With high atomic number, large band gap, and high electron mobility–lifetime product (μ τ), the compound semiconductor of cadmium zinc telluride (CdZnTe, CZT) has been widely used as room-temperature x-ray and γ-ray detectors and other electro-optical devices.^[1–3] The quality of this material is still restricted by various factors, like the point defects, linear defects, bulk defects, and non-uniformity of elemental distribution.^[4] The CZT intentionally doped with the elements of group III (Al, Ga, In) or group VII (Cl, Br) can be incorporated on Cd sites to compensate for Cd vacancies and achieve high resistivity.^[5,6] The inevitable intrinsic defects of dislocations could be formed by the additional stress in the solidification process. Tellurium-rich precipitates due to retrograde solubility and Te inclusions trapped by the growth interface present still severe problems for the commonly used modified vertical Bridgman (MVB) method and travelling heater method (THM).^[7,8] It has been proved that both donor and accepter levels can be introduced by the intrinsic or extrinsic point defects.^[9,10] The low charge collection in ion-beam induced charge (IBIC) images demonstrates transient charge losses around Te inclusions and dislocations.^[11,12] All these defects can significantly degrade the crystal properties and consequently deteriorate the performances of devices. Thus, the comprehensive study of these defects is quite important for the better understanding of the crystal properties and detector performance. However, defects including point defects and Te inclusions in crystal always interact with each other and then are difficult to distinguish each other.

In this study we investigate CdZnTe:In crystal according to the distinctive Te inclusion distribution along the growth direction, and correlate the distribution with the crystal growth process. The photoelectric properties of CZT crystals are investigated using the techniques of Fourier transform infrared (FTIR) transmission spectrum, UV–Vis–NIR transmittance spectrum, I–V and laser beam induced current (LBIC) measurements. Finally, the distribution of defect levels is studied and discussed by low-temperature photo-luminescence (PL) technique and thermally stimulated current (TSC) measurements to better illuminate the nature of the defect behaviors.

Samples with dimensions of 5 × 5 × 2 mm³ were cut from the head (T04) and tail (W02) part of indium doped Cd_0.9Zn_0.1Te crystal, grown by modified vertical Bridgman method in Imdetek Ltd Company. The surface damaged layers after slicing was removed by lapping, mechanical, and chemo-mechanical polish. The samples were then chemically etched using 2% bromine methanol solution for 2 min, prior to the infrared transmittance microscopy, FTIR spectrometer and UV–Vis–NIR spectrometer measurements. For the laser beam induced current (LBIC) measurements, Au electrodes with the thickness of approximate 10–50 nm were fabricated on both surfaces of the samples and the excess carriers were produced by the pulsed laser with a wavelength of 527 nm, pulse width of 893 ps.

For the PL measurement, an argon ion laser with a wavelength of 488 nm was used as an excitation source. A Triax550 tri-grating monochromator and a photomultiplier tube (PMT) were employed to collect the luminescence emissions of the samples at 10 K with a spectral resolution of 0.3 nm. In the TSC measurement, the sample was cooled down to about 50 K in dark. Free carriers were excited by achromatic light from a halogen lamp with a wave length of 650 nm, and captured by defect traps at low temperature. As the temperature increased to 300 K in dark with a constant heating rate, the trapped carriers were thermally emitted and recorded under a volt bias of 10 V with the Keithley 6514 electrometer.

Indium doped Cd_0.9Zn_0.1Te ingot with a diameter of 60 mm and length of 180 mm is grown with the MVB method. The modified crystal growth parameters ensure the large grain size and less grain boundaries of CdZnTe: In crystals as shown in Fig. 1(a). Figures 1(b) and 1(c) show the IRTM images of samples T04 and W02, respectively. It can be seen that Te inclusions are uniformly distributed with an average concentration of ∼10³ cm⁻³ and size of in W02 as shown in Fig. 1(c). However, Te inclusions in T04 are distributed in a belt-like zone with a bandwidth of in the full range of sample thickness and no macroscopic Te inclusions are observed except those in the belt zone. It has been proved that Te-rich boundary layer formed in the melt of CdZnTe crystal could be captured by the growth interface and formed into Te inclusions.^[13] Yang et al. concluded that grain boundaries could have a strong trapping effect of Te inclusions in the annealing process, which probably pass through the high-density zone of dislocation networks in the grain-boundary region.^[14] Thus, we deduce that the Te inclusion belt could probably be formed through the trapping of Te atoms by sub-grain boundaries in the crystal growth process. This could also give an explanation of the “clean” zone near the Te inclusion belt (Fig. 1(b)).

Fig. 1.

	Figure Option ViewDownloadNew Window
	Fig. 1. (color online) Images of CdZnTe: In crystal grown with the MVB method, showing (a) image of asgrown CZT ingot, and ((b), (c)) IRTM images of T04 and W02 samples cut from the head and tail regions of the ingot, respectively.

Typical IR transmission spectra of the samples are shown in Fig. 2(a). The transmittance values of T04 and W02 reach 57.52% and 61.85%, respectively, and remain almost unchanged within the wave number of 1000–4000 cm⁻¹, which represents quite a good lattice perfection and crystalline quality of the ingot.^[15] The same band gap values of the two samples (1.54 eV), as shown in Fig. 2(b), indicate that these two samples possess almost the same compositions. Moreover, good linear and symmetrical I–V results are observed even when subjected to a low bias voltage of ±0.1 V, as shown in Figs. 2(c) and 2(d). The bulk resistivity values are fitted to be and for T04 and W02, respectively. Combining the basic spectra and I–V measurements mentioned above, we can find that although the distribution and concentration of Te inclusions show distinctive difference, the general optical and electrical properties of these two samples are almost the same.

Fig. 2.

	Figure Option ViewDownloadNew Window
	Fig. 2. (color online) (a) Typical FTIR transmission spectra, (b) UV–Vis–NIR transmittance spectra of T04 and W02 samples, ((c), (d)) I–V curves for W02 and T04, measured at room temperature, respectively.

The LBIC waveforms measured with different bias voltages are shown in Figs. 3(a) and 3(c). The typical penetration depth of laser with a wavelength of 527 nm is around 530 nm in the CZT crystal, which is much less than the sample thickness. Thus, when negative bias is applied to the front electrode, the current signals are mainly induced by photo-generated electrons. The photocurrent response under 50 V bias voltage shows an instantaneous rising edge, with delay times of 3.8 ns and 4 ns reaching to the maximum values for T04 and W02, as shown in the inserts in Figs. 3(a) and 3(c), respectively. The linear relationships between the electron drift velocity and electric field strength are obtained in Figs. 3(b) and 3(d). Carrier drift velocity v can be calculated from the transit time t_R under a given bias voltage. The electron mobility is obtained from v=μ E, where E is the applied electric field. The mobility of electron at room temperature in T04 is about 908 ± 18 cm²/vs, which is higher than that in W02 (814 ± 16 cm²/vs). It is considered that the carrier mobility is related to different scattering mechanisms. Xu et al.^[16] concluded that for defect densities higher than 1×10¹⁷ cm⁻³ and lower than 1×10¹⁵ cm⁻³ in CZT, carrier mobility is limited by ionized impurity scattering and polar-optical phonon scattering, respectively. Thus, it is necessary to know the defect species and concentration in crystal for the better understanding of the carrier transport properties of the samples.

Fig. 3.

	Figure Option ViewDownloadNew Window
	Fig. 3. (color online) LBIC measurements by exposing from cathode side, showing current waveforms for samples (a) T04 and (c) W02, and fitting results of electron mobility in the samples (b) T04 and (d) W02. Inserts in panels (a) and (c) are enlarged parts of the dotted blue rectangles.

As a sensitive technique for characterizing the types and distributions of defects and impurities in crystals, the low-temperature photo-luminescence spectrum is used to obtain the information about the recombination property of photo-carriers.^[17–19] The FE peak is the recombination of the peaks of the free exciton, composite particles formed by the coulomb force of opposite charges. The emission peaks of D⁰X and A⁰X in PL spectrum originate from shallow donors (i.e., In , etc.)^[20] and shallow acceptors (i.e., Li_Cd, Na_Cd, N_Te, etc.)^[21,22] respectively. The DAP is emitted by the annihilation of shallow acceptors with shallow donors and the LO peaks are the longitudinal optical (LO) mode of direct electron-phonon interaction in CZT.^[20]

Figure 4 shows the typical PL spectra of T04 and W02, ranging from 1.4 eV to 1.7 eV. For comparison, the measured data are normalized according to the emission peak intensity of D_complex. The sharp emission peaks of D⁰X and A⁰X, as well as the free exciton (FE) recombination are found for T04, indicating a better crystalline quality. In addition, the relative intensity of D⁰X and DAP emission peaks in T04 is higher than that in A⁰X, which shows the higher intensity of shallow donors in the T04 sample. For the asymmetry feature of D_complex, the peak is considered to consist of the structure defects and A-center in crystal.^[12,23] It can be found that the intensity of D_complex relative to other emission peaks from 1.58 eV to 1.65 eV shows that the relative intensity is much higher for W02 than that for T04, which also indicates that sample T04 has much better crystalline quality.

Fig. 4.

	Figure Option ViewDownloadNew Window
	Fig. 4. (color online) Typical PL spectra of the samples T04 (upper) and W02 (lower).

Thermally stimulated current (TSC) measurements are used to investigate the deep-level defects in CZT crystals. Each peak in TSC spectrum corresponds to a certain trap level in the band-gap, and a certain defect state in the crystal lattice. In the experiment, the samples are illuminated for 10 min after the temperature has become stably lower than 50 K for 2 h, to make sure that all the defect levels are fully filled before temperature scanning. A heating rate of 0.2 K/s and bias voltage of 10 V are employed to record current spectra from 50 K to 300 K as shown in Fig. 5. A complete characterization of trap signatures in TSC spectra is simulated with the simultaneous multiple peak analysis (SIMPA) method proposed by Pavlovic et al.,^[24] expressed as

Fig. 5.

	Figure Option ViewDownloadNew Window
	Fig. 5. (color online) Typical TSC spectra of amples (a) T04 and (b) W02.

A comparison of different trap densities between samples T04 and W02 is given in Table 1. We can see that the density of shallow defect levels of W02 before 150 K is one order higher than that of T04. Higher densities of T₅ and T₆ in Table 1 indicate higher Cd vacancies of W02. Correspondingly, higher densities of (5.5×10¹⁵ cm⁻³ for T04 and 3.2×10¹⁶ cm⁻³ for W02) can be formed through : In + + 2e or In + + 3e.^[30] Referring to IRTM images in Fig. 1, it can be concluded that the melt enrichment of Te solutions leads to not only Te inclusions, but also the shallow point defects like T₃ in the crystal growth process. These shallow defects with a total trap density of ∼10¹⁷ cm⁻³ can probably dominate the scattering mechanism of these two samples as discussed above.

Table 1.

Table 1.

Table 1. Comparisons of the trap density in CdZnTe sample between samples T04 and W02. . T1 T2 T3 T4 T5 T6 T7 T04/cm−3 1.6 ×1015 1.5 ×1015 5.5 ×1015 - 1.2 ×1015 2.8 ×1014 1.4 ×1015 W02/cm−3 - - 3.2 ×1016 8.6 ×1015 1.3 ×1015 1.2 ×1015 3 ×1014

Table 1. Comparisons of the trap density in CdZnTe sample between samples T04 and W02. .

The optical and electrical properties of CdZnTe:In samples are intensively studied. The FTIR spectra and UV–Vis–NIR transmittance spectra show almost the same transmittance (57.52% and 61.85%) and the bandgap (1.54 eV), despite distinct difference in Te inclusion distribution between samples T04 and W02. Carriers’ mobility values for T04 and W02 are measured with LBIC technique to be 908 ± 18 cm²/vs and 814 ± 16 cm²/vs, respectively. Sharp emission peaks of D⁰X and A⁰X, as well as the free exciton (FE) recombination in the PL spectrum indicate quite good crystalline quality of the T04 sample. Moreover, the intensity of D_complex relative to other emission peaks between 1.58 eV to 1.65 eV is much higher for sample W02 than for samples T04. Seven trap peaks are identified from the TSC spectrum with the SIMPA method. The results show that higher densities of (5.5 × 10¹⁵ cm⁻³ for T04 and 3.2 × 10¹⁶ cm⁻³ for W02), Cd vacancies and relevant defect complex, could be formed in the crystal growth process, which probably dominates the scattering mechanisms of these two samples.