2.1. MOCVD-Grown HgCdTe
The HgCdTe epitaxial growth was carried out in the horizontal reactor of an AIX-200 Aixtron MOCVD (AIXTRON, Aachen, Germany) unit. The system operated in the laminar flow regime with process pressures from 50 to 1000 mbar, using a butterfly valve for pressure control. The reactor pressure of 500 mbar was used for all successful growth runs. The hydrogen was used as a carrier gas. Dimethylcadmium (DMCd) and diisopropyl telluride (DIPTe) were used as the precursors. Ethyl iodide (EI) and tris(dimethylamino)arsenic (TDMAAs) were used as donor and acceptor dopant sources, respectively. The DMCd and EI were delivered through one channel, while DIPTe and TDMAAs were delivered through another channel over the quartz container, where the elementary mercury was being held. Aixtron’s gas foil rotation technique was applied for better composition uniformity. There were two temperature zones in the reactor: the mercury source zone; and the growth zone with the graphite susceptor. High-temperature annealing was used before each growth run for the reactor and the substrate cleaning. The gas delivery system was additionally equipped with Piezocon ultrasonic precursor concentration monitors. The usage of Piezocons contributed to a better repeatability of the growth processes. Adaptation of the EpiEye reflectometer (ORS Ltd., St Asaph, UK) allowed for in situ monitoring of the thickness and the surface morphology of the growing layer. Laser radiation, incident and reflected from the growing layer, with a length of 650 nm, was passed through the hole in the quartz liner, allowing us to obtain reliable interferograms.
Figure 1 presents a fragment of the precursors’ delivery installation, together with the horizontal reactor cell scheme with an internal mercury source of the Aixtron AIX 200 MOCVD, designed for HgCdTe deposition.
Growth was carried out on 2-inch, epi-ready, semi-insulating (100) GaAs substrates, oriented 2° off toward the nearest <110>. Typically, a 3–4-µm-thick CdTe layer was used as a buffer layer, reducing the stress caused by the crystal lattice misfit between the GaAs substrate and the HgCdTe epitaxial layer structure [
7]. The buffer also prevented gallium diffusion from the substrate to the HgCdTe layer. The GaAs substrate was annealing in a Te-rich atmosphere prior to the growth of the CdTe buffer. The interdiffused multilayer process (IMP) technique was applied for HgCdTe deposition [
8]. The HgCdTe was grown at 350 °C, with the mercury source kept at a temperature of 160–220 °C. The II:VI mole ratio was kept in the range of 1.5–5 during the CdTe cycles of the IMP process. Acceptor and donor doping were examined over the wide range of the compositions, and doping levels of 5 × 10
14–5 × 10
17 cm
−3 were obtained. In order to reduce the mercury vacancies’ concentration, and to increase the uniformity of the HgCdTe heterostructures, the in situ annealing was performed under mercury-rich conditions. However, the obtained heterostructures were not annealed ex situ [
9].
The crystalline quality of the HgCdTe films depends on the MOCVD system purity, the substrate quality, the nucleation, and the growth conditions. The growth of the HgCdTe epilayers with the (111) B orientation has some advantages, such as a lack of large macrodefects, high growth rate, lower consumption of the precursors, and effective n-type doping with iodine up to 10
18 cm
−3. However, the (111) B layers exhibit some drawbacks, which affect the performance of the devices-namely, relatively rough surface morphology, high (>10
15 cm
−3) concentration of residual donor defects, and less efficient p-type doping with arsenic in comparison with (100) HgCdTe. Most of our research, and all of the work presented in this paper, concerns (111) HgCdTe; the selected growth parameters are listed in
Table 1.
2.2. Photodiode Design
MOCVD-grown HgCdTe heterostructures are designed for infrared photodiode construction. In this section, the fundamental rules of infrared photodiode design will be discussed. A classic N
+/p/P
+ structure was enriched with a transient P layer inserted between the absorber and the highly doped P
+ region. This additional P layer helped to shape an interface between the p and the P
+, preventing the formation of unfavorable potential barriers; then, a photodiode took the form of the N
+/p/P/P
+ structure. Furthermore, in order to improve the electrical contact properties of the metallization of the P
+ layer, the structure was upgraded with the P
+/n
+ tunneling junction. Finally, the resulting structure took the form of the N
+/p/P/P
+/n
+. An example of the HOT MWIR HgCdTe photodiode architecture is presented in
Figure 2.
The schematic immersion lens on the top was formed from GaAs substrate. The immersion lens increases the optical area of the detector by approximately 10 times. Under the lens, there is a 3-µm CdTe buffer layer. Below the lens, there is a heavily donor-doped contact N+ layer, with the composition x = 0.48. However, this composition of the N+ layer can be changed, in the range from x = 0.38 to x = 0.57, depending on the cut-on wavelength (λCUT-ON) of the detector. For technological reasons, the thickness of the N+ layer was extended to 10 µm. Next, there is a medium-doped p-type absorber region with a doping level higher than the donor-like background concentration, which is typically 3 × 1015 cm−3 in size. The absorber composition corresponds to the required cut-off wavelength of the detector, and for MWIR devices has a value in the range of 0.27–0.36. The thickness of the absorber should be greater than the minority carrier diffusion length, and is typically about 3 µm. The parameters of the next transient P layer were calculated individually for each device, in order to prevent the formation of a potential barrier, taking into account the interdiffusion processes during growth. The composition of the P+ layer should be significantly larger than that of the absorber, and the doping should be as high as possible in order to attain low concentrations of the minority carriers and low resistance. Our MOCVD system enabled us to achieve an acceptor concentration of around 1017 cm−3. The thickness of the P+ layer was about 1 µm, in order to minimize the in-diffusion of the electrons from the metallic contact. Finally, the n+ layer, with a composition lower than that of the absorber and donor concentration, at about 1018 cm−3, finishes the structure. The thickness of the n+ layer is typically about 1 µm, so as to achieve low resistance.
Following the rules described above, the obtained HgCdTe heterostructures were subjected to the standard processing procedure, which includes, among others: photolithography masking; chemical etching; metallization; cutting; and assembling. The circular photodiodes with mesa diameters ranging from 100 to 900 μm were fabricated and mounted on the thermoelectric coolers. The structures were prepared to be back-side illuminated (irradiated from the substrate side, as was illustrated in
Figure 2). The photodiodes were not passivated, and no anti–reflection coating was applied.
2.4. Heterostructure Characterization
Eight heterostructures of type N
+/p/P/P
+/n
+ were investigated during this study. They differed in the composition and the thickness of the N
+ and absorber regions. The details of the chosen N
+/p/P/P
+/n
+ heterostructures are presented in
Table 2.
Figure 5 presents the scanning electron microscopy (SEM, Hitachi, Tokyo, Japan) image of the N
+/p/P/P
+/n
+ #962 HgCdTe heterostructure cleavage. The sample was broken along the growth direction. This allowed us to specify the individual component layers, and to measure their thickness. The obtained structure was compared with the designed one (
Figure 2). The measured thicknesses were given with 0.01 µm resolution. However, because of the blurred boundaries between the layers, the practical measurement precision was assumed at about 1%, and this was sufficient for our experiments. The graded interfaces between the individual layers were also hardly observed. The measured thickness of the presented MWIR heterostructure was 20.72 µm.
Figure 6 presents the N
+/p/P/P
+/n
+ HgCdTe heterostructure cleavage, taken with an optical microscope at ×1000 magnification. The thickness of the CdTe buffer layer was about 2.5 µm. The total thickness of the HgCdTe structure was estimated at 20.6 µm, and was comparable with the result obtained using the SEM method. The P transition layer is barely visible, due to the interdiffusion process during growth, and because the image resolution is too low. An inset with a diagram of the individual component layers was added to the photo at the bottom for better visualization. This method is less precise than the SEM measurements, but both are prompt and useful, and give better insight into the device performance.
Figure 7 presents the secondary-ion mass spectroscopy (SIMS, Cameca, Gennevilliers, France) profile of the N
+/p/P/P
+/n
+ Hg
1−xCd
xTe heterostructure with the absorbing region composition
x = 0.32. The green line represents iodine counts: 10
5 and 2 × 10
4 in the n
+ and N
+ regions, respectively, which correspond to the designed concentration levels: 10
18 cm
−3 and 2 × 10
17 cm
−3, respectively. In general, abrupt transitions were observed in the iodine profile. As a positive consequence, there was no iodine presence in the adjacent layers, especially in the absorber region. The pink line represents the arsenic-stepped profile, corresponding to abruptly increasing doses of the arsenic in successive layers: p, P, and P
+. We did not observe any undesirable arsenic diffusion to the adjacent layers, suggesting that arsenic is a stable dopant. The mercury, cadmium, and tellurium counts enabled us to calculate a composition (
x) profile (the light blue line) of the whole Hg
1−xCd
xTe heterostructure. We can see that the narrower gap absorber with the composition
x = 0.32 is surrounded by wider energy gap N
+ and P
+ layers with compositions of around
x = 0.46, as assumed in the heterostructure project. Due to the interdiffusive processes in the HgCdTe alloy during growth, a composition profile of the interfaces between the particular layers is not abrupt. Systematic studies of the SIMS results, supported by the theoretical calculations, allow us to make the corrections in the programmed growth recipes, and to obtain the assumed profiles in the real HgCdTe heterostructures.
Another characterization method we used was IR transmission measurement, which was performed using a PerkinElmer Fourier-transform infrared (FTIR, PerkinElmer, Waltham, MA, USA) spectrometer.
Figure 8 presents the transmittance characteristics of particular fragments of the #773 N
+/p/P/P
+/n
+ structure, after selective etching. The selected area of the sample was covered with a photoresist film deposited during the photolithography process. As a result, the part of the sample surface intended for etching was exposed. The selective etching was performed in a solution of Br in HBr diluted in deionized water (50:50:1 Br:HBr:H
2). The green line (
ta) represents the transmission of the total heterostructure just after growth, where the narrow-gap n
+ top layer mainly determines its shape. Additionally, the clear interference fringes were useful in the estimation of the actual thickness of the measured sample. The red line (
tb) was obtained after 4-µm-deep chemical etching. As a result, the n
+ and P
+ layers were removed. The slope of the t
b curve depends on the absorber transmission. This allows us to determine the composition of the absorber, which in this exemplary structure takes a value of
x = 0.302. After removing another 7 µm of the material (11 µm in total), the N
+ layer was exposed, and the blue curve t
c represents its transmission spectral dependence. In this way, the composition (
x) of the obtained N
+ layer was determined at the value of
x = 0.492 (whereas the predicted value was
x = 0.48). The N
+ layer composition determines the λ
CUT-ON value of the photodiode when it is back-side illuminated. In general, the IR transmission measurement results provide valuable feedback for MOCVD growers.