Temperature Dependence of the Defect States in LWIR (100) and (111)B HgCdTe Epilayers for IR HOT Detectors

Abstract

HgCdTe epilayers grown by chemical vapor deposition (MOCVD) on GaAs substrates operating in the long-wave infrared range were characterized by the photoluminescence (PL) method. Photodiode and photoconductor designs, both (100) and (111)B crystallographic, were analyzed. Spectral current responsivity (R_I) and a PL signal approximated by a theoretical expression being the product of the density of states and the Fermi–Dirac distribution were used to determine the fundamental transition (energy gap, E_g). For all the samples, an additional deep-level-related transition associated with mercury vacancies (V_Hg) were observed. The energy distance of about 80 meV above the valence band edge was observed for all the samples. Moreover, measurements at low temperature showed shallow acceptor-level (As_Te and V_Hg as acceptors) transitions. In HgCdTe(100), due to the higher arsenic activation, As_Te was the dominant acceptor dopant, while, in HgCdTe(111)B, the main acceptor level was formed by the neutral V_Hg. The determined activation energies for As_Te and V_Hg dopants were of about 5 meV and 10 meV, respectively.

Introduction

The long-wave infrared (LWIR) range is extremely important for industrial and military applications.^1,2,3,4 Currently, high-operating-temperature (HOT) infrared (IR) detectors are being widely developed and can operate at ambient temperature or be cooled using simple, cheap, and easy-to-use thermoelectric coolers. In this context, mercury cadmium telluride (MCT)-based devices continue to be widely used. However, market requirements are for constant development and the search for new solutions in this field. Current technology uses various growth techniques,^5,6,7,8 crystallographic orientations,^9,10,11,12 and different substrates. Detection devices operating in the LWIR range must meet rigorous standards. Therefore, it is necessary to grow epitaxial layers with high crystallographic quality. Defects that occur in HgCdTe, such as mercury (V_Hg) and tellurium (V_Te) vacancies and tellurium antisites occupying mercury sites (Te_Hg), introduce shallow levels into the energy gap: V_Hg acts as an acceptor, while V_Te and Te_Hg act as donors. Furthermore, the use of arsenic as an acceptor dopant can result in the formation of As_Te (takes the place of Te). When As takes the place of the Hg site (under Te-rich growth conditions), it incorporates as a donor.^13,14 Moreover, due to the existence of V_Hg, a possible acceptor-type V_Hg–As_Hg complex can be formed. V_Hg defects can also be located deep in the band gap. Acting as generation–recombination centers, they influence the detection performance reducing the Shockley–Read–Hall carrier lifetime.

The analysis of defect states can be performed using various measurement techniques,^15,16 e.g., band-to-band and shallow transitions can be determined by photoluminescence (PL). In special cases, deep defect transitions can also be detected.^14,15,17 Therefore, the PL method is also useful for determining the position of the defect state in relation to the energy gap, as it is not entirely consistent in various studies. Some studies claim that the trap level in HgCdTe is at 0.5 E_g or 0.7 E_g and proportionally changes with temperature to E_g.¹⁸ However, there are studies claiming that the trap level does not change its position with temperature, but is fixed at a constant level in relation to the edge of the valence band (VB).^9,15

This paper shows the PL results for four samples with the p-type Hg_1-xCd_xTe absorber. Photodiode and photoconductor designs, both (100) and (111)B crystallographic, have been measured. PL spectra were analyzed for the temperature range from 18 K to 300 K. To correctly determine the energy gap, the current responsivity (R_I) measurements and the theoretical relationship presented by Hansen were used.¹⁹ Furthermore, an analysis of the PL intensity on the excitation power was made.

Sample and Experimental Details

The (100) and (111)B HgCdTe epilayers were grown by an Aixtron AIX-200 MOCVD system on 2-inch (c.5 cm) (100) GaAs substrates. In each sample, the CdTe buffer layer was used in order to compensate the lattice mismatch between the HgCdTe epilayer and the GaAs substrate. The growth technique has been described in detail in Refs. 20, 21, and 22.

Figure 1a shows the design of both the (100) and (111)B N⁺/p/T/P⁺/n⁺ photodiodes (marked as A and B, respectively). Both samples have a similar structure: the wide-bandgap N⁺ contact layer, the p-type active layer, the wide-bandgap P⁺ barrier, and an n⁺ cap-contact layer. The transient layer (T) parameters are conditioned by interdiffusion processes during the MOCVD growth. The assumed doping and Cd molar composition were similar in both samples, but the main difference was the absorber thickness. Figure 1b shows the design of a p-type photoconductor, also grown (100) and (111)B crystallographic (marked as a C and D, respectively). In this case, the main difference was also the absorber thickness.

Fig. 1

Schematic of an LWIR HgCdTe N⁺/p/T/P⁺/n⁺ photodiode (a), and a p-type photoconductor (b).

For the PL measurements, all the structures were chemically etched by a Br:HBr diluted in deionized water solution (50:50:1 Br:HBr:H₂O) to a p-type absorber layer. The PL characteristics were measured using an FTIR Bruker Vertex 70v spectrometer in the step-scan mode,²³ a liquid-nitrogen-cooled MCT DS315 detector, and a lock-in amplifier (Fig. 2). The 640-nm line, mechanically chopped with a frequency of a 10 kHz laser, was used as an excitation. The PL spectra were collected in the temperature range from 300 to 18 K and excitation power from 200 to 10 mW. Due to the small molar composition used in the photoconductors, the energy gap at low temperatures was narrowed and the signal went beyond the system's measurement range; —the measurement was performed in a limited temperature range (from 140 to 300 K). All the characteristics were made in a vacuum.

Fig. 2

PL measurement system.

The R_I measurements were performed using an FTIR-type Spectrum 2000 spectrophotometer, and a blackbody at a temperature of 1000 K. Photodiodes were bottom-illuminated, while the photoresistors were top-illuminated, as shown in Fig. 1. All the characteristics were measured for a bias of − 0.2 V. The electrical area was 7.85 × 10⁻⁵ cm² for the photodiodes and 12.64 × 10⁻⁵ cm² for the photoconductors.

Results and Discussion

The PL signals of the analyzed samples for high temperatures are shown in Fig. 3. At 300 K, the figure additionally shows the spectral current responsivity (R_I) as a second method to determine the E_g, which is the equivalent for the point where the detector response drops to 50% of the peak value.²⁴ The E_g based on the PL measurement was determined using a theoretical approximation which is proportional to the product of the joint density of states and the Fermi–Dirac distribution function.^24,25 For A (Fig. 3a), the estimated E_g value is 195 meV at 300 K. An extra peak for the energy of 162 meV is also visible. As the temperature decreases, this transition disappears and another low-energy peak becomes visible for the energy of 86 meV. The same, most probably defect-related transition with an energy of 82 meV, is visible in B (Fig. 3b). Even though the assumed molar composition of the absorber of B was the same as that of A, the determined E_g has a smaller value of about 175 meV.

Fig. 3

PL and R_I spectra measured at high temperatures for the LWIR HgCdTe A (a), B (b), C (c), and D (d).

Both photoconductor samples (Fig. 3c and d), apart from the main energy transition with 130 meV for C and 128 meV for D at 300 K, have an apparent transition from the defect level with energies of 80 meV and 88 meV, respectively. An defect-related transition with an energy of 115 meV is visible in the PL spectrum of C. This transition disappears at lower temperatures.

Figure 4 shows the PL spectra measured for the photodiodes (sample A and B) at low temperatures. For sample A, three extra transitions can be seen, with the positions of 5 meV, 10 meV, and 18 meV below E_g, respectively. For sample B, only two extra transitions with energies of 4 meV and 8 meV below E_g can be seen.

Fig. 4

PL spectra measured at low temperatures for the LWIR HgCdTe A (a), and B (b).

Figure 5 shows the temperature dependence of all the visible optical transitions in the PL spectrum. Band-to-band transitions were fitted with theoretical Hansen's relationships. It can be seen that, for all the samples, the actual molar composition differs from that assumed in the growth process. The transition with the energy of about 80 meV is visible in all the analyzed samples. This is most likely a transition from the defect level to the VB. The constant with temperature transition energy indicates that the position of the defect level does not change with temperature, but is fixed at a constant distance from the edge of the VB. When temperature decreases and E_g becomes smaller than the defect level position, the defect-related PL peak disappears. This happens because the defect level enters the conduction band (CB), where it creates a so-called resonance state.^15,26 Both samples with the (111)B orientation have one extra defect-related transition, with their temperature behavior analogous to that discussed above.

Fig. 5

Temperature dependence of the PL peaks energy for the LWIR HgCdTe A (a), B (b), C (c), and D (d). The ${E}_{g}$ was determined from the PL and R_I measurements and compared to the Hansen relationship.

Moreover, below 80 K, extra transitions for the photodiodes become visible. Their distance from E_g is constant with temperature for all the transitions. The origin of all the PL peaks will be discussed further by analyzing the PL signal intensity as a function of the excitation laser power.

Figure 6 shows the analysis of the PL intensity dependence on the excitation power for photodiode-based structures at 18 K. Alpha (α) coefficients were determined based on the power law:

$$ I_{\mathrm{PL}} P^{{}} $$

(1)

where ${I}_{\mathrm{PL}}$ is the PL peak intensity for a given excitation power $P$, and the exponent $\alpha$ indicates the transition type.²⁷ The extracted exponent factor $\alpha$ = 1 corresponds to the band-to-band transitions, while $\alpha$ < 1 corresponds to the shallow defect transitions, as well as deep defect transitions. The results for both samples are presented in Table I.

Fig. 6

PL intensity versus excitation power for the LWIR HgCdTe A (a), and B (b) at 18 K.

Table I The exponent factor $\alpha$ for the p-type HgCdTe at 18 K

Deep defect levels with $\alpha$ = 0.75 for sample A and $\alpha$ = 0.4 for sample can be associated with mercury vacancies, which is confirmed by the literature data.²⁸ The distance of about 80 meV above the VB edge agrees with that determined by us in our previous work²⁹ of a defect level originating from the single negatively charged V_Hg :

$$ E_{T} \left( {{\text{V}}_{{{\text{Hg}}}}^{ - 1} } \right) = 144x + 56 $$

(2)

where E_T is in [meV].

The extra peaks at temperatures below 80 K suggest a shallow acceptor transition. Two PL peaks with energies of ∼5 meV and ∼10 meV below the E_g can be assigned to the acceptor states of As_Te and V_Hg, respectively. Their activation energies coincide with the theoretical relationships given by^28,30:

$${E}_{a1}\left({\text{As}}_{\text{Te}}\right)=42x+1.36-1.4\times {10}^{-5}{N}_{a}^{1/3}$$

(3)

and

$$ E_{a2} \left( {{\text{V}}_{{{\text{Hg}}}}^{0} } \right) = 91.1x - 6.93 $$

(4)

where E_a1 and E_a2 are in meV. The experimentally determined and calculated activation energies of the defect and acceptor levels are summarized in Table II.

Table II Summary of the experimental and theoretical data for the p-type HgCdTe at 18 K

As can be seen in Fig. 4, at 18 K, the dominant PL peak is related with the transition from the CB to the acceptor level, in A associated with V_Hg, while in B with As_Te. This may be due to the fact that the As incorporation in HgCdTe is higher in the (100) orientation than in the (111)B one.³¹ When As is not fully activated, it occupies the Hg site (As_Hg). Then, when the inactivated-As exists in sufficient density, V_Hg prefers to form the V_Hg–As_Hg complex. Thus, the remaining PL peak in A with the energy of ∼18 meV below the E_g can be assigned to the V_Hg–As_Hg complex.

Conclusions

The results presented in this paper show PL measurements for p-type HgCdTe LWIR epilayers grown with photodiode and photoconductor designs on GaAs substrates. Two crystallographic orientations typical for HgCdTe were analyzed: (111)B and (100). The PL spectra temperature dependence shows band-to-band and extra defect- and dopant-related transitions. The deep level with the energy distance of ∼80 meV above the VB edge is most likely associated with the single negatively charged V_Hg. It does not change its position with temperature. The measurements at cryogenic temperature show shallow acceptor levels associated with As_Te and V_Hg with the energies of ∼5 meV and ∼10 meV above the VB edge, respectively. In HgCdTe(100), due to higher arsenic activation, As_Te is the dominant acceptor dopant, while, in HgCdTe(111)B, the main acceptor level is formed by the neutral V_Hg. Moreover, due to the existence of an inactive acceptor dopant (As in the Hg sublattice), in HgCdTe(111)B there is a visible transition to the acceptor level created by the V_Hg–As_Hg complex.