J. Phys. Chem. 1985,89, 2283-2285
2283
A Theoretical Study of the Silylboranes Charles W. Bock, Mendel Trachtman, Department of Chemistry, Philadelphia College of Textiles and Science, Philadelphia, Pennsylvania I9144
and Gilbert J. Mains* Department of Chemistry, Oklahoma State University, Stillwater, Oklahoma 74078 (Received: May 17, 1984)
The equilibrium structures and stabilities of the silylboranes, (SiH3),,BH3-",n = 1,2, 3, are studied by ab initio calculations at the HF/6-31G** level. MP2, MP3, and MP4SDQ/6-31G**//HF/6-3IG** calculations show SiH3BH2to be stable with respect to dissociation into SiHz and BH3, thus ruling out a suggestion that this process leads to polysilane and diborane at ambient temperature. The decompositionof silylborane via a homogeneous bimolecular process is also ruled out at ambient temperature. In fact, the reverse reaction, Le., the insertion of singlet silene into borane, appears to have no barrier at the HF/3-21G level or at the HF/6-31G** level. Silylboranesare suggested as possible intermediatas by which boron is incorporated into amorphous silicon.
Introduction
Research to discover and develop an inexpensive method of producing efficient amorphous silicon solar cells is being pursued intensively worldwide. Novel amorphous Si-C alloys' have produced solar cells with efficiencies in excess of 9% and novel modes of deposition2 have been reported with some success. The pdoped layers of these amorphous devices are usually constructed by codeposition of boron and silicon from a mixture containing diborane in silane either by pyrolysis3 (chemical vapor deposition, CVD) or by a glow di~charge.~Although the role of silylboranes as possible intermediates in the deposition or as a possible alternate means of introducing boron into the amorphous silicon film is of interest, experimental studies are difficult because of the instability of these substance^.^ This research reports ab initio calculations of the formation, stability, and structure of silylborane, SiH3BH2, and the structure and stability of disilylborane, (SiH3)2BH,and trisilylborane, (SiH3)3B. Although none of the silylboranes have beem identified by direct observation as an isolated entity, the existence of SiH3BH2has been postulated5 to explain the reactions of potassium silyl (KSiH3) and chloroborane (ClBH2) in the presence and absence of triethylamine. In the presence of the amine a solid substance was formed which was identified as H3SiBH2:N(C2H5)3,the Lewis adduct of silylborane; polysilane and diborane were the observed products in the absence of the amine. The suggested mechanismS involved the rearrangement of silylborane, reaction 1, which was H
H
TABLE I: Geometries of SiH$H2, (SiH,),BH, and (SiH3),B Using the HF/6-31GS*5D Basis Set
bond lengths" Si-B B-H Si-H
SiH,BH,
(SiH,),BH
(SiH,),B
2.0398 1.1896 1.4800 1.4839
2.0420 1.1918 1.4796 1.4836
2.0440
bond angles"
SiH3BH2
(SiH3)2BH
(SiH3)3B
121.3 112.4 107.7
117.5 112.6 107.4 124.9
112.8 106.7 120
LHcBSi LHiSiB LH9SiB LSiBSi dihedral angles" LH9SiBH6 LH9SiBSi LH9BSiH8 LH,Si,BH,
SiH3BH2
(SiH,)2BH
1.4793 1.4835
(SiH43B
89.0 118.8
86.4 118.8
"Bond lengths are in angstroms; angles in degrees. endothermic (vide infra), attention was then directed to alternate modes of decomposition of silylborane and to their possible role in the deposition of amorphous silicon from SiH4/B2H6mixtures. Finally, consideration was given to the structures and stability of disilylborane, (SiH3),BH, and trisilylborane, (SiH3)3B. Computational Methods
All molecular orbital calculations were performed with the program developed by Pople and his colleagues' on a VAX 11/780 computer. Some preliminary calculations were carried out with QCPE program no. 437 which predicted an incorrect linear geometry for SiH, and was not employed further. The Hartree-Fock calculations reported here employed the 63 1G**(5D) basis set8 and include geometry optimizations. These computations were followed by single-point calculations at the HF-optimized geometries using Mraller-Plesset (MP) perturbation theory9 at the MP2, MP3, and MP4SDQ (frozen core) levels to assess the effect of electron correlation on the energy differences. The STO-3G, 3-21G, and 6-31G** basis setslo were used to explore the potential surface for the reaction of SiH2 and BH3 to form silylborane. GAUSSIAN 82
believed to be driven by the well-known electronegativity of the boron hydrides. Dimerization of borane to diborane and polymerization of silene to polysilane were suggested to explain the observed reaction products. While attempting to repeat this research and prepare significant quantities of the amine-stabilized silylborane,6 a concurrent a b initio study was undertaken. The proposed study presumed reaction 1 was correct and computations were to focus on the trajectory of the H atom from the silicon to the boron atom. When it was discovered that reaction 1 was (1) Y. Yamakawa, K. Fujimotoa, K. Okuda, Y. Kashima, S.Nonomura, and H. Okamotm, Appl. Phys. Lett., 43, 644 (1983). (2) T. Saitoh, S. Muramatus, T.Shimada, and M. Migitaka, Appl. Phys. Lett., 42, 678 (1983). (3) S.K. Iya, R. N. Flagella, and F. S.DiPaola, J. Electrochem. Soc., 129, 1531 (1982). (4) N. Hata, A. Matsuda, K. Tanaka, K. Kajiyama, N. Moro, and K. Sajiki, Jpn. J. Appl. Phys., 22, L1 (1983). ( 5 ) E. Amberger and R. Romer, Z . Anorg. Allg. Chem., 345, 1 (1966). (6) H.Berko and G. J. Mains, unpublished.
0022-3654/85/2089-2283$01.50/0
Results and Discussion
The HF/6-31G** optimized geometries of the silylboranes are given in Table I and shown in Figures 1-3. To conserve computer ~~~~~
~
(7) J. S. Binkley, M. J. Frisch, D. J. DeFrees, K. Raghavachari, R. A. Whiteside, H. B. Schlegel E. M. Fluder, and J. A. Pople, GAUSSIAN82 (Release A, 1 Sept, 1983). Carnegie-Mellon University, Pittsburgh, PA 15213. (8) P. C. Hariharan and J. A. Pople, Theor. Chim. Acta, ZS,213 (1973). (9) C. M ~ l l e rand M. S. Plesset, Phys. Reu., 46, 678 (1934).
0 1985 American Chemical Society
2284
The Journal of Physical Chemistry, Vol. 89, No. 11, 1985
Bock et al.
TABLE II: Total Energiesec levele SiH3BH2 (SiH3)2BH (SiH3)3B HF/6-3 1G**(5D)/ /6-3 1G**(5 D) -316.467 905 -606.544 312 -896.621 740 MP2/6-31G**(5D)//6-31G**(5D) -316.649 322 -606.816 162 MP3/6-31G**(sD)//6-31G**(5D) -316.682270 MP4SDQ/6-31G**(5D)//6-31G**(SD) -316.688 156
SiH2 -290.001 887 -290.082 188 -290.100409 -290.105 066
Si2H4 -580.081 201d -580.266 921' -580.298715' -580.304 531'
B2H6
-52.819 292 -53.036 712 -53.069 114 -53.073 877
BH3 -26.392642 -26.485 331 -26.502631 -26.505 748
"Total energies in hartree au. bNot adjusted for zero-point energies. CCompletegeometries for SiH,, Si2H4,B2H6, and BH3can be obtained from C.W.B. dTrans-bent. ePost HF calculations all done with frozen core. $ 2 I
# /
! /' ~
/'
/
!
/' /'
/'
!
/
8'7
#'Y
vI
Figure 1. HF/6-31G**(5D) equilibrium structure of SiH3BH2. See Table I for bond lengths and bond angles.
Figure 2. HF/6-3 1G**(5D)equilibrium structure of (SiH3)2BH.Silyl groups were constrained to be identical. See Table I for bond lengths and bond angles.
time, the two silyl groups in disilylborane were constrained to be identical and the three silyl groups in trisilylborane were similarly constrained. This is reasonable since a rigid rotation of 30° about the Si-B bond in silylborane increases the energy by only 0.1 kcal/mol at both the 6-31G**(5D) and MP4SDQ levels. All remaining variables were allowed to optimize. Contrary to the structure of BH3 the three bonds emanating from the boron atom for all the silylboranes do not lie in the same plane (see Figures 1-3). For silylborane and disilylborane there are also small deviations from C3symmetry for the bonds about the boron atom. There is currently no experimental geometry with which to compare these predictions. The small deviations of the boron bonds from planarity and C3symmetry were observed for each basis set employed and the authors believe they are real. However, these deviations are well within the confidence limits suggested by Pople and his coauthorslh and, hence, will have to await confirmation by future experimental studies and/or more sophisticated computations. Aside from this, there is nothing (10) (a) D. J. DeFrees, B. A. Levi, S. K. Pollack, W. J. Hehre, J. S. Binkley, and J. A. Pople, J . Am. Chem. SOC.,101, 4085 (1979). (b) R. Ditchfield, W. J. Hehre, and J. A. Pople, J . Chem. Phys., 52, 5001 (1970). (c) P. C . Hariharan and J. A. Pople, Mol. Phys., 27,209 (1974). (d) M. M. Francl, W. J. Pietro, W. J. Hehre, J. S. Binkley, M.S. Gordon, D. J. DeFrees, and J. A. Pople, J. Chem. Phys., 77, 3654 (1982). (e) J. S. Binkley, J. A. Pople, and W. J. Hehre, J . Am. Chem. SOC.,102, 939 (1980).
Figure 3. HF/6-31GS*(5D) equilibrium structure of (SiH&B. Silyl groups were constrained to be identical. See Table I for bond lengths and bond angles.
especially noteworthy regarding the geometries. The common bond distances and angles do not change significantly when SiH3 is substituted for H. Table I1 provides total energies for the silylboranes. Complete 6-3 1G**(5D) geometry optimizations were also performed on BH3, B2H6,SiH2,and Si2H4to ensure internal consistency. Since the geometries of these molecules have been reported previously using similar basis we have only included the total energies in Table 11. In Table I11 we report the energy change for selected chemical reactions. The energy change for reaction 1 was calculated to be some 46 kcal/mol endothermic at the HF/6-31G** level, see Table 111. Furthermore, the inclusion of correlation at the MP2, MP3, and MP4SDQ/6-31G**//HF/6-31G** level only served to make the calculated energy change for reaction 1 more endothermic. It is clear from these results that the decomposition of silylborane to borane and silene cannot w u r homogeneously in the gas phase at ambient temperature by reaction 1. There are a number of possible explanations for the experimental observation. One explanation is that the experimental reaction in the absence of the amine does not involve silylborane at all. In order to examine this question one would have to consider the interaction of CIBHz at the surface of KSiH,(s). Other possible explanations involve the formation of silylborane and subsequent homogeneous or heterogeneous decomposition. The heterogeneous processes are not readily amenable to examination by standard ab initio methods. Hence, consideration was given to possible homogeneous mechanisms. In view of the experiment itself, a bimolecular reaction such as reaction 2 seemed worth
-
2SiH3BH2
Si2H4+ B2H6
(2)
consideration. The change in total energy for reaction 2 at the Hartree-Fock level was not encouraging, still some 22 kcal/mol endothermic. See Table 111. Since correlation energy was expected to make a larger contribution to the total energy of the (11) J. S . Binkley and L. R. Thorne, J . Chem. Phys., 79, 2932 (1983). (12) J. D. Dill, P. v. R. Schleyer, and J. A. Pople, J . Am. Chem. SOC., 97, 3402 (1975). (13) M. Colvin, R. S. Grev, H. F. Schaefer, and J. Bicerano, Chem. Phys. Lett., 99, 399 (1983). (14) K. Krogh-Jespersen, Chem. Phys. Lett., 93, 327 (1982). (15) H. Lischka and H. Kohler, Chem. Phys. Letr., 85, 467 (1982).
The Journal of Physical Chemistry, Vol. 89, No. 11, 1985 2285
Theoretical Study of the Silylboranes smaller product fragments of reaction 2 than to the silylborane reactant, the energies were also computed at MP2, MP3, and MP4SDQ/6-31G**//HF/6-31G** levels. Indeed, as can be seen in Table 111,inclusion of correlation energy at the MP2 level drives the total energy change exothermic to the extent of -3.1 kcal/mol. However, higher levels of perturbation theory make the total energy change less exothermic, -2.1 and -1.4 kcal/mol at the MP3 and MP4SDQ levels, respectively. Zero-point energies for diborane16 and disilene14Js have been reported as 39.64 and 20.3 kcal/mol, respectively. A vibrational analysis for silylborane using the HF/6-31G**(5D) basis set yields a zero-point energy of 29.04 kcal/mol, lowering the exothermicity of reaction 2 at the MP2 level to about -1.2 kcal/mol, and actually makes the reaction endothermic at the MP4SDQ level." Furthermore, geometrical consideration^^^ clearly suggest a high activation energy for reaction 2 and in all probability it can be ruled out in any explanation of the experimental observations at ambient temperature. Thus silylborane, if formed in the experimental system at all, decomposes by processes other than reactions 1 and 2. Heterogeneous processes, however, cannot be ruled out. It is important to point out that both reactions 1 and 2 can occur under the conditions of chemical vapor deposition where temperatures in excess of 1000 K are common. Indeed, the computations reported here suggest that the formation of silylborane, i.e., the reverse of reaction 1, is possible under the conditions of chemical vapor deposition, where both SiH2 and BH3 have been proposed as intermediates in the pyrolysis of SiH4 and B2H6. Initially, the STO-3G basis set was used to search for a transition state (TS) for the insertion of singlet silene into borane, i.e. SiH2 + BH3 [SiH2BH3JTS SiH3BH2 (3) The search procedure employed a geometry with a C, framework group (the plane of symmetry included one H atom from BH3 and the B and Si atoms). All parameters were optimized exce t the B-Si distance which was fixed at 2.4, 2.2, and then 2.0 . The energy decreased monotonically. Since 2.0 A is close to the STO-3G equilibrium Si-B distance in silylborane (1 .9633 A), a full (including B-Si) optimization was allowed. A local minimum was found at a B-Si distance of 1.96g4 A with the geometry of BH3 and SiH2 still in evidence.18J9 Next, a TS was found for
-
-
w
(16) L. T. Redmon, G. D. Purvis, and R. J. Bartlett, J . Am. Chem. SOC., 101, 2856 (1979). (17) Zero-point energies computed by using H F calculations are too large and consequently the actual zero-point energy difference is even greater than 1.9 kcal/mol making reaction 2 even more unlikely. (18) Partial STO-3G geometries: BH3: B-H = 1.1600 8, SiH,: adduct:
Si-H = 1.458, and LHSiH = 91.5O B-H = 1.169,8, and 1.1566A B-Si = 1.96948,
TABLE III: Summary of Energy' Changes for Selected Reactionsb reaction HF MP2 MP3 MP4 SiH3BH2 SiHl + BH, 46.1 51.4 49.1 48.5 2SiH3BH2 Si2H4 B2H6 22.1 -3.1 -2.1 -1.4 SiH2 SiH2 SiH2
-+ -
+ +
+
--
+
B2H6 SiH3BH2 BH, SiH3BH2 (SiH,),BH (SiH3)2BH (SiH3),B
-24.7 -46.8 -47.1
-9.9 -53.1
-9.7
-9.4
kcal/mol. bNot corrected for zero-point energy differences.
a 1,2 shift between this adduct and the product silylborane at an energy of only 0.2 kcal/mol above the adduct.'s,20 However, optimizations at the 3-21G level failed to find any adduct and, hence, there appears to be no barrier for singlet SiH2 insertion into BH3. Further calculations at the 6-31G** level support this conclusion. This absence of a barrier is consistent with the essentially zero activation energy computed by Gordon and Gano2I for insertion of SiH2into SiH4. A very low activation energy is consistent with the experimental activation energy of 1.3 f 1.1 kcal/mol reported for the insertion of silene into silane.22 Indeed, SiH2 is known to be very reactive.23 The relative rate of silene insertion has been studied experimentally and related to the hydridic character of the bond broken.24 On this basis one would expect the insertion of SiH2 into SiH4 to dominate in the CVD systems since the Si-H bond is more hydridic than the B-H bond. This, perhaps, explains why the relative concentration of diborane in the gas phase must be significantly higher than the desired boron doping in the solid phase. Thus, we conclude that formation of silylborane is both thermodynamically and kinetically favored in CVD systems. The insertion of SiHz into diborane to form silylborane is also energetically possible as can be seen from Table I11 and is more probable in terms of collisions under most CVD conditions. We have not explored for a transition state for this reaction but, since it involves bond breakage as well as bond making, the TS is probably high.25 Sequential insertions of SiH2 to form disilylborane and trisilylborane deserve consideration since they are also possible energetically (see Table 111). Thus, it would appear possible, even likely, that silylboranes are involved in the mechanism by which boron is incorporated into amorphous silicon. The formation and reactions of silylboranes under CVD conditions are worthy of further theoretical and experimental investigations. Such studies are planned and will be reported later. Acknowledgment. The authors gratefully acknowledge the generous grant of computer time provided by the computer center of the Philadelphia College of Textiles and Science. We also thank Dr. Andrew Komornicki for some helpful suggestions. Registry No. SiH3BH2, 14809-30-4; (SiH,),BH, 95864-35-0; (SiH3)3B, 95864-36-1.
Si-H = 1.43698, and LHSiH = 102.4O TS:
B-H = 1.1526 8, Si-H = 1.422*8, and LHSiH = 110.6O B-Si = 1.829,
A
B-Hbidgc = 1.4731 8, Si-Hbridge= 1.5941 8, SiH3BH2:
B-H = 1.1607 8, Si-H = 1.423, 8, and 1.42498,
= 1.963, 8, Complete geometries can be obtained from C.W.B.
(19) The complete set of STO-3G harmonic force constants was also evaluated for this adduct. The eigenvalues were all positive. (20) The complete set of STO-3G harmonic force constants was also evaluated for this transition state and, as expected, there was one negative eigenvalue of the force constant matrix. (21) M. S. Gordon and D. R. Gano, J . Am. Chem. Soc., 106, 5421 (1984). (22) P. John and J. H. Purnell, J. Chem. SOC.,Faraday Trans. I , 69,1455 (1973). (23) R. West, Science, 225, 1109 (1984). (24) M. D. Sefcik and M. A. Ring, J. Am. Chem. SOC.,95, 5168 (1973). (25) There are numerous systems for which a direct molecular reaction exhibits very high activation energies. For a recent example, see the transition state for the addition of molecular hydrogen to formaldehyde in L. B. Harding, H. B. Schlegel, R. Krishnan, and J. A. Pople, J. Phys. Chem.,84,3398 (1980).
