JEM-1355-S49
6/7/05
8:41 AM
Page 963
Journal of ELECTRONIC MATERIALS, Vol. 34, No. 6, 2005
Special Issue Paper
Growth of Very Low Arsenic-Doped HgCdTe D. CHANDRA,1,2 D.F. WEIRAUCH,1 H.F. SCHAAKE,1 M.A. KINCH,1 F. AQARIDEN,1 C.F. WAN,1 and H.D. SHIH1 1.—DRS Infrared Technologies, Dallas, TX 75374. 2.—E-mail:
[email protected]
Arsenic is known to be an amphoteric impurity that may occupy either sublattice in HgCdTe depending upon sample annealing. As an acceptor in low concentrations, it offers several features that are attractive for the fabrication of certain n-on-p detector diode structures. The epitaxial growth of arsenic-doped HgCdTe from a Te-rich melt can fulfill the requirements for application in a variety of devices where low vacancy concentrations and low defect densities are critical requirements in minimizing dark currents. These devices may include the high operating temperature (HOT) detectors operated in a strong nonequilibrium and reverse bias mode to suppress the Auger-generated dark currents. For the materials’ growth process to be effective, the segregation coefficient determining the incorporation of arsenic from the Te-rich melt needs to be established. This coefficient was measured during these investigations and was observed to vary with arsenic concentration. Within the range of interest, this parameter varied between 8 106 and 1 104. These extremely small values limit the doping that can be achieved to <5 1016 cm3 in the grown epifilm. Furthermore, the large addition of arsenic to the melt, necessitated by the extremely small segregation coefficients, leads to a condition where the concentration of arsenic in the liquid-phase epitaxy (LPE) nutrient melt exceeds that of cadmium. The melt chemistry, phase diagram, and epigrowth process fundamentally change as a result. This new epigrowth process was developed and tuned during these investigations. For acceptor levels at 1 1015 cm3 and lower, the growth of arsenic-doped HgCdTe from a Te-rich LPE melt has been determined to be an extremely reproducible, powerful, and controllable technique. Key words: HgCdTe, As, doping
INTRODUCTION Incorporation of arsenic as an extrinsic acceptor dopant in mercury cadmium telluride offers several powerful advantages recognized and exploited in a wide variety of HgCdTe (MCT)–based infrared devices for a significant period of time. This dopant is stable, with a very low diffusion coefficient under metal-saturated conditions,1,2 and associated with some of the best device performance parameters, with minority carrier lifetimes among the highest among extrinsic dopants.3 Of particular importance is the final equilibrium location of arsenic. Unlike a spectrum of acceptor dopants from group IB or group IA, which occupy sites in the metal sublattice and which are thereby rendered unstable from a very high con(Received October 7, 2004; accepted February 28, 2005)
centration of vacancies in this sublattice, arsenic, belonging to group VA, occupies sites in the tellurium sublattice, with the site vacancy concentrations at levels infinitely lower. However, incorporation of arsenic at very low levels, ranging to 1 1014 cm3 or lower, reproducibly, and without incorporating material damages, required for high operating temperature (HOT) applications remains a challenging proposition. Segregation coefficients for incorporating arsenic employing Hgrich liquid-phase epitaxy (LPE) vary between 0.1 and 10, depending on the growth temperature and the Te atomic fraction in the melt.4 This renders arsenic doping employing Hg-rich LPE progressively easier with increasing arsenic levels, particularly above 1 1016 cm3. With gradually lower targeted doping levels, however, arsenic incorporation employing Hgrich LPE, while still possible, becomes progressively 963
JEM-1355-S49
6/7/05
8:41 AM
Page 964
964
Chandra, Weirauch, Schaake, Kinch, Aqariden, Wan, and Shih
more difficult, since the level of arsenic in the melt itself has to be maintained at a comparable level as in the grown MCT epifilm. Incorporation of very low levels of arsenic in MCT, employing either of the vapor growth techniques, metal organic chemical vapor deposition (MOCVD) or molecular beam epitaxy (MBE), also has been demonstrated,5,6 but with severe potential and real limitations in reproducibility. Indirect introduction of arsenic, employing ion implantation, required for specific applications, is also associated with specific limitations. In particular, most of these methods of introducing arsenic are associated with very high defect concentration levels, ranging to significantly above 1 106 cm2. EXPERIMENTAL PROCEDURES A highly reproducible method of introducing arsenic in mercury cadmium telluride has been developed using Te-rich LPE. The segregation coefficient for incorporating arsenic from a Te-rich melt has been measured during the present investigations. The actual levels of arsenic incorporated in the film, required to determine the segregation coefficient, were determined from the secondary ion mass spectrometry (SIMS) measurements. These were performed by Charles Evans and Associates using a Cs primary ion beam and employing CsAs positive ions. The sensitivity of this technique permitted a determination of arsenic levels present at or above 2 1014 cm3. The segregation coefficient appears to slowly vary with arsenic concentration measured in the epifilms grown. The magnitude of the coefficient appears to increase with increasing arsenic concentration, as measured for an arsenic level at slightly above 2 1014 cm3 in the epitaxial film to 1.1 1015 cm3 in the epitaxial film. These magnitudes are shown in Fig. 1. With increasing segregation coefficients with increasing concentrations of arsenic, as apparent from the measurements during the present investiga-
Fig. 1. Variation of the segregation coefficient of arsenic for growth from the Te-rich LPE nutrient melt with arsenic concentration in the MCT epifilm.
tions, the rate at which the arsenic concentration is required to increase in the nutrient melt to lead to increasing levels of arsenic in the grown epifilm progressively decreases. However, the increase still appears to be insufficient to enable the use of this method to incorporate arsenic for ranges significantly above 5 1016 cm3 in HgCdTe. This process however appears to become increasingly powerful with progressively lower arsenic levels, ranging downward from arsenic concentrations at 5 1015 cm3 to levels as low as 1 1013 cm3 in the grown epifilm. The Te-rich LPE melt appears to undergo basic transformation with the addition of arsenic. The amount of arsenic to be added to the melt may attain levels comparable to the level of cadmium, which is a substituent not a dopant. The arsenic could be added easily as a high-purity element, which dissolved easily and homogeneously in the melt. The distribution coefficient of cadmium appears to have been modified significantly by the addition of arsenic, whereas the distribution coefficient of mercury does not appear to have been measurably influenced. The viscosity of the nutrient melt appears to increase significantly. Basic modifications in growth methods were necessitated and established during these investigations. A more severe etchback of the CdZnTe substrate surface appears to be a necessity to attain a defectfree growth. The LPE process employed remained the infinite melt Te-rich LPE technology developed earlier.7 The growth rate appeared to decrease by more than 50% when compared to rates observed for the growth of MCT epifilms from melts prior to the addition of arsenic. Specular and inclusion-free growth were attained. RESULTS AND DISCUSSION The thickness of the epifilms grown on generally lattice-matched CdZnTe substrates varied between 40 µm and 85 µm. The substrate-epi interdiffused region, determined from a measurement of the compositional profile at the substrate-epi interface, ranged to between 7.5 µm and 12 µm, approximately 40 to 100% wider than widths measured on epifilms grown from melts not containing arsenic. The greater widths appear to stem exclusively from the growth process itself, with durations of the epigrowth process extending to 100% to 150% greater than corresponding periods required to grow epifilms from melts not containing arsenic. The composition of the films was targeted at both the midwavelength infrared (MWIR) band, with the CdTe mole fraction falling between 0.27 and 0.30, and the long-wavelength infrared (LWIR) band, with the CdTe mole fraction falling between 0.21 and 0.23. The targeted arsenic concentration in the epifilms grown was raised in steps to demonstrate the feasibility of attaining arsenic concentrations at 2.5 1014 cm3, 5.0 1014 cm3, 7.5 1014 cm3, and 1.1 1015 cm3, respectively. The presence of arsenic was confirmed directly by SIMS as indicated above, and indirectly by Hall measurements, following an activation anneal regime.
JEM-1355-S49
6/7/05
8:41 AM
Page 965
Growth of Very Low Arsenic-Doped HgCdTe
965
Several aspects appear to be necessary to assess the precise “activation annealing” required to activate the incorporated arsenic as an acceptor, particularly for the very low levels of arsenic under consideration during the present investigations. Some of the observations, however, may be relevant for arsenic present at even higher concentrations, extending over a broader range. Arsenic is an amphoteric impurity in Hg1–xCdxTe: under Te-rich conditions, it tends to occupy metal sublattice sites and may act as a donor, while under mercury-rich conditions, it tends to occupy the tellurium sublattice sites and act as an acceptor.8 An investigation of this equilibrium under selected mercury pressures spanning the entire range from the tellurium-saturated phase limited to the mercury-saturated phase limit9 reveals that a majority of the arsenic remains on the sites in the tellurium sublattice even under tellurium-saturated conditions for arsenic concentrations less than 2 1016 cm3. Quantitatively, more than 50% of the arsenic incorporated during the epigrowth already appears to occupy sites in the Te sublattice immediately following the growth and prior to any “site transfer” activation anneals.9 However, a portion of the total arsenic, ranging to less than 50%, did not appear to be activated as acceptors.9 The true form of this arsenic has not been conclusively established, but can be approximated by assuming it to be behaving similar to a donor.8 A “site transfer” activation anneal was performed to convert this arsenic into acceptors. This could be represented by the relationship8,9 2Hgg TeTe As M ↔ AsTe HgM HgTe 2h
required to attain equilibrium, as determined from the “equilibrium” measurements discussed above. A few were also annealed at 475°C. These films were then stoichiometrically adjusted to eliminate the resultant metal vacancy concentration by a slow cooling “anneal” under metal-saturated conditions.7 This is schematically displayed in Fig. 2. These slow-cooled samples were compared with specimens, which, following the activation anneals, were stoichiometrically adjusted by a constant temperature anneal, at either 225°C or 250°C, under Hg-saturated conditions, and then air quenched. No significant differences were observed in the measured electrical properties between these two groups. Figure 3 displays the arsenic level directly determined by SIMS using a Cs primary ion beam and employing CsAs positive ions. The sensitivity of the technique permitted a determination of arsenic levels
(1)
In this process, mercury from the ambient first displaces an arsenic atom (donor) occupying a metal sublattice site. This arsenic atom in turn displaces a tellurium atom, taking residence on the tellurium sublattice (acceptor). A second mercury atom combines with this excess tellurium atom at a surface or other crystalline discontinuity.8 It is possible that the time to activate this arsenic would depend on the arsenic concentration itself,8 in addition to the usual parameter of temperature. This might also depend on the mercury back pressure employed during this anneal.8,9 Despite some uncertainty in the understanding of arsenic activation, and acknowledging that the description given by Eq. 1 is only approximate, it is reasonable to assume that it would be dependent on (a) the concentration of arsenic, or more precisely the concentration of “unactivated” arsenic; and (b) the rate of Te out-diffusion to a sink,8 which could be either the surface or a defect. Hence, an order of magnitude estimate can be made of the time required to completely activate this arsenic. This time ranged to less than 1 h for anneal temperatures 400°C, in conformity with experimental results.8 The “site transfer” activation anneals were performed at either 400°C or 425°C, under Hg-saturated conditions for time periods 5 to 10 longer than
Fig. 2. Activation anneal path on phase limits for Te-rich LPE-grown MWIR.
Fig. 3. Comparison between arsenic determined by SIMS and net acceptor concentration determined by Hall at 77 K on the same specimen.
JEM-1355-S49
6/7/05
8:41 AM
Page 966
966
Chandra, Weirauch, Schaake, Kinch, Aqariden, Wan, and Shih
at or above 2 1014 cm3, as discussed in details in our earlier investigation.9 Figure 3 also displays the net acceptor concentration determined by Hall at 77 K on the same specimen. The agreement between the two magnitudes confirms 100% activation of arsenic. Defect Density The dislocation density variations in thick HgCdTe epifilms grown by LPE from Te-rich melts have been found to depend on the film growth temperature, substrate dislocation density, epitaxial film thickness, and postgrowth annealing in Hg vapor.7 Use of a high-temperature preanneal in Hg vapor was observed to be necessary to eliminate any possibility of dislocation multiplication during low-temperature annealing.7 During the present investigation studies, results appear to conclusively indicate that the activation anneal under Hg-saturated conditions at or above 400°C itself fulfills the requirement of the preanneal in Hg vapor. No dislocation multiplication was noted at any region of the arsenic-doped epifilms following the anneals. The variation of the dislocation density within the epifilms with increasing distance from the episubstrate interface followed trends very similar to epifilms grown from melts not containing arsenic and reported earlier.7 In fact, the dislocation density measured within the films fell to values below those in the substrates for film thicknesses exceeding 60 µm.7 Figure 4 displays the dislocation density within a representative specimen. The etch pit density corresponds to the status following all the anneals. The dislocation densities ranged between 3 104 cm2 and 7 104 cm2 for all the specimens measured, though the median for the arsenic-doped films determined from a limited sampling at 4.5 104 cm2 appears to be the lowest among the etch pit densities measured on arsenic-doped films grown by any techniques anywhere.
Fig. 5. Lifetimes measured on arsenic-doped films grown from Te-rich LPE at DRS. Comparison with lifetimes measured on arsenic-doped epifilms grown by other methods.
Figure 5 displays the lifetimes measured on the annealed epifilms using the method of microwave reflection. Some of the lifetimes measured during the present work fall among the highest measured on arsenic-doped films, with the best magnitude measured at 32.2 106 s at 77 K. The lifetimes measured during the present investigations can be compared with selected lifetimes measured on arsenic-doped films grown by other techniques, including Hg-rich LPE, MBE, and MOCVD. Postgrowth anneals have reduced the metal vacancy concentration in all these films to magnitudes significantly less than 1 1014 cm3. The lifetimes measured followed generally a Shockley–Read–Hall dependence (CAs1), rather than an Auger 7 dependence (CAs2). The lifetimes measured for individual samples appeared to depend strongly on the surface preparation for a significant number of the samples. Presence of even small surface damages appeared to lead to significant degradation of lifetimes. During our investigation, a selected batch of annealed arsenicdoped films displaying lifetimes between 0.2 µs and 0.4 µs were subjected to specific surface treatments at room temperature, which included etching with bromine-methanol, and did not lead to removals of surfaces exceeding 0.2 µm. The lifetimes of six out of the eight films so treated increased by 10. The lifetimes displayed in Fig. 4 do not reflect these new results, but instead indicate the lifetimes measured on the as-annealed samples. PRELIMINARY DEVICE RESULTS
Fig. 4. Dislocation density (etch pits) for arsenic-doped HgCdTe from Te-rich LPE following all anneals.
Fabrication of HOT detectors has been initiated on these films. These devices would be similar to the excluded/extracted diodes suggested by Elliott.10 A schematic diagram of the unit cell and the corresponding high density vertically integrated photodiode (HDVIP) unit cell are shown in Fig. 6. These
JEM-1355-S49
6/7/05
8:41 AM
Page 967
Growth of Very Low Arsenic-Doped HgCdTe
967
Fig. 6. HOT MWIR structure and corresponding HDVIP unit cell.
arsenic-doped films prepared during the present investigations. The first preliminary results are shown in Fig. 7. This shows the I-V characteristics of an As-doped MWIR device, with the As level at 1 1015 cm3, as a function of temperature. REFERENCES
Fig. 7. Preliminary I-V characteristics as a function of temperature for As-doped MWIR HDVIP diode. The temperatures were increased in 20 K steps to 280 K, and then increased further to 295 K.
devices will be operated in a strong nonequilibrium, reverse bias mode, such that the Auger-generated dark currents are suppressed.10 Hence, the material quality of the low-doped low-defect active area of the device would be critical in the performance of these devices. These have been fabricated from
1. D. Chandra, M.W. Goodwin, M.C. Chen, and J.A. Dodge, J. Electron. Mater. 22, 1033 (1993). 2. D. Chandra, M.W. Goodwin, M.C. Chen, and L.K. Magel, J. Electron. Mater. 24, 599 (1995). 3. Y. Nemirovsky and R. Fastow, Properties of Mercury Cadmium Telluride, EMIS Datareview Series No. 10, ed. P. Capper (INSPEC, U.K.: IEE, 1994), p. 233. 4. T. Tung, J. Cryst. Growth 86, 161 (1988). 5. P. Capper, B.C. Easton, P.A.C. Whiffin, C.D. Maxey, and I. Kenworthy, Mater. Lett. 6, 365 (1988). 6. C.J. Summers, R.G. Benz, B.K. Wagner, J.D. Benson, and D. Rajavel, Proc. SPIE 1106, 2 (1989). 7. D. Chandra, J.H. Tregilgas, and M.W. Goodwin, J. Vac. Sci. Technol. B9, 1852 (1991). 8. H.F. Schaake, J. Appl. Phys. 88, 1765 (2000). 9. D. Chandra, H.F. Schaake, M.A. Kinch, F. Aqariden, C.F. Wan, D.F. Weirauch, and H.D. Shih, J. Electron. Mater. 31, 715 (2002). 10. C.T. Elliott, Properties of Mercury Cadmium Telluride, EMIS Datareview Series No. 10, ed. P. Capper (INSPEC, U.K.: IEE, 1994), p. 339.