Real time in situ composition control of InGaAs lattice matched to InP by an 88wavelength ellipsometer C.-H. Kuo, M. Boonzaayer, M. DeHerrera, T. Kyong, Y.-H. Zhang, B. Johs, and J. S. Hale Citation: Journal of Vacuum Science & Technology B 16, 1484 (1998); doi: 10.1116/1.589971 View online: http://dx.doi.org/10.1116/1.589971 View Table of Contents: http://scitation.aip.org/content/avs/journal/jvstb/16/3?ver=pdfcov Published by the AVS: Science & Technology of Materials, Interfaces, and Processing Articles you may be interested in Optical properties of Si-doped and Be-doped InAlAs lattice-matched to InP grown by molecular beam epitaxy J. Appl. Phys. 114, 103504 (2013); 10.1063/1.4820519 In situ determination of InGaAs and GaAsN composition in multiquantum-well structures J. Appl. Phys. 101, 033533 (2007); 10.1063/1.2435065 Real-time composition control of InAlAs grown on InP using spectroscopic ellipsometry J. Vac. Sci. Technol. B 18, 1435 (2000); 10.1116/1.591398 Isotropic dielectric functions of highly disordered Al x Ga 1−x InP (0x1) lattice matched to GaAs J. Appl. Phys. 86, 2025 (1999); 10.1063/1.371003 Closed-loop control of composition and temperature during the growth of InGaAs lattice matched to InP J. Vac. Sci. Technol. B 17, 1237 (1999); 10.1116/1.590729 Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 209.147.144.20 On: Fri, 13 Feb 2015 01:17:32 Real time in situ composition control of InGaAs lattice matched to InP by an 88-wavelength ellipsometer C.-H. Kuo,a) M. Boonzaayer, M. DeHerrera, T. Kyong, and Y.-H. Zhang Department of Electrical Engineering, Center for Solid State Electronic Research, Arizona State University, Tempe, Arizona 85287-6206 B. Johs and J. S. Hale J. A. Woollam Co., Inc., 645 M Street, Lincoln, Nebraska 68508 ~Received 29 October 1997; accepted 23 March 1998! We have compiled the optical constants database for Inx Ga12x As which covers the composition range from 0.51 to 0.55 and the temperature range from 400 to 525 °C. The InP substrate temperature was monitored by diffusive reflectance spectroscopy during the growth of the epitaxial layer. Ellipsometry was used to monitor the Inx Ga12x As composition over the entire temperature and composition range of the database. The composition monitored by ellipsometry is within 0.002 from the high resolution x-ray data with the exception of growth temperature at 400 °C which is 0.005. We have also demonstrated the real time in situ feedback control of the Inx Ga12x As composition during epitaxial growth by using ellipsometry. The absolute accuracy of the Inx Ga12x As composition from the controlled experiment is 0.002. We can use this database to grow thick Inx Ga12x As layers grown on the InP substrates and can also use this as an in situ tool to fine tune the Inx Ga12x As composition before the growth of the complicated structure. © 1998 American Vacuum Society. @S0734-211X~98!15503-8# I. INTRODUCTION The ideal tool to be used in the molecular beam epitaxial ~MBE! growth experiment should be nondestructive, in situ, and should be able to monitor the growth of the epitaxial layer with the substrate rotation on. The thickness and composition information should be obtained during the growth to be used in the real time feedback control of either composition or thickness. Various optical techniques like reflectance difference spectroscopy,1 normal incidence reflectance spectroscopy,2 and ellipsometry3,4 were used as in situ tools in both MBE and metalorganic chemical vapor deposition ~MOCVD! experiments. From the recent development of the multi-wavelength ~88-wavelength! ellipsometer, ellipsometry turns out to be the best in situ technique to obtain both thickness and composition information during MBE growth. The ellipsometer has been used not only in the growth thickness control3 for AlAs/GaAs distributed Bragg reflectors in the MBE chamber, it has also been used in the thickness control of the etching experiment5–7 of SiO2 and Si3N4 on the Si substrate. This will give the advantage of determining the SiO2 or Si3N4 layer thickness during the etching process in real time and no time consuming sample preparation like cross-section transmission electron microscopy will be needed. We have also demonstrated the strong correlation of ex situ ellipsometric data with excimer laser annealed poly-Si thin film crystallinity and grain size.8 The ellipsometer has also been used to determine the poly-Si crystallinity in real time right after the growth of a-Si in the plasmaenhanced chemical vapor deposition ~PECVD! chamber.9 It is important to be able to monitor the quality of the poly-Si thin film which is used in thin film transistors ~TFTs!. In the a! Present address: Space Vacuum Epitaxy Center, Univ. of Houston, Houston, TX 77204. Electronic mail: chkuo@enuxsa.eas.asu.edu 1484 J. Vac. Sci. Technol. B 16„3…, May/Jun 1998 mass production environment, it is possible to monitor the poly-Si property right after the excimer laser annealing process, which in turn will effect the device performance of the TFT. However, the most challenging work in the real time in situ control experiment is to control the composition of the ternary layer during the epitaxial growth in the MBE chamber. This has been demonstrated by Aspnes et al.10 in their real time in situ composition control of the Inx Ga12x As parabolic quantum well experiment in MOCVD. We are going to demonstrate in this article the real time in situ composition control of the Inx Ga12x As layer lattice matched to InP in MBE. II. EXPERIMENT AND RESULT The schematic of the DCA 450 MBE chamber used in this experiment is shown in Fig. 1. There are three distinct features in this MBE chamber compared to the traditional MBE system. Substrate temperature was measured by direct recoil spectroscopy ~DRS!, growth rate and composition were measured by 88-wavelength ellipsometer, and substrate wobbling during rotation was minimized by applying a negative high voltage on the piezo crystal to adjust the length of the three supporting the manipulator. Detailed theory and operation of DRS is not a subject in this article and can be found elsewhere.11 Since there is no optical constant database available for Inx Ga12x As at MBE growth temperature, we have to compile our own database. Five different samples with composition x50.518, 0.525, 0.532, 0.544, and 0.557 were grown in the MBE chamber with the V/III ratio about 10 from the retractable flux monitoring ion gauge. Each sample was grown at 450 °C with a layer thickness of about 4000 Å. During the growth of each Inx Ga12x As layer, the substrate 0734-211X/98/16„3…/1484/5/$15.00 ©1998 American Vacuum Society 1484 Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 209.147.144.20 On: Fri, 13 Feb 2015 01:17:32 1485 Kuo et al.: Real time in situ composition control of InGaAs 1485 FIG. 1. Implementation of multi-wavelength ellipsometer and diffusive reflectance spectroscopy on a DCA 450 MBE chamber. FIG. 2. Pseudodielectric of Inx Ga12x As at 450 °C and composition x 50.518, 0.532, and 0.557. temperature was kept at constant 450 °C DRS temperature by adjusting the substrate eurotherm temperature setpoint. In order to keep a constant substrate temperature during the Inx Ga12x As layer grown on the InP substrate, the substrate thermocouple temperature has to drop about 45 °C for the growth of 4000 Å Inx Ga12x As layer on the InP substrate. This is due to the radiation heating used to heat up the substrate. The Inx Ga12x As layer lattice matched to InP has a smaller band gap than InP. The infrared radiation from the heater plate passing through the InP substrate was absorbed by the grown Inx Ga12x As layer. The absorption of the infrared radiation by the growing Inx Ga12x As causes the substrate temperature to increase during growth of the Inx Ga12x As layer on the InP substrate. The degree of the extra infrared absorption is also dependent on the Inx Ga12x As layer thickness. In order to achieve precise control of the Inx Ga12x As composition, it is important to keep the substrate temperature constant by monitoring substrate temperature through DRS. After the growth of each of the Inx Ga12x As layers, the substrate temperature was ramped to 525 °C as indicated by DRS and ellipsometric data were measured by an 88wavelength ellipsometer. The substrate temperature was ramped down from 525 to 400 °C in 25 °C steps. Ellipsometric data were taken at each temperature step after the substrate temperature is stabilized and a total of six different temperature ellipsometric data were taken from each sample. The layer thickness used in deriving optical constants for each sample was determined by analyzing the in situ dynamical data during MBE growth. All the temperature and composition optical constants were passed through a series of iteration loops at J. A. Woollam’s Co. to correct for the influence from the change of incident angle. The correction for the incident angle was less than 0.1°. Part of the corrected optical constants are shown in Fig. 2. As can be seen from Fig. 2, the difference of the 450 °C optical constants at x50.518 and x50.557 is very small. It might not be possible to determine the Inx Ga12x As composition in real time with the absolute precision of 0.005 from ellipsometry. A series of samples were grown to verify the validity of the Inx Ga12x As database that we compiled from static ellipsometry experiments. The first set, short term stability test, of five different Inx Ga12x As composition, x 50.518, 0.525, 0.532, 0.544, and 0.557, were grown and monitored by ellipsometry in a period of 2 weeks. Each sample was started at 500 °C growth temperature and ramped down in 50 °C steps and ended at 400 °C growth temperature. The comparison of the Inx Ga12x As composition derived from ellipsometric data and high resolution xray data is shown in Fig. 3~a!. The second set, medium term stability test, of four different Inx Ga12x As composition, x50.502, 0.519, 0.535, and 0.546, were grown and monitored by ellipsometry 1 1/2 months after the short term stability test. The first two Inx Ga12x As samples, DCA130 and DCA131, were grown with the same temperature sequence as the short term stability test which started from 500 °C growth temperature and ramped down to 400 °C in 50 °C steps. The last two Inx Ga12x As samples, DCA132 and DCA133, were started from 400 °C growth temperature and ramped up to 500 °C in 50 °C steps. The results of the derived ellipsometric Inx Ga12x As composition compared to high resolution x-ray data are shown in Fig. 3~b!. The third set of the samples, the long term stability test, were grown after the vacuum break in the MBE chamber to reload As and Ga materials. After the bake out of the MBE chamber, three different samples with compositions of x 50.52, 0.532, and 0.545 were grown to test the static Inx Ga12x As database. The growth temperature was 450 °C for all three samples and the Ga cell temperature was adjusted manually to achieve the target Inx Ga12x As composition. The Inx Ga12x As compositions obtained from high resolution x-ray data were x50.522, 0.532, and 0.547. The precision of the Inx Ga12x As composition monitored by ellipsometry after bake out of the MBE chamber is still within 0.002 of high resolution x-ray data. We have also demonstrated the real time in situ closed JVST B - Microelectronics and Nanometer Structures Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 209.147.144.20 On: Fri, 13 Feb 2015 01:17:32 1486 Kuo et al.: Real time in situ composition control of InGaAs 1486 FIG. 3. Comparison of the Inx Ga12x As composition between high resolution x-ray data and ellipsometry data. The number in parentheses is the standard deviation of the difference of composition between ellipsometry and x-ray data for each sample at three different growth temperatures: ~a! the result from the short term stability test and ~b! the result from the medium term stability test. loop feedback control of Inx Ga12x As/InP. As shown in Fig. 1, the substrate temperature from DRS and composition data from ellipsometry were sent to the control computer with the control program called EPICENTER, which was developed at Arizona State University. EPICENTER analyzed the input data from the ellipsometer and DRS and proper action was taken to maintain a constant substrate temperature and Inx Ga12x As composition. A simple algorithm to control substrate and Ga cell temperature was used in this experiment to achieve both substrate temperature control and Inx Ga12x As composition control. Substrate temperature control was achieved by ramping the substrate temperature down 6 °C in 6 s when both the In and Ga shutter were open to grow the Inx Ga12x As layer. The substrate temperature was updated at EpiCenter by DRS about every 2 s. The EPICENTER checked the substrate temperature every 30 s. If the substrate is not within 0.5 °C of the desired value, the substrate eurotherm setpoint was set to the proper temperature according to the difference of the DRS temperature and the desired growth temperature. No action will be taken again at the substrate temperature control within 30 s of each substrate temperature change. A similar algorithm was used in the Ga cell temperature control. The Inx Ga12x As composition was controlled by using a fixed In cell temperature and ramped the Ga cell temperature to achieve the target Inx Ga12x As composition. Details of the ellipsometric model used in the real time in situ composition control will be discussed in a later publication. The Inx Ga12x As composition value sent to EpiCenter was neglected for the first 8 min of growth of the Inx Ga12x As layer to allow a stable composition value from the ellipsometry. After 8 min, the Ga cell temperature was ramped up or down according to the difference in ellipsometric determined composition and the desired composition. For each 10.01 change in Inx Ga12x As composition, the Ga cell was changed by 21 °C. The composition value at EpiCenter was updated by the 88-wavelength ellipsometer every 9 s and the Ga cell temperature was set to the proper temperature according to J. Vac. Sci. Technol. B, Vol. 16, No. 3, May/Jun 1998 Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 209.147.144.20 On: Fri, 13 Feb 2015 01:17:32 1487 Kuo et al.: Real time in situ composition control of InGaAs FIG. 4. High resolution x-ray data for sample DCA 150. The Inx Ga12x As layer composition is lattice matched to InP. the difference in ellipsometer return composition and the desired composition. The Ga cell temperature was not changed within 30 s from the previous Ga cell temperature change or the ellipsometry determined Inx Ga12x As composition is within 0.002 of the desired composition. The temperature control algorithm used in the substrate and Ga cell temperature is very simple and effective. Two real time in situ Inx Ga12x As composition controlled samples, x50.54 and 0.532, were used to demonstrate the absolute precision of the InGaAs composition control. The high resolution x-ray data indicated the Inx Ga12x As was x 50.542 and 0.532. The x-ray data for x50.532 is shown in Fig. 4 and the analyzed Inx Ga12x As composition from ellipsometric data and Ga cell temperature is shown in Fig. 5. III. DISCUSSION We have compiled a database for Inx Ga12x As lattice matched to InP. Although the database was compiled from the Inx Ga12x As layer after the growth, it is still valid to use FIG. 5. Solid line is the time plot of the ellipsometry determined Inx Ga12x As composition after the shutter was opened. The dash line is the time plot of the Ga cell temperature controlled by the EPICENTER program to achieve x 50.532 for Inx Ga12x As composition. The Ga cell temperature is not changed if the Inx Ga12x As composition is within 60.002 of the target value which is 0.532 for sample DCA 150. 1487 it to monitor the Inx Ga12x As composition during the growth by using a proper ellipsometric model. The short term, medium term test of the Inx Ga12x As composition determined from ellipsometry was consistent with high resolution x-ray data as shown in Figs. 2 and 3. It is clear that the Inx Ga12x As composition from 500 and 450 °C growth temperature can be in situ measured by ellipsometry with the absolute precision of 0.002. However, the Inx Ga12x As composition at 400 °C growth temperature always shows a larger deviation, 0.005, from the high resolution x-ray data. As indicated in the second set of the experiment, part of the sample was grown with the reverse growth temperature sequence. If the problem is coming from the inability of the substrate temperature determination from DRS with the presence of the thick Inx Ga12x As layers, we will expect to have a larger deviation of ellipsometric composition compared to high resolution x-ray data at 500 °C growth temperature from samples DCA132 and DCA133. We observed from the medium term stability test that the growth temperature sequence does not affect the precision of the ellipsometric determined Inx Ga12x As composition at 500 and 400 °C growth temperatures. It is clear from the medium term stability test that the substrate temperature from DRS is not the reason for lager deviation at 400 °C as shown in Fig. 3. One possible explanation for the larger deviation of composition at 400 °C is the strong anisotropy in the broadening parameters of the E 1 , E 1 1D 1 transition at 400 °C, or lower growth temperature observed by Philips et al.1 from their in situ RDS experiment. Since the database was collected in the static condition after the growth of the Inx Ga12x As layer, the surface structure at 400 °C probably doesn’t show the same strong anisotropy effect as in the dynamical condition during growth. The difference of the dielectric response between the dynamical and static condition causes the ellipsometer measured Inx Ga12x As composition to drift away from the real value. The ellipsometric model used in the real time data analysis just cannot take care of the problem in the database and the composition determined by the ellipsometer was as accurate as the higher growth temperature. We have demonstrated from the short term, medium term, and long term stability the absolute accuracy of Inx Ga12x As composition determined from the ellipsometry to be better than 0.002 by using the static database. We have also demonstrated that we can use this database to do real time in situ feedback control of the Inx Ga12x As composition. The absolute precision of the composition is within 0.002 of the target value as shown in Fig. 5. The Ga cell temperature was oscillating in the controlled samples as shown in Fig. 5. This is caused by the time constant to get a stable Ga flux out of the Ga cell and time delay from the Inx Ga12x As composition value determined by signal averaging of ellipsometric data. We can apply a proportional integral differential ~PID! type of algorithm to solve this problem. The other drawback is the inability of the ellipsometry to determine the Inx Ga12x As composition in the first 8 min of the thin film growth which is about 500 Å thick. We have demonstrated the absolute accuracy in determin- JVST B - Microelectronics and Nanometer Structures Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 209.147.144.20 On: Fri, 13 Feb 2015 01:17:32 1488 Kuo et al.: Real time in situ composition control of InGaAs ing the Inx Ga12x As composition lattice matched to InP by using our database to be better than 0.002 from the in situ ellipsometry experiment. This can also apply to other ternary systems as soon as the database is available. This makes ellipsometry a useful tool in MBE and MOCVD growth for both composition control and thickness control. ACKNOWLEDGMENTS The authors would like to thank the Defense Advanced Research Agency for their support of ULTRA Grant No. MDA972-95-1-0016 as part of the Integrated Multi-Sensor Control Consortium, agreement No. MDA972-95-3-0046, between Hughes Research Laboratories, J. A. Woollam Company, Arizona State University, and The University of Colorado. 1488 1 B. A. Philips, I. Kamiya, K. Hingerl, L. T. Florez, D. E. Aspnes, S. Mahajan, and J. P. Harbison, Phys. Rev. Lett. 74, 3640 ~1995!. 2 K. P. Killeen and W. G. Breiland, J. Electron. Mater. 23, 179 ~1994!. 3 C. H. Kuo, M. D. Boonzaayer, M. F. DeHerrerea, D. K. Schroder, G. N. Maracas, and B. Johs, J. Cryst. Growth 175/176, 281 ~1997!. 4 I. P. Herman, Optical Diagnostics for Thin Film Processing ~Academic, New York, 1996!. 5 W. M. Duncan and S. A. Henck, Appl. Surf. Sci. 63, 9 ~1993!. 6 S. Nafis, N. J. Ianno, P. G. Snyder, and W. A. McGahan, Thin Solid Films 233, 253 ~1993!. 7 B. Drevillon, Appl. Surf. Sci. 63, 27 ~1993!. 8 C. H. Kuo, I. H. Hsieh, D. K. Schroder, S. Chen, T. W. Sigmon, and G. N. Maracas, Appl. Phys. Lett. 71, 359 ~1997!. 9 K. Tachibana, T. Shirafuji, Y. Hayashi, S. Maekawa, and T. Morita, Jpn. J. Appl. Phys., Part 1 7B, 4191 ~1994!. 10 D. E. Aspnes, W. E. Quinn, M. C. Tamargo, M. A. A. Pudensi, S. A. Schwarz, M. J. S. P. Brasil, R. E. Nahory, and S. Gregory, Appl. Phys. Lett. 60, 1244 ~1992!. 11 S. R. Johnson and T. Tiedje, J. Cryst. Growth 175/176, 273 ~1997!. J. Vac. Sci. Technol. B, Vol. 16, No. 3, May/Jun 1998 Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 209.147.144.20 On: Fri, 13 Feb 2015 01:17:32