1: \documentclass[prb,twocolumn,showpacs,superscriptaddress]{revtex4}
2: \usepackage{graphicx,dcolumn,amsmath}
3:
4: \begin{document}
5:
6: \title{\textbf{ Optical properties of structurally-relaxed Si/SiO$_2$
7: superlattices: the role of bonding at interfaces. }}
8:
9: \author{\firstname{Pierre} \surname{Carrier}} \affiliation{D\'{e}partement
10: de Physique et Groupe de Recherche en Physique \\et Technologie des
11: Couches Minces (GCM), Universit\'{e} de Montr\'{e}al, \\ Case Postale
12: 6128, Succursale~Centre-Ville, Montr\'{e}al, Qu\'{e}bec, Canada H3C 3J7}
13:
14: \author{\firstname{Laurent J.} \surname{Lewis}} \email[Author to whom
15: correspondence should be addressed: Email address:\ ]
16: {laurent.lewis@umontreal.ca} \affiliation{D\'{e}partement de Physique et
17: Groupe de Recherche en Physique \\et Technologie des Couches Minces (GCM),
18: Universit\'{e} de Montr\'{e}al, \\ Case Postale 6128,
19: Succursale~Centre-Ville, Montr\'{e}al, Qu\'{e}bec, Canada H3C 3J7}
20:
21: \author{\firstname{M. W. C.} \surname{Dharma-wardana}}
22: \affiliation{Institute for Microstructural Sciences, National Research
23: Council, Ottawa, Canada K1A 0R6}
24:
25: \date{\today}
26:
27: \begin{abstract}
28:
29: We have constructed microscopic, structurally-relaxed atomistic models of
30: Si/SiO$_2$ superlattices. The structural distortion and oxidation-state
31: characteristics of the interface Si atoms are examined in detail. The role
32: played by the interface Si suboxides in raising the band gap and producing
33: dispersionless energy bands is established. The suboxide atoms are shown
34: to generate an abrupt interface layer about 1.60 \AA\ thick. Bandstructure
35: and optical-absorption calculations at the Fermi Golden rule level are
36: used to demonstrate that increasing confinement leads to (a) direct
37: bandgaps, (b) a blue shift in the spectrum, and (c) an enhancement of the
38: absorption intensity in the threshold-energy region. Some aspects of this
39: behaviour appear not only in the symmetry direction associated with the
40: superlattice axis, but also in the orthogonal plane directions. We
41: conclude that, in contrast to Si/Ge, Si/SiO$_2$ superlattices show clear
42: optical enhancement and a shift of the optical spectrum into the region
43: useful for many opto-electronic applications.
44:
45: \end{abstract}
46:
47: \pacs{78.66.Jg, 68.65.+g, 71.23.Cq}
48:
49: \maketitle
50:
51: \section{Introduction}\label{Intro}
52:
53: The initial interest in light-emitting Si-based nanostructures has lead to
54: a number of important experiments establishing that Si/SiO$_2$
55: superlattices (SLs) show enhanced, blue-shifted luminescence.
56: \cite{LuLockBar,Kanemitsu,Novikov,Khriachtchev,Mulloni,Kanemitsu_bis}
57: While the luminescence pattern is more complex in these systems than in
58: others, the blue shift was found to correlate with decreasing Si-layer
59: thickness. This simple relation between the silicon-layer thickness and
60: the luminescence peak is of great interest for applications such as
61: Si-based light-emitting diodes (\mbox{Si-LEDs}). All reported SL energy
62: peaks are in the lower part of the visible spectrum --- the highest
63: reported value being \mbox{2.3 eV} (540 nm), i.e., green,\cite{LuLockBar}
64: and the lowest \mbox{1.2 eV} (1030 nm), in the near
65: infrared.\cite{Novikov} These SLs are really multiple Si quantum wells
66: (MQW), the silicon oxide layers playing the role of barriers. The
67: thickness of the Si quantum wells, $L_{\rm Si}$, is the critical
68: parameter. A fixed $L_{\rm Si}$ would simply set the colour of the
69: \mbox{Si-LED}, while MQWs with a range of $L_{\rm Si}$ would span a range
70: of colours in the luminescence.
71:
72: Other systems containing confined silicon structures (besides MQWs) are,
73: e.g., porous Si\cite{Canham,LehmanGosele} (consisting of quasi
74: one-dimensional structures), silicon nanoclusters in SiO$_2$
75: matrices,\cite{Tsybeskov} nanocrystals\cite{Cheylan} or dislocation loops,
76: and quantum-dot structures made from implantation of boron\cite{Ng} or
77: other ions.\cite{ericson} The choice of a particular structure, e.g., for
78: \mbox{Si-LED} applications, depends on many factors: stability over time,
79: optical efficiency at room temperature, experimental reproductiveness,
80: facility to accept \mbox{n-} or \mbox{p-} type dopants (e.g., for n-p
81: junctions), ease of incorporation in ultra-large-scale-integration
82: technology, and production costs. \mbox{Si/SiO$_2$} SLs are stable
83: structures, as opposed to porous-Si. In addition, the silicon layer
84: thicknesses in SLs are directly related to the energy peaks in the
85: Si/SiO$_2$ luminescence spectra. This is not straighforward in porous
86: silicon, Si clusters or nanocrystals, where pore dimensions as well as
87: hydrogen concentrations play an uncontrolled role on the energy
88: shift.\cite{Tsybeskov,porous}
89:
90: Silicon-based structures have many advantages over structures made of
91: other semiconductors: low-cost (as compared to any of the \mbox{III-V's}
92: or \mbox{II-VI's}), non-toxicity, practically unlimited availability (in
93: contrast to germanium) and benefits from decades of experience in
94: purification, growth, and device fabrication. However, the indirect energy
95: gap (\mbox{$\sim$1.1 eV}) in bulk crystalline silicon (\mbox{c-Si}) makes
96: it unsuitable for optoelectronic applications. Silica (SiO$_2$) is another
97: key material of the microelectronics industry; it has a bandgap of
98: \mbox{$\sim$9 eV}. Optical-fiber technologies and
99: metal-oxide-semiconductor field-effect transistors (MOSFETs) are based on
100: (high-quality) silica. Molecular-beam epitaxy and chemical vapor
101: deposition provide the needed growth technology for Si and SiO$_2$. It is
102: possible to combine crystalline
103: silicon\cite{Kanemitsu,Luprivate,Zacharias,ZachariasAPL} and SiO$_2$ to
104: produce structured materials having chemically pure, sharp, defect-free
105: interfaces. The degree of advancement of the fabrication technology is
106: such that the enhanced luminescence in \mbox{Si/SiO$_2$} SLs cannot be
107: explained only in terms of defects\cite{Kanemitsu_bis} and/or residual
108: hydrogen atoms filling unsaturated Si dangling bonds --- the latter being
109: referred to as $P_b$-type centers\cite{Stirling} --- located at the SL
110: interfaces, as in the well-understood enhanced luminescence of
111: hydrogenated {\em a-}Si.
112:
113: The objective of the present article is to consider the detailed
114: microscopic structure of the \mbox{SiO$_2$/Si/SiO$_2$} double interface
115: structure and provide a first-principles understanding of the emission of
116: light from \mbox{Si/SiO$_2$} SLs. Only a few atomistic DFT calculations on
117: \mbox{Si/SiO$_2$} SLs have been
118: reported.\cite{Delley,Kageshima,Pukkinen,Degoli,Agrawal,Carrier} Many more
119: would have been available were it not of the (naturally-occuring)
120: amorphous structure of SiO$_2$: amorphous structures require large
121: supercells to get physically relevant results. Another important issue is
122: the need to model the Si/SiO$_2$ interface so as to correctly incorporate
123: the known experimental details. Recent core-level shift
124: measurements\cite{Fuggle} provide details of the suboxide (partially
125: oxidized) Si atoms.\cite{SiegerLuHimpsel} Further, the abruptness of the
126: Si/SiO$_2$ interface has been established from transmission electronic
127: microscopy (TEM) experiments; values of the interface width as low as 5
128: \AA\ have been reported.\cite{Lockwood,Kanemitsu} Based on these
129: observations, realistic Si/SiO$_2$ interface models have been designed by
130: several
131: workers.\cite{Pasquarello,Neaton,Tu,NgInterface,Stirling,KageshimaInterface}
132: Suboxide Si atoms in most of these models are distributed within three
133: atomic layers of the Si/SiO$_2$ interface, corresponding to the lowest
134: experimental interface thickness. Such interface models are needed for
135: first-principles modelling of MOSFETs, and for SL structures, where
136: \textit{multiple} Si/SiO$_2$ interfaces are present.
137:
138: An early (and tractable) Si/SiO$_2$ interface model was proposed by Herman
139: and Batra.\cite{HermanBatra} The large lattice mismatch between SiO$_2$
140: and Si was accommodated by setting the \mbox{$\beta$-cristobalite} SiO$_2$
141: unit cell diagonally on the diamond-like \mbox{c-Si} unit cell. An oxygen
142: atom was included in the interface to saturate the dangling bonds,
143: resulting in a crystalline model of the interface. Another simple
144: crystalline model, involving a bridge oxygen at the interface, was
145: introduced by Tit and Dharma-wardana.\cite{TitDharmaDBMBOM} These models
146: were studied using a variety of methods --- tight-binding
147: (TB),\cite{HermanBatra,TitDharmaDBMBOM} full-potential,
148: linear-muffin-tin-orbitals (FP-LMTO)\cite{Pukkinen,Degoli} and
149: linearized-augmented-plane-waves (FP-LAPW).\cite{Carrier} The
150: dispersionless character of the bandstructure in the growth axis has been
151: confirmed within all three theoretical approaches, thus demonstrating the
152: existence of strong confinement. However, the nature of the energy gap ---
153: from both LMTO and LAPW calculations --- is still indirect. Furthermore,
154: these crystalline models show that the light absorption is quite dependent
155: on the details of bonding and interface structure (bondlengths, angles and
156: chemical species), emphasizing the need for \emph{more realistic},
157: \emph{structurally-relaxed} models.\cite{Carrier}
158:
159: Several structurally-relaxed models have been investigated by Kageshima
160: and Shiraishi using first-principles methods. Their models were
161: constructed starting from $\beta$-cristobalite as well as $\alpha$-quartz
162: SiO$_2$ layers superposed onto c-Si layers with different possibilities
163: for the Si/SiO$_2$ interface (such as hydrogen atoms at dangling
164: bonds);\cite{Kageshima} the atomic sites were then structurally relaxed.
165: The calculations indicate that the energy gaps are indeed direct, and that
166: interfacial Si--OH bonds are possible candidates for the light-emitting
167: enhancement in SLs. However, the models are not consistent with the
168: observed suboxide Si atomic distributions at the Si(001)/SiO$_2$
169: interfaces. For instance, no Si atoms bonded to three oxygens are present.
170: Tit and Dharma-wardana\cite{TitDharma} have constructed a
171: partially-relaxed model (PRM) starting from a structurally relaxed
172: Si(001)/SiO$_2$ interface structure due to Pasquarello, Hybertsen, and Car
173: (PHC).\cite{Pasquarello} This interface model contains all three suboxide
174: Si atomic species. Hydrogen atoms were used by PHC to terminate the
175: surface. In the Tit and Dharma-wardana model, the H atoms were removed and
176: the Si/SiO$_2$ interface structure was converted into a SiO$_2$/Si/SiO$_2$
177: double interface SL structure while preserving local tetrahedral bonding
178: to obtain the PRM structure. Within the TB approach, the energy gap of the
179: PRM was shown to be direct and enhancement of the optical absorption (as
180: compared to c-Si) was confirmed.\cite{tran} This is further described in
181: the next section, since the fully relaxed models (FRMs) discussed in the
182: present work are based on this PRM.
183:
184: The results presented here go beyond the TB approach and were obtained
185: within the projector-augmented wave (PAW) theoretical framework. A brief
186: description of the theory is given in section \ref{CompDet}. The SL models
187: were all structurally relaxed, and contain no hydrogen atoms at the
188: interfaces. The interface suboxide Si atoms observed in
189: experiment\cite{SiegerLuHimpsel} arise naturally in the model.
190: Calculations were carried out for different Si-layer thicknesses, in order
191: to assess the effect of confinement on the electronic and optical
192: properties. Supplementary models have been constructed to clarify the role
193: of the suboxide atoms on the SL optical properties. We first review the
194: theoretical methods (section \ref{CompDet}), then focus on the four
195: central issues needed to understand the Si/SiO$_2$ luminescence
196: properties: construction of realistic interface models
197: (section~\ref{Ionic}), quantum confinement (section~\ref{confinement}),
198: role of the suboxide atoms (section~\ref{interface}), and optical effects
199: associated with increased confinement (section~\ref{blueshift}). Our study
200: thus provides a complete, microscopic picture of the luminescence
201: properties of Si/SiO$_2$ SLs.
202:
203: \section{Computational details}\label{CompDet}
204:
205: The electronic-structure calculations were carried out using the Vienna
206: \textit{ab initio} simulation package ({\sf VASP})\cite{VASPref} using the
207: ``frozen-core'' PAW approach.\cite{KresseJoubert} The overall framework is
208: density-functional theory\cite{HohenbergKohn,KohnSham} (DFT) within the
209: local-density approximation (LDA).\cite{Payne}
210:
211: The ``frozen-core'' PAW is a simpler form of the general PAW method
212: introduced by Bl\"ochl.\cite{BlochlPAW} Bl\"ochl's method is an extension
213: of the usual LAPW\cite{Singh} approach. Hence, the PAW method formally
214: bridges the LAPW to the ultrasoft-pseudopotentials (US-PP) in order to
215: combine the precision of the former and the rapidity (for larger systems)
216: of the latter. The PAW method has another advantage over the usual
217: implementation of US-PP, essential for optical calculations: it avoids
218: correcting for spatial nonlocality effects in typical pseudopotentials
219: when evaluating the momentum operator \textbf{p}. Further details can be
220: found in the article of Adolph \emph{et al}.\cite{Adolph}
221:
222: Matrix elements of the momentum operator \textbf{p} are needed for
223: calculating interband optical effects. We describe the main steps that
224: lead to the calculation of the absorption coefficient, starting from PAW
225: solutions of the Kohn-Sham equations. The PAW approach rests on the
226: following linear transformation:
227: $$
228: \vert{\Psi}_N\rangle = \vert\tilde{\Psi}_N\rangle +
229: \sum_i \left(\vert{\phi}_N\rangle-\vert\tilde{\phi}_N\rangle\right)
230: \langle\tilde{p}_i\vert\tilde{\Psi}_N\rangle.
231: $$
232: This relates the (calculated) pseudo wave function
233: $\vert\tilde{\Psi}_N\rangle$ to the (now corrected) all-electrons wave
234: function $\vert{\Psi}_N\rangle$. The index specifies the atomic site,
235: angular momentum numbers and reference energy. The two functions
236: $\vert\tilde{\phi}_N\rangle$ and $\vert{\phi}_N\rangle$ are respectively
237: the pseudo, and all-electron wave function of a reference atom. They are
238: forced to overlap outside a given core region. The functions
239: $\vert\tilde{p}_i\rangle$ are the \textit{projector} functions
240: characteristic of the PAW method. Thus, the three functions
241: $\vert{\phi}_N\rangle$, $\vert\tilde{\phi}_N\rangle$ and
242: $\vert\tilde{p}_i\rangle$ constitute the frozen-core PAW data, being set
243: prior to self-consistent field calculations. The projector functions
244: $\vert\tilde{p}_i\rangle$ are constructed so as to remain dual to the
245: pseudo wave functions, to fulfill generalized orthogonality constraints
246: and to remain (approximately) complete (see Ref.~\onlinecite{Adolph} for
247: full definitions).
248:
249: The application of the above linear transformation to any operator $A$
250: within the PAW approach has been described by Bl\"ochl [cf.\ eq.\ (11) of
251: Ref.~\onlinecite{BlochlPAW}]. If $A$ is the momentum operator \textbf{p}
252: (in a certain direction defined by the polarization vector), we have
253: \begin{eqnarray*}
254: \lefteqn{
255: \langle{\Psi}_N\vert\mbox{{\textbf{p}}}\vert{\Psi}_M\rangle =
256: \langle\tilde{\Psi}_N\vert
257: \mbox{{\textbf{p}}}\vert\tilde{\Psi}_M\rangle
258: + } \\ & & \sum_{i,j} \langle
259: \tilde{\Psi}_N\vert\tilde{p}_i\rangle\left(
260: \langle{\phi}_i\vert\mbox{{\textbf{p}}}\vert{\phi}_j\rangle -
261: \langle\tilde{\phi}_i\vert
262: \mbox{{\textbf{p}}}\vert\tilde{\phi}_j\rangle
263: \right) \langle \tilde{p}_j\vert\tilde{\Psi}_M\rangle.
264: \end{eqnarray*}
265: The imaginary part of the dielectric function, $\epsilon_i$, can be
266: determined by using the Fermi Golden rule within the Coulomb gauge; the
267: expression becomes:\cite{Cardona}
268: \begin{eqnarray*}
269: \displaystyle
270: \epsilon_i(E) = & \frac{\kappa^2}{E^2}
271: \displaystyle
272: \sum_{M,N} \int_{BZ} \frac{2d^3\vec{k}}{(2\pi)^3}
273: \displaystyle
274: \vert\langle{\Psi}_N\vert
275: \mbox{{\textbf{p}}}\vert{\Psi}_M\rangle\vert^2
276: \\ \nonumber
277: & \!\!\times f_{N}(1\!\!-\!\!f_{M})
278: \delta(E_{N} - E_{M} -
279: E), \\ \nonumber
280: \end{eqnarray*}
281: where $\kappa=2\pi e/m$. The function $f_{n}$ is the Fermi distribution
282: and $\langle{\Psi}_N\vert\mbox{{\textbf{p}}}\vert{\Psi}_M\rangle$ are the
283: PAW matrix elements. The whole expression corresponds to the probability
284: per unit volume for a transition of an electron from the valence band
285: state $\vert{\Psi}_N\rangle$ to the conduction band state
286: $\vert{\Psi}_M\rangle$ to occur.
287:
288: The tetrahedron method\cite{BlochlTET} is used to evaluate
289: $\epsilon_i(E)$. The joint density of states, which determines the
290: interband transitions $\delta(E_{N} - E_{M} - E)$ and the optical matrix
291: elements
292: $\vert\langle{\Psi}_N\vert\mbox{{\textbf{p}}}\vert{\Psi}_M\rangle\vert^2$,
293: are computed on each tetrahedron (i.e., 1/4 $\times$ the sum of the matrix
294: elements on the four corners of each tetrahedron). The real part
295: $\epsilon_r$ is then obtained using the Kramers-Kronig
296: relation.\cite{Cardona} Since the dielectric function is the square of the
297: complex refractive index, $(\epsilon_r + i \epsilon_i) = (n_r + i n_i)^2$,
298: the absorption coefficient becomes
299: $$
300: \alpha(E)= 4\pi \frac{E}{hc} n_i =
301: 4\pi \frac{E}{hc} \left[\frac{(\epsilon_r^2 + \epsilon_i^2)^{1/2} -
302: \epsilon_r}{2}\right]^{1/2}
303: $$
304: with $c$ the speed of light in vacuum, $h$ Planck's constant, and $E$ the
305: photon energy.
306:
307: Electron-hole (e-h, excitonic) interactions were not included in the
308: calculations, as they would be in, say, solutions of the Bethe-Salpeter
309: equation. The size of our systems prohibits such complete optical
310: calculations, which are feasible only for very small systems (a few
311: atoms). These additional effects would enhance the results from interband
312: transitions since e-h interactions generally increase the absorption at
313: the onset.\cite{Chang}
314:
315: \section{Construction of the structural models}\label{Ionic}
316:
317: Recent core-level shift experiments\cite{SiegerLuHimpsel} have revealed
318: the presence of all possible oxidation states for Si atoms in Si/SiO$_2$
319: structures such as SLs, that is Si$^{+n}$, where $n= 0,1,2,3,4$ is the
320: charge found within each Si Wigner-Seitz sphere. Si$^{0}$ and Si$^{+4}$
321: are the charge states of Si found in bulk-Si and bulk-SiO$_2$. The
322: suboxide (subO) Si atoms with $n=1,2,3$ are found at the interface.
323: Slightly larger distributions for the subO Si$^{+3}$ densities, as
324: compared to those for Si$^{+1}$ and Si$^{+2}$, were reported in these
325: experiments. Microscopic Si/SiO$_2$ interface models should be consistent
326: with experiments in closely reproducing the density distributions of
327: \emph{all} subO Si atoms.
328:
329: As mentioned above, the Si/SiO$_2$ SL model structures discussed in the
330: present article are based on the Si(001)/SiO$_2$ interface structures
331: obtained by PHC,\cite{Pasquarello} who used the Car-Parrinello method to
332: relax the models to their energy minima. It is important to note that the
333: PHC models were \emph{not} designed for the double interface structure
334: found in SiO$_2$/Si/SiO$_2$ SLs but, rather, for a single Si/SiO$_2$
335: interface which terminates into the vacuum; this is done by saturating
336: dangling bonds with H atoms. Thus, these models {\it de facto} contain the
337: essential details of atomic positions and charge states at the Si/SiO$_2$
338: interface.
339:
340: Tit and Dharma-wardana\cite{TitDharma} have generated a Si/SiO$_2$ SL
341: structure starting from one of the PHC models that contains an
342: \emph{equal} distribution of the three subO atoms, in (almost complete)
343: accord with experiment. This SL model has been constructed by first
344: operating a mirror transformation and then a partial rotation of the
345: Si/SiO$_2$ section of the relaxed interface structure, leading to an
346: intermediate Si/SiO$_2[\mbox{mirror}]$SiO$_2$/Si SL structure. Second,
347: some of the Si layers were inverted in order to meet the $sp^3$-bonding
348: topology. The resulting Si/SiO$_2$ SLs structure, fully described in the
349: article of Tit and Dharma-wardana,\cite{TitDharma} has the following final
350: configuration,
351: $$
352: \overbrace{[O]C^{'}O\underline{D}^{'}O}^{\mbox{Si$^{+4}$}}
353: \underbrace{A^{'}OB^{'}}_{\mbox{Si}^{+1,2,3}}
354: \underbrace{\bf C^{'}\stackrel{\Downarrow}{\displaystyle
355: \bf DA}\bf BC^{'}}_{\mbox{Si$^0$}}
356: \underbrace{D^{'}OA^{'}}_{\mbox{Si}^{+1,2,3}}
357: \overbrace{O\underline{B}^{'}OC^{'}}^{\mbox{Si$^{+4}$}}.
358: $$
359: The connection between the symbols in this configuration and the specific
360: atomic layers in the model is shown in Fig.~\ref{modelSL}. The letters
361: $A$, $B$, $C$ and $D$ correspond to silicon atomic layers, while the $O$s
362: are oxygen layers. The primes denote layers that depart from the
363: diamond-like-Si crystalline arrangement, and $[O]$ corresponds to the
364: layer where the mirror operation has been performed. $\underline{B}^{'}$
365: and $\underline{D}^{'}$ are the Si layers which have been inverted in
366: order to satisfy the $sp^3$-bonding topology. The subO Si$^{+n}$ atoms,
367: with $n=1,2,3$, are distributed within only two Si layers, while the
368: Si$^{0}$ atoms are distributed within five atomic layers. The embryonic
369: PHC interface model corresponds roughly to one side of the above
370: configuration, starting from the arrow up to the right.
371:
372: Of course, this construction induces artificial symmetries in the middle
373: of the SiO$_2$ layer (more precisely, upon and around the $[O]$ layer);
374: this model is thus in essence \emph{partially} relaxed --- the PRM
375: referred to earlier. Significant information on the electronic and optical
376: properties of this model have been extracted, within the TB approach, by
377: Tit and Dharma-wardana,\cite{TitDharma} who obtained direct energy gaps as
378: well as dispersionless bandstructures. Furthermore, the imaginary part of
379: the dielectric function was calculated, and then the absorption
380: coefficient was deduced. From this calculation, enhancement of absorption
381: as well as blueshift with confinement have been demonstrated.\cite{tran}
382:
383: The next obvious step is to relax the PRM, i.e., determine the set of
384: positions which leads, via the Hellman-Feynman forces, to the lowest total
385: energy. We have used the PAW approach described in the previous section to
386: obtain a first fully-relaxed model (FRM1), which contains approximately
387: one unit cell of confined-Si. The supercell contains 52 Si and 44 O atoms,
388: and has dimensions \mbox{7.675 $\times$ 7.675 $\times$ 24.621 \AA$^3$}.
389: The relaxation procedure has been performed with five \textbf{k} points in
390: the reduced Brillouin zone (BZ). The energy cutoff was 25.96 Ryd in all
391: calculations. The total energy was found to decrease by 30.84 eV (0.32 eV
392: per atom) during relaxation. Figure \ref{PRM2FRM} shows the bondlength
393: distributions before (PRM) and after (FRM1) relaxation; the bondlengths
394: are centered around the expected values, viz., $\sim$ 1.61 \AA\ for Si--O
395: and 2.35 \AA\ for Si--Si bonds. The shaded boxes in Fig.~\ref{PRM2FRM} are
396: the distributions of the interface subO Si atoms, while the empty boxes
397: are the total distribution bondlengths, including subO Si atoms; the
398: shaded boxes remain relatively unchanged upon relaxation since both
399: interfaces were already at their energy minimum, after PHC. The main
400: atomic drift during relaxation occurs in the center of the Si and SiO$_2$
401: layers, i.e., near the [O] and the $A$ layer of the configuration
402: discussed above. Interfacial Si--O bondlengths of all subO Si atoms depart
403: from the values of Si$^{+4}$ in the SiO$_2$ layer. The broadening of the
404: Si--Si bondlengths is in general much larger than that of Si--O; the
405: distortion of the bondlengths are thus mainly within the Si layer and at
406: the Si/SiO$_2$ interface, i.e., not inside the silica layer. This is
407: further discussed below. The resulting FRM1 is shown in
408: Fig.~\ref{modelSL}.
409:
410: Additional models having thicker Si wells were constructed in order to
411: examine the role of subO Si layers and the effect of confinement on the
412: electronic and optical properties. As noted earlier, the FRM1 structure
413: contains approximately one unit cell of confined-Si, i.e., the set of
414: layers with charge state Si$^0$ (bulk-like Si). By inserting one, then
415: two, additional $ABCD$ Si atomic planes (i.e., one Si unit cell, thickness
416: 5.43 \AA), and relaxing all atoms, we generated two additional models --
417: FRM2 and FRM3. The FRM2 contains 68 Si atoms while the FRM3 has 84 Si
418: atoms; both have 44 oxygen atoms, as in the FRM1. The {\em total} energy
419: variation for the FRM2 during relaxation was found to be only 0.15 eV
420: (i.e., 0.0013 eV/atom) while for the FRM3, this change is a minuscule
421: 0.051 eV (i.e., 0.00040 eV/atom). These numbers imply that both FRM2 and
422: FRM3 have essentially crystalline Si layers. The FRM2 has nine Si$^{0}$
423: atomic planes while the FRM3 has thirteen.
424:
425: Figure \ref{couches} shows the distributions of the Si--Si bondlengths in
426: the FRM3, starting from the Si(001)/SiO$_2$ interfaces (at the bottom of
427: Fig.~\ref{couches}) and going towards the centre of the silicon layer
428: along the growth axis. The standard deviation ($\sigma$) from the mean
429: value ($\bar{x}$=2.34 \AA\ in all atomic layers except at the interface,
430: where $\bar{x}$=2.29\AA\ ) is also given. The diamond-shaped symbols
431: correspond to the Si--Si bondlengths for Si$^0$ atoms while the filled
432: circles are subO Si--Si bondlengths at the interfaces. The Si--Si
433: bondlengths depart significantly from their crystalline counterparts at
434: the interfaces up to about three atomic layers, where $\sigma$=0.019; the
435: standard deviation is four times higher at the interfaces than in the
436: fifth atomic layer. This deviation of the bondlengths at the Si/SiO$_2$
437: interfaces shows that it is important to take relaxation aspects into
438: account. Indeed, amorphous SiO$_2$ layers in realistic SLs induce strain
439: and disorder in the Si layer, as confirmed in Fig.~\ref{couches}. This
440: effect could generate localized defects giving rise to efficient radiative
441: electron-hole recombination. However, the strain fields only contribute to
442: the small \emph{quasi-momenta} regime and cannot easily supply the
443: momentum deficit involved in the indirect transition of c-Si. Moreover,
444: the relatively small size of our supercell models in the $x$-$y$
445: directions prevents firm conclusions being drawn about the influence of
446: this strain on the optical properties. Further aspects of the role of
447: interfaces are discussed in section \ref{interface}.
448:
449: The three subO Si configurations at the interfaces of the FRMs are shown
450: in Fig.~\ref{tetra} with their corresponding bondlengths. Note that the
451: left and right interfaces in the SiO$_2$/Si/SiO$_2$ SLs are not
452: \emph{exactly} equivalent; they remain independent (during structural
453: relaxation, for instance). However, a subO Si on the left interface has a
454: locally equivalent subO Si on the right interface, by construction. As a
455: consequence, each pair of equivalent subO Si atoms have approximately the
456: same bondlengths and angles in all the FRMs. As seen in Fig.~\ref{tetra},
457: the bondlengths depart from their bulk values, which are 2.35 \AA\ for
458: c-Si and 1.61 \AA\ for SiO$_2$. In addition, the angles of the subO Si
459: tetrahedra vary considerably: the Si--Si--Si angles vary from 99$^{\circ}$
460: to 125$^{\circ}$, the O--Si--Si angles vary from 96$^{\circ}$ to
461: 126$^{\circ}$, while all O--Si--O angles remain around 106$^{\circ}$. It
462: is thus clear that the subO Si tetrahedra at the interfaces are distorted
463: as compared to bulk-Si tetrahedra.
464:
465: As discussed later on, the role of subO Si atoms was further studied using
466: the following variations of the FRM2: First, we removed all Si$^{+4}$
467: atoms and attached the proper number of hydrogen atoms to neutralize the
468: excess charges. The H positions were then relaxed while keeping all the
469: silicon and oxygen atoms fixed. This structure thus contains Si$^0$ atoms
470: and Si$^{+n}$ subOs, where $n=1,2,3$ (i.e., $n \neq 4$). A variation of
471: this structure was generated by removing all oxygen atoms and filling the
472: Si dangling bonds with H atoms, and again relaxing the H atoms. The final
473: Si-H bondlengths vary from 1.47 \AA\ to 1.53 \AA, after relaxation. This
474: final structure is thus subO-free and contains only Si$^{0}$ atoms, except
475: at the interface with the vacuum, where hydrogen atoms fill the dangling
476: bonds. These three confinement models are shown in Fig.~\ref{ConfinMod};
477: they will be referred to as ``FRM2'', ``FRM2/O-H/vacuum'' and
478: ``FRM2/H/vacuum'', respectively.
479:
480: In Fig.~\ref{BZcSi} we show the BZ of the supercell, the standard c-Si
481: diamond BZ, and the high symmetry axes used for the bandstructure
482: calculations. We also constructed the bulk c-Si structure in a supercell
483: of dimensions similar to that of the FRM SLs, so that comparisons can be
484: done within the same \textbf{k} space zone scheme. This will be used in
485: the next three sections for comparisons of bandstructure as well as
486: absorption calculations.
487:
488: \section{Quantum confinement}\label{confinement}
489:
490: In this section, we discuss the nature of the confined states in the SLs.
491: We calculated the bandstructures of the three SL models --- FRM1, FRM2 and
492: FRM3 --- as well as the supplementary FRM2/O-H/vacuum structure, cf.\
493: Fig.~\ref{ConfinMod}(b). The latter is the ``ultimate'' in terms of
494: confinement, as the two interfaces with vacuum constitute infinite
495: potential walls. All bandstructures are analysed and compared within
496: equivalent supercell BZ. The growth axis of the SLs being the $z$-axis,
497: confinement effects are expected to take place in the \mbox{$X$--$R$} and
498: \mbox{$Z$--$\Gamma$} axis of the BZ (see Fig.~\ref{BZcSi} for axis
499: definitions).
500:
501: The bandstructures of the three SLs and their total density of states
502: (DOS) are shown in Fig.~\ref{bands}. We find that the bandstructures in
503: the growth axis ($X$-$R$ and $Z$-$\Gamma$) are dispersionless, for all
504: models and all energies. In physical terms, dispersionless bandstructures
505: imply infinite effective masses, reflecting the strong confinement. The
506: DOS have a more abrupt variation in the valence bands than in the
507: conduction bands. However, DOS alone are not enough to fully understand
508: the optical processes involving the interband transitions, since the
509: weighting of the optical matrix elements is needed. This is further
510: discussed in Section \ref{blueshift}.
511:
512: Let us consider in more details one of the SL models, namely the FRM2. We
513: select the \mbox{$R$--$Z$--$\Gamma$--$M$} high-symmetry axis where the
514: major features, viz.\ the relevant energy gaps, appear. The
515: bandstructures of the folded c-Si structure, the FRM2 SLs structure as
516: well as the FRM2/O-H/vacuum structure have been calculated and compared.
517: Figure \ref{bandSiOH}(a) and \ref{bandSiOH}(b) contain a synopsis of all
518: calculations for \textbf{k} points along \mbox{$R$--$Z$--$\Gamma$--$M$}.
519: Several conclusions can be drawn from these figures.
520:
521: By comparing the bands for c-Si, Fig.~\ref{bandSiOH}(a), with those for
522: the FRM2 SL, Fig.~\ref{bandSiOH}(b) (solid lines), we see that folding
523: effects cannot by themselves explain the new band-structure. The bands in
524: the \mbox{$Z$--$\Gamma$} directions are totally modified by the
525: confinement. Although c-Si always has an optically indirect bandgap, this
526: bandgap becomes almost direct in its folded configuration, as can be seen
527: from Fig.~\ref{bandSiOH}(a), while for the SL the bandgap is
528: \emph{unequivocally} direct, and significantly increased. Moreover,
529: comparing the bands in a direction orthogonal to the growth axis, for
530: instance around $M$ in both Fig.~\ref{bandSiOH}(a) and
531: Fig.~\ref{bandSiOH}(b), we see that the valence bands are raised; the
532: lowest conduction band in the \mbox{$\Gamma$--$M$} direction is pushed to
533: higher energies. In addition, they exhibit less dispersion in the SL than
534: in c-Si, in general. Hence, confinement modifies the electronic properties
535: in the growth axis as well as in directions orthogonal to it. This is
536: further analysed from optical absorption calculations, below.
537:
538: We compare in Fig.~\ref{bandSiOH}(b) the bandstructures of the FRM2 and
539: FRM2/O-H/vacuum models. The positions of the Si$^0$ atoms, as well as the
540: subO Si in the two structures, are identical. The solid lines displays the
541: bandstructures of the FRM2, while the dots display the bands of the
542: FRM2/O-H/vacuum structure. It is clear that the two bandstructures are
543: nearly identical. This calculation shows that the SiO$_2$ layers act as
544: virtually impenetrable barriers. The electronic properties of hypothetical
545: SL structures having only subO Si -- i.e., no Si$^{+4}$ of SiO$_2$ -- and
546: positioned at the subO Si sites, would give nearly identical electronic
547: properties as Si/SiO$_2$ SLs, for energies close to the band gap. Our
548: calculations show that the electronic wavefunctions die out at the
549: suboxide Si atoms of the interfaces. Thus the barrier is sharply located
550: just behind the subO Si atoms, and therefore just two atomic layers could
551: be used as a barrier without altering the electronic properties, when
552: energies involved (in the device) remain close to the energy gap. The
553: influence of the interface subO Si atoms is further analysed next.
554:
555: \section{Role of interfaces}\label{interface}
556:
557: In order to assess the role of the subO ions on the electronic properties,
558: we calculated the bandstructures of the FRM2/H/vacuum structure,
559: Fig.~\ref{ConfinMod}(c), which contains {\it no} subO Si; the dangling
560: bonds have been filled by hydrogen atoms and hydrogen atoms (only) have
561: been structurally relaxed. Figure \ref{bandH} summarizes the results. The
562: LDA band gap is still direct but significantly lowered, from 0.81 eV in
563: the FRM2 to 0.67 eV in the FRM2/H/vacuum structure. Interface
564: reconstruction (as discussed in section \ref{Ionic} and
565: Fig.~\ref{couches}) thus has significant impact on the electronic
566: properties. The bands in the plane orthogonal to the growth axis are quite
567: different from FRM2; e.g., the valence band is lowered near the $M$ and
568: $R$ points, becoming similar to the c-Si bandstructure. We thus conclude
569: that the subO Si atoms have two effects: (i) increase the bandgap and (ii)
570: produce dispersionless valence bands. The charge states in the subO are
571: responsible for the increase in the bandgap, while the strongly increased
572: valence-band offset leads to essentially dispersionless bands.
573:
574: \section{Blueshift and optical enhancement}\label{blueshift}
575:
576: The matrix elements entering the calculations of the optical properties
577: are often approximated as a constant in a given range of energies.
578: However, such an approximation is inadequate for elucidating the enhanced
579: luminescence in Si/SiO$_2$ SLs. This section deals with calculating the
580: absorption coefficient within the Fermi Golden rule and
581: interband-transition theory.
582:
583: Since the Si-layer thickness $L_{\rm Si}$ is relevant to the energy-shift
584: in SLs, this quantity needs first to be defined. This involves some
585: uncertainty associated with the interface thicknesses $L_{\rm subO}$.
586:
587: The interface thickness $L_{\rm subO}$ was estimated to be $\sim$1.60 \AA\
588: from our calculations of the subO Si region. This is the largest distance
589: (projected onto the $z$ axis) between any two subO Si atoms. Hence the
590: upper-bound to $L_{\rm Si}$ are, 11.17 \AA, 16.58 \AA\ and 22.01 \AA\ for
591: the FRM1, FRM2 and FRM3 respectively, while the lower-bound thicknesses
592: are simply $L_{\rm Si}-2L_{\rm subO}$. We define the Si thickness in the
593: SLs to include the interface subO Si atoms as well (corresponding to the
594: upper bound). This choice is made since subO Si (Si$^{+1}$, Si$^{+2}$ or
595: Si$^{+3}$ ) atoms contribute to the electronic properties, as do bulk-Si
596: atoms (Si$^0$ atoms); for instance, we showed above [see e.g.
597: Fig.~\ref{bandSiOH}(b)] that the bandstructures of the FRM2 SLs and the
598: FRM2/O-H/vacuum systems overlap, and indicated where the effective barrier
599: begins.
600:
601: The band gaps of the FRM1, FRM2 and FRM3 (see the bandstructures in
602: Fig.~\ref{bands}) are direct except for the FRM1 where the band gap is
603: \emph{nearly} direct with only 0.12 eV between the direct and indirect
604: transitions. The values of the gap are \mbox{0.99 eV}, \mbox{0.81 eV} and
605: \mbox{0.68 eV} for the FRM1, FRM2 and FRM3, respectively. The direct
606: transition at $\Gamma$ for the FRM1 equals \mbox{1.11 eV}. For the FRM2
607: and the FRM3, the band gaps are direct and located on the whole
608: $Z$--$\Gamma$ axis [see Fig.~\ref{bands}(b) and (c)]. Direct transitions
609: can thus be achieved between the valence band and the conduction band,
610: along the $Z-\Gamma$ line of the BZ. We thus obtain, under the LDA, a
611: blue shift
612: $$
613: \mbox{0.68 eV} \longrightarrow \mbox{0.81 eV} \longrightarrow
614: \mbox{0.99 eV}
615: $$
616: with increased confinement
617: $$
618: \mbox{2.2 nm} \longrightarrow \mbox{1.7 nm} \longrightarrow
619: \mbox{1.1 nm}.
620: $$
621: However, these values for the energy gaps are much lower than the
622: experimental ones. It is well known that DFT within the LDA underestimates
623: the energy gaps of semiconductors and insulators. For c-Si, the DFT-LDA
624: gap is approximately 0.6 eV less than the experimental value. For the
625: $\beta$-cristobalite phase of SiO$_2$, in the group $I\bar{4}2d$, the
626: DFT-LDA energy gap is \mbox{5.8 eV} while the experimental value is about
627: 3 eV higher, i.e., \mbox{$\sim$9 eV}. Approximate, but realistic, band
628: gaps can be obtained by adding 0.6 eV overall:
629: $$
630: \begin{array}{ccccc}
631: \mbox{\bf 1.28 eV} & \longrightarrow & \mbox{\bf 1.41 eV} &
632: \longrightarrow & \mbox{\bf 1.59 eV}. \\
633: \mbox{(2.2 nm)} & & \mbox{(1.7 nm)} & & \mbox{(1.1 nm)}
634: \end{array}
635: $$
636: These energy gaps correspond to the lower bound of the
637: experiments\cite{Novikov} and lie in the visible spectrum. However, these
638: gaps are still somewhat lower than the experimental
639: values.\cite{LuLockBar,Kanemitsu,Khriachtchev,Mulloni,Kanemitsu_bis} The
640: discrepancy can be explained by recrystallization processes, which lead to
641: the formation of nanoclusters that would increase the confinement, and
642: correspondingly the measured energy gaps.\cite{ZachariasMRS} The analysis
643: of such a behaviour, which is beyond the scope of the present work, would
644: require zero-dimensional-confined model structures.
645:
646: The absorption of the three SL models and c-Si have been calculated both
647: in the diamond-like BZ and in the SLs BZ. For c-Si in the diamond-like
648: BZ, we used \mbox{(20 $\times$ 20 $\times$ 20)} \textbf{k}
649: points,\cite{Adolph} while in the SLs BZ, \mbox{(7 $\times$ 7 $\times$ 2)}
650: \textbf{k} points are used. Calculations using more \textbf{k} points,
651: viz.\ \mbox{(8 $\times$ 8 $\times$ 3)}, show that \mbox{(7 $\times$ 7
652: $\times$ 2)} is quite sufficient to recover the form of the absorption
653: curve for all the SL models. The purpose of calculating the absorption of
654: c-Si in two different BZ is to estimate errors, first due to zone folding
655: effects (which cause round-off errors, leading to non-absolutely-null
656: transitions at the onset)\cite{Furthmuller} and, second, to the
657: tetrahedron method itself which needs large amounts of \textbf{k} points.
658: The broadening in the absorption curves has been fixed to \mbox{0.015 eV}
659: for all the absorption curves discussed below, as suggested by
660: Fuggle.\cite{Fuggle}
661:
662: Figure \ref{Absorp} shows the absorption results. Panels (a) and (b) give
663: an overall view of the absorption curves for the $z$ axis in (a) and the
664: $x$-$y$ plane in (b). Panels (c) and (d) show the absorption at the onset,
665: for the $z$ axis in (c) and the $x$-$y$ plane in (d). In all cases, we
666: included the absorption of c-Si calculated in the diamond-like BZ, as well
667: as the one in the SLs BZ. Direct comparison of the two c-Si absorption
668: curves give an estimate of imprecisions due to zone foldings and
669: intermediate number of \textbf{k} point effects. It shows that the
670: absorption is slightly underestimated in the SL calculations; e.g., in
671: Fig.~\ref{Absorp}(c), the onset of absorption of c-Si in the SLs BZ take
672: place at 2.0 eV while in the diamond-like BZ the onset happens at the
673: correct value of 2.52 eV (which is for c-Si the LDA direct transition at
674: $\Gamma$). This numerical effect cannot be avoided and will arise, as
675: well, in any Si/SiO$_2$ supercell. Hence, all comparison of the SLs
676: absorption must be made with c-Si calculated in the equivalent SLs BZ,
677: i.e., within equivalent {\bf k} space zone schemes.
678:
679: Since the SLs are fabricated with the objective of changing the indirect
680: gap to a direct gap, we now discuss the absorption threshold region.
681: Comparison at the onset of absorption from Fig.~\ref{Absorp}(c) or (d)
682: shows that all absorption curves have a lower energy threshold than
683: \emph{both} the c-Si absorption curves, and especially below the one
684: calculated in equivalent SLs BZ having equal number of \textbf{k} points.
685: That is, {\it the SLs show absorption (and emission) in the spectral
686: region above the indirect gap of c-Si and below the direct gap of c-Si}.
687: This shows that the absorption in \emph{all} confined Si/SiO$_2$ SL models
688: is enhanced, compared to \mbox{c-Si}; the transitions are direct in SLs
689: and have an active oscillator strength. For the folded c-Si energy bands,
690: the lower bands above the Fermi level, and the corresponding oscillator
691: strength, remain dark; in other words, the optical matrix elements of the
692: SL BZ of c-Si are null up to $\sim$2.0 eV. This result clearly
693: demonstrates the enhancement of the absorption (and emission) mechanisms
694: in confined Si structures. Furthermore, upon inspection of the absorption
695: curves in the plane orthogonal to the growth axis [Fig.~\ref{Absorp}(d)],
696: we note that the energy thresholds of the absorption are all below c-Si:
697: thus, the $x$-$y$ absorption of the FRMs takes place approximately at
698: their respective direct energy gaps, and then behave in a similar manner,
699: as expected from the similarity of the SLs bandstructures, in this plane.
700:
701: We examine, finally, the higher-energy region which corresponds to the
702: usual direct transition (3 - 6 eV) in c-Si. Even here,
703: Fig.~\ref{Absorp}(a) demonstrates a blueshift with increased confinement
704: in the $z$ axis. The overall absorption maxima for FRM1-FRM3 are at 5.28
705: eV, 4.83 eV, and 4.71 eV, with intensities of 136, 155 and 162 ($\times
706: 10^4$/cm) respectively. For c-Si in the SLs BZ (to ensure comparable
707: precision in the calculations) the second peak, i.e., the maximum, take
708: place at 4.70 eV (with absorption equal to \mbox{231 $\times$ 10$^4$/cm}),
709: while the first peak is at 4.0 eV (and with absorption equal to \mbox{229
710: $\times 10^4$/cm}). We emphazise that there is still a slight blueshift in
711: the $x$-$y$ plane orthogonal to the growth axis, but less pronounced than
712: in the growth axis.
713:
714: \section{Concluding remarks}
715:
716: In this work, the structural, electronic and optical properties of
717: Si/SiO$_2$ superlattices have been studied on the basis of
718: structurally-relaxed models. These SL models, which contain no hydrogen
719: atoms at the Si/SiO$_2$ interfaces, exhibit enhanced optical
720: absorption/emission, as observed in experiment; this can be attributed to
721: the presence of silicon dioxide barriers. In experiments performed under
722: ultra-high vacuum conditions, the oxidization process would predominantly
723: give rise to oxide bonds at the interfaces, but still, few hydrogen atoms
724: are expected to be present and fill some of the remaining dangling bonds.
725:
726: Our calculations show that the oxide barriers are central to the optical
727: enhancement in SLs. It is well known that hydrogen atoms play a similar
728: role in amorphous silicon by filling dangling bonds. This suggests that it
729: might also be the case in Si/SiO$_2$ SLs, where hydrogen atoms fill extra
730: dangling bonds, and thus would amplify the optical enhancement effect
731: already exerted by the oxide barriers. Further studies are needed to
732: ascertain this.
733:
734: We have shown that suboxide Si atoms at the interfaces act as virtually
735: impenetrable barriers. The active barrier thickness thus corresponds to
736: the suboxide Si layer --- only 1.6 \AA\ in our models. The confined Si
737: layer thus consists of bulk Si \emph{and} suboxide Si atoms. Suboxide Si
738: atoms at the interfaces modify the electronic properties in two manners:
739: they (i) increase the energy gap and (ii) lead to dispersionless band
740: structures, which increases the transition probabilities.
741:
742: Other confinement models (zero-dimensional structures) and interface
743: effects will be considered in future work. For instance, inclusion of
744: other atomic species --- such as nitrogen that would generate subnitric Si
745: atoms at the Si/SiO$_2$ interface --- are expected to modify the
746: electronic properties, and hence enhance the optical absorption/emission
747: spectra.
748:
749: \vspace{0.5cm}
750:
751: {\it Acknowledgments} -- It is a pleasure to thank Dr.\ Jurgen
752: Furthm\"uller for help with the optical calculations in VASP. We
753: gratefully acknowledge Dr.\ Zheng-Hong Lu for helpful discussions and
754: suggestions. This work is supported by grants from the Natural Sciences
755: and Engineering Research Council (NSERC) of Canada and the ``Fonds pour la
756: formation de chercheurs et l'aide {\`a} la recherche'' (FCAR) of the
757: Province of Qu{\'e}bec. We are indebted to the ``R\'eseau qu\'eb\'ecois de
758: calcul de haute performance'' (RQCHP) for generous allocations of computer
759: resources.
760:
761: \newpage
762:
763: \begin{references}
764: \bibitem{LuLockBar} Z.H. Lu, D.J. Lockwood, and J.-M. Baribeau, Nature
765: \textbf{ 378}, 258 (1995).
766: \bibitem{Kanemitsu} Y. Kanemitsu and S. Okamoto, Phys. Rev. B
767: \textbf{56}, R15561 (1997).
768: \bibitem{Novikov} S.V. Novikov, J. Sinkkonen, O. Kilpel\"a, and S. V.
769: Gastev, J. Vac. Sci. Technol. B \textbf{ 15}, 1471 (1997).
770: \bibitem{Khriachtchev} L. Khriachtchev, M. R\"as\"anen, S. Novikov,
771: O. Kilpel\"a, and J. Sinkkonen, J. Appl. Phys \textbf{ 86}, 5601
772: (1999).
773: \bibitem{Mulloni} V. Mulloni, R. Chierchia, C. Mazzeleni, G. Pucker, L.
774: Pavesi, and P. Bellutti, Philos. Mag. B \textbf{ 80}, 705 (2000).
775: \bibitem{Kanemitsu_bis} Y. Kanemitsu, M. Liboshi, and T. Kushida, Appl.
776: Phys. Lett. \textbf{ 76}, 2200 (2000).
777: \bibitem{Canham} L.T. Canham, Appl. Phys. Lett. \textbf{ 57}, 1046
778: (1990).
779: \bibitem{LehmanGosele} V. Lehman and U. G\"osele, Appl. Phys. Lett.
780: \textbf{58}, 856 (1991).
781: \bibitem{Tsybeskov} L. Tsyberkov, K.L. Moore, D.G. Hall, and P.M.
782: Fauchet, Phys. Rev. B \textbf{ 54}, R8361 (1996).
783: \bibitem{Cheylan} S. Cheylan and R.G. Elliman, Appl. Phys. Lett.
784: \textbf{78}, 1912 (2001).
785: \bibitem{Ng} W.L. Ng, M.A. Louren\c{c}o, R.M. Gwilliam, S. Ledain, G.
786: Shao, and K.P. Homewood, Nature \textbf{ 410}, 192 (2001).
787: \bibitem{ericson} P. Schmuki, L.E. Erickson, and D.J. Lockwood,
788: Phys. Rev. Lett. {\bf 80}, 4060 (1998).
789: \bibitem{porous} S. Schuppler, S.L. Friedman, M.A. Marcus, D.L. Adler,
790: Y.-H. Xie, F.M. Ross, Y.J. Chabal, T.D. Harris, L.E. Brus, W.L.
791: Brown, E.E. Chaban, P.F. Szajowski, S.B. Christman, and P.H. Citrin,
792: Phys. Rev. B {\bf 52}, 4910 (1995); D.J. Lockwood, A. Wang, and
793: B. Bryskiewicz, Solid State Comm. {\bf 89}, 587 (1994).
794: \bibitem{Luprivate} Z.H. Lu (private communication).
795: \bibitem{Zacharias} M. Zacharias, J. Bl\"asing, K. Hirschman, L.
796: Tsybeskov, P.M. Fauchet, J. Non-Crys. Solids {\bf 266-269}, 640 (2000).
797: \bibitem{ZachariasAPL} M. Zacharias, J. Bl\"{a}sing,
798: P. Veit, L. Tsybeskov, K. Hirschman, and P. M. Fauchet, Appl. Phys.
799: Lett. \textbf{74}, 2614 (1999).
800: \bibitem{Stirling} A. Stirling, A. Pasquarello, J. -C. Charlier, R. Car,
801: Phys. Rev. Lett. \textbf{85}, 2773 (2000).
802: \bibitem{Delley} B. Delley and E.F. Steigmeier, Appl. Phys. Lett.
803: {\bf 67}, 2370 (1995).
804: \bibitem{Kageshima} H. Kageshima and K. Shiraishi, in \emph{Materials and
805: Devices for Silicon-Based Optoelectronics}, edited by J.E. Cunningham,
806: S. Coffa, A. Polman, and R. Soref, Mater. Res. Soc. Symp.
807: Proc. No.\textbf{ 486} (Material Research Society, Pittsburgh, 1998),
808: p. 337.
809: \bibitem{Pukkinen} M.P.J. Pukkinen, T. Korhonen, K. Kokko, and I.J.
810: V\"ayrynen, Phys. Stat. Sol. {\bf 214}, R17 (1999).
811: \bibitem{Degoli} E., Degoli, S. Ossicini, Surf. Sci. {\bf 470}, 32
812: (2000).
813: \bibitem{Carrier} P. Carrier, L.J. Lewis, and M.W.C. Dharma-wardana,
814: Phys. Rev. B. {\bf 64}, 195330-1 (2001).
815: \bibitem{Agrawal} B.K. Agrawal and S. Agrawal, Appl. Phys. Lett.
816: {\bf 77}, 3039 (2000).
817: \bibitem{Fuggle} J.C. Fuggle, in \textit{Unoccupied Electronic States},
818: edited by J.C. Fuggle and J.E. Inglesfield,
819: Topics in Applied Physics Vol.\ \textbf{69}
820: (Springer-Verlag, Berlin, 1992).
821: \bibitem{SiegerLuHimpsel}
822: M.T. Sieger, D.A. Luh, T. Miller, and T.-C. Chiang,
823: Phys. Rev. Lett. {\bf 77}, 2758 (1996);
824: Z.H. Lu, M.J. Graham, D.T. Jiang, and K.H. Tan,
825: Appl. Phys. Lett. {\bf 63}, 2941 (1993);
826: F.J. Himpsel, F.R. McFeely, A. Taleb-Ibrahimi,
827: J.A. Yarmoff, and G. Holliger, Phys. Rev. B {\bf 38}, 6084 (1988).
828: \bibitem{Lockwood} D.J. Lockwood, Z.H. Lu, and J.-M. Baribeau, Phys. Rev.
829: Lett. {\bf 76}, 539 (1996).
830: \bibitem{Pasquarello} A. Pasquarello, M.S. Hybertsen, and R. Car, Appl.
831: Surf. Sci. \textbf{104/105}, 317 (1996);
832: A. Pasquarello, M.S. Hybertsen, and R. Car, Appl. Phys. Lett.
833: {\bf 68}, 625 (1996).
834: \bibitem{Neaton} J.B. Neaton, D.A. Muller, and N.W. Ashcroft, Phys. Rev.
835: Lett. {\bf 85}, 1298 (2000).
836: \bibitem{Tu} Y. Tu and J. Tersoff, Phys. Rev. Lett. {\bf 84}, 4393
837: (2000).
838: \bibitem{NgInterface} K.-O. Ng and D. Vanderbilt, Phys. Rev. B {\bf 59},
839: 10132 (1999).
840: \bibitem{KageshimaInterface} H. Kageshima and K. Shiraishi,
841: Phys. Rev. Lett. {\bf 81}, 5936 (1998).
842: \bibitem{HermanBatra} F. Herman and I.P. Batra, in {\it The Physics of
843: SiO$_2$ and its interfaces} edited by S.T. Pantelides (Pergamon,
844: Oxford, 1978).
845: \bibitem{TitDharmaDBMBOM} N. Tit and M.W.C. Dharma-wardana, Physics
846: Letters A \textbf{ 254}, 233 (1999).
847: \bibitem{TitDharma} N. Tit and M.W.C. Dharma-wardana, J. Appl. Phys.
848: \textbf{ 86}, 1 (1999).
849: \bibitem{tran} M. Tran, N. Tit, and M.W.C. Dharma-wardana, Appl. Phys.
850: Lett. {\bf 75}, 4136 (1999).
851: \bibitem{VASPref} G. Kresse and J. Furthm\"uller, computer code
852: \textbf{VASP 4.4} (Vienna University of Technology, Vienna, 1997)
853: [Improved and updated Unix version of
854: the original copyrighted VASP/VAMP code, which was published by G.
855: Kresse and J. Furthm\"uller, Comput. Mater. Sci. {\bf 6}, 15 (1996)].
856: \bibitem{KresseJoubert} G. Kresse and D. Joubert, Phys. Rev. B {\bf 59},
857: 1758 (1999).
858: \bibitem{HohenbergKohn} P. Hohenberg and W. Kohn, Phys. Rev. {\bf 136},
859: B864 (1964).
860: \bibitem{KohnSham} W. Kohn and L.J. Sham, Phys. Rev. {\bf 140}, A1133
861: (1965).
862: \bibitem{Payne} M.C. Payne, M.P. Teter, D.C. Allan, T.A. Arias, and
863: J.D. Joannopoulos, Rev. Mod. Phys. {\bf 64}, 1045 (1992).
864: \bibitem{BlochlPAW} P.E. Bl\"ochl, Phys. Rev. B {\bf 50}, 17953 (1994).
865: \bibitem{Singh} D.J. Singh, \textit{Planewaves, pseudopotentials and the
866: LAPW method}, (Kluwer Academic Publishers, Norwell, 1994).
867: \bibitem{Adolph} B. Adolph, J. Furthm\"uller, and F. Bechstedt, Phys.
868: Rev. B {\bf 63}, 125108-1 (2001).
869: \bibitem{Cardona} P.Y. Yu and M. Cardona, \emph{Fundamentals of
870: Semiconductors} (Springer, New-York, 1996);
871: G. Baym, \emph{Lectures on Quantum
872: mechanics}, (Addison-Wesley, Reading, 1993); L. Ley in \emph{The
873: Physics of Hydrogenated Amorphous Silicon II}, edited by J.D.
874: Joannopoulos and G. Lucovsky, Topics in Applied Physics, Vol.\
875: {\bf 56} (Springer-Verlag, Berlin, 1984).
876: \bibitem{BlochlTET} P.E. Bl\"ochl, Phys. Rev. B {\bf 49}, 16223 (1994).
877: \bibitem{Chang} E.K. Chang, M. Rohlfing, and S.G. Louie, Phys. Rev. Lett.
878: {\bf 85}, 2613 (2000); S. Albrecht, L. Reining, R. Del Sole, and G.
879: Onida, Phys. Rev. Lett. {\bf 80}, 4510 (1998).
880: \bibitem{ZachariasMRS} M. Zacharias, J. Heitmann, and U. G\"{o}sele,
881: MRS Bulletin \textbf{26}, 975 (2001).
882: \bibitem{Furthmuller} J. Furthm\"uller (private communication).
883: \end{references}
884:
885:
886: %FIGURE_1
887: \begin{figure}[htb]
888: \includegraphics*[width=6cm]{FRM1.ps}
889: \caption{ The unit cell of the fully relaxed SL model (FRM1).
890: The configuration of
891: the bulk-like and suboxide Si atomic planes is also depicted.
892: The white and black circles are respectively the positions of
893: Si and O atoms. }
894: \label{modelSL}
895: \end{figure}
896:
897: %FIGURE_2
898: \begin{figure}[htb]
899: \includegraphics*[width=8.5cm]{distPRM.eps}
900: \includegraphics*[width=8.5cm]{distFRM1.eps}
901: \caption{ Evolution of the Si--O and Si--Si bondlengths from (a) the PRM
902: construction by Tit and Dharma-wardana, to (b)
903: the fully relaxed structure (FRM1) described in the text. }
904: \label{PRM2FRM}
905: \end{figure}
906:
907: %FIGURE_3
908: \begin{figure}[htb]
909: \includegraphics*[width=8.5cm]{couches.eps}
910: \caption{ Si--Si bondlength distribution in the FRM3, from the interface
911: (bottom of figure)
912: towards the center (top of figure) of the Si layer. (There are thirteen
913: Si$^0$ layers in the FRM3 and thus six interplanar Si--Si bondlengths
914: starting from both interfaces towards the center of the Si layer.}
915: \label{couches}
916: \end{figure}
917:
918: %FIGURE_4
919: \begin{figure}[htb]
920: \includegraphics*[width=6cm]{Si1.ps}
921: \includegraphics*[width=6cm]{Si2.ps}
922: \includegraphics*[width=6cm]{Si3.ps}
923: \caption{Structure of the three suboxide interfacial
924: Si atoms.}
925: \label{tetra}
926: \end{figure}
927:
928: %FIGURE_5
929: \begin{figure}[htb]
930: \includegraphics*[width=3.0cm]{FRM2SL.ps}\includegraphics*[width=3.0cm]{FRM2oh.ps}\includegraphics*[width=3.0cm]{FRM2h.ps}
931: \caption{ (a) The FRM2 SLs (b) the FRM2/O-H/vacuum model
932: (c) the FRM2/H/vacuum model.}
933: \label{ConfinMod}
934: \end{figure}
935:
936: %FIGURE_6
937: \begin{figure}[htb]
938: \includegraphics*[width=6cm]{bzV.ps}
939: \caption{Definition of the SLs BZ superposed to the
940: diamond-like BZ. The principal axis used for bandstructure
941: calculation are also depicted.}
942: \label{BZcSi}
943: \end{figure}
944:
945: %FIGURE_7
946: \begin{figure}[htb]
947: \includegraphics*[width=8.2cm]{bandFRM1.eps}
948: \includegraphics*[width=8.2cm]{bandFRM2.eps}
949: \includegraphics*[width=8.2cm]{bandFRM3.eps}
950: \caption{ Band structures and density of states (DOS)
951: of the three SL models. The DOS are
952: calculated using
953: \mbox{(7 $\times$ 7 $\times$ 1)} \textbf{k}-point grid,
954: which corresponds to 144 irreducible tetrahedra.}
955: \label{bands}
956: \end{figure}
957:
958: %FIGURE_8
959: \begin{figure}[htb]
960: \includegraphics*[width=4.2cm]{bandSi.eps}\includegraphics*[width=4.2cm]{bandOH.eps}
961: \caption{ Band structures
962: of (a) c-Si in the (folded) SL BZ (b) comparison between the FRM2 SLs
963: and the FRM2/O-H/vacuum bandstructures.}
964: \label{bandSiOH}
965: \end{figure}
966:
967: %FIGURE_9
968: \begin{figure}[htb]
969: \includegraphics*[width=8.2cm]{FRM2hVSFRM2.eps}
970: \caption{ Comparison of the bandstructures for
971: FRM2/H/vacuum and FRM2 SLs. }
972: \label{bandH}
973: \end{figure}
974:
975: %FIGURE_10
976: \begin{figure}[htb]
977: \includegraphics*[width=7cm]{absZtot.eps}\includegraphics*[width=7cm]{absXYtot.eps}
978: \includegraphics*[width=7cm]{absZzoom.eps}\includegraphics*[width=7cm]{absXYzoom.eps}
979: \caption{ Absorption coefficient of the SL models as compared to c-Si.
980: (a) and (c) are the absorption in the growth axis; (b) and (d)
981: are the absorption in the plane orthogonal to the growth axis. }
982: \label{Absorp}
983: \end{figure}
984:
985: \end{document}
986:
987: