t THE EFFECTS OF RAPID RECRYSTALLIZATION AND ION - /

IMPLANTED CARBON ON THE SOLlD PHASE EPITAXIAL REGROWTH OF Sil ,xGEx ALLOY LAYERS ON SILICON

M.J. Antoneil". T.E. Haynes"" and K.S. Jones" * University of Florida, Gainesville. FL 326 11

Oak Ridge National Laboratory, Oak Ridge, TN 37831 **

ABSTRACT

Transmission electron microscoDv * . has been combined with time-resolved reflectivity and ion channeling to study the effects of regrowth temperature and carbon introduction by ion implantation on the solid phase epitaxial regrowth (SPER) GI strained 2000A. Sio.g&eo &i alloy films grown by molecular-beam epitaxy (MBE). Relativz to the undoped layers, carbon incorporation in the MBE grown SiGe layers prior to regrowrn at moderate temperatures (500- 700°C) has three main effects on SPER: these include a reducrion in SPER rate, a delay in the onset of strain-relieving defect formation, and a sharpening cr' the amorphous-crystalline ( a/c) interface, Le., promotion of a two-dimensional (planar) growh front.' Recrystallization of amorphized SiGe layers at higher temperatures (1 100°C) substanually modifies the defect structure in samples both with and without carbon. At these elevated temperatures threading dislocations extend completely to the SUSiGe interface. Stacking faults are eliminated in the high temperature regrowth, and the threading dislocation density is slightly higher with carbon implantation.

INTRODUCTION

Sil-,Ge,/Si alloys have been used for a variety of high performance electronic device applications such as heterojunction bipolar transistors (HBT),'-6 high mobility transistors (HEMT),' heterojunction field effect transistors (H-FET)'-' and photodetector^.^^^^^ The addition of germanium to silicon allows the resulting SiGe layer to have a reduced bandgap. When this layer is used as the base region of bipolar devices, the resulting band gap reduction provides several advantages, including higher emitter injection efficiency. lower base resistance, a shorter RC time constant and improved performance at low temperaturs. However, a major drawback in the use of such structures for high performance electronic devices is the relatively large lattice mismatch between Si and Ge (-4%). Such a mismatch leads to the formation of strain-relieving misfit dislocations if the layer thickness exceeds a critical value. These dislocations are undesirable since they adversely affect the device properties of the material. Several methods have been proposed to alleviate this problem. For instance, the thickness of the layer can be maintained below the critical thickness so that the layer remains pseudomorphic in nature, or the addition of a

DISCLAIMER

Portions of this document may be illegible in electronic image products. Images are produced from the best available original document.

r

samples using either 1.0 or 2.0 MeV He+ ions. with the detector at 160" or 110" scattering angle. XTEM samples were prepared by lapping thickness to 50-lOOpm and ion milling at room temperature. Microscopy was performed on a JEOL 200 CX, or. for high resolution, a JEOL 4000FX.

RESULTS AND DISCUSSION

Figure 1 shows the Si portion of the ion channeling spectra cornparins three fully regrown Sio.g8Geo.~2/Si samples. The C-implanted samples were regrown at 560 and 1 100°C, respectively, and are compared to a sample without Lhe C implant that was also regrown at 1100°C. Both of the 1 100°C regrowths have a lower channeling yield than the 560°C regrown layer, which indicates a layer that is less defective after the hlgher temperature recrystallization. Comparing the two spectra regrown at 1 100°C, the best quality layer is found to be in the sample which did not receive the carbon implant. This can be determined by comparing the slope of the back edges of the silicon peaks in both of the 1100°C regrowth . The difference in slope between the two samples implies a larger defect density in the carbon implanted layer. Thus the higher regrowth temperature improves the quality of the regrown layer relative to the lower temperature anneals and the addition of carbon at this higher recrystallization temperature appears to degrade the layer quality.

I I 1 1 I 0.40 0.50 0.60

Energy (MeV)

Figure 1. Si portion of RBS-C spectra of 2000A thick Si0.12Geo.g8/Si; implanted with a 30 keV carbon and regrown at 56OoC, implanted with 30 keV carbon and regrown at 1 loO°C, and an amorphized layer regrown at 1100°C (without carbon).

.h XTElI m:,-ragraph of a 2000A thick. Si-. 53Geo.lz iayer resrown at 5OO'C is snmm in - - 1 y c -- 3. T-iis smpi .? .::as mplantea wirn c m o n L: I;! keV ana corresmnas to [he ion c n m e h g specua snonn in Figur? i (560°C with Ci. Tile pr2;xted range of carbon is SOO,% Tnere :s a Tpicai deiay in ~e o r s x of normai stram reiieving ixkts. stacking fauiu and threading dislocations. fi ZoO-~:;:;c)A~ al'ter the arnorphous/crxdiine interr'ace enters the SiGs. This is consistsnr ni in yw:i!::s rqorts of non irnpianted SiGt regrou-th. l8 T ! e aislocxions are =T.':nly spacca hroughout ti.,? 1al;er occurring sver; 1 ~ O O A = ~ooA. ~n some c ~ s s s . near h e top o i h e breading disiGcauons. t$ra..iai regron.th smps f o A x ~ i g areas o i polycrystalline mterial bcund by ;1 large dcnsiF ;r stacking raults. These poiycrystatiine areas are found in the top 250-3OO.i a i the regrown lay?:.

-.

Figure 2 . Bright field XTEM (g=220) image of a 2OOOA thick S$).ggGert.12 /Si layer with a 30 keV carbon mplant regrown at 560°C.

Figure 3 show-s a XTEM micrograph from the carbon implanted sample regrown at 1100°C. Three main differences are observed when rhe 1100°C regrown sample is comparcd to the carbon-implanted sample regrown at 560°C. First. during the 1100°C regrowth. a higher density of threading dislocauons can been seen throughout the 2000A SiGe layer. Second, the defects in the 1100°C regronn sample extend cornpleteiy from the surface down to the Si/SiGe heterointerface. This is in contrast to the 560°C regrowth, where the onset of defects is delayed on average 300-400A bqond the heterointerface and terminate at the surface. Third, the types of defects are different. In the 560°C regrown sample. the defects consist primarily of stacking fauit bundles. These stacking fault bundles occur in the last 1WOA of regrowth. In the 1100°C regrown material implanted with carbon the, defects consist of threading dislocations that extend from the Si/SiGe heterointerface to the surface of the alloy layer and the layer shows no evidence of stacking faults and no areas of polycrystalline material.

Figure 3. Bright field XTEM (g=220) image of a 2WOA thick Si~88Ge0.12 /Si layer with a 30 keV carbon implant regrown at llOO°C.

Figure 4 shows a STEM micrograph of the unirnpiantea specimen rcgrown at 1100°C. The primary difference benvcen the unimplanted and the carbon impianted 1100'C regrown sampie (Figure 3) is that the dislocation density is decreased. This is consistent with the ion channeling spectra shown in Figure I which shows the best quality layer to bc the 1100°C regrowth without carbon. This was confirmed by PTEM which found the dislocation density for the carbon implanted sample to be 3.25 x 109 cm-2 while the dislocation density for the unimplanted SiGe sample was 7.8 x 108 cm-2. The unimplanted sample does not exlubit stacking faults, just as in the C-implanted case. This substantiates the interpretation that the main improvement h quality of the regrown layer comes by using a higher temperature, rather than by implanting carbon for strain compensation. One must also keep in mind when examining Figure 1 that RBS-C is less sensitive to threadin3 dislocations as compared to structural defects such as stacking faults.

Figure 4. Bright field XTEM (g=220) image of a 20WA thick S&&e0.12 /si layer regrown at 1 100°C (without carbon).

The importance of the regrowth temperature has been shown to have a substantial effect on rhe Cziiity or' the regrown epitaxial layer =a offers ;i ;;lossibls expianation ror discrepancies 01

XCCT: addition ior stran compcnsauon in rne SiGe aiio! layers. The results o i this jiudy suggest that LX rapid recrystallization rate in the dicy layer may be preventing compiete reiasation by nor Alon-:ng enough time for defects to form.

COKCLUSIONS

Regrowth temperature 1s a vital s t q Lor predictmz defect suvcture ana determining epllayr quai;:;;. Stacking fault formation is effecavdy suppressed during high temperature regrowth. Base2 on evidence presented it is believed that this is due to the rapid growth rate n.hich does not give mcking faults enough time to nucleate and grow and reducing the likeliness of polycrystalline fomxrion. Subsequently the lattice forms s m e strain rdieving defects and these defxts extend to the Si SiGe heterointerface. In addition i t has been reported that carbon can no imger occupy subsxutional lattice sites above 900°C l 9 [herefore. my strain compensation prxided by the carbcn would be lost during the higher tempatwe anneals.

ACKSOWLEGDEMENTS

Research performed at Oak Ridge National Laboratory was sponsored by the U.S. Depanment of Energy, Division of Matenals Sciences, under contract DE-AC05-840R2 1400 with Marun Marietta Energy Systems, Inc. Work at the University of Florida was supported by NASA and the SURA/ORNL Summer Cooperative Research Program.

REFERENCES

1. 2. 3. 4. 5 .

6.

7 .

8 . 9.

10.

11. 12.

13.

Antonell, M.J.. Haynes. T.E., Jones. K.S. J. .4ppl. Ph!*s. in press (1995). Bean. J.C. Proc. of the I€E€ 80. 571-587 (1992). Patton. G.L., Comfort, J.H. IEEE Electron Device Letters 11. 171-173 (1990). RouIston, D.J., McGregor, J.M. Solid State Electronics 35, 1019- 1020 (1992). Stiffler. S.R.. Comfort, J.H., Stanis. C.L., Harame, D.L., Fresart, E.D. J. ;Ipp 70. 1461-1420 (1991).

Phys.

K&njoo, K., Nayak, D.K., Park. J.S., Woo, J.C.S., Wang, K.L. J. Appl. Phys. 69,

Shaffler, F., Tobben, D., Herzog, H.J., Abstreiter, G.. Hollander, B. Sernicond. Sci. Technol. 7, 260-266 (1992). Gronet, C.M. J. Appl. Phys. 61, 2107-2409 (1987). Xie, Y.H., Fitzgerald, E.A., Monroe, D., Silverman, P.J., Watson, G.P. J. Appl. Phys.

Vetter, A., Kovacic, S.J., Simmons. J.G., Noel. J.P., Houghton, D.C. Solid State Electronics 36, 1247-1250 (1993). Wang, K.L., Karunasiri, R.P. J.Vuc. Sci. Technol. 11, 1159-1167 (1993). Kesan, V.P., May, P.G., Bassous, E., Iyer, S.S. IEEE EZectron Device Letters 90,

Lin, T.L., Jones, E.W. IEEE Electran Device Letters 90, 641-643 (1990).

6674-6678 (1991).

73, 8364-8370 (1993).

637-640 (1990).

I ' i4. * 3 . 16.

Fukami. .A.. Shoir. K.. Nagano. T. Appl. Phys. Lett. 57. 2345-2347 11990). Lombarac, 5.. Pr:cio. F.. Cmpisano. S.U. .iDpi. ?hs. Lrrt. 62. 2335-2337 (1993). Erbel. K.. !T;er. S.S.. Zollne:. S.. Tsang. J.C.. L:Gou?s. F.K. ADP/. P h ~ s . Lett. 60.

7 -

3033-3035 i1992:. Nishikawa. S.. Yamaji. T. Appi. Phss. Lett. 62. 303-305 ( 1992). Lee, C.. Eaynes. T.E., Jones. K.S. Appl. Phss. Let t . 62. 501-503 i 1993). Isomae. 5.. Ishiba. T.. Xndo. T.. Tamura. M. J. &pi. ? h a . 74. 3815-3819 (1993).

i7. 18. 19.

DISCLAIMER

This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsi- bility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Refer- ence herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recom- mendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.