MBE Growth Techniques for InAs-based nbn IR Detectors

Transcription

1 MBE Growth Techniques for InAs-based nbn IR Detectors Running title: MBE growth techniques for InAs-based nbn IR detectors Running Authors: Sidor et al. D.E. Sidor a) The Institute of Optics, University of Rochester, Rochester, New York, G.R. Savich Air Force Research Laboratory, Kirtland AFB, New Mexico, B.T. Marozas The Institute of Optics, University of Rochester, Rochester, New York, X. Du The Institute of Optics, University of Rochester, Rochester, New York, T.A. O Loughlin The Institute of Optics, University of Rochester, Rochester, New York, G.D. Jenkins Air Force Research Laboratory, Kirtland AFB, New Mexico, W.D. Hughes The Institute of Optics, University of Rochester, Rochester, New York, C.P. Morath Air Force Research Laboratory, Kirtland AFB, New Mexico, V.M. Cowan Air Force Research Laboratory, Kirtland AFB, New Mexico, G.W. Wicks The Institute of Optics, University of Rochester, Rochester, New York, a) Electronic mail: sidor@optics.rochester.edu 1

2 This manuscript describes an investigation of the effects of growth temperature on InAs epitaxial layers and InAs-based nbn detectors grown by molecular beam epitaxy (MBE). The motivation for this work is to improve the overall performance of InAsbased nbn detectors, which depends both on the bulk material quality of the individual device layers, particularly the infrared absorbing layer, as well as on the quality of the layer interfaces, particularly the interface between the absorber and barrier layers. Absorber layer bulk quality and absorber/barrier interface quality are presumably optimized by performing InAs growth at different temperatures, thus the preferred MBE growth strategy is not immediately apparent. 2µm thick InAs epitaxial layers are grown at several temperatures ranging from 420ºC to 490ºC, and are examined by differential interference contrast microscopy, atomic force microscopy, steady-state photoluminescence, and time-resolved photoluminescence measurements. 2µm thick absorber layers in nbn detectors are also grown at the same temperatures as the InAs single layers, and the resulting devices are evaluated on the basis of dark current density. Competitively high InAs material quality and low nbn dark current densities have been achieved across the range of investigated growth temperatures. The material quality of the InAs single epitaxial layers is found to improve monotonically with growth temperature over the investigated range, and likewise, the reverse saturation dark current density of the nbn detectors is found to decrease monotonically with growth temperature. nbn detectors with dark current density within a factor of 5 of Rule 07 are reported. Finally, it is noted that this work uses an InAs growth rate of 0.9 µm/hour, whereas many other studies have chosen to use InAs growth rates in the range of µm/hour. The results of this study demonstrate that high performance InAs-based detectors can be grown at this more convenient rate. I. INTRODUCTION Semiconductor-based infrared (IR) detector performance is commonly limited by noise that results from dark current mechanisms generated within the small-bandgap absorbing material of the detector. Thus, a direct approach to improving IR detector performance is to reduce the amount of dark current in the device. The InAs-based nbn 1 2

3 is an advanced IR detector structure that exhibits competitively low dark current, within an order of magnitude of the state-of-the-art described by Rule 07. 2, 3 While this is among the best performance demonstrated to date by III-V IR detectors, further improvement is expected with refinement of the molecular beam epitaxy (MBE) growth techniques used to create these devices. InAs-based nbn structures consist of an AlAsSb epitaxial barrier layer located between two InAs layers. As often happens in MBE growth of heterostructures, the two materials used in this device have different optimum growth conditions, and compromises have to be made to optimize the growth of the entire structure. In the present case, high-quality InAs and AlAsSb have been successfully grown at temperatures of ~420ºC and ~500ºC, respectively. This work therefore examines two growth options for the InAs absorber: (1) prioritize its bulk quality, but presumably sacrifice the quality of its top interface, by growing it at 420 C and including a growth interruption to heat to 500 C for AlAsSb growth; or (2) presumably sacrifice bulk quality in order to prioritize the quality of the top interface, by growing the entire structure at 500 C. The quality of the absorber layer and absorber/barrier interface both impact nbn detector performance, thus the more beneficial growth strategy is not immediately apparent. II. MBE GROWTH The growth reported in this study was conducted in a Riber 32P solid source molecular beam epitaxy reactor equipped with standard effusion sources for group III materials (Al, Ga and In) and dopants (Be and Te), and valved cracker sources for group V materials (As and Sb). The substrates were epi-ready InAs (100) exact-orientation wafers. After removal from their inert packaging, 2-inch wafers were immediately cleaved into quarters and loaded into the vacuum system. Substrates intended for subsequent growth were mounted indium-free in quarter-wafer molybdenum substrate holders, while others were stored under ultrahigh vacuum to prevent further surface oxidation. All substrates were outgassed in a dedicated prep chamber for a period of 1 hour under a constant current sufficient to heat the substrate to a final temperature of ~250ºC, and were then allowed to return to room temperature before transfer into the 3

4 growth chamber. Upon transferring substrates into the growth chamber, following only this gentle outgassing procedure and prior to heating, reflection high-energy electron diffraction (RHEED) interrogation of the substrate surface indicated the presence of only a thin native oxide layer remaining. Substrates were heated from room temperature to growth temperature at a constant rate of 10ºC/minute. Care was taken to prevent temperature overshoot and oscillations by limiting the maximum substrate heater current to the amount eventually required to maintain stability at growth temperature. A slight As 2 overpressure was supplied as the substrate temperature exceeded 250ºC, which was the maximum temperature encountered during outgassing. That overpressure was held constant until the substrates reached their maximum temperature for congruent sublimation, T CS, which is approximately 420ºC for InAs. 4 The exact substrate temperature at which As 2 overpressure is first supplied does not appear to be critical, however it is critical that the As 2 overpressure be present before the thin native oxide layer is fully removed from the substrate surface. As the substrate temperature is increased above T CS, the As 2 overpressure is also increased in a gradual manner, in an attempt to match the increased rate of As desorption from the substrate surface. With some substrate materials, oxide desorption occurs abruptly at a fixed and reproducible temperature, and observing the oxide desorption temperature by RHEED provides a convenient temperature reference. However the InAs substrates, prepared in the manner described here, did not exhibit this behavior. Rather, a gradual evolution of the RHEED pattern was observed as the substrates were heated. A (1 3) surface reconstruction generally became discernable at a temperature of 350ºC-375ºC, and continued to sharpen until the substrates reached approximately 470ºC. Others have reported similar observations and speculated that the (1 3) reconstruction is due to residual Sb present in the growth chamber, 5 a possible explanation which is reasonable in the present context. Heating beyond 470ºC leads to the removal of the (1 3) pattern and the appearance of the expected (2 4) surface reconstruction. After removal of any anomalous (1 3) pattern, substrates were returned to 420ºC. A 10 minute surface anneal was performed, followed by 10 minutes of buffer layer growth at a growth rate of 0.9 µm/hour, which was the growth rate used for all InAs layers in this study. For samples 4

5 intended for growth at T CS, layer growth was initiated immediately upon completion of the buffer; for samples intended for growth above T CS, growth was paused upon completion of the buffer layer, and the substrate temperature was increased to the desired layer growth temperature at a rate of 10ºC/minute. InAs growth is performed with an As 2 flux just sufficient to maintain stable growth in the (2 4) surface reconstruction. Informal observations suggest that this minimally sufficient As 2 flux results in the highest quality material, a sentiment which has persisted since the early days of InAs growth by MBE. 6 The procedure used to determine the minimally sufficient As 2 flux at a given substrate temperature begins by initiating InAs growth with a more-than-sufficient As 2 overpressure. The As 2 flux is then steadily decreased by closing the needle valve on the cracker source, and the surface reconstruction is closely monitored until the (2 4) (4 2) transition is observed. The needle valve position corresponding to the transition is recorded as x 1. The valve is then progressively opened again while still monitoring the surface reconstruction, and the valve position at which the original surface reconstruction returns is recorded as x 2. As 2 flux varies nearly linearly with valve position for positions in the vicinity of x 1 and x 2, so the minimally sufficient arsenic flux is taken to be provided by the needle valve position halfway between, at (x 1 +x 2 )/2. The As 2 :In beam equivalent pressure ratios determined in this way were approximately 4.8:1 for growth at 420ºC, 5.7:1 for growth at 450ºC, and 7.6:1 for growth at 490ºC. The substrate temperatures reported in this work are based on readings of the substrate thermocouple. Remote thermocouples of the type required for substrate rotation during growth are notorious for providing inaccurate measurements of the actual substrate temperature. However, observations of temperature-dependent RHEED transitions together with corresponding thermocouple readings are sufficient to correct for this inaccuracy, at least in the present case of InAs epitaxial growth on InAs substrates. Bracker et al. have reported a reconstruction phase diagram that describes the boundary between the (2 4) and (4 2) surface reconstructions as a function of substrate temperature and As 2 flux. 7 Determination of As 2 flux, e.g., by performing As-limited RHEED oscillations, and subsequently varying the substrate temperature until the surface reconstruction transition is observed, provides a means for correcting the thermocouple 5

6 readings to yield actual temperatures. The thermocouple offset was found to deviate from the actual substrate temperature determined in this way by a constant factor across the range of growth temperatures explored in this work. For example, the maximum temperature for congruent sublimation, which occurs at an actual substrate temperature of 420ºC, was consistently observed at a substrate thermocouple temperature of 295ºC; and the temperature boundary for the (2 4) (4 2) transition under an As 2 overpressure of 0.6 monolayers/second, expected to occur at an actual substrate temperature of 525ºC, was consistently observed at a substrate thermocouple temperature of 364ºC. Based on the consistency in the observed ratio of substrate temperature to substrate thermocouple readings, the actual temperatures reported here derived from thermocouple readings are believed to be accurate to within ±5ºC. III. EXPERIMENTAL Single InAs layers, and InAs-based nbn detectors, have been grown and characterized to investigate the effects of InAs growth temperature on material quality and, ultimately, on nbn device performance. InAs growths were performed at 420ºC, 450ºC, and 490ºC. Material quality is judged by inspection of the single InAs epitaxial layers: physical inspection of the top surfaces is conducted by differential interference contrast (DIC) microscopy and atomic force microscopy (AFM), and is expected to provide an indication of interface quality. Optical quality of the bulk layer material is investigated with steady-state and time-resolved photoluminescence spectra. Finally, nbn dark current is measured to determine the aggregate effect. A. InAs Epitaxial Layer Characterization The single InAs epitaxial layers were grown identically to the InAs absorber layers of the nbn detectors, that is, 2µm thick, and unintentionally n-type with an electron concentration n~ cm -3, as previously determined from room temperature Hall measurements performed on a thick (~7µm) InAs layer grown on a semi-insulating GaAs substrate, at a substrate temperature of 420ºC. These layers were first compared on the basis of the physical characteristics of their top surfaces. Large-scale surface defects of the type shown in Fig. 1, characterized by an easily observed central feature surrounded 6

7 by a dim, roughly elliptical outline with major and minor axes oriented parallel to cleavage planes, were observed by DIC microscopy. While the nature of these defects has not been characterized in any further detail here, they resemble the hexagonal pits previously described by Wang et al 8. The surface area between large-scale defects appeared mirror-smooth under all magnifications, with no discernable haze or texture (non-uniformity in the image shown is the result of digital camera noise, and was not present when viewed directly through the microscope). FIG. 1. Typical large-scale surface defects on an as-grown InAs surface. Several locations on the surface of each sample were selected at random and the densities of such large-scale defects were counted. Surface photomicrographs of the three samples are shown in Fig. 2, and defect densities are summarized in Fig. 3. The mean defect density was observed to decrease with increasing growth temperature, from a density of approximately defects/cm 2 at a growth temperature of 420ºC to defects/cm 2 at a growth temperature of 490ºC. The variability in defect density across the sample surface was also observed to decrease with increasing growth temperature. It should be noted that although a trend of improving surface quality with increasing growth temperature is shown, defect densities over the entire range of investigated growth temperatures are consistently lower than previously reported values, 9 which are on the order of 10 4 defects/cm 2. The reason for this is unknown, however differences in growth rate and As 2 overpressure could plausibly be related. 7

8 FIG. 2. Representative photomicrographs of InAs epitaxial layers grown at 420ºC (a), 450ºC (b), and 490ºC (c) showing large-scale surface defects. FIG. 3. (Color online) Large-scale surface defect density on 2µm thick InAs epitaxial layers. Error bars represent 95% confidence intervals about the mean values. Samples were also investigated by AFM in order to examine surface roughness in the regions that appeared mirror smooth under DIC microscopy. 2µm 2µm scans, pictured in Fig. 4, showed favorable step growth characteristics and root-mean-square (RMS) surface roughness values approximately equal to 0.13nm, independent of growth temperature. AFM scans of larger 20µm 20µm fields revealed a trend resembling that of the large-scale surface defects: RMS surface roughness across this larger field size decreased with increasing growth temperature, as did the variation between measurements performed at several locations over the sample surface. RMS roughness at this scale decreased from a mean value of 2.1nm on the InAs layer grown at 420ºC to 0.57nm on the layer grown at 490ºC. The results of these RMS roughness measurements over both 2µm 2µm and 20µm 20µm fields are plotted in Fig. 5. 8

9 FIG. 4. 2µm 2µm AFM scans of InAs single epitaxial layers grown at 420ºC (a), 450ºC (b), and 490ºC (c). The vertical scale extends from 0nm (black) to 2nm (white). FIG. 5. (Color online) RMS surface roughness obtained from AFM scans of InAs single epitaxial layers over 20µm 20µm fields (blue suqares) and 2µm 2µm fields (red circles). Error bars represent 95% confidence intervals about the mean values. Next, optical characteristics of the InAs layers were investigated by analyzing photoluminescence (PL) spectra. Both steady-state and time-resolved photoluminescence measurements were performed. Steady-state PL spectra obtained at a temperature of 15K are shown in Fig. 6 for the InAs layers grown at 420ºC and 490ºC. Both PL spectra show strong features centered at energies of 415 mev, corresponding to the excitonic bandgap, and 402 mev, likely corresponding to a donor-acceptor-pair (DAP) band 10. The observation of multiple spectral features in low-temperature PL spectra is generally indicative of high material quality, as defect-related states in lower-quality material will tend to broaden single peaks and obscure multiple peaks. Several comparisons can be 9

10 made between the two spectra to suggest that higher material quality has been obtained at higher growth temperature: the intensity of the excitonic bandgap peak, the ratio of intensities between the bandgap and DAP peaks, and the integrated spectral intensity are all greater for the sample grown at 490ºC than for the sample grown at 420ºC. FIG. 6. (Color online) Steady-state photoluminescence spectra of 2µm thick InAs epitaxial layers grown at 420ºC (dashed blue line) and 490ºC (solid red line), measured at 15K. Time-resolved photoluminescence (TRPL) measurements were also performed and analyzed. Under conditions of low-level optical injection, the time dependence of the TRPL signal intensity was described well by a shifted and scaled single exponential decay model of the form I PL (t)=a exp[-(t-t 0 )/τ], where A and t 0 are fit parameters and τ is the PL recombination lifetime 11. The lifetime parameter τ generally represents minority carrier recombination through Shockley-Read-Hall (SRH), radiative, and Auger processes. Under low injection the dependence of radiative and Auger recombination rates on optically generated excess carrier density leads to a reduction of those mechanisms. The minority carrier recombination lifetime determined here is therefore expected to be predominantly a SRH recombination lifetime. Recombination lifetimes obtained from TRPL spectra recorded at various temperatures are shown in Fig. 7. At measurement temperatures 150K, the recombination lifetimes vary approximately as T -1/2, as shown by the dashed fit lines. This is the expected temperature dependence for minority carrier recombination lifetimes 10

11 limited by Shockley-Read-Hall (SRH) processes 12. As SRH recombination proceeds through defect-related trap states, the generally longer recombination lifetimes observed from higher growth temperatures provides further evidence that this material is of higher crystalline quality. FIG. 7. (Color online) Minority carrier recombination lifetimes determined by fitting a single exponential decay model to low-level injection TRPL spectra obtained from InAs epitaxial layers grown at 420ºC (blue triangles), 450ºC (green circles), and 490ºC (red squares). Dashed lines are fits proportional to T -1/2. B. InAs-based nbn performance InAs-based nbn detectors have also been grown by MBE and characterized on the basis of dark current density. nbn detectors were produced with InAs layers grown at the same three substrate temperatures investigated for InAs growth in the previous section. The epitaxial structure of the nbn devices consisted of an InAs absorber layer identical to the single InAs epitaxial layers described above, that is, 2.0µm thick and unintentionally doped n-type, with an electron concentration of approximately cm -3. The absorber layer was followed by 0.1µm of lattice-matched AlAs 0.16 Sb 0.84 which was lightly Tedoped to maintain an n-type carrier concentration of ~ cm -3. The doping level is estimated from calibrations performed on InAs Hall layers grown on semi-insulating GaAs substrates, and is supported by comparison of the resulting device dark current characteristics (see Fig. 8) to recently modeled results 13. The AlAsSb barrier layer was 11

12 grown at a substrate temperature of 500ºC in all cases. Finally, an additional 0.2µm thick InAs contact layer was grown on top of the barrier layer, under the same conditions as the absorber. After each InAs absorber layer was completed, growth was paused and As 2 overpressure was maintained while the substrate temperature was increased at a rate of 10ºC/minute in preparation for AlAsSb growth. The group V fluxes were adjusted to the values required for the desired AlAsSb composition once a substrate temperature of 500ºC was achieved, and AlAsSb growth was initiated as soon as the background pressure in the growth chamber stabilized. The reverse of this process was used when switching from the barrier layer to the contact layer. The duration of the growth interruptions varied from approximately 9 minutes for InAs growth at 420ºC, to approximately 5 minutes for InAs growth at 490ºC. Dark current density was measured as a function of applied voltage at several different temperatures for each of the three nbn detectors. J-V-T characteristics for the three devices were qualitatively similar, resembling the set of measurements obtained from the nbn with InAs layers grown at 490ºC, shown in Fig. 8. All three devices developed nearly diffusion-limited behavior under moderate reverse voltage, and the dark current densities exhibited full-bandgap thermal activation energy, as shown in Fig. 9. In an attempt to make the observed trend in dark current density with absorber layer growth temperature more apparent, the measurements obtained at 200K are plotted in Fig. 10 on an absolute scale and also as multiples of Rule 07. These two figures indicate that nbn dark current density displayed the same trend of improvement with InAs layer growth temperature as did the physical and optical comparisons of the single InAs epitaxial layers. 12

13 FIG. 8. (Color online) Dark current density measured as a function of applied voltage and sample temperature for an nbn device with InAs layers grown at 490ºC. FIG. 9. (Color online) Arrhenius analysis of nbn detectors with InAs layers grown at 420ºC (blue triangles), 450ºC (green circles), and 490ºC (red squares), compared to Rule 07 (dashed line). 13

14 FIG. 10. (Color online) nbn dark current density measured at 200K. The right vertical axis shows the ratio of measured dark current density to that of Rule 07. IV. RESULTS AND DISCUSSION This work was undertaken from a starting point of having previously observed desirably high performance and low dark current in infrared detector structures employing InAs absorber layers grown by MBE at or near T CS. It was assumed that increasing the InAs growth temperature was likely to alter the bulk quality of the material, yet might improve nbn dark current characteristics by enabling a higher quality interface between the absorber and barrier layers. Interestingly, all employed methods of inspection suggest that both the bulk material quality of the InAs absorber layer as well as the quality of the interface formed with the barrier layer improved monotonically with increasing growth temperature, up to the highest investigated temperature, which was 490ºC. This result is essentially consistent with the findings of several others who have studied InAs growth in various contexts. This includes heteroepitaxial growth on mismatched GaAs substrates 8, 14 as well as more recent reports concerning homoepitaxial growth. 5, 9, These studies generally report successful InAs growth in the temperature range of 470ºC-510ºC. It should be noted that this report can not recommend a particular set of optimum growth conditions for InAs, or for InAs-based nbn detectors, because critical points for the various quality indicators used in this study were not identified in the range of 14

15 investigated InAs growth temperatures. The bulk zone temperature setting chosen for the As valved cracker source at the outset of this study eventually proved to be the limiting factor in increasing InAs growth temperature, as the maximum available As 2 flux was insufficient for stable (2 4) growth above 490ºC. The observed trends suggest that further improvement in InAs material quality and further reduction in nbn dark current density may be achievable at an InAs growth temperature above 490ºC. It is also noted that the preceding results on InAs material quality should be interpreted in the context of the particular InAs growth rate used throughout this study, which was 0.9 µm/hour. High-quality results have been obtained by growing InAs at this higher-than-usual growth rate, yet this growth rate was chosen primarily as a matter of convenience, and not as the result of systematic study. It seems reasonable to expect that any future investigations into optimal InAs growth conditions should include growth rate as an optimizable parameter. V. SUMMARY AND CONCLUSIONS InAs epitaxial layers and InAs-based nbn detectors have been grown by MBE in order to assess variations in InAs bulk material and interface quality, and in nbn dark current performance, with InAs growth temperature. Investigations of the single InAs epitaxial layers through differential interference contrast microscopy, atomic force microscopy, steady-state photoluminescence, and time-resolved photoluminescence spectra all indicate monotonic improvement in material quality with growth temperature spanning the range of 420ºC-490ºC. Likewise, nbn dark current density decreased monotonically with increasing InAs growth temperature over the same temperature range, and nbn detectors with dark current density within a factor of 5 of Rule 07 are reported. ACKNOWLEDGMENTS The authors would like to acknowledge the support of the United States Air Force Research Laboratory, the Universities Space Research Association, the Army Research Office, and a NASA Space Technology Research Fellowship. 15

16 1 S. Maimon and G. Wicks, Appl. Phys. Lett. 89, 15 (2006). 2 W. Tennant, D. Lee, M. Zandian, E. Piquette and M. Carmody, J. Electron. Mater. 37, 9 (2008). 3 W. Tennant, J. Electron. Mater. 39, 7 (2010). 4 C. Sasaoka, Y. Kato and A. Usui, Appl. Phys. Lett. 62, 19 (1993). 5 R. Contreras-Guerrero, S. Wang, M. Edirisooriya, W. Priyantha, J. Rojas-Ramirez, K. Bhuwalka, G. Doornbos, M. Holland, R. Oxland and G. Vellianitis, J. Cryst. Growth 378 (2013). 6 A. Milnes and A. Polyakov, Mater. Sci. Eng. B 18, 3 (1993). 7 A. Bracker, M. Yang, B. Bennett, J. Culbertson and W. Moore, J. Cryst. Growth 220, 4 (2000). 8 H. Wang, T. Fan, J. Wu, Y. Zeng, J. Dong and M. Kong, J. Cryst. Growth 186, 1 (1998). 9 H. Ye, L. Li, R. T. Hinkey, R. Q. Yang, T. D. Mishima, J. C. Keay, M. B. Santos and M. B. Johnson, J. Vac. Sci. Technol. B 31, 3 (2013). 10 Y. Lacroix, S. Watkins, C. Tran and M. Thewalt, Appl. Phys. Lett. 66, 9 (1995). 11 G. D. Jenkins, C. P. Morath and V. M. Cowan, Proc. SPIE 9226 (2014). 12 B. C. Connelly, G. D. Metcalfe, H. Shen and M. Wraback, Appl. Phys. Lett. 97, 25 (2010). 13 D. Z. Ting, A. Soibel, A. Khoshakhlagh, S. A. Keo, C. J. Hill, A. M. Fisher, E. M. Luong, J. K. Liu, J. M. Mumolo and B. Rafol, Proc. SPIE 9973 (2016). 14 A. Wolkenberg, T. Przesławski, J. Kaniewski, J. Bąk-Misiuk and K. Regiński, Mater. Sci. Eng. B 77, 3 (2000). 15 K. Bhatnagar, J. Rojas-Ramirez, M. Caro, R. Contreras, B. Henninger and R. Droopad, J. Cryst. Growth 425 (2015). 16 W. Xiang, G. Wang, H. Hao, Y. Liao, X. Han, L. Zhang, Y. Xu, Z. Ren, H. Ni and Z. He, J. Cryst. Growth 443 (2016). 17 Y. H. Zhang, Appl. Phys. Lett. 66, 2 (1995). 16