Optimization of ClusterCarbon™ process parameters for strained Si lattice

Optimization of ClusterCarbon™ process parameters for strained Si lattice

Materials Science and Engineering B 154–155 (2008) 122–125 Contents lists available at ScienceDirect Materials Science and Engineering B journal hom...

771KB Sizes 0 Downloads 19 Views

Materials Science and Engineering B 154–155 (2008) 122–125

Contents lists available at ScienceDirect

Materials Science and Engineering B journal homepage: www.elsevier.com/locate/mseb

Optimization of ClusterCarbonTM process parameters for strained Si lattice Karuppanan Sekar a,∗ , Wade A. Krull a , Thomas N. Horsky a , Thomas Feudel b , Christian Krüger b , Stefan Flachowsky b , Ina Ostermay b a b

SemEquip, Inc., 34 Sullivan Road, North Billerica, MA 01862, USA AMD Saxony, LLC & Co. KG, Wilschdorfer Landstr 101, D01109 Dresden, Germany

a r t i c l e

i n f o

Article history: Received 6 May 2008 Received in revised form 18 September 2008 Accepted 23 September 2008 Keywords: ClusterCarbon Strain HRXRD Ion implant SiC Molecular implants

a b s t r a c t We present here the substitutional carbon dependence of ClusterCarbon implant energy and dose, and anneal parameters such as solid phase epitaxial regrowth (SPER) temperature and various high temperature millisecond flash anneal conditions. With a multiple implant sequence of carbon implants one can obtain a fairly uniform carbon profile and we show that it provides better carbon substitution [C]sub when compared to a single implant. It is been established that optimizing the percentage of [C]sub requires an SPER anneal temperature <850 ◦ C followed by a millisecond anneal. We show that carbon substitution increases with SPER temperature up to 800 ◦ C and decreases beyond this temperature supporting the fact that carbon has a significant probability of being excited out of its substitutional site beyond 850 ◦ C. We report here that SPER anneal with an additional millisecond flash anneal that provides highest carbon substitution, [C]subs > 2%. For a given millisecond anneal and for implants with various energies and doses we show that the percentages of [C]sub increases linearly with the fraction of carbon dopants within the amorphous layer. © 2008 Elsevier B.V. All rights reserved.

1. Introduction As devices approach the 32-nm regime, scaling becomes more challenging and enhancing the performance more difficult. New materials are required to overcome fundamental physical limitations imposed by existing materials. The amazing advancements achieved to date in Si complementary metal-oxide-silicon (CMOS) technology have come primarily from scaling, i.e. from reducing the critical dimensions of the transistors. It is increasingly difficult to further reduce critical dimensions, hence alternative methods of improving transistor performance are also being employed. One important approach is to increase the electron and hole mobility in the transistors. Several approaches for achieving enhanced mobility in CMOS devices are reviewed. Methods for achieving defect-free strained Si structures have also been discussed elsewhere [1]. Such strained Si is a viable option to sustain continual drive current increases not achieved by traditional scaling. In such devices, Si:C layers are created in the source and drain regions of the device and act as stressors, producing lateral tensile strain. Addition of strained Si to conventional MOSFET devices is also compatible with existing mainstream CMOS process technology.

∗ Corresponding author. Tel.: +1 978 262 3628; fax: +1 978 262 0950. E-mail address: [email protected] (K. Sekar). 0921-5107/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.mseb.2008.09.039

For the NMOS transistor, strain induced by epitaxially grown Si:C is an option but the difficulty comes in the form of lower throughput and non-repeatable process performance. Use of such epitaxially grown Si:C also has other process and integration issues. Monomer carbon implant alone does not result in obtaining the appropriate percentage of substitutional carbon and requires an extra Ge pre-amorphization (Ge-PAI) step [2] to form Si:C layer. With the self-amorphizing capability of ClusterCarbon implant (thus avoiding an extra Ge-PAI step) it has been shown that one can introduce about 2% of substitutional carbon after suitable anneal. Substitutional carbon fraction in implanted samples approaches the ideal 100% limit if the optimized anneal process includes a millisecond anneal process like flash or laser anneal. Thus Si:C formation by ClusterCarbon implant is more appealing and a promising contender for high performance device applications. Carbon ion implantation into Si acts as a barrier to boron diffusion too. Carbon traps interstitial silicon thus hindering the dopant diffusion [3]. It has been shown that ClusterCarbon implant also hinders boron diffusion and thereby providing an enhanced activation and abrupt junction depths [4,5]. With these added advantages, it will be interesting to look at other benefits of ClusterCarbon implant too. In this study we focus on Si:C layer formation by cluster ion implantation with low temperature solid phase epitaxial regrowth (SPER) and high temperature millisecond (ms) anneals. Stress/strain and the percentage of substitutional carbon will be discussed from data obtained using high resolution X-ray diffrac-

K. Sekar et al. / Materials Science and Engineering B 154–155 (2008) 122–125

Fig. 1. Si amorphous layer thickness for C7 H7 implants at 10 keV for various doses. The amorphous layer thickness for multiple implants is slightly higher than a single implant for a given highest energy and dose implant.

123

implant when compared to C7 H7 implant. Extrapolating from the experimental results of these two cluster molecules, the monomer carbon implant is expected to show partial amorphization and a corresponding very rough interface [9]. Such partial amorphization will lead to incomplete re-crystallization. This illustrates that a PAI implant is necessary to use with a monomer implant to obtain a recrystallized layer and thus adds an extra step in device fabrication flow. Fig. 3 shows SIMS and simulation profiles (SRIM code) for (3 keV 1.5e15 + 6 keV 3.0e15 + 10 keV 3e15 atoms/cm2 ) multiple implant. The simulated profile shown reproduces fairly accurately the peak concentration (∼2.0e21 atoms/cm3 ) and width of the actual experimental profile. The simulation profile match very well with the actual SIMS peak concentration (2e21 atoms/cm3 ). This roughly corresponds to 4% atomic carbon at the peak. The simulated profile assumes an amorphous Si

tion (HRXRD) method and secondary ion mass spectrometry (SIMS) measurements. The amorphous layer thickness by ClusterCarbon implant using XTEM will be correlated to the percentage of substitutional carbon obtained from HRXRD measurements. We will show that self-amorphizing property of ClusterCarbon [6] plays an important role in such enhanced carbon substitutionality in Si. Dependence of substitutional carbon percentage on various implant and anneal parameters will be presented and discussed. 2. Experimental The starting substrates used in this study were 200 mm, n-type Si(1 0 0) silicon substrates. The wafers were implanted with ClusterCarbon at different energies and doses using C7 H7 (from C14 H14 material) from a ClusterIon® source. We selected C7 H7 part of the mass spectrum that provided a cluster implant with 7 carbon atoms. All the energies and doses mentioned in the text correspond to the monomer equivalent values. SPE anneals in this work were performed with an RTP 200 mm SUMMIT XT tool, while flash RTP (fRTP) anneals were done with Mattson Technology Canada’s Millios flash anneal system [7]. SIMS and HRXRD profiles were carried out using commercially available facilities. XTEM measurements were carried out using JEOL 2010 FEG TEM using on-axis multibeam imaging conditions. 3. Results and discussion Fig. 1 shows the amorphous layer thickness as measured by XTEM for various doses of 10 keV C7 H7 implant. For a given energy, the amorphous layer thickness (␣-Si) increases more or less linearly with the implant dose. At 10 keV, 4e15 and 8e15 atoms/cm2 the ␣-Si layer thickness values are around 36.5 nm and 44 nm. It is observed that the amorphous layer thickness plays an important role in determining the carbon substitutional percentage. This will be discussed in the later part of this article. In the same figure, we show the thickness of the ␣-Si layer for multiple implants (triangles) for (3 keV 1.5e15 + 6 keV 3e15 + 10 keV 4e15 and 6e15 atoms/cm2 ) at two different combinations of doses. The thickness of the ␣-Si layer at single implant 10 keV, 8e15 atoms/cm2 dose is similar to the ␣-Si layer produced by the multiple implants as shown above at 10 keV, 6e15 atoms/cm2 dose. The XTEM images in Fig. 2(a) shows the formation of the self-amorphized layer for the implantation of C7 H7 for 10 keV at 8e15 atoms/cm2 and Fig. 2(b) shows XTEM image of multiple energy C7 H7 implant [3 keV 1.5e15 + 6 keV 3e15 + 10 keV 6e15 atoms/cm2 ]. In an earlier publication [8] it has been shown that the a–c interface roughness is smoother and lower in the case of the heavier C16 H16 ClusterCarbon

Fig. 2. (a) XTEM image shows the formation of the self-amorphized layer for the implantation of C7 H7 for 10 keV at 8e15 atoms/cm2 . The amorphous layer thickness is ∼44 nm. (b) XTEM image shows amorphous layer thickness in the case of a multiple energy C7 H7 implant with [3 keV 1.5e15 + 6 keV 3e15 + 10 keV 6e15 atoms/cm2 ]. The amorphous layer thickness is ∼44 nm.

124

K. Sekar et al. / Materials Science and Engineering B 154–155 (2008) 122–125

Fig. 3. The figure shows SIMS and simulation profiles (SRIM code) for (3 keV 1.5e15 atoms/cm2 + 6 keV 3.0e15 atoms/cm2 + 10 keV 3e15 atoms/cm2 ) multiple implant. For comparison purpose SRIM simulation profile for 10 keV, 3e15 atoms/cm2 single implant. With multiple implant one can achieve a more or less a flat profile by tuning the energy and dose of the implants.

substrate, and so shows a shallower profile than the experimental SIMS profile. The wafers were implanted at zero degree tilt and twist so that any contribution from channeling should be evident in the profile. For comparison purpose SRIM simulation profile for 10 keV, 3e15 atoms/cm2 single implant is shown. With multiple implants one can achieve flat profile by tuning the energy and dose of the implants. Fig. 4 shows HRXRD spectrum of a multiple C7 H7 implant sequence: (3 keV + 6 keV + 10 keV) at doses (1.5e15 + 3e15 + 3e15) atoms/cm2 respectively, flash annealed at 1275 ◦ C. A 750 ◦ C SPER anneal for 60 s was carried out prior to the flash anneal step. The profile shows a bulk Si peak (lower angle) and a strained layer peak to the right of Si peak. The separation between the bulk Si peak and the strained Si peak (higher angles) provides information about the substitutional carbon and its substitutional percentage is determined using Kelires Model [10,11]. The estimated [C]subs percentage value for this implant sequence is greater than 2.20%. Table 1 summarizes the percentage [C]subs values for the above sequential implant for both SPE and flash anneal temperatures. Highest percentage [C]subs value is obtained with high temperature flash anneal. The high temperature millisecond anneal not only provides higher substitutional carbon but also helps in better re-crystallization of the damaged Si lattice.

Fig. 4. The figure shows HRXRD spectrum of a multiple C7 implant sequence: (3 keV + 6 keV + 10 keV) at doses (1.5e15 + 3e15 + 3e15) atoms/cm2 respectively. The sample was flash annealed (fRTP) at 1275 ◦ C.

Fig. 5. The figure shows high resolution diffraction curves of the Si(4 0 0) reflections for samples implanted with multiple energy. ClusterCarbon implants annealed at various temperatures. The implant sequence condition is: (2 keV + 5 keV + 8 keV) at doses (3e15 + 3e15 + 3e15) atoms/cm2 respectively.

Fig. 5 shows high resolution diffraction curves of the Si(4 0 0) reflections for samples implanted with multiple energy ClusterCarbon implants annealed at various SPE and flash anneal temperatures. The ClusterCarbon implant energies are 2 keV, 5 keV and 8 keV. The implant dose for each implant energy is 3e15 atoms/cm2 . With accurate placement of carbon dopants one can realize a controlled profile with multiple sequence implants [12]. The strained Si:C peaks appear to the right of Si bulk peak at a larger angle. An increase in tensile strain (due to the formation of Si:C with a smaller lattice constant) yields a lattice spacing reduction, causing the diffraction peak position to shift to a higher angle. The volume concentration of C and the strain of the Si lattice vary in depth according to the carbon concentration profile, resulting in different local conditions for nucleation and growth of Si–C complexes and SiC particles. Hence one obtains a strained gradient in depth from the region with highest concentration. The shoulder in the left hand side of the bulk Si peak is due to the presence of interstitial carbon. Looking at the profiles, the strained peak slowly shifts towards higher angle with increasing temperature except at 850 ◦ C. The substitutional carbon percentage as determined using Kelires Model [10,11] is estimated to be ∼1.55% at 750 ◦ C, ∼1.66% at 800 ◦ C and ∼1.75% using fRTP at 1060 ◦ C. However, at 850 ◦ C it drops to 1.35% indicating diffusion of carbon from substitutional sites to interstitial sites. This is also evident from the broadening of the profile for 850 ◦ C to the left of the bulk Si peak. Beyond 850 ◦ C formation of SiC precipitates hinder the

Fig. 6. The figure shows the percentage of [C]subs versus fraction of implanted carbon in the amorphous Si layer in the case of flash annealed samples. The data include various implant energies, doses and flash anneal temperatures.

K. Sekar et al. / Materials Science and Engineering B 154–155 (2008) 122–125

125

Table 1 HRXRD results for % of substitutional carbon for various anneals for the following ClusterCarbon implant condition. Implant energy dose (atoms/cm2 )

3 keV + 6 keV + 10 keV 1.5e15 + 3.0e15 + 3.0e15

Anneal condition % of [C]subs

SPE 750 ◦ C 1.66

stability of carbon atoms in substitutional sites [13,14]. To trap the carbon atoms in substitutional sites one requires a high temperature millisecond anneal, where the atoms could be trapped in their positions in a non-equilibrium condition. Fig. 6 shows the percentage of [C]subs versus fraction of implanted carbon in the ␣-Si layer in the case of flash annealed samples. The results represent data from various implant energies, doses and anneal conditions. The anneal temperature ranged from 1000 ◦ C to 1300 ◦ C. The percentage of [C]subs is higher if the fraction of implanted carbon in the ␣-Si layer is higher. It is known that the percentage of carbon in substitutional site increases in the presence of an amorphous layer. To get to 2% [C]subs , our flash anneal data shows that it requires more than 70% of dopant carbon atoms within an amorphous layer. There is no scientific explanation why one needs 70% of carbon dopants to achieve 2% substitutional carbon. It is well known that carbon solubility limit is very low (around 2%) in Si. This leads to a point where if one wants more [C]subs higher carbon concentration is required. This data set from flash anneal just suggests that higher the % of carbon profile within an amorphous Si layer larger will be the % [C]subs . 4. Conclusions We have shown that ClusterCarbon implant could be used both as an self-amorphizing implant and as well as to fabricate Si:C layers. Formation of such Si:C layers with substitutional carbon greater than 2% could be achieved using ClusterCarbon approach thus providing better tensile strained layer thereby enhancing carrier mobility. Obtaining tensile strained Si:C with monomer carbon implant approach suffers from partial incomplete amorphization resulting in lower carbon subsitutionality. With controlled dopant

SPE 850 ◦ C 1.55

fRTP 1060 ◦ C 1.82

fRTP 1275 ◦ C 2.20

placement using a sequence of multiple ClusterCarbon implants and using a SPER anneal followed by a high temperature millisecond anneal, one can achieve greater than 2% substitutional carbon that is otherwise difficult to achieve with a Si:C epi-process. Acknowledgements We would like to thank Dr. D.C. Jacobson for his help in some of the data analysis. We thank SemEquip GSD team Brian Haslam and Dennis Klesel and Jeff Gelpey, Steve McCoy and Jason Chan of Mattson Technology, Canada, for performing flash anneals. References [1] P.M. Mooney, Mater. Sci. Eng. B 134 (2006) 133–137. [2] Y. Liu, et al., Symposium on VLSI Technology Digest of Technical Papers, vol. 44, 2007. [3] A. Cacciato, et al., J. Appl. Phys. 79 (1996) 2314. [4] W. Krull, B. Haslam, T. Horsky, T. Verheyden, K. Funk, Proceedings of the International Conference on Ion Implant Technology, 2006, p. 142. [5] K. Sekar, et al., IEEE International Conference on Advanced Thermal Processing of Semiconductors, RTP, 2006, p. 251. [6] K. Sekar, et al., Proceedings of the International Workshop on INSIGHT in Semiconductors Device Fabrication, Metrology and Modeling, 2006, p. 141. [7] J. Gelpey, et al., Proceedings of the Electro-chemical Society (ECS) Meeting, 2002, p. 313. [8] K. Sekar, et al., Proceedings of the Materials Research Society Symposium, vol. 1070, 2008. [9] A. Li-Fatou, et al., Proceedings of the Electro-chemical Society Meeting, vol. 212, 2007, p. 1305. [10] J. Hornstra, W.J. Bartels, J. Cryst. Growth 44 (1978) 513. [11] M. Berti, et al., J. Appl. Phys. 72 (1998) 1602. [12] K.M. Kramer, M.O. Thompson, J. Appl. Phys. 79 (1996) 4118. [13] M.S. Goorsky, et al., Appl. Phys. Lett. 60 (1992) 2758. [14] J.W. Stane, et al., J. Appl. Phys. 76 (1994) 3656.