Evaluation of CZT Detector Performance in Alpha and Gamma Spectrometry
Extracted source text · arXiv 2509.26437
\documentclass[a4paper,11pt]{article}
\usepackage{jinstpub}
\usepackage{lineno}
\title{\boldmath Evaluation of CZT Detector Performance in Alpha and Gamma Spectrometry}
\author[a,b,c,1]{N.~Kramarenko,\note{Corresponding author.}}
\author[a,b,c]{M.~Väänänen,}
\author[a,c]{M.~Kalliokoski,}
\author[a,b,c]{M.~Bezak,}
\author[a,c]{R.~Turpeinen}
\author[d]{and A.~Winkler}
\affiliation[a]{Helsinki Institute of Physics,\\Gustaf Hällströmin katu 2, FI-00014 University of Helsinki, Finland}
\affiliation[b]{Lappeenranta-Lahti University of Technology LUT,\\Yliopistonkatu 34, FI-53850 Lappeenranta, Finland}
\affiliation[c]{University of Helsinki, Department of Physics,\\Gustaf Hällströmin katu 2, FI-00014 University of Helsinki, Finland}
\affiliation[d]{Detection Technology Plc,\\A Grid, Otakaari 5a, FI-02150 Espoo, Finland}
\emailAdd{nikita.kramarenko@helsinki.fi}
\abstract{The present work reports on the performance studies of Cadmium Zinc Telluride (CZT) planar detector structures. To better understand the correlation between the physical properties and the electrical and spectroscopic behavior of the devices, we carried out current–voltage (IV) and capacitance–voltage (CV) measurements, in addition to gamma- and alpha-spectroscopy. For each tested sensor, spectroscopic responses were recorded under varying applied bias voltage. The described suite of measurements provides the parameters needed to evaluate bulk resistivity, signal efficiency, and energy resolution for characteristic peaks at different energies. Readout configurations and data processing are discussed in related sections.}
\keywords{Solid state detectors, Gamma detectors (scintillators, CZT, HPGe, HgI etc), Electronic detector readout concepts (solid-state)}
%\arxivnumber{}
\begin{document}
\maketitle
\flushbottom
\section{Introduction}
\label{sec:Intro}
Cadmium zinc telluride is a room-temperature semiconductor used in demanding applications such as medical physics, nuclear security, and X-ray, gamma-ray, and, less commonly, alpha spectroscopy~\cite{a}. Despite this wide range of uses, processing and characterization of CZT detectors remain complex. To trace the connection between the physical properties of CZT and its ionizing radiation response, we evaluated the performance of five planar sensors grown by the vertical gradient freeze (VGF) method. This suite of measurements provides a solid basis for future studies aimed at understanding and improving the CZT crystal growth process.
\section{Setup and Electronics}
\label{sec:Setup}
The samples were placed on simple PCBs. The PCBs were designed in such a way that both sides of the pad detectors were accessible. The PCBs were then placed inside a metal enclosure, as shown in figure \ref{fig:Box}. The signal line from the Detector Under Test (DUT) to the first amplification stage was kept as short as possible to reduce induced noise.
\begin{figure}[htbp]
\centering
\includegraphics[width=.55\textwidth]{Box.png}
\caption{Detector enclosure design. The cover has a 1 mm in diameter round centered collimator opening in a 3 mm thick lead layer. PCB hosting the DUT is placed on the 3D printed holder, keeping the detector lined up underneath the collimator.\label{fig:Box}}
\end{figure}
The electrical behaviors reflected in the IV and CV characteristics were studied by using Keithley 2410 source meter, Keithley 6487 picoammeter and Agilent E4980A LCR meter.
To test the gamma response of the samples, the enclosure was connected to the readout amplification chain featuring the Cremat CR-150-R5 CSP evaluation board with the Cremat CR-111-R2.1 charge sensitive preamplifier (CSP) module and CAEN DT5781 Multi-Channel Analyzer (MCA) acting as a shaper along its MCA functions. The MCA parameters (rise time, decay time, trigger threshold, trigger hold-off) were tuned based on the preamplifier step-like pulse response. Schematic of the setup is shown in figure \ref{fig:BD1}. CAEN MCA gain was set to the highest value of 10 and the range was set to 0.3 V peak to peak (Vpp).
\begin{figure}[htbp]
\centering
\includegraphics[width=.85\textwidth]{BD1.png}
\caption{Block diagram illustrating data acquisition setup for characteristic peak position - voltage dependence and spectra measurements with gamma radiation sources.\label{fig:BD1}}
\end{figure}
Signal Efficiency (SE) and waveform measurements with a laboratory alpha source were carried out using the Cividec C6 fast charge amplifier, which includes a built-in shaper. The Cividec amplifier was connected to the Amptek MCA-8000D, as shown in figure \ref{fig:BD2}. The internal logic of the Amptek MCA-8000D is optimized for positive peaks and expects unipolar pulses. Thus, a positive offset is required to detect negative signals. The measurement setup and bias configuration produced positive pulses at the Amptek MCA-8000D input.
\begin{figure}[htbp]
\centering
\includegraphics[width=.85\textwidth]{BD2.png}
\caption{Block diagram illustrating data acquisition setup for the alpha source SE and waveform measurements.\label{fig:BD2}}
\end{figure}
Tri-alpha source with an effective activity of $A(t) \approx 10 \, \text{kBq}$ was used. The source contains three isotopes: $^{239}\text{Pu}$ (alpha emission at 5.24 MeV), $^{241}\text{Am}$ (5.49 MeV), and $^{244}\text{Cm}$ (5.81 MeV). The alpha source has the shape of a thin disk and was collimated with a 2 mm round opening in a 1 mm thick lead cover.
Gamma sources were collimated with a 1 mm round opening in a 3 mm thick lead cover. $^{241}\text{Am}$ with the main photopeak emission at 59.5 keV and activity of $A(t) \approx 397.0 \, \text{kBq}$; $^{133}\text{Ba}$ (356.0 keV, $A(t) \approx 40.9 \, \text{kBq}$); and $^{137}\text{Cs}$ (661.7 keV, $A(t) \approx 98.5 \, \text{kBq}$) were used.
\section{Electrical characteristics}
\label{sec:IVCV}
For each sample, we recorded the IV and CV characteristics under increasing and decreasing bias voltage (bidirectional scans) in the range from -650 V to 650 V. The current values recorded with the Keithley 6487 picoammeter from the front electrode are shown in figure \ref{fig:IV}. Bias voltage was applied to the back electrode. For the CV measurements shown in figure \ref{fig:CV}, the bias configuration remained the same.
\begin{figure}[htbp]
\centering
\includegraphics[width=.8\textwidth]{IV.png}
\caption{IV curves measured for the samples under positive and negative bias voltages in both directions. The curves are plotted using a logarithmic scale for y-axis. Microscope images of the tested samples show the differences in electrode shape configurations.\label{fig:IV}}
\end{figure}
\begin{figure}[htbp]
\centering
\includegraphics[width=.9\textwidth]{CV.png}
\caption{CV measurement results for the samples under positive and negative bias voltages.\label{fig:CV}}
\end{figure}
It can be seen that the electrical characteristics of the tested samples are similar except for one outlier. Sample 1 showed unusual behavior during both IV and CV tests. Hysteresis, as well as non-stability, can be observed, especially when a negative potential is applied to the bottom electrode. From the IV curves, the resistivity values are calculated. These are shown in table \ref{tab:parameters} along with the other measured parameters.
\section{Response to gamma irradiation}
\label{sec:Gamma}
\paragraph{Energy calibration.} Energy calibration was performed by using the characteristic peaks of $^{133}\text{Ba}$\\ (81 keV and 356 keV) and $^{137}\text{Cs}$ (662 keV) gamma emissions, confirming a linear response over this energy range. The energy resolution (ER) for emissions above 100 keV deteriorated significantly, which can be attributed to cross-section attenuation, especially for structures with relatively small bulk thicknesses. The bulk thickness measured values of the studied samples are shown in table \ref{tab:parameters}.
By fitting known characteristic peaks of $^{133}\text{Ba}$ and $^{137}\text{Cs}$ with Gaussian functions, we obtained the calibration points to convert the MCA channels to energy. The calibration linear equation $E \approx 0.6829 \cdot \text{channel} + 1.357$ was used for samples 1, 2, 3, 4. $E \approx 2.339 \cdot \text{channel} + 2.024$ was applied to sample 5 data due to the altered MCA range configuration (from 0.3 Vpp to 1 Vpp).
\paragraph{$^{241}\text{Am}$ spectra.}
Recorded with the setup described in section \ref{sec:Setup}, $^{241}\text{Am}$ spectrum histograms are shown in figure \ref{fig:Gamma}. The voltage dependence of the main $^{241}\text{Am}$ photopeak position was analyzed for the tested structures by calculating the mean and standard deviation of the Gaussian function fits. The resulting curves are shown in figure \ref{fig:PeakPosition}. The MCA response to the pulse generator signals was recorded and shown to be linear over the 0.3 Vpp and 1 Vpp voltage ranges.
\begin{figure}[htbp]
\centering
\includegraphics[width=.9\textwidth]{Gamma.png}
\caption{Recorded $^{241}\text{Am}$ spectra with peaks fit with a Gaussian function corresponding to the 59.51 keV emission. A 10 mVpp pulse generator signal is shown as a reference.\label{fig:Gamma}}
\end{figure}
\begin{figure}[htbp]
\centering
\includegraphics[width=.9\textwidth]{Peak_position.png}
\caption{Peak position - voltage dependence and MCA response to generated pulses over the 0.3 Vpp voltage range.\label{fig:PeakPosition}}
\end{figure}
The response to the 59.51 keV characteristic $^{241}\text{Am}$ gamma emission for sample 1 is poor, making it challenging to measure the peak position - voltage dependence at low bias voltage values. Electronic noise introduced to the lower MCA channels for bias voltages below 90 V alters the emission response histogram shape, thus lowering the MCA channel value during the fit. With decreasing bias voltage, it becomes progressively harder to resolve the signal coming from gamma emissions and electronic noise. There are a number of ways to counteract this adverse effect. One of the methods employed is changing the Least Significant Bit (LSB) value and then tuning the filter settings in CAEN MCA such as decay and rise pulse times. Lowering the LSB value leads to an increased sensitivity for low-amplitude signals, including electronic noise. One has to find the balance between acceptable noise level and processable detector response.
\section{Response to alpha irradiation}
\label{sec:Alpha}
Figures \ref{fig:Alphas} and \ref{fig:Alphas13} show the tri-alpha source spectra recorded with the tested samples. The difference in MCA channel values is due to the set Amptek MCA binning range; otherwise, no changes were made to the setup.
\begin{figure}[htbp]
\centering
\includegraphics[width=.9\textwidth]{Alphas.png}
\caption{Alpha radiation spectra measured with samples 2, 4 and 5.\label{fig:Alphas}}
\end{figure}
\begin{figure}[htbp]
\centering
\includegraphics[width=.9\textwidth]{Alpha_1_3.png}
\caption{Alpha radiation spectra measured with samples 1 and 3.\label{fig:Alphas13}}
\end{figure}
The inter-peak distances were calculated relative to samples’ middle-peak centroids. Resulting values are shown to be consistent throughout the batch, yielding \text{11.42 ± 0.32\%} for the left peak and \text{9.99 ± 0.33\%} for the right one. The ratio of these values corresponds to the ratio of isotope relative energies. Sample 3 alpha radiation response is substantially worse compared to the rest of the samples. This is reflected in the calculated ER values. Surface or close-to-surface defects, as well as electrode thickness, might be a potential reason for deteriorated performance. Evaluating the defect concentration impact on detection performance is planned as a continuation study.
\section{Signal efficiency and \texorpdfstring{$\mu\tau$}{mu-tau} estimation with alpha source}
\label{sec:SE}
To estimate the SE we have recorded the waveforms averaged over 2000 events for a set of bias voltage values. The waveforms are collected with a trigger of set amplitude, typically from 2 mV to 4.5 mV. The peak amplitude of the signal waveforms is bias voltage dependent and was recorded to be in the range from 5 mV at 30 V bias to 250 mV at 650V bias. The three isotope alpha source requires a complicated data acquisition process. With a sufficiently large data set, the averaged waveform shape remains stable, allowing further data processing. Baseline correction was implemented for all waveforms.
The signal detected with the CSP is dominated by the electron contribution, as the positive potential was applied to the bottom electrode, capacitively disconnected from the signal readout line. Such biasing geometry minimizes the hole-induced signal because positive charge carriers are collected shortly after the alpha particle penetrates the bulk surface levels~\cite{b}.
Calculated from the obtained waveforms, $\mu\tau$ values shown in table \ref{tab:parameters}, are typical for a CZT material~\cite{a,c}. Another drawback of the data acquisition method used was observed for waveforms recorded at low bias values. Some waveforms tend to be partially negative due to data averaging which directly affects the SE calculation, despite the applied baseline correction. The resulting SE plots for all characterized samples are shown in figure \ref{fig:SE}.
\begin{figure}[htbp]
\centering
\includegraphics[width=.9\textwidth]{SE.png}
\caption{Normalized SE - voltage plot for studied samples with Hecht equation fits used to estimate $\mu\tau$ values. Error bars propagated from $\pm 5\%$ integration limits.\label{fig:SE}}
\end{figure}
\begin{table}[htbp]
\centering
\caption{Sample thickness and calculated parameters from the measurement data.\label{tab:parameters}}
\smallskip
\begin{tabular}{c|c|c|c}
\hline
Sample&Thickness, mm&Resistivity, $\Omega \cdot \mathrm{m}$&$\mu\tau$, $\mathrm{cm^2/V}$\\
\hline
1&1.33&$8.19 \times 10^{6}$&$2.65 \times 10^{-4}$\\
\hline
2&1.70&$3.85 \times 10^{8}$&$3.21 \times 10^{-4}$\\
\hline
3&2.00&$3.72 \times 10^{8}$&$4.41 \times 10^{-4}$\\
\hline
4&1.60&$9.90 \times 10^{7}$&$3.09 \times 10^{-4}$\\
\hline
5&1.45&$6.31 \times 10^{8}$&$3.47 \times 10^{-4}$\\
\hline
\end{tabular}
\end{table}
A large difference in signal shape formation between the samples is observed with increasing bias voltage. It is valid to assume that the physical properties of sensors, such as resistivity, defect configuration and density alter the response to specific types of ionizing radiation~\cite{d}.
\section{Conclusions}
From the lifetime product values obtained, we can roughly estimate the performance of the characterized CZT crystals. However, in order to have more certainty in the estimation, infrared (IR) imaging and follow-up charge-transport studies are required. Defect density and qualitative composition are performance-defining for CZT-based devices~\cite{c}. The results of the charge-transport and IR defect studies will be reported in forthcoming publications.
We also observe a large qualitative difference in response to gamma and alpha ionizing radiation for samples 1 and 3. Based on the IV/CV characteristics and resistivity values, it was expected to see sample 3 perform similarly to the rest of the batch, and the sample 1 to be the outlier. Nonetheless, the sample 1 shows better energy resolution under alpha source irradiation, but fails to properly convert the gamma radiation energy to a detectable charge.
One further step to improve the SE and $\mu\tau$ estimation results is to record a set of waveforms yielding sufficient statistics. Filtering out the waveforms of certain shape and amplitude would allow for a more precise estimation of the SE and underlying physics.
\begin{thebibliography}{99}
\bibitem{a}
Y. Xiang, X. Jiang, C. Wang, F. Meng, Y. Han, X. Huang and L. Wei,
\emph{Alpha particle detection with a planar CdZnTe detector and relative simulations},
\emph{Radiation Detection Technology and Methods} {\bf 5} (2021) 609–617, doi:10.1007/s41605-021-00296-z.
\bibitem{b}
P. Praus, E. Belas, J. Franc, R. Grill, P. Höschl and J. Pekárek,
\emph{Electronic Pulse Shape Formation in Transient Charge and Transient Current Detection Approach in (CdZn)Te Detectors},
\emph{IEEE Trans. Nucl. Sci.} {\bf 61} (2014) 2333–2337, doi:10.1109/TNS.2014.2330070.
\bibitem{c}
R. Gul, Z. Li, R. Rodriguez, K. Keeter, A. Bolotnikov and R.B. James,
\emph{Defect Measurements in CdZnTe Detectors Using I-DLTS, TCT, I-V, C-V and $\gamma$-ray Spectroscopy},
\emph{Proc. SPIE} {\bf 7079} (2008) 70790U, doi:10.1117/12.797865.
\bibitem{d}
A. Hossain, A.E. Bolotnikov, G.S. Camarda, Y. Cui, A. Prishivalko, G. Yang, R. Gul, R.B. James and L. Li,
\emph{Direct observation of influence of secondary-phase defects on CZT detector response},
\emph{J. Cryst. Growth} {\bf 470} (2017) 99–103, doi:10.1016/j.jcrysgro.2016.10.023.
\end{thebibliography}
\end{document}