a negative potential. As a
result, selective electrons (produced by laser-enhanced ionization of Cs atoms) were
accelerated toward the probe and detected. All electrons from the primary combustion
zone of the flame were removed, via the burner head, to ground. Thus, the three-
electrode scheme formed or produced a "silent" (low thermal electron) zone in the upper
part of the flame providing a favorable environment for avalanche amplification of the
selective, laser-induced electrons.
In a weakly ionized plasma such as a flame, very often the electric field between
the electrodes is not uniform and is modified by a positive ion space charge or "sheath",
adjacent to the cathode [61, 66]. When the concentration of ions is high enough and
sufficiently low voltages are applied between electrodes, the electric field strength has a
maximum at the cathode, decreasing linearly with the distance from the cathode to zero at
the edge of the sheath. However, when higher voltages are applied, a condition of
"electrical saturation" with a non-zero field and a positive space charge at all positions in
the flame, is observed . At higher voltages, the sheath can extend to the primary
combustion zone with its non-uniform temperature and composition. As discussed above,
the electrons produced in the primary combustion zone of the flame were the main source
of noise and the cause of breakdown in our experiments and was eliminated by using the
three-electrode scheme. In the work of Travis, the sheath length was evaluated
experimentally for Na atom LEI in an air/acetylene flame using a 1 ptg/ml solution .
The field was formed between a 5 cm single slot burner head and a 0.635 cm diameter
stainless steel tube. At applied voltages of up to the 1200 V, the sheath length was 11-13
mm. In the experiments reported here, up to 5 kV was applied to electrodes separated by
only 10 mm. Moreover, the very dilute Cs solutions (100 pg/ml) aspirated into the low
background hydrogen/air flame assured the presence of high electric field at all points
between the electrodes. The small diameter (0.4 mm) of the Pt wire probe also should be
emphasized because that is the key to having a very strong electric field, which is
essential for avalanche amplification. The field map in the three-electrode scheme is
complex, but a field between the Pt wire probe and the planar electrode could be
approximated as an idealized field between a cylinder and a plane . The field
intensity is inversely proportional to the wire probe diameter.
96-a) 3 b)
1.88 / *
A 200- 2/
1.84- 'Ai E 1 "....50. CS(100pg/ml)
/ r A Li (1 g/ml) /
A_ 8o Cs (100 pg/ml)
S' 7 Li (1 plg/ml) 500o.
0.0 04 08 12 1.6 20 20 2.4 2.8 3.2 3.6 4.0
HV, kV HV, kV
Figure 7-9. LEI signal vs probe voltage: a) ion, and (b) electron signals for Cs and Li
Evidence of avalanche multiplication is seen from the comparison of the electron
and ion signals obtained with 100 pg/ml Cs and 1 Vtg/ml Li solutions under the same
experimental conditions (Figure 7-9 a, b). Ions were detected by switching the polarity of
the planar electrode to positive so that both the probe and the burner head were negative
relative to the planar electrode. When ions were detected, the signal increased linearly
with the electrode voltage until it reached a plateau showing no avalanche behavior. In
contrast, when electrons were detected, the exponential increase in signal as a function of
voltage shows the typical shape of an avalanche process, continuing until breakdown. In
the strong electric field between the probe and the burner head, electrons can gain enough
kinetic energy to liberate electrons from inner shells of gas atoms and molecules leading
to de-excitation by UV-photon emission. These energetic UV-photons will produce
further electrons by ionizing the gas atoms. This results in a transition from proportional
avalanche to a streamer followed by electrical breakdown. This transition process is well
known and has been extensively studied and described in the nuclear particle detector
literature. The breakdown time is very small, of the order of a few tens of nanoseconds
. In this particular experiment, the breakdown occurred at voltages close to 4 kV. The
breakdown voltage varied with the distance between the electrodes and the solution
concentration aspirated into the flame. The planar electrode should be kept just outside
the flame, but as close as possible in order to provide the maximum possible electric
field. If the planar electrode is immersed in the flame, then electrons which are thermally
emitted from the surface of the electrode will reach the Pt wire probe, increasing the
noise and the risk of breakdown. A maximum avalanche amplification factor of
approximately 105 was observed. The avalanche amplification factor was estimated as the
ratio of the signal obtained just before breakdown (at 3.8 kV) to the smallest detected
signal (at 2.4 kV) detectable when the avalanche process was absent. The experiments
were repeated with a 1 gg/ml Li solution in order to prove that the avalanche effect is not
due to any unique properties of Cs. The ionization rate of Li atoms in the flame was
enhanced by tuning the dye laser to the 2d2D -> 3/F resonance transition of Li (610 nm).
The behavior of the ion and electron signals was exactly the same as in the case of Cs.
Three types of flames were studied: hydrogen/air, hydrogen/20% oxygen/argon and
propane/air. The hydrogen/air flame gave the highest proportional ionization signal
compared to the other two flames (Figure 7-10). The signal in hydrogen/air flame was
approximately two orders of magnitude larger than that in the propane/air flame, and one
order of magnitude larger than that in the hydrogen/oxygen/argon flame. As known from
the literature, all three flames are expected to provide similar temperatures, and
consequently similar thermal ionization efficiencies. As can be seen in Figure 7-10, the
three flames exhibited largely different signal amplitudes at the critical voltage just below
the breakdown point. This different behavior reflects the different chemical composition
of the flame mixtures and the resulting different species responsible for the avalanche
-m- Hydrogen + Air
2.0 -*- Hydrogen + (Ar/O2)
A Propane + Air
> 1.5 /
0.5 U- /
0.0 *-- -- *- A A
0.5 1.0 1.5 2.0 2.5 3.0
Figure 7-10. Signal vs. voltage behavior in flames showing the occurrence of the
avalanche effect. Cs: 1 [tg/ml.
effect. Since the signal decreased substantially when argon was substituted for nitrogen,
i.e., when a H2 /02 /Ar flame was substituted for the hydrogen/air flame, it can be
concluded that nitrogen is the most probable source of avalanche electrons. This
conclusion is supported by the fact that (i) its ionization potential (15.5 eV) is slightly
less than that of Ar (15.7 eV); (ii) nitrogen molecules in Hz flames are usually more
efficient collisional partners, compared to argon atoms, and the ionization rates are
therefore higher in N2 than in Ar; and, (iii) Ar-diluted flames usually exhibit lower
temperatures than the corresponding N2-diluted flames . The poorest results were
obtained with the use of a propane-air flame, which can be attributed to the avalanche
quenching properties of the organic molecules in the flame. Additionally, hydrocarbon-
fueled flames are known to have a much higher background current compared to
hydrogen flames . This background current has to be kept as low as possible for both
conventional LEI and avalanche-LEI, since it will ultimately determine the detection
limit and the selectivity of the technique.
F_ -36kW] 1200-
> Boo/ BO- . .
0" 001 U)
1E-3- 4 0
1E-4 -, ., ., . , -, .. .. .. ,.,
i.3 001 0o i 10 100 1000 10000 10oo6ooo 3.4 3.6 3.8 40 42 44 46 48 50 52
Concentration, pg/ml HV, kV
Figure 7-11. a) Calibration curves for Cs at two different electrode voltages in the H2 Air
flame b) Signal-to-background vs applied voltage. Cs: 100 pg/ml
Once the effect of avalanche amplification of the LEI signal was confirmed, the
limit of detection of the system was estimated in the hydrogen/air flame. The practical
LOD achieved in this work was 100 pg/ml (Figure 7-1 la), limited by memory effects of
the burner/nebulizer system. As mentioned above, the breakdown voltage varied with the
distance between electrodes. At the optimal distance between electrodes, with a voltage
of up to 5 kV, the signal-to-noise characteristics of the system were studied and the
highest S/B was observed at 4.4 kV (Figure 7-1 lb). The drop of the signal-to-background
ratio at higher voltages indicates that when the avalanche amplification is very high, the
system is getting close to the transition from proportional avalanche to a streamer. The
system was further optimized in terms of flow rate of the flame gases in order to obtain
the maximum possible signal at 4.4 kV. Five measurements of the signal and
background were carefully taken at the optimal voltage. Background measurements were
taken with no lasers and aspirating only de-ionized water into the flame, which was used
as a solvent for all solutions. Each measurement was an average of 1200 pulses at 20 Hz.
Sample measurements of averaged signal and noise are shown in Figure 7-12.
600- a)01 b)
50 0 -0 .2
100- CO) -0.5-
0 kY 06-
-0.05 0.00 0.05 0.10 0.15 0.20 -50 0 50 100 150 200
Time (pis) Time (ps)
Figure 7-12. Temporal variation of (a) signal and (b) noise; Cs: 100 pg/ml.
For each measurement, the signal was calculated as the area under the curve, calculated
as the sum of intensities of 8 points (from -1 to 2.5 pts). For the noise calculation, the
entire time scale (500 points) was divided into 62 sections, each with 8 points. The area
under each 8 point segment was then calculated and the standard deviation of those 62
areas was taken as the noise. The signal-to-noise ratio was calculated based on 3y
criteria and found to be more than 104. Based on this S/N, an extrapolated, limit of
detection of 30 fg/ml was estimated (average of 5 measurements). For comparison, the
LOD was 40 fg/ml when calculated using the peak height. Table 7-1 compares the LODs
for Cs obtained by selected analytical techniques as well as previously obtained LODs
Table 7-1. LOD comparison.
Element Method LOD (ppt, pg/ml)
Cs AAS* 1000
Cs ICP-MS 1
Cs LEI in Acetylene/Air  2
Li LEI in Acetylene/Air  0.3
Cs Experiment 0.03
* The data from the LOD table of VHG LABS (www.vhglabs.com)
In this chapter, the concept of electrodeless optical detection of laser resonance
ionization signal with avalanche amplification has been discussed. Based on such
concept, the design of the resonance ionization imaging detector with a buffer gas has
been illustrated and the experimental results proving the concept has been shown. The
buffer gas RIID could be especially useful in applications requiring a broadband but
sensitive imaging detector, such as chemical and biological Raman imaging
spectroscopy. Also possible is construction of very thin imaging cells, with high spectral
and spatial resolution.
The avalanche effect was applied to the fundamental laser-enhanced ionization
technique. The LEI signal of Cs atoms in a hydrogen/air flame was detected using
avalanche amplification. Three types of flames were studied: hydrogen/air,
hydrogen/20% oxygen/argon and propane/air. The hydrogen/air flame gave the highest
signal-to-noise ratio, showing that nitrogen molecules might be the best environment for
avalanche amplification of the LEI signals in flames. The LOD of 30 fg/ml obtained in
the experiments was approximately 70 times lower than the best previously obtained
LOD for Cs in flame LEI (2 pg/ml) and 10 times lower than the lowest LEI detection
limit ever reported (Li, 0.3 pg/ml).
CONCLUSIONS AND FUTURE WORK
The resonance ionization imaging detector with Cs vapors has been constructed and
evaluated. The detector, based on narrowband absorption of signal radiation in an atomic
medium and subsequent laser stepwise ionization of the excited atoms, offers
unsurpassed performance as a two-dimensional imaging detector compared to existing
ones. The superiority of the detector is determined by its high Luminosity Resolving
power (LR) product, which is orders of magnitude greater than any other spectroscopic
imaging technique. The cesium RIID is the second existing RIID, in addition to the
mercury RIID developed previously in our laboratory. Cs is one of the few elements that
is suitable for the construction of the RIID and can offer certain advantages, such as high
sensitivity due to strong signal transition of Cs atoms. Also, the infrared detection
wavelength of Cs RIID (852 nm) allows important biological applications of the detector,
because radiation in this spectral range is safe for biological tissues and has good
penetration depth. Two-step (2/ = 852, 22 = 508 nm) and three-step (2A = 852, A2 = 917,
23 = 1064 nm) ionization schemes of Cs have been investigated using various lasers,
including XeCl excimer, Nd:YAG, dye and color-center lasers. The efficiency of the
three-step ionization scheme has been shown in Cs the atoms resonance ionization
detection (RID). The imaging capabilities of the Cs RIID have been evaluated using two-
step ionization of Cs. Images were obtained with a conventional CCD camera, varying
the RIID parameters such as applied high voltage, the number density of Cs atoms and