1: % INJ
2: \documentclass[12pt]{article}
3: \usepackage{graphicx}
4:
5: \advance\textwidth by 3truecm
6: \advance\oddsidemargin by -1.5truecm
7:
8: \newcommand\HF{{\mathrm{HF}}}
9: \newcommand\BOCE{{\mathrm{BOCE}}}
10:
11: \begin{document}
12: \title{\normalsize\bf
13: Bond-order correlation energies for small Si-containing molecules
14: compared with \emph{ab initio} results from low-order
15: M\o{}ller-Plesset perturbation theory}
16: \author{\normalsize A. Grassi, G. M. Lombardo, G. Forte\\
17: \normalsize\emph{Dipartimento di Scienze Chimiche, Facolt\`a di Farmacia,
18: Universit\`a di Catania,}\\
19: \normalsize\emph{Viale A. Doria, 6, I-95126 Catania, Italy}\\[\baselineskip]
20: \normalsize G. G. N. Angilella, R. Pucci\\
21: \normalsize\emph{Dipartimento di Fisica e Astronomia, Universit\`a di
22: Catania, and}\\
23: \normalsize\emph{Istituto Nazionale per la Fisica della Materia, UdR Catania,}\\
24: \normalsize\emph{Via S. Sofia, 64, I-95123 Catania, Italy}\\[\baselineskip]
25: \normalsize N. H. March\\
26: \normalsize\emph{Department of Physics, University of Antwerp,}\\
27: \normalsize\emph{Groenenborgerlaan 171, B-2020 Antwerp, Belgium, and}\\
28: \normalsize\emph{Oxford University, Oxford, UK}}
29:
30: \date{}
31:
32: \maketitle
33:
34: \abstract{%
35: The present study of small molecules containing silicon has been
36: motivated by (a) the considerable interest being shown currently in
37: the kinetics and reactivity of such molecules, and (b) the
38: biotechnological potential of silicon-derivate surfaces as
39: substrates in the adsorption of, for instance, amino acids and
40: proteins.
41: Therefore, we have studied by (i) a semi-empirical approach and (ii)
42: an \emph{ab initio} procedure employing low-order M\o{}ller-Plesset
43: perturbation theory, the molecular correlation energies of some
44: neutral closed and open shell silicon-containing molecules in the
45: series SiX$_n$Y$_m$.
46: Procedure (i) is shown to have particular merit for the correlation of
47: the ionic members studied in the above series, while the \emph{ab
48: initio} procedures employed come into their own for neutral
49: species.
50: }
51:
52:
53: \section{Introduction}
54:
55:
56: In the last twenty years, much work has been carried out in order to
57: calculate the correlation energy in atoms and molecules.
58: In particular, in atoms, March and Wind \cite{March:92a} showed how it
59: is possible to recover the main trend of the semi-empirical
60: correlation energies obtained long ago by Clementi
61: \cite{Clementi:63}, using simple arguments based on the correlation
62: energy density (see also \cite{Alonso:03}).
63:
64: Within the framework of Density Functional Theory (DFT), many studies
65: have been made to develop new correlation energy functionals.
66: These are now available as a result of the work of Vosko, Wilk and
67: Nusair (VWN) \cite{Vosko:80}, Lee, Yang and Parr (LYP)
68: \cite{Lee:88}, and Perdew and Wang (PW) \cite{Perdew:92}, and have
69: all been introduced into some quantum chemistry packages
70: ({\sc gaussian}, {\sc nwchem}, {\sc gamess}, etc).
71:
72: In a previous study (hereafter denoted as Paper~I) \cite{Grassi:96},
73: some of the present authors have introduced a new
74: semi-empirical technique in order to calculate the electron
75: correlation energy in molecules.
76: Essential to this work was the division of the total electron density
77: of the molecule into parts belonging to specific nuclei, and a part
78: belonging solely to the chemical bond and therefore, by using
79: semi-empirical modelling, strictly correlated with the molecular
80: bond-order (BO).
81: One important conclusion of this work was that the molecular
82: correlation energy is generally larger, in absolute value, than the
83: sum of the correlation energies of the separated atoms.
84: It is known from the pioneering work of Pauling, Mulliken, Coulson and
85: other authors \cite{Coulson:39,Pauling:35,Mulliken:59}, that bond
86: order and molecular bond length are strongly correlated.
87: In this sense our work is somewhat related to the work of Fulde
88: \emph{et al.} \cite{Pfirsh:85,Oles:86}, where an analytical model
89: of molecular correlation energies based on bond lengths was
90: proposed.
91:
92: In Paper~I we developed a parameterization for compounds containing
93: only hydrogen, oxygen and carbon atoms and the related bonds X--Y
94: (X, Y = C, H, O).
95: The very good results obtained in the calculation of the correlation
96: energy for some molecules containing these elements, compared with
97: the experimental values, encouraged us to extend this approach to
98: second-row elements and in particular to silicon-containing
99: compounds, like fluorine and chlorine-substituted silylenes, silyl
100: radicals and silanes.
101: It is relevant in this context to note that Schlegel \emph{et al.}
102: \cite{Schlegel:84,Ignacio:90,Su:93} have calculated the heats of
103: formation of these compounds by \emph{ab initio} methods,
104: introducing part of the correlation energy using the standard
105: M\o{}ller-Plesset perturbative expansion technique.
106:
107: We stress at this point that the kinetics and reactivity of small
108: silicon-containing molecules are attracting considerable interest,
109: due to the quite different reaction mechanism of the silicon
110: compounds with respect to the carbon compounds.
111: For example, pyrolysis of silanes yields silylenes rather than silyl
112: radicals, whereas photolysis produces silyl radicals
113: \cite{Gaspar:81}.
114: Moreover, silylenes and silyl radicals with various degrees of
115: fluorination (or chlorination) play an important role in chemical
116: vapor deposition \cite{Scott:81,Haller:83,Robertson:84}.
117: But the main area of interest of these compounds are the
118: silicon-derivate surfaces which are widely used as substrates in
119: the adsorption process of biological compounds, such as
120: amino~acids, proteins etc, in view of their application in
121: different biotechnology areas such as biomaterials, biosensors, and
122: bioseparation \cite{Gambino:04}.
123:
124: In this study we have investigated the correlation energies of
125: SiX$_n$Y$_m$ (X, Y = H, F, Cl; $n+m=4$, 3, 2) compounds using both
126: standard \emph{ab initio} techniques, at various levels of
127: approximation, and the bond-order method developed in Paper~I.
128: In section~\ref{sec:methods} we describe briefly the theoretical
129: methods used to estimate the molecular correlation energy.
130: M\o{}ller-Plesset theoretical techniques are summarized in
131: subsection~\ref{ssec:abinitio} while the Bond-Order Correlation
132: Energy (BOCE) method is described in subsection~\ref{ssec:boce}.
133: Section~\ref{sec:results} is concerned with the calculated correlation
134: energies of SiX$_n$Y$_m$ compounds and the ionization potentials
135: (IP) of some of these molecules together with a comparison with the
136: experimental values.
137: Our final conclusions are recorded in Section~\ref{sec:summary}.
138:
139: \section{Methods}
140: \label{sec:methods}
141:
142: \subsection{\emph{Ab initio} methods}
143: \label{ssec:abinitio}
144:
145: \emph{Ab initio} molecular orbital Hartree-Fock calculations (HF) were
146: performed with the {\sc gaussian~03} package \cite{Frisch:03} using
147: the polarization 6-31G$^{\ast\ast}$ basis sets.
148: The 6-31G$^{\ast\ast}$ basis contains a set of six $d$ Cartesian
149: functions on each of the heavy atoms, and a set of three $p$
150: function on each hydrogen.
151: Equilibrium geometries were obtained by full optimization using
152: analytical gradient methods \cite{Schlegel:82}.
153: Electron correlation energy was estimated with M\o{}ller-Plesset (MP)
154: perturbation theory \cite{Moeller:34} up to fourth order (MP4),
155: including all electrons and considering single, double excitations.
156: Full molecular geometry optimization was performed up to third order
157: M\o{}ller-Plesset level (MP3).
158: Quadruple excitations (MP4SQDT) were introduced by fixing the
159: molecular geometry obtained from MP3 calculations.
160: For both HF and MP calculations, restricted Hartree-Fock (RHF) and
161: unrestricted Hartree-Fock (UHF) methods were applied to closed and
162: open shell systems, respectively.
163:
164: We stress, at the outset, that the emphasis of this work is the
165: correlation energy.
166: To present such results on a large number of Si-containing compounds,
167: we have deliberately restricted the basis sets used.
168: To extend these is straightforward for the future, should it be deemed
169: instructive to do so.
170:
171: \subsection{Bond-Order Correlation Energy (BOCE) Method}
172: \label{ssec:boce}
173:
174: The `experimental' correlation energy $E_c^{\exp}$ is defined as
175: \begin{equation}
176: E_c^{\exp} = E_S - E_\HF,
177: \label{eq:S-HF}
178: \end{equation}
179: where $E_\HF$ denotes the (molecular) Hartree-Fock energy and $E_S$ is
180: the so-called Schr\"odin\-ger (or ``true'') energy.
181: In the case of atoms, the latter is obtained from the sum of the
182: experimental ionization energies, plus Bethe-Salpeter and
183: relativistic corrections.
184: In the case of molecules, $E_S$ also includes smaller contributions,
185: taking into account for the other molecular degrees of freedom, such
186: as vibrational and rotational motions.
187: In both cases, $E_S$ can be constructed from available data
188: \cite{Lide:94} following the standard procedure described in
189: Cremer's papers \cite{Cremer:82,Cremer:82a}.
190: By using this procedure and the available experimental data, we have
191: obtained $E_S$ for all molecules in the series SiX$_n$Y$_m$ (X, Y =
192: H, F, Cl; $n+m=4$, 3, 2).
193: Although $E_c^{\exp}$ in Eq.~(\ref{eq:S-HF}) is usually referred to as
194: an `experimental' correlation energy, it obviously contains some
195: theoretical input through the Hartree-Fock energy $E_\HF$, whose
196: degree of approximation can be controlled by choosing a
197: sufficiently large basis set.
198: Moreover, for some molecules, whenever the experimental data for the
199: heat of formation, or the vibrational molecular frequencies were
200: unavailable, we have made recourse to available theoretical
201: estimates of these quantities, as indicated in the table captions.
202: Fig.~\ref{fig:schema} summarizes the definitions of the experimental
203: correlation energies, as well as of other various energies
204: addressed in turn below.
205:
206:
207: Within the bond-order approach \cite{Grassi:96}, the correlation
208: energy can be estimated as follows.
209: Given the formation reaction of a generic molecule,
210: \begin{equation}
211: A + B \longrightarrow C + \mbox{binding energy},
212: \end{equation}
213: the molecular energy can be partitioned as
214: \begin{equation}
215: E(C) = E(A) + E(B) - \mbox{binding energy} = E(A) + E(B) + E(AB),
216: \label{eq:binding}
217: \end{equation}
218: where $E(AB)$ is the energy due to the $A$--$B$ bond formation.
219: Then, in general, the Schr\"odinger energy $E_S$ of a polyatomic
220: molecule can be written as
221: \begin{equation}
222: E_S = \sum_A E_S (A) + \sum_{{\mbox{\scriptsize all $AB$}}} E_S (AB).
223: \label{eq:S}
224: \end{equation}
225: Likewise, in the HF approximation, the total molecular energy can be
226: expressed in the form
227: \begin{equation}
228: E_\HF = \sum_A E_\HF (A) + \sum_{\mbox{\scriptsize all $AB$}} E_\HF (AB).
229: \label{eq:HF}
230: \end{equation}
231: Subtracting Eq.~(\ref{eq:HF}) from Eq.~(\ref{eq:S}), one obtains the
232: theoretical correlation energy as
233: \begin{eqnarray}
234: E_c^{\mathrm{theo}} &=& \sum_A [ E_S (A) - E_\HF (A)] +
235: \sum_{\mbox{\scriptsize all $AB$}} [ E_S (AB) - E_\HF (AB) ] \nonumber\\
236: &=& \sum_A E_c^{\mathrm{exp}} (A_{\mbox{\scriptsize in molecule}} ) +
237: \sum_{\mbox{\scriptsize all $AB$}} E_c^{\mathrm{theo}} (AB),
238: \label{eq:theo}
239: \end{eqnarray}
240: where we take as known the Schr\"odinger atomic energy from
241: experimental data, as explained in the next section.
242: In Eq.~(\ref{eq:theo}), $E_c^{\mathrm{exp}} (A_{\mbox{\scriptsize in
243: molecule}} )$
244: is the effective correlation energy for the atom $A$ \emph{in the
245: molecule.}
246: This quantity is obtained from $E_c^{\mathrm{exp}} (A)$ relative to an
247: isolated atom $A$ and rescaled taking into account the effective
248: number of non-bonding electrons of the atom $A$ in the molecule.
249: In particular, if $n$ is the total number of electrons of $A$ involved
250: in the bonds and $Z$ its atomic number, $E_c^{\mathrm{exp}}
251: (A_{\mbox{\scriptsize in molecule}} )$ is obtained as
252: \begin{equation}
253: E_c^{\mathrm{exp}} (A_{\mbox{\scriptsize in molecule}} ) =
254: E_c^{\mathrm{exp}} (A) \frac{Z-n}{Z} .
255: \label{eq:Zn}
256: \end{equation}
257:
258: The next step in the present approximation scheme is to consider
259: $E_c^{\mathrm{theo}} (AB)$ as an analytical function of the
260: bond-order $P_{AB}$ between the two atoms $A$ and $B$ of the
261: molecule.
262: In practice, we have assumed a linear form for $E_c^{\mathrm{theo}}
263: (AB)$ as a function of $P_{AB}$, namely
264: \begin{equation}
265: E_c^{\mathrm{theo}} (AB) = a_{AB} P_{AB} ,
266: \label{eq:Loewdin}
267: \end{equation}
268: where $a_{AB}$ is a parameter which depends on the $A$--$B$ bond.
269: Moreover, the $P_{AB}$ bond-order, according to L\"owdin's definition
270: \cite{Loewdin:50}, can be written as
271: \begin{equation}
272: P_{AB} = \sum_\mu^A \sum_\nu^B D_{\mu\nu} D_{\mu\nu} ,
273: \end{equation}
274: with $D= S^{1/2} T S^{1/2}$, $S$ being the overlap matrix and $T$ the
275: first-order density matrix obtained from the HF calculation.
276: One then obtains:
277: \begin{equation}
278: E_c^{\mathrm{theo}} = \sum_A E_c^{\mathrm{exp}}
279: (A_{\mbox{\scriptsize in molecule}} ) + \sum_{\mbox{\scriptsize all
280: $AB$}} a_{AB} P_{AB} .
281: \end{equation}
282: The parameters $a_{AB}$ can be obtained from a model system which
283: contains the $A$--$B$ bond and for which is it possible to
284: calculate the Schr\"odinger molecular energy $E_S$ and then the
285: experimental correlation energy $E_c^{\mathrm{exp}}$.
286: After obtaining the best values of the $a_{AB}$ parameters for each
287: $A$--$B$ bond, application of Eq.~(\ref{eq:Loewdin}) then
288: permits to calculate the theoretical correlation energy
289: $E_c^{\mathrm{theo}}$.
290: In Appendix~\ref{app:bond} we report a specific example to illustrate the
291: procedure used to obtain the $a_{AB}$ parameters.
292:
293:
294: \section{Results and discussion}
295: \label{sec:results}
296:
297: \subsection{Neutral closed and open shell SiX$_n$Y$_m$ compounds}
298: \label{ssec:neutral}
299:
300: In order to calculate the correlation energy of fluorine and
301: chlorine-substituted silylenes compounds, we have obtained the bond
302: parameter $a_{AB}$ of Eq.~(\ref{eq:Loewdin}) using the experimental
303: correlation energy $E_c^{\mathrm{exp}}$ of some model molecules and
304: the experimental atomic correlation energies.
305: In Tab.~\ref{tab:HF} we report the Schr\"odinger and the HF atomic
306: energies of the constituent atoms in their fundamental state and
307: the related correlation energies.
308: In Tab.~\ref{tab:bond} we record results for some model systems
309: together with the bond parameter extracted from
310: $E_c^{\mathrm{exp}}$.
311: All energy values are in Hartrees.
312:
313: In the second column of Tab.~\ref{tab:theo} we record the
314: Schr\"odinger energy of the SiX$_n$Y$_m$ compounds under study,
315: obtained from experimental values of the heat of formation and of
316: the vibrational frequencies of these molecules
317: \cite{Schlegel:84,Ignacio:90,Su:93}.
318: For the compounds for which no such experimental data were available,
319: the theoretical values of these quantities were used to calculate
320: the Schr\"odinger energy.
321: Columns 3--6 of Tab.~\ref{tab:theo} report the total energy obtained
322: at the HF, MP2, MP3 and MP4SDTQ4 (MP4) levels.
323: In the last column, the BOCE energy is reported.
324: For the sake of simplicity, Tab.~\ref{tab:theo} does not include the
325: molecules employed in the BOCE treatment to calculate the
326: bond-order parameter $a_{AB}$.
327: Tab.~\ref{tab:exptheo} reports the experimental and the calculated
328: correlation energies for the same molecules listed in
329: Tab.~\ref{tab:theo}.
330:
331: From Tab.~\ref{tab:theo}, it is evident that using the \emph{ab
332: initio} methods, the successive corrections to the molecular energy
333: are very small.
334: The differences between the MP2 and MP3 results range in general
335: between $0.02-0.04$~a.u., while the difference between the next
336: successive corrections, MP3 and MP4, is even smaller
337: ($0.01-0.02$~a.u.).
338: For some molecules (SiHF$_3$, SiHF$_2$, SiF$_2$), the total energy
339: increases on passing from the MP2 to the MP3 correction.
340: This trend is better evidenced by the correlation energies in
341: Tab.~\ref{tab:exptheo}.
342: There, the change in the values of $E_c$ ranges between 0.02 and
343: 0.05~a.u., on going from the MP2 to the MP4 correction.
344:
345:
346: The very good agreement of the calculated molecular energy using the
347: BOCE technique with the Schr\"odinger value appears clearly from
348: the values in Tabs.~\ref{tab:theo} and \ref{tab:exptheo}.
349: The highly accurate values of the correlation energy obtained from
350: BOCE for all molecules of the series are confirmed from the
351: calculated percentage error with respect to the experimental
352: correlation energy, ranging within $0.002-4.06$~\%.
353: On the contrary, \emph{ab initio} correlation energies yield very high
354: percentage errors.
355: In Tab.~\ref{tab:theo}, for some molecules, the calculated correlation
356: energy with the BOCE method gives a value higher than the
357: experimental one, and the corresponding molecular energy is lower
358: than the Schr\"odinger energy.
359: The absolute percentage error for these molecules has a value in the
360: range between 0.002~\% and 0.4~\%.
361: However, it is important to add some comments here.
362: The first one is that the Schr\"odinger energy is obtained from
363: experimental data and any experimental measurement is subject to an
364: error that, in general, is about $\pm 5$~\%.
365: The second comment, pertaining to the compounds corresponding to a
366: molecular energy lower than the Schr\"odinger energy, is that one
367: needs to note that in closed shell systems, such as those
368: considered here, this happens only in two cases, where we have used
369: some theoretical values (vibrational frequencies, or molecular heat
370: of formation, or both) in order to determine the Schr\"odinger
371: molecular energy.
372:
373: The results of Tab.~\ref{tab:exptheo} are shown in
374: Fig.~\ref{fig:exptheo}.
375: From this figure it can be seen that the trends of the calculated
376: correlation energies using \emph{ab initio} methods are close to
377: the experimental behaviour, but the values are on average
378: 1.5~a.u. higher than the experimental values.
379: Moreover, the increase in the calculated correlation energy on passing
380: from MP2 to MP4 is negligible and, taking into account that the
381: computational effort increases considerably from MP2 to MP4, we can
382: conclude that it is not necessary to make corrections at higher
383: order than MP2.
384: This is confirmed by Fig.~\ref{fig:percentage}, where the absolute
385: percentage error between the experimental and theoretical
386: correlation energies is almost constant for each compound on
387: passing from MP2 to MP4.
388:
389: The extent of the agreement between the experimental and the
390: calculated BOCE correlation energies is shown in
391: Fig.~\ref{fig:exptheo} (last panel), where the experimental and the
392: BOCE curves are almost superimposed and hardly distinguishable.
393: Consequently, as shown in Fig.~\ref{fig:percentage}, the percentage
394: error between the experimental and the calculated BOCE correlation
395: energies is nearly constant, and varies between 0.002~\% and 4.0~\%.
396:
397:
398: \subsection{Ionization potentials (IP)}
399: \label{ssec:ip}
400:
401: We have also calculated the ionization potentials (IP) of some
402: SiX$_n$Y$_m$ compounds, using both \emph{ab initio} and BOCE
403: methods.
404: These results have then been compared with the experimental values.
405: Within the so-called $\Delta$SCF procedure, the theoretical values of
406: IP (in eV) are defined as the difference between the total energies
407: of the neutral and the ionized molecule, \emph{i.e.}
408: \begin{equation}
409: \mathrm{IP} = E - E^+
410: \end{equation}
411: (see also Fig.~\ref{fig:schema}).
412: Table~\ref{tab:delta} reports the experimental and calculated IP for
413: the molecules under study.
414: In these calculations, the geometries of both the neutral and the ionic
415: species have been optimized.
416:
417: In the second column of Tab.~\ref{tab:delta} we record the IP values
418: obtained as the difference $E-E^+$, both calculated at HF level
419: (without correlation), while in columns 5, 7, and 9 this difference
420: refers to the calculated values at the MP2, MP3, and MP4 levels,
421: respectively.
422: Finally, column~11 reports the calculated IP values using the BOCE
423: method.
424: Fig.~\ref{fig:delta} displays the data in Tab.~\ref{tab:delta}.
425:
426: From Tab.~\ref{tab:delta} it is clear that the MP corrections,
427: contrary to the case of the neutral compounds, are important in the
428: IP calculation.
429: In fact, in all compounds of this series (with the exception of
430: SiH$_4$), the absolute percentage error of the IP at the HF level
431: is very high ($\sim 7$~\%; see Fig.~\ref{fig:ip}), and the
432: introduction of correlation effects at MP2 level reduces this error
433: to 3.26~\%, on the average.
434: Then the introduction of correlation in the estimate of IP yields a
435: calculated value closer to the experimental one.
436: The trend of the next corrections, both at the MP3 and at the MP4
437: levels, is not constant.
438: For some compounds of this series, SiH$_4$, SiH$_2$F$_2$, SiHCl$_3$,
439: and SiF$_4$, the percentage error decreases on passing from MP2 to
440: MP4.
441: On the contrary, for other compounds (SiCl$_4$, SiH$_3$Cl) the error
442: increases, while for SiHF$_3$ and SiH$_2$Cl$_2$ the best estimate
443: of IP is obtained at the MP3 level.
444: The average absolute percentage errors for the MP2, MP3, and MP4
445: corrections are 3.26~\%, 3.26~\%, and 3.44~\%, respectively, thus
446: showing that all these low order MP corrections give
447: almost equivalent approximations of the IP.
448: Coming now to the BOCE approach, from the absolute percentage errors
449: plotted in Fig.~\ref{fig:ip} one may conclude that this method
450: yields more accurate estimates of the IP, on the average, than the
451: \emph{ab initio} methods, although the MP methods occasionally
452: produce lower absolute percentage errors in IP than the BOCE
453: method.
454:
455: A final consideration concerns the calculated differences, $\Delta$,
456: between the experimental and calculated IP, reported in
457: Tab.~\ref{tab:delta}.
458: Fig.~\ref{fig:schema} schematically defines the experimental and
459: theoretical correlation energies for both neutral and ionic
460: molecules.
461: According to Fig.~\ref{fig:schema}, we may then deduce that
462: \begin{equation}
463: \mathrm{IP}^{\mathrm{exp}} - \mathrm{IP}^{\mathrm{theo}} =
464: (E_c^{\mathrm{exp}} - E_c^{\mathrm{theo}} ) -
465: (E_c^{+\,\mathrm{exp}} - E_c^{+\,\mathrm{theo}} ) = \Delta -
466: \Delta^+ ,
467: \end{equation}
468: where a superscript $+$ refers to the same quantity in the ionic
469: molecule.
470: In Tab.~\ref{tab:delta} we record the experimental and theoretical
471: ionization potentials (IP), along with their difference, $\Delta$.
472: The fact that the average error in the calculation of the IP is not
473: very different within the MP and BOCE methods, is an indication
474: that there are large cancellations between the correlation energies
475: of the neutral and the ionic molecules.
476: In turn, this means that within the BOCE approach also the correlation
477: energy of the ions is well approximated by using the same bond
478: parameters $a_{AB}$, as derived for the same model molecules.
479:
480: \section{Summary and concluding remarks}
481: \label{sec:summary}
482:
483: The biotechnological possibilities of silicon-containing molecules in
484: substrate layers, plus much current interest in understanding the
485: quite different reaction mechanism of silicon compounds in
486: comparison with carbon compounds, has motivated us to extend our
487: earlier work \cite{Grassi:96} on the latter class to the case of
488: Si.
489:
490: What seemed to us important in this different series of small
491: molecules was to assess the utility of our earlier semi-empirical
492: use of molecular bond-order to estimate electron correlation
493: energies in some members of C-containing molecules when applied to
494: SiX$_n$Y$_m$, where X, Y = H, F, Cl, and $n+m = 4$, 3, 2.
495: In this process, we have found it valuable to compare and contrast our
496: bond-order approach with some ab initio results we have obtained
497: using low-order M\o{}ller-Plesset perturbation theory.
498: While, by these two approaches, we record quite a number of useful
499: results for molecular correlation energies of Si-containing
500: molecules, we wish especially to stress that the semi-empirical
501: bond-order approach proposed in \cite{Grassi:96} and applied there
502: to C-containing molecules continues to be valuable for molecules of
503: the class SiX$_n$Y$_m$.
504: It seems particularly useful when ionicity plays a role, whereas the
505: low-order M\o{}ller-Plesset approach comes into its own for
506: essentially neutral members of this series.
507:
508: In conclusion, we emphasize the essence of the present approach which
509: is to yield an impressive empirical correlation for obtaining bond
510: additivity corrections to the energy based on bond orders
511: calculated from a L\"owdin population analysis.
512: In order to present rather extensive results for a series of
513: Si-containing molecules, we have, in the present work, accepted the
514: limitations of the small basis set used.
515: However, should it prove instructive for future purposes, it is a
516: straightforward matter to extend the basis, even if somewhat
517: time-consuming.
518: Plainly, then, some quantitative changes will occur and also some
519: specific technical points can be examined such as the numerical
520: modifications in the L\"owdin population analyses, and the
521: parameters obtained from them, due to changes in the basis set.
522: Another matter deserving attention if eventually larger basis sets are
523: used concerns specifically the Si-F bond, which present indications
524: suggest is somewhat problematic in the calculations, or experiment,
525: or perhaps both.
526:
527: We are encouraged by the results of the present investigation to
528: attempt a pilot study of the relevance of our bond-order
529: considerations to the still more difficult area of the kinetics and
530: reactivity of small Si-containing molecules.
531: We hope to report on this area, approached via our semi-empirical
532: procedures, at a later date.
533:
534:
535: \appendix
536:
537: \section{Calculation of the bond-order parameter $a_{\mathrm{SiH}}$}
538: \label{app:bond}
539:
540: In this Appendix we describe the procedure employed to estimate the
541: bond parameter $a_{AB}$ in Eq.~(\ref{eq:Loewdin}), when
542: $\mathrm{AB}=\mathrm{SiH}$.
543: According to Eqs.~(\ref{eq:binding}) and (\ref{eq:Loewdin}), the
544: experimental correlation energy of the SiH$_4$ molecule can be
545: written as
546: \begin{equation}
547: E_c^{\mathrm{exp}} (\mathrm{SiH_4} ) = E_c^{\mathrm{exp}}
548: (\mathrm{Si}_{\mbox{\scriptsize in molecule}} )
549: + \sum_{j=1}^4 E_c^{\mathrm{exp}} (\mathrm{H}^{(j)}_{\mbox{\scriptsize in
550: molecule}} )
551: + a_{\mathrm{SiH}} \sum_{i=1}^4 P_{\mathrm{SiH}}^{(i)}
552: + a_{\mathrm{HH}} \sum_{i=1}^6 P_{\mathrm{HH}}^{(i)} ,
553: \label{eq:SiH4}
554: \end{equation}
555: where $P_{\mathrm{SiH}}$ and $P_{\mathrm{HH}}$ and the
556: bond-orders for each Si--H and H--H bonds in SiH$_4$, respectively.
557: From the H$_2$ molecule one immediately obtains the bond parameter for
558: the H--H bond as
559: \begin{equation}
560: a_{\mathrm{HH}} = \frac{E_c^{\mathrm{exp}} (\mathrm{H_2} ) -2E_c^{\mathrm{exp}}
561: (\mathrm{H}_{\mbox{\scriptsize in molecule}} )}{P_{\mathrm{HH}}} .
562: \end{equation}
563: From Eq.~(\ref{eq:Zn}), with $Z=n=1$, one has $E_c^{\mathrm{exp}}
564: (\mathrm{H}_{\mbox{\scriptsize in molecule}} )=0$.
565: Since in the hydrogen molecule $P_{\mathrm{HH}} = 1$, one finds
566: \begin{equation}
567: a_{\mathrm{HH}} = E_c^{\mathrm{exp}} (\mathrm{H_2} ) =
568: 4.28\cdot10^{-2} .
569: \end{equation}
570: In Tab.~\ref{tab:bom} we report the symmetric matrix of the bond-order
571: for SiH$_4$.
572: The diagonal elements are the sum over the off-diagonal elements, and
573: represent the total electrons involved for each atom in the bonds.
574:
575: Using the experimental value $E_c^{\mathrm{exp}} (\mathrm{SiH_4} ) =
576: 2.1742$~a.u., and employing the results of Tables~\ref{tab:exptheo}
577: and \ref{tab:bom}, from Eq.~(\ref{eq:SiH4}) we eventually find
578: \begin{equation}
579: a_{\mathrm{SiH}} = 6.238\cdot 10^{-2} ,
580: \end{equation}
581: as quoted by Tab.~\ref{tab:bond}.
582:
583: \bibliographystyle{molphys}
584: \bibliography{a,b,c,d,e,f,g,h,i,j,k,l,m,n,o,p,q,r,s,t,u,v,w,x,y,z,zzproceedings,Angilella}
585:
586: \newpage
587:
588: \begin{table}[t]
589: \centering
590: \begin{tabular}{|c|c|r@{E}lr@{E}lr@{E}l|}
591: \hline
592: Atom & $S$ & \multicolumn{2}{c}{$E_\HF$} & \multicolumn{2}{c}{$E_S$} &
593: \multicolumn{2}{c|}{$E_c^{\mathrm{exp}}$} \\
594: \hline
595: Si & 3 & $-2.888318$ & $+02$ & $-2.893116$ & $+02$ & $4.798144$ & $-01$ \\
596: H & 2 & $-4.982329$ & $-01$ & $-4.993000$ & $-01$ & $1.067100$ & $-03$ \\
597: F & 2 & $-9.936496$ & $+01$ & $-9.972500$ & $+01$ & $3.600440$ & $-01$ \\
598: Cl & 2 & $-4.594480$ & $+02$ & $-4.600301$ & $+02$ & $5.821385$ & $-01$ \\
599: \hline
600: \end{tabular}
601: \caption{Hartree-Fock energy, $E_\HF$, estimated with the
602: 6-31G$^{\ast\ast}$ basis set, Schr\"odinger energy, $E_S$,
603: and experimental correlation energy, $E_c^{\mathrm{exp}}$, for Si, H, F and
604: Cl atoms.
605: The second column lists the values of the spin multiplicity, $S$.
606: All values are in a.u.}
607: \label{tab:HF}
608: \end{table}
609:
610:
611:
612: \begin{table}[c]
613: \centering
614: \begin{tabular}{|c|c|l@{E}r|}
615: \hline
616: $A$--$B$ & $AB_n$ & \multicolumn{2}{c|}{$a_{AB}$} \\
617: \hline
618: H--H & H$_2$ & 4.287390 & $-02$ \\
619: Si--H & SiH$_4$ & 6.237950 & $-02$ \\
620: F--F & F$_2$ & 1.674987 & $-01$ \\
621: Cl--Cl & Cl$_2$ & 1.314397 & $-01$ \\
622: H--F & HF & 1.137677 & $-01$ \\
623: H--Cl & HCl & 7.700011 & $-02$ \\
624: Si--F & SiF$_4$ & 1.273342 & $-01$ \\
625: Si--Cl & SiCl$_4$ & 1.024566 & $-01$ \\
626: F--Cl & FCl & 1.586152 & $-01$ \\
627: \hline
628: \end{tabular}
629: \caption{L\"owdin bond parameters $a_{AB}$ for $A$--$B$ bonds in
630: several $AB_n$ model molecules.
631: See Appendix~\ref{app:bond} for the derivation of $a_{\mathrm{SiH}}$.
632: For the FCl molecule (last row), the experimental vibrational
633: frequency being unavailable, a theoretical value has been used in
634: the calculation of $E_S$.}
635: \label{tab:bond}
636: \end{table}
637:
638: \begin{table}[c]
639: \centering
640: \begin{small}
641: \begin{tabular}{|c|r@{.}lr@{.}lr@{.}lr@{.}lr@{.}lr@{.}l|}
642: \hline
643: Molecule & \multicolumn{2}{c}{$E_S$} & \multicolumn{2}{c}{$E_\HF$} &
644: \multicolumn{2}{c}{MP2} & \multicolumn{2}{c}{MP3} &
645: \multicolumn{2}{c}{MP4} & \multicolumn{2}{c|}{BOCE} \\
646: \hline
647: SiH$_3$F & $-$391 & 149094 & $-$390 & 152840 & $-$390 & 438375 & $-$390 & 450934 &
648: $-$390 & 463879 & $-$391 & 136150 \\
649: SiH$_2$F$_2$ & $-$490 & 474680 & $-$489 & 084835 & $-$489 & 538902 & $-$489 & 543176 &
650: $-$489 & 563139 & $-$490 & 460715 \\
651: SiHF$_3$ & $-$589 & 799528 & $-$588 & 019851 & $-$588 & 642790 & $-$588 & 639082 &
652: $-$588 & 665926 & $-$589 & 792771 \\
653: SiH$_3$Cl & $-$751 & 377125 & $-$750 & 187745 & $-$750 & 434135 & $-$750 & 461123 &
654: $-$750 & 469475 & $-$751 & 374545 \\
655: SiH$_2$Cl$_2$ & $-$1210 & 934066 & $-$1209 & 143833 & $-$1209 & 520705 & $-$1209 & 553336
656: & $-$1209 & 564921 & $-$1210 & 929943 \\
657: SiHCl$_3$ & $-$1670 & 491322 & $-$1668 & 097366 & $-$1668 & 607261 & $-$1668 & 644964
658: & $-$1668 & 660441 & $-$1670 & 486756 \\
659: SiH$_2$FCl$^{a,b}$ & $-$850 & 697705 & $-$849 & 113982 & $-$849 & 529345 & $-$849 & 547673 &
660: $-$849 & 563687 & $-$850 & 694892 \\
661: SiHF$_2$Cl$^{a,b}$ & $-$950 & 024261 & $-$948 & 045333 & $-$948 & 630412 & $-$948 & 640346 &
662: $-$948 & 663824 & $-$950 & 024303 \\
663: SiHFCl$_2$$^{a,b}$ & $-$1310 & 255828 & $-$1308 & 071123 & $-$1308 & 610000 & $-$1308 & 642327
664: & $-$1308 & 662046 & $-$1310 & 255383 \\
665: SiF$_2$Cl$_2$$^{a,b}$ & $-$1409 & 578850 & $-$1406 & 998802 & $-$1407 & 716386 & $-$1407 & 731725
666: & $-$1407 & 759092 & $-$1409 & 583792 \\
667: SiF$_3$Cl$^a$ & $-$1049 & 355060 & $-$1046 & 974456 & $-$1047 & 728506 & $-$1047 & 730616
668: & $-$1047 & 761047 & $-$1049 & 353836 \\
669: SiFCl$_3^a$ & $-$1769 & 814012 & $-$1767 & 022899 & $-$1767 & 704115 & $-$1767 & 732799
670: & $-$1767 & 756713 & $-$1769 & 813079 \\
671: SiH$_2$ & $-$290 & 561953 & $-$290 & 002631 & $-$290 & 093976 & $-$290 & 111950 &
672: $-$290 & 117816 & $-$290 & 539281 \\
673: SiHF & $-$389 & 887310 & $-$388 & 933412 & $-$389 & 140967 & $-$389 & 205213 &
674: $-$389 & 219847 & $-$389 & 884142 \\
675: SiF$_2$ & $-$489 & 218209 & $-$487 & 884672 & $-$488 & 321054 & $-$488 & 320264 &
676: $-$488 & 343220 & $-$489 & 206212 \\
677: SiHCl$^b$ & $-$750 & 122200 & $-$748 & 9442 & $-$749 & 195368 & $-$749 & 219098 &
678: $-$749 & 228569 & $-$750 & 077830 \\
679: SiCl$_2$ & $-$1209 & 699563 & $-$1207 & 943683 & $-$1208 & 301785 & $-$1208 & 330378 &
680: $-$1208 & 344114 & $-$1209 & 678803 \\
681: \hline\hline
682: SiH$_3^a$ & $-$291 & 173027 & $-$290 & 610579 & $-$290 & 709229 & $-$290 & 726369 &
683: $-$290 & 731504 & $-$291 & 175489 \\
684: SiH$_2$F$^a$ & $-$390 & 493513 & $-$389 & 525900 & $-$389 & 791422 & $-$389 & 800054 &
685: $-$389 & 812671 & $-$390 & 480470 \\
686: SiHF$_2^a$ & $-$489 & 813614 & $-$488 & 452373 & $-$488 & 886854 & $-$488 & 886530 &
687: $-$488 & 906703 & $-$489 & 800833 \\
688: SiH$_2$Cl$^{a,b}$ & $-$750 & 723808 & $-$749 & 566110 & $-$749 & 792607 & $-$749 & 815941 &
689: $-$749 & 823949 & $-$750 & 688190 \\
690: SiHCl$_2^{a,b}$ & $-$1210 & 279079 & $-$1208 & 521554 & $-$1208 & 878868 & $-$1208 & 907693
691: & $-$1208 & 919212 & $-$1210 & 281674 \\
692: SiF$_3$ & $-$589 & 132214 & $-$587 & 381225 & $-$587 & 984340 & $-$587 & 975648 &
693: $-$588 & 002862 & $-$589 & 126715 \\
694: SiCl$_3^a$ & $-$1669 & 840418 & $-$1667 & 474563 & $-$1667 & 964920
695: & $-$1667 & 998734 & $-$1668 & 014351 & $-$1669 & 838233 \\
696: \hline
697: \end{tabular}
698: \end{small}
699: \caption{Schr\"odinger and theoretical energies (in a.u.) for open and
700: closed shell of SiX$_n$Y$_m$ compounds.
701: The upper table refers to relevant energies of closed-shell silicon
702: compounds: especially the bond-order correlation energy.
703: The lower table refers to the energies of open-shell silicon
704: compounds.
705: {\sl \underline{Notes}:}
706: (a) The Schr\"odinger energies have been calculated using theoretical
707: values of the vibrational frequencies.
708: (b) The Schr\"odinger energies have been calculated using theoretical
709: values of the molecular heat of formation.
710: }
711: \label{tab:theo}
712: \end{table}
713:
714:
715:
716: \begin{table}[t]
717: \centering
718: \begin{small}
719: \begin{tabular}{|c|r@{.}lr@{.}lr@{.}lr@{.}lr@{.}l|}
720: \hline
721: Molecule & \multicolumn{2}{c}{$E_c^{\mathrm{exp}}$} &
722: \multicolumn{2}{c}{MP2} &
723: \multicolumn{2}{c}{MP3} &
724: \multicolumn{2}{c}{MP4} &
725: \multicolumn{2}{c|}{BOCE} \\
726: \hline
727: SiH$_3$F & 0 & 996255 & 0 & 285535 & 0 & 298094 & 0 & 310939 & 0 & 983310 \\
728: SiH$_2$F$_2$ & 1 & 389846 & 0 & 454068 & 0 & 458341 & 0 & 478385 & 1 & 375880 \\
729: SiHF$_3$ & 1 & 779678 & 0 & 622939 & 0 & 619231 & 0 & 646075 & 1 & 772920 \\
730: SiH$_3$Cl & 1 & 189380 & 0 & 246389 & 0 & 273378 & 0 & 281730 & 1 & 186800 \\
731: SiH$_2$Cl$_2$ & 1 & 790233 & 0 & 376872 & 0 & 409502 & 0 & 421088 & 1 & 786110 \\
732: SiHCl$_3$ & 2 & 393956 & 0 & 509894 & 0 & 547597 & 0 & 563074 & 2 & 389390 \\
733: SiH$_2$FCl$^{a,b}$ & 1 & 583723 & 0 & 415363 & 0 & 433691 & 0 & 449706 & 1 & 580910 \\
734: SiHF$_2$Cl$^{a,b}$ & 1 & 978928 & 0 & 585079 & 0 & 595013 & 0 & 618492 & 1 & 978970 \\
735: SiHFCl$_2^{a,b}$ & 2 & 184705 & 0 & 538877 & 0 & 571204 & 0 & 590923 & 2 & 184260 \\
736: SiF$_2$Cl$_2^{a,b}$ & 2 & 580048 & 0 & 717584 & 0 & 732924 & 0 & 760291 & 2 & 584990 \\
737: SiF$_3$Cl$^a$ & 2 & 380604 & 0 & 754049 & 0 & 756160 & 0 & 786592 & 2 & 379380 \\
738: SiFCl$_3^a$ & 2 & 791113 & 0 & 681216 & 0 & 709900 & 0 & 733814 & 2 & 790180 \\
739: SiH$_2$ & 0 & 559322 & 0 & 091345 & 0 & 109319 & 0 & 115185 & 0 & 536650 \\
740: SiHF & 0 & 953899 & 0 & 207555 & 0 & 271801 & 0 & 286436 & 0 & 950730 \\
741: SiF$_2$ & 1 & 333537 & 0 & 436382 & 0 & 435591 & 0 & 458548 & 1 & 321540 \\
742: SiHCl$^b$ & 1 & 177958 & 0 & 251168 & 0 & 274897 & 0 & 284368 & 1 & 133630 \\
743: SiCl$_2$ & 1 & 755880 & 0 & 358102 & 0 & 386695 & 0 & 400432 & 1 & 735120 \\
744: \hline
745: \hline
746: SiH$_3^a$ & 0 & 562448 & 0 & 098650 & 0 & 115791 & 0 & 120925 & 0 & 564910 \\
747: SiH$_2$F$^a$ & 0 & 967613 & 0 & 265521 & 0 & 274154 & 0 & 286771 & 0 & 954570 \\
748: SiHF$_2^a$ & 1 & 361241 & 0 & 434481 & 0 & 434157 & 0 & 454329 & 1 & 348460 \\
749: SiH$_2$Cl$^{a,b}$ & 1 & 157698 & 0 & 226497 & 0 & 249830 & 0 & 257838 & 1 & 122080 \\
750: SiHCl$_2^{a,b}$ & 1 & 757525 & 0 & 357315 & 0 & 386139 & 0 & 397659 & 1 & 760120 \\
751: SiF$_3$ & 1 & 750989 & 0 & 603115 & 0 & 594423 & 0 & 621637 & 1 & 745490 \\
752: SiCl$_3^a$ & 2 & 365855 & 0 & 490357 & 0 & 524171 & 0 & 539788 & 2 & 363670 \\
753: \hline
754: \end{tabular}
755: \end{small}
756: \caption{Experimental and theoretical correlation energies (in a.u.)
757: for open and closed shell of SiX$_n$Y$_m$ compounds.
758: The upper table refers to closed-shell, whereas the lower
759: table refers to open-shell compounds.
760: {\sl \underline{Notes}:}
761: (a) The Schr\"odinger energies have been calculated using theoretical
762: values of the vibrational frequencies.
763: (b) The Schr\"odinger energies have been calculated using theoretical
764: values of the molecular heat of formation.}
765: \label{tab:exptheo}
766: \end{table}
767:
768:
769:
770:
771: \begin{table}[t]
772: \centering
773: \begin{small}
774: \begin{tabular}{|c|r@{.}l|r@{.}lr@{.}l|r@{.}lr@{.}l|r@{.}lr@{.}l|r@{.}lr@{.}l|r@{.}lr@{.}l|}
775: \hline
776: Molecule & \multicolumn{2}{c|}{IP$^{\mathrm{exp}}$} &
777: \multicolumn{2}{c}{HF} &
778: \multicolumn{2}{c|}{$\Delta$} &
779: \multicolumn{2}{c}{MP2} &
780: \multicolumn{2}{c|}{$\Delta$} &
781: \multicolumn{2}{c}{MP3} &
782: \multicolumn{2}{c|}{$\Delta$} &
783: \multicolumn{2}{c}{MP4} &
784: \multicolumn{2}{c|}{$\Delta$} &
785: \multicolumn{2}{c}{BOCE} &
786: \multicolumn{2}{c|}{$\Delta$} \\
787: \hline
788: SiH$_4$ & 11 & 60 & 12 & 38 & $-$0 & 78 & 10 & 70 & 0 & 90 & 10 & 75 & 0 & 85 & 10 & 78 &
789: 0 & 82 & 11 & 16 & 0 & 44 \\
790: SiH$_3$F & 11 & 70 & 12 & 20 & $-$0 & 5 & 11 & 52 & 0 & 18 & 11 & 67 & 0 & 03 & 11 & 99 &
791: $-$0 & 29 & 11 & 49 & 0 & 21 \\
792: SiH$_3$Cl & 11 & 40 & 10 & 37 & 1 & 03 & 11 & 40 & 0 & 00 & 11 & 08 & 0 & 32 & 11 & 08 &
793: 0 & 32 & 11 & 53 & $-$0 & 13 \\
794: SiH$_2$F$_2$ & 12 & 20 & 11 & 77 & 0 & 43 & 12 & 07 & 0 & 13 & 12 & 24 & $-$0 & 04 & 12 & 11
795: & 0 & 09 & 12 & 45 & $-$0 & 25 \\
796: SiH$_2$Cl$_2$ & 11 & 70 & 11 & 36 & 0 & 34 & 11 & 52 & 0 & 18 & 11 & 56 & 0 & 14 & 11 & 43 &
797: 0 & 27 & 11 & 86 & $-$0 & 16 \\
798: SiHF$_3$ & 14 & 00 & 12 & 43 & 1 & 57 & 12 & 78 & 1 & 22 & 13 & 02 & 0 & 98 & 12 & 88 &
799: 1 & 12 & 13 & 51 & 0 & 49 \\
800: SiHCl$_3$ & 11 & 40 & 12 & 04 & $-$0 & 64 & 11 & 67 & $-$0 & 27 & 11 & 68 & $-$0 & 28 & 11 & 56
801: & $-$0 & 16 & 12 & 02 & $-$0 & 62 \\
802: SiF$_4$ & 15 & 70 & 17 & 82 & $-$2 & 12 & 15 & 12 & 0 & 58 & 14 & 93 & 0 & 77 & 15 & 35 &
803: 0 & 35 & 16 & 54 & $-$0 & 84 \\
804: SiCl$_4$ & 11 & 80 & 12 & 53 & $-$0 & 73 & 11 & 48 & 0 & 32 & 11 & 44 & 0 & 36 & 11 & 34 &
805: 0 & 46 & 11 & 85 & $-$0 & 05 \\
806: \hline
807: \end{tabular}
808: \end{small}
809: \caption{Experimental and theoretical ionization potentials (in eV).
810: The $\Delta$ columns report the differences between the experimental
811: and the theoretical values.}
812: \label{tab:delta}
813: \end{table}
814:
815:
816:
817:
818: \begin{table}[t]
819: \centering
820: \begin{tabular}{|c|ccccc|}
821: \cline{2-6}
822: \multicolumn{1}{c|}{} & Si & H & H & H & H \\
823: \hline
824: Si & 3.970845 & & & & \\
825: H & 0.993248 & 1.015563 & & & \\
826: H & 0.993248 & 0.006901 & 1.015563 & & \\
827: H & 0.991655 & 0.007873 & 0.007873 & 1.014972 & \\
828: H & 0.992695 & 0.007541 & 0.007541 & 0.007571 & 1.015348 \\
829: \hline
830: \end{tabular}
831: \caption{Bond-order matrix of SiH$_4$.}
832: \label{tab:bom}
833: \end{table}
834:
835: \newpage
836:
837: \begin{figure}[t]
838: \centering
839: \includegraphics[width=0.5\columnwidth]{fig1.eps}
840: \caption{Schematic determination of the experimental and theoretical
841: correlation energies for both neutral and ionic molecules.}
842: \label{fig:schema}
843: \end{figure}
844:
845: \begin{figure}[t]
846: \centering
847: \includegraphics[height=0.9\columnwidth,angle=-90]{fig2.ps}
848: \caption{Correlation energies $E_c$ (in a.u.) of closed and
849: open shell silicon compounds, for the 24 molecules in
850: Tab.~\ref{tab:exptheo}.
851: The abscissa is the row index in Tab.~\ref{tab:exptheo}.
852: The vertical dashed line separates closed from open shell molecules.
853: In all insets, open squares refer to experimental values, while open
854: circles refer to calculated values.}
855: \label{fig:exptheo}
856: \end{figure}
857:
858: \begin{figure}[t]
859: \centering
860: \includegraphics[height=0.9\columnwidth,angle=-90]{fig3.ps}
861: \caption{Absolute percentage errors, $100|(E_c^{\mathrm{exp}} -
862: E_c^{\mathrm{theo}})/E_c^{\mathrm{exp}} |$, of the theoretical correlation
863: energies with respect to the experimental correlation
864: energy in Tab.~\ref{tab:exptheo} and Fig.~\ref{fig:exptheo}.
865: The abscissa is the row index in Tab.~\ref{tab:exptheo}.
866: The vertical dashed line separates closed from open shell molecules.
867: }
868: \label{fig:percentage}
869: \end{figure}
870:
871: \begin{figure}[t]
872: \centering
873: \includegraphics[height=0.9\columnwidth,angle=-90]{fig4.ps}
874: \caption{Ionization potentials IP (in eV) for the 9 Si-containing
875: molecules in Tab.~\ref{tab:delta}.
876: The abscissa is the row index in Tab.~\ref{tab:delta}.
877: Open squares refer to experimental values, while open circles refer to
878: calculated values.
879: }
880: \label{fig:delta}
881: \end{figure}
882:
883: \begin{figure}[t]
884: \centering
885: \includegraphics[height=0.9\columnwidth,angle=-90]{fig5.ps}
886: \caption{Absolute percentage errors,
887: $100|(\mathrm{IP}^{\mathrm{exp}} -
888: \mathrm{IP}^{\mathrm{theo}})/\mathrm{IP}^{\mathrm{exp}} |$, of the
889: theoretical ionization potentials with respect to the experimental
890: values in Tab.~\ref{tab:delta} and Fig.~\ref{fig:delta}.
891: The abscissa is the row index in Tab.~\ref{tab:delta}.
892: }
893: \label{fig:ip}
894: \end{figure}
895:
896:
897:
898:
899:
900: \end{document}
901:
902:
903:
904:
905:
906:
907:
908:
909:
910:
911:
912:
913:
914:
915:
916:
917:
918:
919:
920:
921:
922:
923:
924:
925:
926:
927:
928:
929:
930:
931:
932:
933:
934: \begin{table}[t]
935: \centering
936: \begin{small}
937: \begin{tabular}{|c|r@{.}lr@{.}lr@{.}lr@{.}l|}
938: \hline
939: Molecule & \multicolumn{2}{c}{MP2} &
940: \multicolumn{2}{c}{MP3} &
941: \multicolumn{2}{c}{MP4} &
942: \multicolumn{2}{c|}{BOCE} \\
943: \hline
944: SiH$_3$F & 71 & 339182 & 70 & 078550 & 68 & 789211 & 1 & 299320 \\
945: SiH$_2$F$_2$ & 67 & 329636 & 67 & 022177 & 65 & 580010 & 1 & 004826 \\
946: SiHF$_3$ & 64 & 997082 & 65 & 205468 & 63 & 697105 & 0 & 379711 \\
947: SiH$_3$Cl & 79 & 284226 & 77 & 015111 & 76 & 312855 & 0 & 216931 \\
948: SiH$_2$Cl$_2$ & 78 & 948432 & 77 & 125748 & 76 & 478586 & 0 & 230326 \\
949: SiHCl$_3$ & 78 & 700760 & 77 & 125831 & 76 & 479361 & 0 & 190722 \\
950: SiH$_2$FCl$^{a,b}$ & 73 & 773022 & 72 & 615705 & 71 & 604518 & 0 & 177638 \\
951: SiHF$_2$Cl$^{a,b}$ & 70 & 434575 & 69 & 932575 & 68 & 746135 & $-$0 & 002119 \\
952: SiHFCl$_2$$^{a,b}$ & 75 & 334108 & 73 & 854407 & 72 & 951804 & 0 & 020372 \\
953: SiF$_2$Cl$_2$$^{a,b}$ & 72 & 187189 & 71 & 592639 & 70 & 531918 & $-$0 & 191527 \\
954: SiF$_3$Cl$^a$ & 68 & 325294 & 68 & 236635 & 66 & 958321 & 0 & 051425 \\
955: SiFCl$_3$$^a$ & 75 & 593390 & 74 & 565699 & 73 & 708916 & 0 & 033432 \\
956: SiH$_2$ & 83 & 668575 & 80 & 455132 & 79 & 406381 & 4 & 053528 \\
957: SiHF & 78 & 241375 & 71 & 506245 & 69 & 972110 & 0 & 332164 \\
958: SiF$_2$ & 67 & 276343 & 67 & 335636 & 65 & 614146 & 0 & 900000 \\
959: SiHCl$^b$ & 78 & 677680 & 76 & 663228 & 75 & 859226 & 3 & 763130 \\
960: SiCl$_2$ & 79 & 605556 & 77 & 977114 & 77 & 194817 & 1 & 182325 \\
961: \hline
962: {\sl Average} & 74 & 33626 & 72 & 841641 & 71 & 757966 & 0 & 802483 \\
963: \hline
964: \hline
965: SiH$_3$$^a$ & 82 & 460641 & 79 & 413125 & 78 & 500259 & $-$0 & 437701 \\
966: SiH$_2$F$^a$ & 72 & 559115 & 71 & 666970 & 70 & 363009 & 1 & 347928 \\
967: SiHF$_2$$^a$ & 68 & 082023 & 68 & 105766 & 66 & 623875 & 0 & 938924 \\
968: SiH$_2$Cl$^{a,b}$ & 80 & 435608 & 78 & 420073 & 77 & 728364 & 3 & 076591 \\
969: SiHCl$_2$$^{a,b}$ & 79 & 669428 & 78 & 029366 & 77 & 373938 & $-$0 & 147649 \\
970: SiF$_3$ & 65 & 555751 & 66 & 052168 & 64 & 497954 & 0 & 314049 \\
971: SiCl$_3$$^a$ & 79 & 273597 & 77 & 844342 & 77 & 184247 & 0 & 092366 \\
972: \hline
973: {\sl Average} & 75 & 433738 & 74 & 21883 & 73 & 181664 & 0 & 740644 \\
974: \hline
975: \end{tabular}
976: \end{small}
977: \caption{Absolute percentage error of the theoretical correlation
978: energies (in a.u.) with respect to the experimental correlation
979: energy for open (upper table) and closed shell
980: (lower table) of SiX$_n$Y$_m$ compounds.
981: {\sl \underline{Notes}:}
982: (a) The Schr\"odinger energies have been calculated using theoretical
983: values of the vibrational frequencies.
984: (b) The Schr\"odinger energies have been calculated using theoretical
985: values of the molecular heat of formation.}
986: \label{tab:percentage}
987: \end{table}
988:
989: \begin{table}[t]
990: \centering
991: \begin{small}
992: \begin{tabular}{|c|r@{.}lr@{.}lr@{.}lr@{.}lr@{.}l|}
993: \hline
994: Molecule & \multicolumn{2}{c}{HF} &
995: \multicolumn{2}{c}{MP2} &
996: \multicolumn{2}{c}{MP3} &
997: \multicolumn{2}{c}{MP4} &
998: \multicolumn{2}{c|}{BOCE} \\
999: \hline
1000: SiH$_4$ & 6 & 72 & 7 & 76 & 7 & 33 & 7 & 07 & 3 & 79 \\
1001: SiH$_3$F & 4 & 27 & 1 & 54 & 0 & 26 & 2 & 48 & 1 & 79 \\
1002: SiH$_3$Cl & 9 & 03 & 0 & 00 & 2 & 81 & 2 & 81 & 1 & 14 \\
1003: SiH$_2$F$_2$ & 3 & 52 & 1 & 06 & 0 & 33 & 0 & 74 & 2 & 05 \\
1004: SiH$_2$Cl$_2$ & 2 & 90 & 1 & 54 & 1 & 20 & 2 & 31 & 1 & 37 \\
1005: SiHF$_3$ & 11 & 21 & 8 & 71 & 7 & 00 & 8 & 00 & 3 & 50 \\
1006: SiHCl$_3$ & 5 & 61 & 2 & 37 & 2 & 46 & 1 & 40 & 5 & 44 \\
1007: SiF$_4$ & 13 & 50 & 3 & 69 & 4 & 90 & 2 & 23 & 5 & 35 \\
1008: SiCl$_4$ & 6 & 17 & 2 & 71 & 3 & 05 & 3 & 90 & 0 & 42 \\
1009: \hline
1010: {\sl Average} & 6 & 99 & 3 & 26 & 3 & 26 & 3 & 44 & 2 & 76 \\
1011: \hline
1012: \end{tabular}
1013: \end{small}
1014: \caption{Absolute percentage error of the theoretical ionization
1015: potentials IP$_{\Delta SCF}$ (in eV) with respect to the
1016: experimental value.}
1017: \label{tab:ip}
1018: \end{table}
1019: