194 IEEE ELECTRON DEVICE LETTERS, VOL. 32, NO. 2, FEBRUARY 2011
Dopant Activation in Single-Crystalline Germanium
by Low-Temperature Microwave Annealing
Yao-Jen Lee, Shang-Shiun Chuang, Fu-Kuo Hsueh, Ho-Ming Lin,
Shich-Chuang Wu, Ching-Yi Wu, and Tseung-Yuen Tseng
Abstract—Phosphorus activated in germanium epitaxy atop
Si wafer by low-temperature microwave annealing technique was investigated in this letter. Compared to the conventional RTA process, the temperature of phosphorus activation could be 120◦C to 140 ◦C which is an improvement in temperature re-duction at the same sheet resistance. According to the SRP, up to 150◦C reduction in maximum temperature at the same activation concentration (about 2× 1019cm−3) could be achieved. Through
adjusting the microwave power and process time, sheet resistance could be decreased while suppressing dopant diffusion. In addi-tion, the inserted susceptor wafers above and below the processing wafer also suppressed the dopant diffusion and improved film roughness.
Index Terms—Germanium, low temperature, microwave
an-neal, phosphorus, rapid thermal anneal (RTA). I. INTRODUCTION
G
ERMANIUM (Ge) HAS been considered a potentialmaterial as replacement for silicon because of its higher mobility for electron and hole than Si. However, the thermal instability of Ge and GeOx during the fabrication process
lim-its the fabrication temperature of Ge-based devices. Recently, high-k materials and metal gates have become a solution for continuous scaling down of Ge-based devices. The deposition of high-k material can be achieved by atomic layer deposition at very low temperature (170◦C–340◦C). This spurs consider-ation of germanium as a replacement material for the Si channel [1]. However, the ability of dopant activation is another issue for the fabrication of Ge-based devices [2], [3]; dopant diffusion in Ge film is observed in rapid thermal processing (RTP) method.
Manuscript received September 21, 2010; revised October 16, 2010; accepted October 24, 2010. Date of publication December 13, 2010; date of current version January 26, 2011. This work was supported in part by the National Science Council, Taiwan, under Contracts NSC-99-2221-E-491-031-MY2 and NSC-98-2221-E-212-033-MY3. The review of this letter was arranged by Editor C.-P. Chang.
Y.-J. Lee is with the National Nano Device Laboratories, Hsinchu 30010, Taiwan, and also with the Department of Physics, National Chung Hsing University, Taichung 402, Taiwan (e-mail: [email protected]).
S.-S. Chuang and T.-Y. Tseng are with the Department of Electronics En-gineering & Institute of Electronics, National Chiao-Tung University, Hsinchu 300, Taiwan.
F.-K. Hsueh is with the National Nano Device Laboratories, Hsinchu 30010, Taiwan, and also with the Department of Electrophysics, National Chiao Tung University, Hsinchu 30010, Taiwan.
H.-M. Lin and S.-C. Wu are with the National Nano Device Laboratories, Hsinchu 30010, Taiwan.
C.-Y. Wu is with the Department of Electronics Engineering, Dayeh Univer-sity, Changhua 51591, Taiwan (e-mail: [email protected]).
Color versions of one or more of the figures in this letter are available online at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/LED.2010.2090937
According to Satta et al. [4], rapid thermal annealing, even at 500◦C for 60 s, leads to severe implanted dopant diffusion for high-dose phosphorus in Ge film. However, lower annealing temperature is used; the lower activation percentage causes higher contact resistance, reducing the drain current.
Low-temperature dopant activation method has been devel-oped by Park et al. [5]. They have successfully used the metal-induced crystallization (MIC) process to activate dopants in amorphous germanium at about 360 ◦C [5]. However, for n-type doped Ge film, good results leading to low resistivity may be obtained only with cobalt. Other metal impurities work as acceptor-like traps which capture the electrons, resulting in resistivity increases. Further, the low metal diffusion coefficient in n-type doped germanium is a disadvantage for cobalt MIC processing. Accordingly, microwave annealing may be an alter-native to other rapid thermal processing methods in silicon pro-cessing [6], [7]. Microwaves could repair the damage caused by high-dose boron implantation. Further, the dopant is activated by the microwave initiation of solid-phase epitaxial regrowth at 500◦C. In addition, we have demonstrated boron activation on the Si/Ge/Si structure by microwave annealing [8].
In this letter, we used microwave annealing to activate the phosphorus in single-crystalline Ge epitaxial film at low tem-perature. The major factors in the microwave process are power, frequency, and loads inside the chamber, while a less important factor is temperature. We demonstrated microwave annealing by controlling the power magnitude and the loads inside the chamber in order to compare it with the conventional RTP method.
II. EXPERIMENTS
Several Ge epitaxial films were prepared for dopant activa-tion experiments. A 10-nm Si layer was deposited with SiH4
as the buffer layer on 6-in p-type Si (100) substrates with resistivity of 15–25 Ω− cm by UHVCVD at 550 ◦C. Then, a 200-nm Ge epitaxial layer was deposited by UHVCVD at 420◦C, followed by P31(25 keV, 1× 1015cm−2) implantation at room temperature. Additional wafers with a layer of AlSiCu and tungsten thin film, both 200 nm thick, deposited on thermal oxide, were also prepared for microwave testing.
The splits underwent microwave annealing at 5.8 GHz with a power of about 600–1400 W in a 5.57× 106cm3chamber for 100 to 600 s. A 10-min N2purge from the top of the chamber
was performed before the microwave was started, and the N2
flow was maintained until the process was completed. The processing time was measured from when the microwave power 0741-3106/$26.00 © 2010 IEEE
LEE et al.: ACTIVATION IN SINGLE-CRYSTALLINE GERMANIUM 195
Fig. 1. Temperature versus process time during microwave annealing in different setting conditions; S is the represented susceptor, and L is the represented load wafer.
TABLE I
CONDITIONS OFMICROWAVEANNEALING ANDRTAFORINVESTIGATING THEDOPANTDIFFUSIONMECHANISM
was turned on until the power was turned off, without regard to temperature, different from the approach used in rapid thermal anneal (RTA). The temperature profile versus time by micro-wave is shown in Fig. 1. As one added susceptors or loading wafers above and below the process wafer inside the microwave chamber, the maximum temperature and temperature ramping slope would decrease because the susceptor and loading wafers also soak up the microwave power inside the chamber [9]. Therefore, the heating slope would be lower with susceptors. Finally, SIMS, SRP, and sheet resistance (Rs) were measured,
and cross-sectional transmission electron microscope (TEM) images were also compared. The splits without microwave treatment were activated by RTA in a heat-pulse AG601i RTA system at 450◦C and 550◦C, respectively.
III. RESULTS ANDDISCUSSIONS
The conditions of the different microwave processes and Rs
are displayed in Table I. The microwave magnetron emitted 600–700 W of power when it was fully turned on (100%). The Rsof sample (i) is 316.3 Ω/.
The Rsby RTA was also summarized in Table I. Comparing
to RTA at 550◦C for 60 s (RTA2), the microwave power used on sample (i) appears to be too low to activate sufficient dopants. After increasing the power to 150% for advanced activation, the Rs of sample (ii) is much lower than that of sample (i).
Fig. 2. SIMS profiles of implanted P in Ge before and after microwave annealing in different conditions and RTA at 550◦C for 60 s.
However, the SIMS profile in Fig. 2 shows that the diffusion of phosphorus in sample (ii) is nearly as severe as under RTA at 550 ◦C for 60 s, although the Rs of sample (ii) is low.
Hence, two approaches to suppressing dopant diffusion while maintaining lower Rs are investigated. First, in sample (iii),
the time that the wafers stay in the saturated temperature is reduced by splitting the total process time into two cycles, of 300 s each. Second, in sample (iv), the load wafers are added to soak up the power inside the chamber. As a result, sample (iv) has a higher Rsbecause of the insufficient power applied to the
wafer, as with sample (i). Sample (iii) gets an acceptable Rs
value and less dopant diffusion than sample (ii). Higher power and a shorter processing time thus appear to be necessary for good activation and less dopant diffusion.
With magnetron power at 200% fully turned on for 100 s, as shown in Table I, sample (v) has less dopant diffusion and a lower sheet resistant value than sample (iii). Only about 10.7% of dopants in sample (v) diffused into the bulk region, and the peak of the phosphorus concentration of sample (v) remains at the level of the as-implanted sample. In addition, from Fig. 3 of the SRP depth profiles, one can conclude that the lower Rsmagnitude of sample RTA2 among all splits was
due to dopant diffusion. The maximum electrical concentration between microwave annealing and RTA was almost the same (2× 1019cm−3). In addition, the depth of the electrical dopants by microwave anneal was 50 nm shallower than that by RTA at 550 ◦C for 60 s, and the minimum resistivity near the Ge surface is about 10−3 ohm-cm. Therefore, microwave could well activate the P31dopant inside the Ge film, and the dopant
diffusion could be suppressed due to its lower temperature anneal process.
The comparison of the dopant profile of samples (v) and (vi), in Fig. 2, shows that the susceptor can significantly sup-press dopant diffusion. Two important issues are suggested by the temperature profiles of the samples with and without the susceptors (Fig. 1): reduction in the maximum temperature of about 10 ◦C to 20◦C and the rate of temperature increase at the beginning of the process. The susceptor and loading wafers appear to cause a slower temperature increase. Based on the experiments above, there appear to be three possible mechanisms which lead to dopant diffusion which include the
196 IEEE ELECTRON DEVICE LETTERS, VOL. 32, NO. 2, FEBRUARY 2011
Fig. 3. SRP depth profiles of P in Ge from sample (v) and RTA2.
Fig. 4. TEM of (a) as-implanted, (b) RTA2, (c) sample (v), 200% for 100 s with susceptors, and (d) sample (vi) without susceptors.
following: 1) the power magnitude; 2) the processing duration time at the saturated temperature; and 3) the temperature ramp rate at the beginning of the process.
In Fig. 4(a), a 60-nm amorphous region was created by ion implantation. In Fig. 4(b), after RTA at 550◦C for 60 s, the surface was damaged because of germanium out diffusion. It caused a critical issue of device fabrication. The microwave process has the ability to repair the bombardment damage by ion implantation and drove dopants into the lattice site. An amorphous region of 2–3 nm deep is found on the surface of the germanium of sample (v), as shown in Fig. 4(c). In addition, without susceptors, there are pin holes on the germa-nium surface, as shown in Fig. 4(d). This may be due to the rate of temperature increase at the beginning of the process. Susceptors soak up part of microwave power. It causes a slower temperature increase at the beginning of the process. There-fore, by adding susceptors, the process wafer was treated in more uniform temperature ambient because of thermal coupling effect [9].
Finally, the AlSiCu and tungsten films also underwent mi-crowave treatment, and the resistivity value remained the same after the microwave annealing process. These metals are not damaged by microwave annealing.
IV. CONCLUSION
This work has demonstrated that phosphorus in single-crystalline germanium thin films can be activated by microwave annealing. The SRP and SIMS analyses showed that the severe dopant diffusion and the dopant activation level by microwave annealing, as compared to that by RTA, can be suppressed. Moreover, the TEM images showed that radiation damage occurring during the implantation process can be repaired by microwave annealing. This appears to be due to the lower tem-perature during processing than in conventional RTA process-ing. In addition, higher power may be used to obtain a sufficient activation level, while shorten the process time, and susceptors are needed for suppressing dopant diffusion and stimulating germanium out diffusion. Microwave annealing also appears to be compatible with metallic processing.
ACKNOWLEDGMENT
The authors would like to thank DSG Technologies for the useful suggestions and discussions.
REFERENCES
[1] C. C. Cheng, C. H. Chien, G. L. Luo, C. L. Lin, H. S. Chen, J. C. Liu, C. C. Kei, C. N. Hsiao, and C. Y. Chang, “Junction and device characteris-tics of gate-last Ge p- and n-MOSFETs with ALD- Al2O3gate dielectric,” IEEE Trans. Electron Devices, vol. 56, no. 8, pp. 1681–1689, Aug. 2009. [2] C. O. Choui, K. Gopalakrishnan, P. B. Griffin, J. D. Plummer, and
K. C. Saraswat, “Activation and diffusion studies of ion-implanted p and n dopants in germanium,” Appl. Phys. Lett., vol. 83, no. 16, pp. 3275–3277, Oct. 2003.
[3] Y. S. Suh, M. S. Carroll, A. Levy, G. Bisognin, D. D. Salvador, M. A. Sahiner, and C. A. King, “Implantation and activation of high con-centrations of boron in germanium,” IEEE Trans. Electron Devices, vol. 52, no. 11, pp. 2416–2421, Nov. 2005.
[4] A. Satta, E. Simoen, R. Duffy, T. Janssens, T. Clarysse, A. Benedetti, M. Meuris, and W. Vandervorst, “Diffusion, activation, and regrowth be-havior of high dose P implants in Ge,” Appl. Phys. Lett., vol. 88, no. 16, pp. 162 118-1–162 118-3, Apr. 2006.
[5] J. H. Park, M. Tada, W. S. Jung, H. S. Wong, and K. C. Saraswat, “Metal-induced dopant (boron and phosphorus) activation process in amorphous germanium for monolithic three-dimensional integration,” J. Appl. Phys., vol. 106, no. 7, pp. 074 510-1–074 510-6, Oct. 2009.
[6] T. L. Alford, D. C. Thompson, J. W. Mayer, and N. D. Theodore, “Dopant activation in ion implanted silicon by microwave annealing,” J. Appl. Phys., vol. 106, no. 11, pp. 114 902-1–114 902-8, Dec. 2009.
[7] H. Zohm, E. Kasper, P. Mehringer, and G. A. Muller, “Thermal processing of silicon wafers with microwave co-heating,” Microelectron. Eng., vol. 54, no. 3–4, pp. 247–253, Dec. 2000.
[8] Y. J. Lee, F. K. Hsueh, S. C. Huang, J. M. Kowalski, J. E. Kowalski, T. Y. Cheng, A. Koo, G. L. Luo, and C. Y. Wu, “A low-temperature microwave anneal process for boron-doped ultrathin Ge epilayer on Si substrate,” IEEE Electron Device Lett., vol. 30, no. 2, pp. 123–125, Feb. 2009.
[9] H. Bosman, Y. Y. Lau, and R. M. Gilgenbach, “Microwave absorption on a thin film,” Appl. Phys. Lett., vol. 82, no. 9, pp. 1353–1355, Mar. 2003.
