You are on page 1of 3

APPLIED PHYSICS LETTERS 94, 111910 2009

Semiconductor point defect concentration proles measured using coherent acoustic phonon waves
A. Steigerwald,1,a Y. Xu,1 J. Qi,1 J. Gregory,1 X. Liu,2 J. K. Furdyna,2 K. Varga,1 A. B. Hmelo,1 G. Lpke,3 L. C. Feldman,1,4 and N. Tolk1
1 2

Department of Physics, Vanderbilt University, Nashville, Tennessee 37203, USA Department of Physics, University of Notre Dame, Notre Dame, Indiana 46556, USA 3 Department of Applied Science, College of William & Mary, Williamsburg, Virginia 23187, USA 4 Institute of Advanced Materials, Devices, and Nanotechnology, Rutgers University, New Brunswick, New Jersey 08901, USA

Received 29 January 2009; accepted 19 February 2009; published online 19 March 2009 Coherent acoustic phonon interferometry is used to quantitatively measure depth-dependent point defect concentrations in semiconductor systems with a depth range of the order of tens of microns. Using time-resolved pump-probe techniques, the optical response of ion-beam irradiated GaAs crystals is analyzed as a function of defect concentration ranging over four orders of magnitude. Varying the ion dose quantitatively relates changes in the optical response to local defect concentrations. Thermal annealing is shown to reduce the effect on the optical response, indicating recovery of the crystal lattice through self-interstitial-vacancy recombination. 2009 American Institute of Physics. DOI: 10.1063/1.3099341 Accurate determination of defect concentrations and their effects is an important and often formidable task in characterizing doped, ion-implanted, or radiation-damaged semiconducting devices. Techniques to detect and quantify depth-dependent concentrations can be technically difcult, have depth-limits, and may be inherently destructive in nature. Here we use an optical technique known as coherent acoustic phonon CAP interferometry to nondestructively measure absolute defect proles in semiconductors. CAP interferometry, also known as picosecond ultrasonics, is an optical technique wherein a traveling acoustic phonon wave is utilized to sample, layer by layer, the optical and mechanical properties of bulk semiconductors and nanostructures. Typically, samples of interest are deposited with a thin 10 nm capping layer, which acts to absorb an intense, ultrafast optical pulse, launching a strain wave consisting of CAPs into the sample, with frequencies near 100 GHz Refs. 1 and 2 and spatial dimensions comparable to the thickness of the capping layer. A time-delayed probe pulse partially reects from both the surface of the sample and the region where the strain wave instantaneously perturbs the optical properties of the sample. The CAP wave travels at the speed of sound through the sample, causing the reected probe components to act as an interferometer and resulting in a long-lasting oscillatory component in the optical response, whose amplitude, phase, period, and attenuation are analyzed. CAP has been used to study the optical,35 mechanical,6,7 and electronic8 properties of different materials. In semiconductors, the amplitude is sensitive to the probe photon energy, peaking strongly when probed near the band-gap,4 and the acoustic waves have been shown to persist hundreds of microns with limited dispersion.2,3 Large differences in the characteristic oscillatory response amplitude, attenuation, etc. have been observed between similar materials,7 reecta

Author to whom correspondence should be addressed. Electronic mail: andrew.d.steigerwald@vanderbilt.edu.

ing the sensitivity of this technique to the complex index of refraction N and the speed of sound of a material Vs. In this letter we exploit the sensitive dependence on N to measure continuously varying defect concentrations in GaAs crystals as a function of depth, using near-band-gap photon energies to maximize the sensitivity to small perturbations in local optical properties caused by defects. Defect concentrations in GaAs specimens were created through He+ ion implantation over a wide range of doses, reliably creating damage proles consisting of vacancy and interstitial defects,9 which can be simulated using the transport of ions in matter TRIM code.10 Experiments were carried out using GaSb20 nm/GaAs heterostructures grown via MBE. The GaSb capping layers were grown at 480 C on GaAs 100 semi-insulating substrates, where the GaAs substrate was rst heated to 600 C to remove oxidation. The process was carried out in a Riber 32 MBE system, equipped with a Veeco SUMO source and a valved cracker source for the Sb, at a rate of approximately 0.8 ML/s. Each sample was irradiated by a helium He4 ion beam generated using a HVE AN-2000 Electrostatic Van de Graaff accelerator with currents near 10 nA, and total doses varying between 1012 and 1014 ions / cm2, and incident energies of 300600 keV. A 75 MHz Mira 900 Ti:sapphire laser with a pulse width of 120 fs was used in a standard time-resolved pump-probe setup with typical power ratios of 30:3 mW to optically excite charge carriers in the GaSb layers. Figure 1a shows the experimental setup. Figure 1b shows the transient reectivity response at probe wavelengths of = 880 nm for an undamaged sample and a sample implanted with 325 keV He+ at an irradiation dose of 3.5 1013 cm2. Both signals exhibit a typical fast initial rise due to electronic excitation and relaxation, with a superimposed with a long-lived oscillatory tail. Figure 1c shows the subtracted oscillatory response, whose period can be described by T = / 2nVs cos , with oscillation period T, probe wavelength , refractive index n,
2009 American Institute of Physics

0003-6951/2009/9411/111910/3/$25.00

94, 111910-1

Downloaded 17 May 2012 to 129.59.117.79. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions

111910-2

Steigerwald et al.

Appl. Phys. Lett. 94, 111910 2009

FIG. 1. Color online Showing the a experimental setup, b total timeresolved pump-probe optical response of the undamaged and damaged samples, and c the subtracted oscillatory responses displaying strong amplitude modulation observed in the damaged sample in the rst 400 ps, as compared to the undamaged response.

FIG. 3. Color online a Experimental defect concentration prole for a sample irradiated at 325 keV at a dosage of 7 1013 ions / cm2. Peak defect concentration observed to be near 1.1 1020 cm3, agreeing well with simulated proles for total damage black and helium ion distribution red, enhanced 25 from TRIM code. b Experimentally measured defect proles before blue and after red 2 h of thermal annealing at 300 C.

and probe beam angle here 0.1 Both samples exhibit amplitude attenuation due to natural attenuation of the probe light in the bulk GaAs, with absorption coefcient of = 210 cm1 for = 880 nm.11 The period and phase remain identical for both samples, with a clear difference in oscillation amplitudes. A strong amplitude modulation in the irradiated sample at early delay times of up to tdelay = 350 ps is seen, corresponding to 1.6 m into the GaAs for 100 GaAs, Vs 4.73 nm ps1.11 The two responses become concurrent after tdelay = 350 ps, indicating that only a region near the surface was perturbed. The peak difference in amplitude Rmax between the samples is roughly 35% occurring at tdelay = 240 ps or 1.1 m. From the TRIM code, the simulated defect concentration at this point is estimated to be of the order of 5 1019 cm3. To extract a prole of the inuence of defect populations on the optical CAP response, the difference between the undamaged and damaged amplitudes is plotted versus depth. Figure 2a plots the change in oscillation amplitudes R / R for the sample irradiated at 325 keV as a function of depth,

FIG. 2. Color online a Difference in oscillatory amplitudes between irradiated sample and undamaged samples, with comparison to the simulated peak defect concentration solid black curve. b Experimental dependence of peak amplitude modulation Rmax vs peak defect concentration max as predicted from TRIM calculations. This dependence may be used to transform raw data in a into absolute defect concentration proles.

with the predicted total defect simulated by the TRIM code. The experimental and simulated data peak at the same depth, with a maximum amplitude difference near R / R = 35% and a FWHM of about 750 nm. In Fig. 2a the experimental prole is broader than the simulated damage proles, indicating a nonlinear dependence between the optical response and the inuence of local defect concentrations. An experimental correlation changes in the optical response, and peak defect concentrations was established by varying irradiation doses, resulting in predicted peak defect concentrations between 5 1018 and 1 1021. The optical CAP response for each sample was then collected, and the peak amplitude difference e.g., 0.35 for the sample from Fig. 2a, denoted as Rmax, was plotted versus the maximum simulated defect concentration e.g., 5 1019 cm3 in Fig. 2a, denoted as max for each sample. The result is plotted in Fig. 2b and displays a logarithmic dependence of R versus defect concentration. This dependence may be used to transform the raw data such as those in Fig. 2a into quantitative depth-dependent defect concentration proles. Figure 3a shows the experimental defect concentration prole for a sample irradiated with 325 keV He+ ions at a total dosage of 7 1013 ions / cm2. The measured maximum defect concentration is near 1 1020 defects / cm3, peaking near 1100 nm into the GaAs substrate. After calibration the FWHM of the experimental prole has been reduced to 300 nm. The calibrated experimental prole agrees with simulations in peak depth, shape, and absolute defect concentration. Experimental defect proles for other irradiation dosages yielded similar comparisons with simulations. To test the effect of thermal annealing, another sample with an irradiated dose of 7 1013 ions / cm2 was placed in an annealing furnace with an argon ow. The sample was annealed for 2 h at 300 C. Figure 3b shows the experimental defect concentrations obtained before and after the annealing step. The annealed defect concentration is almost an order of magnitude lower than the unannealed sample. The above analysis assumes that local defect concentrations modify the strength of the optical response as the probe beam reects off the traveling CAP wave. The reectivity

Downloaded 17 May 2012 to 129.59.117.79. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions

111910-3

Steigerwald et al.

Appl. Phys. Lett. 94, 111910 2009

response may be characterized as R / R Aet/ sin2t / T + , with time t, period T, damping time proportional to penetration depth at probe, phase , and oscillation amplitude A1. A is then proportional to A

N N Eg Eg


Eg
2

k Eg

Eg ,
1

where N is the complex refractive index, is the z component of the strain tensor, Eg is the band-gap, and n and k are the real and imaginary parts of the refractive index, respectively.1,5 The reduction in oscillatory amplitudes caused by radiation damage in the GaAs lattice can be attributed to a modication of the absorption coefcient induced by lattice defects or b changes through the derivative terms n , k / Eg in the above equation. In Fig. 1c the optical response of the damaged sample is seen to become concurrent with the undamaged sample after roughly tdelay = 350 ps in both amplitude and period, indicating that any difference in the absorption coefcient in the damaged layer is below experimental resolution. Linear absorption measurements were performed, though no differences could be resolved between the two samples, likely due to the small thickness of the damaged layer compared to the bulk. Therefore it can be concluded that amplitude modulation may be wholly attributed to changes in the derivative terms. This effect may be explained by the introduction of an impurity band during the irradiation process, which overlaps with the valence and conduction bands and leads to the formation of a band tail near the 1.43 eV GaAs electronic transition.12 Band tailing broadens the 1.43 eV band edge, which directly decreases the magnitude of the n , k / Eg terms, as seen in the decrease in amplitude in the damaged region in Fig. 1c. In conclusion, we have demonstrated the ability of CAP interferometry to detect and measure depth-dependent defect proles in radiation-damaged GaAs samples. The extracted proles show excellent comparison with simulations. By correlating changes in the band-structure caused by lattice defects with the measured optical response, quantitative defect

proles were obtained. After thermal annealing, the defect proles were reduced by an order of magnitude, indicating healing of the crystal lattice. This technique is both noninvasive and nondestructive, is only limited by the transparency of the material to the probe pulse, spans at least four orders of magnitude in defect concentration, and has a depth range of the order of tens of micrometers and 30 nm depth resolution throughout the probed range. The upper depth limits achieved here surpass the conventional use of MEIS and ion channeling measurements, which also create similar damage during analysis. The sensitivity of CAP makes it a robust defect detection technique, and from a materials standpoint, the optical nature of the CAP technique allows it to be used in a wide variety of semiconductor and insulator materials. This work was supported at Vanderbilt by DOE through Grant No. DE-FGO2-99ER45781 and at Notre Dame by NSF Grant No. DMR06-03752.
C. Thomsen, H. T. Grahn, H. J. Maris, and J. Tauc, Phys. Rev. B 34, 4129 1986. 2 B. C. Daly, T. B. Norris, J. Chen, and J. B. Khurgin, Phys. Rev. B 70, 214307 2004. 3 A. Devos and R. Cote, Phys. Rev. B 70, 125208 2004. 4 J. K. Miller, J. Qi, Y. Xu, Y. J. Cho, X. Liu, J. K. Furdyna, I. Perakis, T. V. Shahbazyan, and N. Tolk, Phys. Rev. B 74, 113313 2006. 5 S. Wu, P. Geiser, J. Jun, J. Karpinski, and R. Sobolewski, Phys. Rev. B 76, 085210 2007. 6 H. Tanei, N. Nakamura, H. Ogi, H. Hirao, and R. Ikeda, Phys. Rev. Lett. 100, 016804 2008; A. Devos, M. Foret, S. Ayrinhac, P. Emery, and B. Rufe, Phys. Rev. B 77, 100201R 2008. 7 H. Y. Hao and H. J. Maris, Phys. Rev. B 63, 224301 2001. 8 T. Dehoux, N. Chigarev, C. Rossignol, and B. Audoin, Phys. Rev. B 77, 214307 2008; O. B. Wright, B. Perrin, O. Matsuda, and V. E. Gusev, ibid. 64, 081202 2001. 9 J. S. Williams, Mater. Sci. Eng., A 253, 8 1998; J. Tesner and M. Natasi, Handbook of Modern Ion Beam Analysis Materials Research Society, Pittsburgh, PA, 1995. 10 J. F. Ziegler, J. P. Biersack, and U. Littmark, The Stopping and Ranges of Ions in Solids Pergamon, New York, 2000. 11 M. R. Brozel and G. E. Stillman, Properties of Gallium Arsenide, 3rd ed. Institution of Engineering and Technology, London, United Kingdom, 1996. 12 M. O. Manasreh, D. C. Look, K. R. Evans, and C. E. Stutz, Phys. Rev. B 41, 10272 1990; P. Van Mieghem, Rev. Mod. Phys. 64, 755 1992.
1

Downloaded 17 May 2012 to 129.59.117.79. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions

You might also like