1: \documentclass[prb,twocolumn,showpacs,superscriptaddress]{revtex4}
2: \usepackage{graphicx,dcolumn,amsmath}
3:
4: \begin{document}
5:
6: \title{\bf
7: Electron confinement and optical enhancement in Si/SiO$_2$ superlattices
8: }
9:
10: % Authors and affiliations
11: \author{\firstname{Pierre} \surname{Carrier}}
12: \affiliation{D\'{e}partement de Physique et Groupe de Recherche en
13: Physique \\et Technologie des Couches Minces (GCM), Universit\'{e} de
14: Montr\'{e}al, \\ Case Postale 6128, Succursale~Centre-Ville, Montr\'{e}al,
15: Qu\'{e}bec, Canada H3C 3J7}
16:
17: \author{\firstname{Laurent J.} \surname{Lewis}}
18: \email[To whom correspondence should be addressed; \\
19: e-mail: ]{laurent.lewis@umontreal.ca}
20: \affiliation{D\'{e}partement de Physique et Groupe de Recherche en
21: Physique \\et Technologie des Couches Minces (GCM), Universit\'{e} de
22: Montr\'{e}al, \\ Case Postale 6128, Succursale~Centre-Ville, Montr\'{e}al,
23: Qu\'{e}bec, Canada H3C 3J7}
24:
25: \author{\firstname{M. W. Chandre} \surname{Dharma-wardana}}
26: %\email[E-mail: ]{chandre@cm1.phy.nrc.ca}
27: \affiliation{Institute for Microstructural Sciences, National Research
28: Council, Ottawa, Canada K1A 0R6}
29:
30: \date{\today}
31:
32: \begin{abstract}
33: We have performed first-principles calculations of Si/SiO$_2$ superlattices
34: in order to examine their electronic states, confinement and optical
35: transitions, using linearized-augmented-plane-wave techniques and
36: density-functional theory. Two atomic models having fairly simple interface
37: structure are considered; they differ in the way dangling bonds at
38: interfaces are satisfied. We compare our first-principles band structures
39: with those from tight-binding calculations. The real and imaginary parts of
40: the dielectric function are calculated at the Fermi-Golden-rule level and
41: used to estimate the absorption coefficients. Confinement is confirmed by the
42: dispersionless character of the electronic band structures in the growth
43: direction. Optical enhancement is shown to exist by comparing the direct and
44: indirect transitions in the band structures with the related transitions in
45: bulk-Si. The role played by the interface on the optical properties is
46: assessed by comparing the absorption coefficients from the two models.
47: \end{abstract}
48:
49: \pacs{78.66.Jg, 68.65.+g, 71.23.Cq}
50:
51: \maketitle
52:
53:
54:
55: \section{Introduction}\label{Intro}
56:
57: Informatics and telecommunications are the driving forces behind the
58: development of new photonic devices.\cite{lampro} Such devices can be
59: assembled from a wide variety of materials; among these, silicon is probably
60: the most prevalent, although it suffers from having an indirect band gap.
61: Direct band-gap materials such as GaAs, InP, and many other III-V and II-VI
62: semiconductors, are efficient electron-photon energy converters. However they
63: are costly for large-scale applications, and would be less interesting if
64: full Si-based photonics were possible. These considerations have lead to
65: considerable efforts towards the development of new types of Si-based
66: materials having direct energy gap in the visible (viz.\ 1.5 to 3
67: eV),\cite{Zhang} suitable for photo-diodes and lasers, as well as solar-cell
68: applications.
69:
70: The simultaneous discovery by Canham\cite{Canham} and Lehmann and
71: G\"{o}sele\cite{LehmannGosele} of intense luminescence in porous silicon
72: ($\pi$-Si) has opened up new horizons for Si-based materials. $\pi$-Si can be
73: regarded as consisting of long and thin nanowires,\cite{ChuangCollinsSmith}
74: forcing the electronic states to be confined within the finite dimension of
75: the nanostructures. The shift of the luminescence spectrum towards the blue
76: with decreasing nanowire size\cite{Canham} suggests that the confinement of
77: the electronic states might be responsible for the visible luminescence.
78: Unfortunately, most forms of $\pi$-Si are unstable, lasting only a few hours
79: before a strong decrease of the luminescence efficiency is observed. This
80: instability has sometimes been attributed to the volatile Si--H bonds at the
81: surface.\cite{LehmannGosele} Annealing under nitrogen and oxygen has been
82: suggested as a means of stabilizing the structures; however, the annealed
83: material has weaker luminescence than the parent $\pi$-Si. Layered
84: polysilanes (Si$_{6n}$H$_6$) also exhibit optical properties similar to
85: $\pi$-Si. Uehara et al\cite{Uehara} have performed first-principles
86: calculations of layered polysilanes structures and showed that porous Si may
87: be akin to thin Si$_6$H$_6$ sheets.
88:
89: As a perhaps more promising avenue, confinement can also be achieved in
90: superlattices (SLs) --- or quantum wells --- as well as quantum wires and
91: quantum dots. Lu et al have reported enhanced luminescence in the visible
92: part of the spectrum of Si/SiO$_2$ SLs fabricated by molecular beam epitaxy
93: (MBE),\cite{LuLockwoodBaribeau} and a blue shift was observed when the
94: thickness of the Si layers was reduced from 6 to 2 nm. In contrast to
95: $\pi$-Si, Si/SiO$_2$ SLs present the advantage of being stable, with no
96: significant decrease of the luminescence observed with time.
97:
98: Motivated by these promising experimental observations, and to provide
99: insight into the microscopic physics associated with the luminescence
100: efficiency of SL-based devices, we undertook a detailed, first-principles
101: investigation of the electronic and optical properties of Si/SiO$_2$ SLs.
102: Silicon and SiO$_2$ are both already standard components of MOSFETs and other
103: devices. SiO$_2$ has an energy gap of $\sim$9 eV; Si has an indirect gap of
104: $\sim$1.1 eV. Thus, the electrons are confined by the SiO$_2$ barriers within
105: the silicon quantum wells. Luminescence, however, is not determined solely by
106: confinement since the nature of the energy gap is also important. The
107: possibility of creating a direct energy gap in a SL can be discussed in
108: simplified terms within the concept of Brillouin zone
109: folding:\cite{gnutzmann} The periodicity of a SL, $d$, being larger than that
110: of the constituent lattices, $a_0$, the Brillouin zone gets folded to
111: dimension $\pi/d$ instead of $\pi/a_0$, which may bring the indirect minimum
112: of the conduction band to the $\Gamma$-point and thus produce an effectively
113: direct energy-gap material. ``Quasi-direct'' energy gaps were so obtained in
114: Si/Ge superlattices;\cite{pearsall} however, there was little enhancement in
115: the optical properties.
116:
117: In this work, we examine the luminescence properties of Si/SiO$_2$ SLs using
118: model structures and first-principles methodology. Transmission electron
119: microscopy reveals that in Si/SiO$_2$ SLs, both Si and SiO$_2$ layers are
120: amorphous.\cite{LuLockwoodBaribeau} Amorphous structures are however not
121: easily amenable to quantum calculations. Here, we assume crystalline phases
122: for both the Si and SiO$_2$ slabs. This is further motivated by the need to
123: study simple ideal systems as benchmarks, and in particular to determine how
124: the difference in the interface affects the optical properties. We however
125: note that crystalline-Si based SLs have very recently been
126: fabricated.\cite{Lu_private}
127:
128: Two simple model structures for Si/SiO$_2$ SLs are considered: (i) The
129: double-bonded model (DBM) of Herman and Batra,\cite{HermanBatra} shown in
130: Fig.~\ref{modelFIG}(a); here the Si and SiO$_2$ layers are arranged so as to
131: minimize the lattice mismatch and have relatively high symmetry, and the
132: dangling bonds at the interfaces are saturated by the addition of a single
133: double-bonded oxygen atom. (ii) A bridge-oxygen model (BOM), shown in
134: Fig.~\ref{modelFIG}(b), where the interface dangling bonds are saturated by
135: adding an oxygen atom to bridge two Si bonds. Full details of the models are
136: given in Section~\ref{StructSec}.
137: %FIGURE_HERE
138:
139: The electronic properties of both the DBM and the BOM have been studied
140: already within an empirical tight-binding framework by Tit and
141: Dharma-wardana,\cite{TitDharma} who observed the bands along the
142: high-symmetry lines of the BZ parallel to the $k_z$ axis to be essentially
143: dispersionless, in agreement with the present calculations. However, the band
144: gaps were found to be direct in both models, in contradiction with our
145: results. The DBM has been studied from first principles by Punkkinen et
146: al.\cite{Punkkinen} using a full-potential linear-muffin-tin orbital
147: approach. However, no other interface-structure was examined and thus the
148: effect of changing the interface could not be assessed; further, the optical
149: properties were not calculated. Kageshima and Shiraishi have also used
150: first-principles methods to examine different models for the Si--SiO$_2$
151: interface in oxyde-covered Si slabs,\cite{Kageshima} but the SL geometry was
152: not examined. Their calculations indicate that the interface plays an
153: important role in the light-emitting properties of the system and that, in
154: particular, interfacial Si-OH bonds are possibly related to the observed
155: luminescence. Here we present results for the electronic structure, the
156: dielectric function and the absorption coefficient of both the DBM and the
157: BOM, and focus on a comparison between the two models.
158:
159:
160: \section{Computational details}\label{CompDet}
161:
162: \subsection{Model structures}\label{StructSec}
163:
164: SLs can be fabricated by MBE;\cite{HermanSitter} a major advantage of this
165: technique, as compared to others, is that it gives sharper interfaces. In the
166: case of Si/SiO$_2$ SLs, Lu et al.\cite{LuLockwoodBaribeau} report an
167: interface thickness of about 0.5 nm, while the thickness of the SiO$_2$ and
168: Si layers vary between 2 and 6 nm. Here we assume that the interfaces are
169: perfectly sharp.
170:
171: The model Si/SiO$_2$ SL structures are constructed by alternating layers of
172: crystalline (diamond) silicon and crystalline SiO$_2$ in the ideal
173: $\beta$-cristobalite structure, viz. with a Si--O--Si bond angle of
174: 180$^{\circ}$ (space group Fd$\bar{3}$m). The (experimental) lattice
175: constants are 5.43 and 7.16 \AA, respectively, with therefore a lattice
176: mismatch of 32\%. The mismatch can be reduced to less than 7\% by rotating
177: the Si unit cell by an angle of $\pi/4$ so that the diagonal of the
178: Si-diamond structure fits, approximately, the cubic edge of the
179: $\beta$-cristobalite unit cell. The superposition of the Si and SiO$_2$
180: layers gives rise to dangling bonds on some of the interface Si atoms. These
181: can be satisfied in different ways. In the DBM, this is done by adding a
182: single O atom onto one of the Si atoms. The resulting unit cell has 23 atoms
183: [cf.\ Fig.\ref{modelFIG}(a)] and dimensions 5.43 \AA\ in the $x-y$ plane and
184: 5.43~$\times$~(1~+~$\sqrt{2}$)~=~13.11~\AA\ in the $z$ (growth) direction. In
185: this model, the Si--Si and Si--O distances, 2.35 and 1.66 \AA, respectively,
186: are realistic, but the double-bonded oxygen Si=O is not found in naturally
187: occuring silicates. In the BOM, Fig.~\ref{modelFIG}(b), an oxygen atom
188: saturates two dangling bonds provided by two different Si atoms; this model
189: contains 21 atoms. Here the Si--O--Si angle is the usual 144$^\circ$ and the
190: length of the Si--O bonds at the interface is 2.02 \AA, i.e., somewhat longer
191: than found in nature. No energy relaxation of the models have been carried
192: out. However, as will be discussed in section~\ref{gapSec}, we found gap
193: states to be present in the BOM; in order to understand this, we investigated
194: two other variants of the BOM, having Si--O--Si angles of 109 and
195: 158$^\circ$, respectively.
196:
197: The Brillouin zone for the SL, along with that for c-Si, is depicted in
198: Figure~\ref{bz}. The principal symmetry axes are indicated: the $Z-\Gamma$,
199: $X-R$ and $M-A$ directions correspond to possible growth directions where
200: confinement is expected to occur, as discussed below.
201: %FIGURE_HERE
202:
203:
204: \subsection{First-principles calculations}\label{NumericSec}
205:
206: The electronic-structure calculations were carried out with the {\sf WIEN97}
207: code,\cite{WIENref} which uses the all-electron, full-potential,
208: linearized-augmented-plane-wave (LAPW) method,\cite{Singh} within the
209: framework of density-functional theory\cite{HohenbergKohn,KohnSham} in the
210: local-density approximation (LDA).\cite{Payne} Thus, the core states in the
211: (spherical) atomic region are described using an atomic basis, while the
212: delocalized states in the interstitial region are expanded in plane waves
213: which are correctly formulated to account for the core region. The Kohn-Sham
214: (KS) basis inside the atomic spheres is expressed as an angular-momentum
215: expansion:
216: $$
217: \psi(\vec{k}_n,\vec{r}) =
218: \sum_{l,m} \left[ A_{lm}^n u_{l}(r,E_l)+B_{lm}^n
219: \left.\frac{\partial u_{l}}{\partial E}\right|_{E=E_l}
220: \right] Y_{lm}(\hat{\boldmath r}).
221: $$
222: Here the energy $E_l$ is calculated and fixed at the first cycle of the
223: self-consistent-field calculation. The KS equations are solved for a grid of
224: $\vec {k}$ points as discussed below.
225:
226: The integration of the radial secular equations was performed on a radial
227: mesh containing 581 points. The radius of the spheres for silicon and oxygen
228: atoms were set at 0.87 and 0.79 \AA, respectively, and an energy cutoff of
229: $-7.1$ Ry was used for separating core from valence electrons. This generated
230: 11~149 plane waves and the matrix size for the eigenvalue problem contained
231: 5581 elements. These two parameters correspond to an angular momentum cutoff
232: of 7.5 (see Singh\cite{Singh} for further details), a value which ensures
233: proper energy convergence of the solution.
234:
235: The DBM has nine non-equivalent atoms while the BOM has eight. The
236: irreducible wedge of the Brillouin zone is obtained after four symmetry
237: operations, which reduces considerably the computational load. The supercell
238: in the $z$ direction being 3 times longer than in the $x$ or $y$ directions,
239: only 181 $\vec {k}$ points were necessary to achieve convergence of the
240: electronic energies. However, in order to ensure convergence of the wave
241: functions needed to set up the optical matrix, the density of $\vec {k}$
242: points had to be doubled. Integration of the Brillouin zone was performed
243: using the tetrahedron method and Broyden density mixing. The self-consistent
244: calculations were performed in parallel mode (using four processors) on a
245: Silicon Graphics Origin 2000 computer system. About 10 cycles were necessary
246: to get convergence of the energy to about 10$^{-4}$ Ry.
247:
248:
249:
250: \section{Results and discussion}\label{Results}
251:
252:
253: \subsection{Band structures}\label{BandSec}
254:
255: The electronic band structures along the high symmetry axes of the tetragonal
256: Brillouin zone (cf.\ Fig.\ref{bz}) for the two models are plotted in
257: Fig.~\ref{DBMBOMband}, together with the densities of states (DOS). For the
258: BOM, unless otherwise noted, the interfacial Si-O-Si angle is the normal
259: 144$^{\circ}$. The confinement in the $z$-direction can already be inferred
260: from the essentially dispersionless character of the bands along the three
261: symmetry axes which are parallel to the axis of the SL, namely $X-R$,
262: $Z-\Gamma$ and $M-A$; this was also observed by Punkkinen et
263: al.\cite{Punkkinen} Thus, optical transitions between bands lying in the
264: growth direction are favored by the confinement.
265: %FIGURE_HERE
266:
267: The DOS of the two models differ strongly in the region of the gap, with an
268: isolated conduction band clearly visible in the gap of the BOM which is
269: absent in the DBM. The only difference between the two models lies in the way
270: that the interfacial dangling bonds are fixed; the gap state in the BOM is
271: thus connected to the bridge-bonded Si atom. The results, evidently, are
272: sensitive to the way that the interface is formed, either in experiment, or
273: in a theoretical model. This question will be discussed in more detail in
274: section~\ref{gapSec}.
275:
276: The effect of the confinement in both models is clearly visible from the DOS.
277: If we ignore the gap state in the BOM for the moment, the DOS at the Fermi
278: level has a step-function-like threshold, as opposed to bulk-silicon which is
279: known\cite{Cardona} to have a threshold depending on the photon energy
280: $\hbar\omega$ as $(\hbar\omega-E_g)^{1/2}$, where $E_g$ is the gap energy.
281: Higher transition probabilities are thus expected in Si/SiO$_2$ SLs.
282:
283: Figure~\ref{fermiband} zooms on the band structure of the two models near the
284: Fermi level. The numerical values of the various possible transition energies
285: are given in Table~\ref{gaps}. For the DBM, the valence-band maximum (VBM) is
286: at the $\Gamma$ point but the conduction band minimum (CBM) is in between the
287: $\Gamma$ and $X$ points (labeled ${\Gamma X/2}$ in Table~\ref{gaps}), giving
288: an indirect gap of 0.81 eV. For the BOM, the VBM is at $X$ while the CBM is
289: at $Z$, giving an almost null (0.01 eV) indirect band gap. However, it should
290: be remembered that the energy gaps are significantly underestimated in the
291: LDA and other implementations of the DFT. Thus the bulk-Si LDA energy gap is 0.49
292: eV, about 0.6 eV below the true value of 1.1 eV. Better (but non-rigorous)
293: estimates for the DBM and the BOM energy gaps would thus be 1.4 and 0.6 eV,
294: respectively. The unusually low value for the energy gap of the BOM is an
295: indication that the lowest conduction band is really a gap state. If this
296: state [``CB1'' in Fig.\ref{fermiband}(b)] is assumed to be an unphysical
297: artifact of the model, and temporarily ignored, the indirect gap of 1.49 eV
298: gets corrected to $\sim$2.1 eV. Thus, the energy gap would be in the near
299: infrared or in the red, consistent with
300: experiment.\cite{LuLockwoodBaribeau,Mulloni,Novikov}
301: %FIGURE_HERE
302:
303: The dispersionless character of the bands in the growth direction provides
304: evidence of the confinement effect, as noted above. However, confinement
305: alone is not sufficient to explain the enhancement of the optical properties;
306: comparison of the direct and indirect transitions for the two models can
307: clarify this question. For the DBM, the direct transitions at $\Gamma$ and
308: $\Gamma X/2$ have energies of 0.98 eV and 0.84 eV, respectively, while the
309: $\Gamma-\Gamma X/2$ indirect transition (i.e., the band gap) costs 0.81 eV.
310: Thus the {\em smallest direct--indirect gap} (SDIG) in the LDA for the DBM is
311: a mere 0.03 eV ($0.84-0.81$ eV). For the BOM (if the ``CB1'' gap state is
312: ignored), the direct transitions at $X$ and $Z$ have energies of 2.19 and
313: 1.87 eV, respectively, compared to 1.49 eV for the $Z$--$X$ indirect
314: transition, for a SDIG of only 0.35 eV. These numbers can be compared to the
315: corresponding ones for bulk silicon (in the LDA): the direct transition at
316: the $\Gamma$ point has energy 2.52 eV while the direct transition at the CBM
317: (near $3/4\Gamma X$) costs 3.20 eV. The indirect band gap is 0.49 eV, so that
318: the SDIG in this case is $\sim$2.0 eV. This is {\em much larger} than that
319: for the DBM and the BOM. The latter, as a consequence, has much better
320: optical properties than bulk Si. It is also clear that the DBM will have
321: better optical properties than the BOM (see section~\ref{OpticSec}) since the
322: SDIG of the DBM is smaller than that of the BOM.
323:
324:
325: \subsection{Comparison with tight-binding calculations}\label{TBsec}
326:
327: It is instructive to compare the present results to those obtained for the
328: same model systems by Tit and Dharma-wardana\cite{TitDharmaDBMBOM} within a
329: tight-binding (TB) framework. In a TB description,\cite{Papaconstantopoulos}
330: the wavefunctions are expressed as linear combinations of atomic orbitals and
331: the exact many-body Hamiltonian is replaced by a parametrized Hamiltonian
332: matrix whose eigenvalues and eigenvectors yield the energies and the
333: wavefunctions of the corresponding electron levels. The size of the numerical
334: problem is determined by the number of atomic orbitals chosen to describe the
335: valence electrons, and about 4 or 5 orbitals per atom are typically used.
336: This allows the study of much larger systems than is possible in
337: first-principles calculations. It is therefore important to assess the
338: validity of TB models in order to re-calibrate and improve the TB
339: parametrization. Indeed, the parameters are typically fitted to experimental
340: data or first-principles calculations for the {\em pure} crystalline phases.
341: For mixed systems, e.g., a SL, the parameters are assumed to be transferable
342: except for simple adjustments of energy reference or scaling to bond lengths,
343: etc.
344:
345: An important parameter entering TB calculations is the valence-band offset
346: (VBO), which is the difference in energy between the VBM of the two materials
347: constituting the SL.\cite{Williams} In their calculations, Tit and
348: Dharma-wardana assumed a VBO of 3.75 eV, based on the experimental offset
349: between Si and amorphous SiO$_2$. However, this value need not be appropriate
350: to the idealized DBM and BOM models.
351: %FIGURE_HERE
352:
353: In Fig.~\ref{tbvswien}(a) we compare the LDA band structure for the DBM with
354: the TB results of Tit and Dharma-wardana,\cite{TitDharmaDBMBOM} who used a
355: VBO value of 3.75 eV; Fig.~\ref{tbvswien}(c) is the corresponding plot for
356: the BOM. The agreement with this particular value of the VBO is, at best,
357: qualitative; this may indicate that the choice of VBO needs revision. In
358: order to assess this, we have done new TB calculations using different VBO
359: values. By adjusting as closely as possible the valence band structures to
360: the TB ones, better values of the VBO were found to be 0.0 eV for the DBM and
361: 1.0 eV for the BOM; the results are indicated in Fig.~\ref{tbvswien}(b) and
362: (d), where the band gaps from LDA are adjusted to the TB values to help
363: comparison. For the BOM, the gap state ``CB1'' from LDA coincides well with
364: the first conduction band obtained from TB, except in the $X$--$Z$ region of
365: the BZ. The first-principles ``CB2'', in contrast, is somewhat higher than
366: found from the TB calculations. Thus, while not perfect, the agreement with
367: the first-principles calculations improves significantly with these new VBO
368: values.
369:
370:
371: \subsection{Gap states in the BOM}\label{gapSec}
372:
373: In view of the presence of a gap state in the BOM --- labeled ``CB1'' in
374: Fig.~\ref{fermiband} --- and in order to assess the role of the interface on
375: the electronic properties, other possibilities for the interfacial Si--O--Si
376: angle were considered: (i) 109$^{\circ}$, corresponding to positioning the
377: oxygen atom on a normal silicon site of the Si lattice; here, $d_{\rm Si-O}=
378: 2.35$ \AA; (ii) 144$^{\circ}$ which corresponds to the experimental value of
379: the Si--O--Si angle, used in the calculations discussed above, yielding
380: $d_{\rm Si-O} = 2.02$ \AA; and finally (iii) 158$^{\circ}$, a value obtained
381: by relaxing the oxygen atom position, and which corresponds to a (local)
382: energy minimum; in this case, $d_{\rm Si-O} = 1.96$ \AA. The ``normal'' Si--O
383: distance in silica is 1.61 \AA, which cannot be accomodated by the
384: crystalline silicon lattice in the BOM.
385:
386: Figure~\ref{gapstatefig} shows the band structures for the three cases; here
387: we consider only the $X-R-Z$ direction, wherein lies the energy gap.
388: Evidently, the precise value of the interfacial Si--O--Si bond angle, and
389: corresponding Si--O bondlength, have a sizeable effect on the band structure.
390: In fact, the 158$^{\circ}$--BOM is, within the LDA, a metal if CB1 is not
391: assumed to be a gap state, as discussed earlier. Likewise, the
392: 144$^{\circ}$--BOM is nearly metallic. These results indicate that the
393: acceptable range of interfacial Si--O--Si angles, and the consequent longer
394: than ``normal'' Si--O distances, are probably essential issues in explaining
395: the luminescence in Si/SiO$_2$ SLs.
396: %FIGURE_HERE
397:
398:
399: \subsection{Optical properties}\label{OpticSec}
400:
401: The Kohn-Sham calculations provide matrix elements and joint densities of
402: states necessary for the calculation of the complex dielectric function
403: $\vec{\epsilon} = \vec{\epsilon}_r +i\vec{\epsilon}_i$. The absorption
404: coefficient $\alpha$, which can then be deduced from $\vec{\epsilon}$ as a
405: function of the photon energy (see below), provides a detailed picture of one
406: aspect of the optical properties of the material. In principle, the
407: luminescence requires a knowledge of the excited states of the system, with
408: electrons occupying the conduction band. Such states are not available from
409: DFT calculations. Further, the electrons in the conduction band are
410: associated with holes in the valence band. For low carrier concentrations,
411: the screening is weak and hence exciton formation occurs.\cite{knox} This
412: many-body effect is also required for a complete description of luminescence.
413: Such calculations require, e.g., the solution of the Bethe-Salpeter equation
414: for the electron-hole pair (see Chang et al., Ref.\ \onlinecite{Chang}) and
415: are still prohibitive even for our model structures. Excitonic effects will
416: thus be neglected here.
417:
418: As already noted, the band gaps are underestimated by LDA-DFT and this has to
419: be corrected if the absorption thresholds are to be realistic. In practice,
420: this can be done to a reasonable approximation by rigidly shifting all the
421: conduction bands to the appropriate energy. However, we remain within the DFT
422: framework and calculate the absorption in the Fermi-golden-rule
423: approximation. Also, we assume that the emission spectrum is similar to the
424: absorption spectrum and neglect excitonic effects. In spite of these
425: approximations, the main mechanisms of luminescence enhancement --- the
426: effects arising from the joint density of states and the matrix elements ---
427: would be correctly captured.
428:
429: Using the program written by Abt et al.\cite{AbtDraxlKnoll} as part of the
430: {\sf WIEN97} package,\cite{WIENref} the imaginary part of the dielectric
431: function has been determined; it is given by:
432: \begin{eqnarray*}
433: \epsilon_i^{\alpha}(\omega) = & \left(\frac{4\pi e^2}{m^2\omega^2}\right)
434: \sum_{v,c} \int \frac{2d\vec{k}^3}{(2\pi)^3} |\!\!<\!\!c\vec{k}
435: | {\cal H}^{\alpha} | v\vec{k}\!\!>\!\!|^2 \\ \nonumber
436: & \!\!<\!\! f_{v\vec{k}}(1\!\!-\!\!f_{c\vec{k}}) \delta(E_{c\vec{k}} - E_{v\vec{k}} -
437: \hbar \omega), \\ \nonumber
438: \end{eqnarray*}
439: where $f_{v\vec{k}}$ is the Fermi distribution and ${\cal H}^{\alpha}$ is the
440: $\alpha$-component of the electron-radiation interaction Hamiltonian in the
441: Coulomb gauge; it corresponds to the probability per unit volume for a
442: transition of an electron in the valence band state $|v\vec{k}\!\!>$ to the
443: conduction band state $|c\vec{k}\!\!>$ to occur. From Kramers-Kronig
444: relations one then deduces the real part $\epsilon_r$.\cite{Cardona} The
445: dielectric function $\vec{\epsilon}$ is defined as the square of the complex
446: refractive index $\vec{n} = n_r + i n_i$, so that
447: $$
448: \alpha(E)= 4\pi \frac{E}{hc} n_i =
449: 4\pi \frac{E}{hc} \left[\frac{(\epsilon_r^2 -\epsilon_i^r)^{1/2} -
450: \epsilon_r}{2}\right]^{1/2},
451: $$
452: with $c$ the speed of light in vacuum, $h$ Planck's constant, and $E$ the
453: photon energy.
454:
455: Figure~\ref{tousabs} shows the $z$-component of the absorption coefficients
456: for the DBM and for the three different variations of the BOM. We also give,
457: for comparison, the corresponding curves for silicon and for the ideal
458: $\beta$-cristobalite SiO$_2$ structures. The energy gaps are those directly
459: from the LDA; correcting the gaps would only affect the position of the
460: onset, but not the general aspect. The absorption coefficient for the two
461: models are similar. This similarity is, to a large extent, a ``zone folding''
462: effect, as the two model structures have identical $z$-dimensions (the growth
463: axis). However, closer inspection reveals that the DBM has better absorption
464: properties than the three BOMs: the onset of absorption is sharper, and
465: occurs earlier in the DBM than in all the BOM. This is consistent with the
466: band structure analysis above that showed the SDIG to be smaller in the DBM
467: than in the BOM ($\sim$ 0.03 vs $\sim$0.35 eV), implying a larger transition
468: probability for the former than the latter.
469:
470: In addition, the calculated SDIG for the BOM with an interfacial Si--O--Si
471: angle of 109$^{\circ}$ is 0.18 eV --- the band gap (or the indirect
472: transition) being 1.53 eV while the direct transition between $X$ and $Z$ are
473: 2.06 eV and 1.71 eV respectively --- in between the SDIG for the DBM and the
474: BOM with 144$^{\circ}$. This is in agreement with the absorption curves of
475: Fig.~\ref{tousabs}, where the absorption onset is higher for the DBM, lower
476: for the BOM with 109$^{\circ}$ and then even lower for the BOM with
477: 144$^{\circ}$ or 158$^{\circ}$. The last two give similar absorption curves
478: as can be seen from Fig.~\ref{tousabs}.
479: %FIGURE_HERE
480:
481: Thus, the four models (the DBM and the BOM with bond angles 109$^{\circ}$,
482: 144$^{\circ}$ and 158$^{\circ}$), which differ only by their interface
483: definition, give rather different optical properties. The onset of the
484: absorption curve in the BOM with Si--O--Si angles of 144$^{\circ}$ and
485: 158$^{\circ}$ do not differ that much from the one obtained from bulk-Si,
486: while the DBM and the BOM with 109$^{\circ}$, viz. the BOM with higher Si--O
487: bondlengths at the interface, give better optical properties. This is
488: additional indication that the interfacial atomic structure has to be
489: connected to the optical enhancement in the SLs, in agreement with
490: Kageshima's analysis.\cite{Kageshima} Likewise, it can be concluded from
491: Fig.\ \ref{tousabs} that bulk Si has poor optical properties compared to the
492: SLs, since the onset of absorption lags behind that for the two SLs and the
493: SDIG is much larger.
494:
495:
496: \section{Concluding Remarks}\label{ConcluSec}
497:
498: We have used first-principles calculations to (i) study the electronic and
499: optical properties of Si/SiO$_2$ SL models and (ii) examine the applicability
500: of tight-binding calculations for these systems.
501:
502: Concerning the second point, our results indicate that, for both models, the
503: VBO value of 3.75 eV used in TB calculations is excessive. In addition, the
504: direct-transition nature of the gap from the TB calculations is incorrect.
505: The present calculations indicate that the band gap in both models is
506: indirect (albeit quasi-direct), in the red or infrared region of the
507: spectrum. Confinement is further confirmed by the essentially dispersionless
508: character of the electronic band structures in the growth direction (cf.
509: Figs~\ref{DBMBOMband} and \ref{fermiband}). Our calculations indicate, also,
510: that the SLs have enhanced optical properties as compared to pure Si.
511:
512: The influence on the optical properties of the Si--SiO$_2$ interface has been
513: assessed from the absorption coefficient. This quantity depends critically on
514: the details of the interfacial structure. Realistic models are thus essential
515: for determining the electronic and optical properties. We are presently
516: studying improved models for this system.\cite{Tran}
517:
518: \vspace{0.5cm}
519:
520: {\it Acknowledgments} --
521: It is a pleasure to thank Gilles Abramovici, Ralf Meyer and Michel C\^ot\'e for
522: useful discussions. This work is supported by grants from the Natural
523: Sciences and Engineering Research Council (NSERC) of Canada and the ``Fonds
524: pour la formation de chercheurs et l'aide {\`a} la recherche'' (FCAR) of the
525: Province of Qu{\'e}bec. We are indebted to the ``R\'eseau qu\'eb\'ecois de calcul de
526: haute performance'' (RQCHP) for generous allocations of computer resources.
527:
528:
529: \begin{references}
530:
531: \bibitem{lampro} G.A. Lampropoulos and R.A. Lessard,
532: \emph{Applications of Photonic Technology}, SPIE, {\bf 3491} (1998).
533:
534: \bibitem{Zhang} P. Zhang, V.H. Crespi, E. Chang, S.G. Louie, and M.L. Cohen,
535: Nature {\bf 409}, 69 (2001).
536:
537: \bibitem{Canham} L.T. Canham, Appl. Phys. Lett. {\bf 57}, 1046 (1990).
538:
539: \bibitem{LehmannGosele} V. Lehmann, U. G\"{o}sele, Appl. Phys. Lett.
540: {\bf 58}, 856 (1991).
541:
542: \bibitem{ChuangCollinsSmith} S.-F. Chuang, S.D. Collins, R.L. Smith,
543: Appl. Phys. Lett. {\bf 55}, 675 (1989).
544:
545: \bibitem{Uehara} K. Uehara, J.S. Tse, Chem. Phys. Lett. {\bf 301},
546: 474 (1990).
547:
548: \bibitem{LuLockwoodBaribeau} Z.H. Lu, D.J. Lockwood, J.-M. Baribeau,
549: Nature {\bf 378}, 258 (1995).
550:
551: \bibitem{gnutzmann} U. Gnutzmann, K. Clausecker, Appl. Phys. {\bf 3},
552: 9 (1974).
553:
554: \bibitem{pearsall} T.P. Pearsall, J. Bevk, L.C. Feldman, J.M. Bonar,
555: J.P. Mannaerts, Phys. Rev. Lett. {\bf 58}, 729 (1987).
556:
557: \bibitem{Lu_private} Z.H. Lu, private communication.
558:
559: \bibitem{HermanBatra} F. Herman, I.P. Batra, {\it The Physics of SiO$_2$
560: and its interfaces} ed. by S.T. Pantelides, (Pergamon, Oxford, 1978).
561:
562: \bibitem{TitDharma} N. Tit, M.W.C. Dharma-wardana, J. Appl. Phys.
563: {\bf 86}, 1 (1999).
564:
565: \bibitem{Punkkinen} M.P.J. Punkkinen, T. Korhonen, K. Kokko,
566: I.J. V\"ayrynen, Phys. Stat. Sol. (b) {\bf 214}, R17 (1999).
567:
568: \bibitem{Kageshima} H. Kageshima, K. Shiraishi, Mat. Res. Soc. Symp.
569: Proc. {\bf 486}, 337 (1998).
570:
571: \bibitem{HermanSitter} M.A. Herman, H. Sitter, \emph{Molecular Beam
572: Epitaxy} (Springer-Verlag, Berlin, 1989).
573:
574: \bibitem{WIENref} P. Blaha, K. Schwarz, and J. Luitz, {\sf WIEN97},
575: Vienna University of Technology 1997. [Improved and updated Unix
576: version of the original copyrighted WIEN code, which was published by
577: P. Blaha, K. Schwarz, P. Sorantin, S.B. Trickey, Comput. Phys. Commun.
578: {\bf 59}, 399 (1990).]
579:
580: \bibitem{Singh} D.J. Singh, \emph{Planewaves, pseudopotentials and the
581: LAPW method} (Kluwer Academic, Norwell, 1994).
582:
583: \bibitem{HohenbergKohn} P. Hohenberg, W. Kohn, Phys. Rev. {\bf 136},
584: B864 (1964).
585:
586: \bibitem{KohnSham} W. Kohn, L.J. Sham, Phys. Rev. {\bf 140}, A1133 (1965).
587:
588: \bibitem{Payne} M.C. Payne, M.P. Teter, D.C. Allan, T.A. Arias, and
589: J.D. Joannopoulos, Rev. Mod. Phys. {\bf 64}, 1045 (1992).
590:
591: \bibitem{Cardona} P.Y. Yu, M. Cardona, \emph{Fundamentals of
592: Semiconductors} (Springer, New-York, 1996).
593:
594: \bibitem{Mulloni} V. Mulloni, R. Chierchia, C. Mazzoleni, G. Pucker,
595: L. Pavesi, P. Bellutti, Philo. Mag. B {\bf 80}, 705 (2000).
596:
597: \bibitem{Novikov} S.V. Novikov, J. Sinkkonen, O. Kilpel\"{a},
598: S.V. Gastev, J. Vac. Sci. Technol. B {\bf 15}, 1471 (1997).
599:
600: \bibitem{TitDharmaDBMBOM} N. Tit, and M.W.C. Dharma-wardana, Physics Letters
601: A {\bf 254}, 233 (1999).
602:
603: \bibitem{Papaconstantopoulos} D.A. Papaconstantopoulos, \emph{Handbood
604: of the Band Structure of Elemental Solids} (Plenum, New York, 1986).
605:
606: \bibitem{Williams} R. Williams, J. Vac. Sci. Technol. {\bf 14}, 1106
607: (1977).
608:
609: \bibitem{knox} R.S. Knox, \emph{Theory of excitons} in Solid State
610: Physics, (Academic Press, New-York, 1963).
611:
612: \bibitem{Chang} E.K. Chang, M. Rohlfing, S.G. Louie, Phys. Rev. Lett.
613: {\bf 85}, 2613 (2000).
614:
615: \bibitem{AbtDraxlKnoll} R. Abt, C. Ambrosch-Draxl, P. Knoll, Physica B
616: {\bf 194-196}, 1451 (1994).
617:
618: \bibitem{Tran} M. Tran, N. Tit, M.W.C. Dharma-wardana, Appl. Phys. Lett.
619: {\bf 75}, 4136 (1999).
620:
621: \end{references}
622:
623: \newpage
624:
625: \vspace{1cm}
626:
627: \begin{table}
628: \caption
629: {
630: Possible (LDA-DFT) transition energies (in eV) for the two models (DBM and
631: BOM with Si--O--Si interfacial angle of 144$^{\circ}$). CB1, CB2, and VB refer
632: to the first and second conduction band, and the valence band, respectively.
633: For the DBM, $\Gamma X/2$ designates the conduction band minimum, which lies
634: approximately half-way between $\Gamma$ and $X$ points (cf.\ Fig.\
635: \protect\ref{fermiband}). The lowest direct and indirect transitions are
636: listed; the energy gaps, which are indirect, are indicated in boldface. For
637: the BOM, CB1 can be viewed as a gap state (see text), and the ``true''
638: LDA-gap is therefore 1.49 eV; for the DBM, this gap state is absent, i.e.,
639: CB1 is the true lowest band, and the LDA gap is 0.81 eV.
640: }
641: \begin{tabular}{l|lr|lr}
642: \hline
643: & \multicolumn{2}{c|}{BOM} & \multicolumn{2}{c}{DBM} \\ \hline
644: & \mbox{} \hfill $X$ \hfill \mbox{}
645: & \mbox{} \hfill $Z$\hfill \mbox{}
646: & \mbox{} \hfill $\Gamma$\hfill \mbox{}
647: & \mbox{} \hfill $\Gamma X/2$\hfill \mbox{} \\ \hline
648: CB2 & \hspace{3pt}2.1870 & 1.4932 &
649: \mbox{} \hfill ----- & \mbox{} \hfill ----- \\
650: CB1 & \hspace{3pt}1.7447 & 0.0143 &
651: 0.9781 & 0.8050 \\
652: VB & \hspace{3pt}0.0000 & -0.3754 &
653: 0.0000 & -0.0369 \\ \hline
654: CB2$-$VB dir. & \hspace{3pt}2.1870 & 1.8686 &
655: \mbox{} \hfill ----- & \mbox{} \hfill ----- \\ \hline
656: CB2$-$VB ind. & \multicolumn{2}{c|}{CB2$_Z$--VB$_X$: {\bf 1.4932}} &
657: \mbox{} \hfill ----- &\mbox{} \hfill ----- \\ \hline
658: CB1$-$VB dir. & \hspace{3pt}1.7447 & 0.3897 &
659: 0.9781 & 0.8419 \\
660: CB1$-$VB ind. & \multicolumn{2}{c|}{CB1$_Z$--VB$_X$: {\bf 0.0143}} &
661: \multicolumn{2}{c}{CB1$_{\Gamma X/2}$--VB$_\Gamma$:
662: {\bf 0.8050}} \\ \hline
663: %\hline
664: \end{tabular}
665: \label{gaps}
666: \end{table}
667:
668:
669: \begin{figure}[t]
670: \includegraphics*[width=9cm]{Figs/DBM_BOM144.ps}
671: \caption{
672: The two model structures considered in the present study: (a) the DBM,
673: which contains 23 atoms (13 Si and 10 O), and (b) the BOM, which contains 21
674: atoms (11 Si and 10 O atoms). Both unit cells are rectangular, of size
675: $5.43\times5.43\times13.11$ \AA$^3$. The two models differ only by
676: their Si-SiO$_2$ interface.
677: }
678: \label{modelFIG}
679: \end{figure}
680:
681: \begin{figure}[t]
682: \includegraphics*[width=7cm]{Figs/bz.eps}
683: \caption{
684: Brillouin zone for the SL compared to that for the diamond structure. The
685: principal axes of symmetry used in the electronic structures calculations are
686: also shown.
687: }
688: \label{bz}
689: \end{figure}
690:
691: \begin{figure}[t]
692: \includegraphics*[width=8.7cm]{Figs/bandDBM.eps}
693: \includegraphics*[width=8.7cm]{Figs/bandBOM144.eps}
694: \caption{
695: Band structure and DOS for (a) the DBM and (b) the BOM.
696: }
697: \label{DBMBOMband}
698: \end{figure}
699:
700: \begin{figure}[t]
701: \includegraphics*[width=7cm]{Figs/ZOOMband.eps}
702: \caption{
703: Detail of the band structure near the Fermi level for the DBM and the BOM.
704: }
705: \label{fermiband}
706: \end{figure}
707:
708: \begin{figure}[t]
709: \includegraphics*[width=8.7cm]{Figs/VBO_DBM.eps}
710: \hspace{1cm}
711: \includegraphics*[width=8.7cm]{Figs/VBO_BOM144.eps}
712: \caption{
713: Comparison between TB (dotted lines) and first-principles (full lines) band
714: structures for the two models, for two different values of the VBO: (a) 3.75
715: eV and (b) 0.0 eV for the DBM; (c) 3.75 eV and (d) 1.0 eV for the BOM. The
716: LDA energy gaps were ``manually'' set to the gap from TB calculations in
717: order to facilitate the comparison.
718: }
719: \label{tbvswien}
720: \end{figure}
721:
722: \begin{figure}[t]
723: \includegraphics*[width=8.7cm]{Figs/gapSTATE.eps}
724: \caption{
725: Band structure of the BOM having three different interfacial Si--O--Si angles
726: (as indicated).
727: }
728: \label{gapstatefig}
729: \end{figure}
730:
731: \begin{figure}[t]
732: \includegraphics*[width=8.7cm]{Figs/tousabs.eps}
733: \caption{
734: Absorption curves for the two models compared to bulk Si and ideal
735: $\beta$-cristobalite SiO$_2$. The three BOM angles correspond to the
736: interfacial Si--O--Si angle in the model. The BOM curves for 144$^{\circ}$
737: and 158$^{\circ}$ overlap almost exactly; the full line corresponds to an
738: angle of 144$^{\circ}$ while the dotted line corresponds to an angle of
739: 158$^{\circ}$.
740: }
741: \label{tousabs}
742: \end{figure}
743:
744: \end{document}
745:
746: