1:
2: %\documentclass[preprint,aps]{revtex4}
3: %\documentclass[aps,prl,preprint,groupedaddress]{revtex4}
4: %\documentclass[twocolumn,showpacs,preprintnumbers,
5: %amsmath,amssymb,groupedaddress]{revtex4}
6: \documentclass[twocolumn,showpacs,preprintnumbers,
7: amsmath,amssymb,superscriptaddress]{revtex4}
8:
9: %\newcommand{\mwcd}[1]{\marginpar{\protect\scriptsize \protect\setlength
10: % {\baselineskip}{8pt} to editor- MS revision: #1}}
11: %\setlength {\marginparwidth}{7.0cm}
12:
13: \usepackage{graphicx}% Include figure files
14:
15: \begin{document}
16:
17: \title
18: {Molecular electronics exploiting sharp structure in the
19: electrode density-of-states.
20: Negative differential resistance and
21: Resonant Tunneling in a poled molecular layer on
22: Al/LiF electrodes.
23: }
24:
25:
26:
27: %
28: \author
29: {
30: Z.H. Lu}
31: \affiliation{Department of Materials Science and Engineering, University of Toronto\\
32: 184 College Street, Toronto, Ontario M5S 3E4
33: }
34: %
35: \author
36: { M.W.C. Dharma-wardana$^*$
37: }
38: %
39: \affiliation{
40: National Research Council of Canada, Ottawa,Canada. K1A 0R6\\
41: }
42:
43:
44: \author
45: { R.S. Khangura}
46: %
47: \affiliation{Department of Materials Science and Engineering, University of Toronto\\
48: 184 College Street, Toronto, Ontario M5S 3E4
49: }
50: %
51: \author
52: {Marek Z. Zgierski}
53: \affiliation{
54: National Research Council of Canada, Ottawa,Canada. K1A 0R6\\
55: }
56: %
57: \author
58: {Douglas Ritchie}
59: \affiliation{
60: National Research Council of Canada, Ottawa,Canada. K1A 0R6\\
61: }
62: %\date{01 October 2002}
63: \date{\today}
64: %
65: %
66: \begin{abstract}
67: Density-functional calculations are used to clarify the role of an
68: ultrathin $LiF$ layer on $Al$ electrodes used in molecular
69: electronics. The $LiF$ layer creates a sharp density of states (DOS),
70: as in a scanning-tunneling microscope
71: (STM) tip. The sharp DOS,
72: coupled with the DOS of the molecule leads
73: to negative differential resistance (NDR).
74: Electron transfer between oriented molecules
75: occurs {\it via}
76: resonant tunneling. The $I-V$ characteristic for a
77: thin-film of tris (8-hydroxyquinoline)-
78: aluminum ($AlQ$) molecules, oriented using
79: electric-field poling, and sandwiched between two $Al/LiF$ electrodes
80: is in excellent agreement with theory.
81: This molecular device presents a new paradigm
82: for a convenient, robust, inexpensive alternative to
83: STM or mechanical
84: break-junction structures.
85: \end{abstract}
86: \pacs{PACS Numbers: 05.30.Fk, 71.10.+x, 71.45.Gm}
87: %\vspace{0.5in}
88: %\hspace{0.5in} see file /usr/people/chandre/tekstuff/xc2d/ms1.tex
89: %
90: \maketitle
91: %
92: %\section{Introduction.}
93: {\it Introduction.}
94: An intense effort has been directed to
95: the practical realization of molecular electronics (ME)
96: \cite{gimzew,datta,ratner},
97: since conventional silicon technology is
98: approaching the limit of ``Moore's
99: law''~\cite{moore}. The advantages of ME are
100: related to properties unique to
101: molecules. Molecules are
102: nanostructures
103: with near-perfect topological linkages. The disruption
104: in surface topology involved in microfabrication
105: is a limiting factor in using inorganic
106: materials such as GaAs\cite{surfacedis}.
107: The energy scales associated with molecules (unlike, say, quantum dots)
108: make them capable of operating at room temperature.
109: Mechanically controllable break junctions\cite{read} and molecular contacts
110: using scanning-tunneling microscope (STM) tips have been
111: considered\cite{gimzew}.
112: However, molecule-atom contacts
113: may sometimes take disruptive pathways that lead to fragmentation
114: of the metal contacts or the
115: molecule itself\cite{critique,turak}. Assuming non-disruptive
116: contacts could be made, an additional step of coherently
117: connecting many molecules together
118: requires exotic tools.
119: Hence the future of ME depends on surmounting the
120: theoretical and practical
121: challenge of realizing robust, reproducible, inexpensive devices.
122:
123: In this study we show both theoretically and experimentally that
124: a practical, robust realization of
125: molecular-NDR characteristics is possible by exploiting two important
126: features, {\it viz}., (i) the
127: special characteristics of the density of states (DOS) of $LiF$
128: thin films deposited on $Al$, and
129: (ii) The possibility of aligning molecules by
130: electric poling, i.e, cooperatively aligning the molecules along the
131: field direction when
132: an external electric field is applied.
133: Examples of electric poling are common in the field of
134: liquid
135: crystals and non-linear optical polymers\cite{polling}.
136: The experimental $I-V$ characteristic
137: inclusive of NDR effects is found to be in agreement
138: with our calculations.
139: %
140: %\begin{figure}
141: %$\includegraphics*[width=8.0cm, height=5.0cm]{m1new.ps}
142: %caption
143: %{
144: %Schematics of the device structure
145: %}
146: %\label{device}
147: %fig 1
148: %$\end{figure}
149: %
150: %
151:
152: {\it The device structure.--}
153: The experimental structure
154: was fabricated in a cluster tool having a central
155: distribution chamber, a load-lock chamber, a plasma cleaning chamber, a
156: sputtering chamber, an organic molecular-growth chamber, and a
157: metallization chamber.
158: Figure 1 shows the layer sequence used.
159: A 2"x2" $Si$ substrate with $\sim$ 70 nm of
160: oxide was
161: cleaned by oxygen plasma and then transferred to the
162: metallization chamber
163: for electrode deposition. The bottom $Al$ electrode (1 mm wide and 2 mm
164: spacing between adjacent electrodes) was fabricated by thermal
165: evaporation through a shadow mask in the metallization chamber. The top
166: electrode was
167: similarly deposited but the electrode
168: lines run orthogonal to the bottom ones. The interception area
169: (1mm x1mm) forms one individual device. One hundred such testing devices
170: were fabricated on a 2"x2" silicon substrate. After the
171: deposition of the bottom electrode,
172: the wafer was transferred to the organic molecular-growth
173: chamber for $AlQ$ deposition. Electronic grade $AlQ$ was
174: purchased from Kodak.
175: $AlQ$ films were fabricated
176: using a low-temperature K-cell with an alumina crucible. Thicknesses of
177: 50-200 nm have been tried and 100 nm $AlQ$
178: was optimal with good electrical properties and high
179: yield ($>$90\%). The 100 nm $AlQ$ thin-film
180: devices used $Al$ electrodes having
181: an ultrathin layer of ( $\sim 0.2$-0.3 nm) $LiF$ between
182: $Al$ and $AlQ$ \cite{grosea}.
183: The crucial role of the
184: $LiF$ will be discussed below.
185: %The left-electrode (Al/LiF/AlQ) interface may be some what different
186: % from the right
187: %electrode as the LiF is deposited on the AlQ layer followed by the Al layer.
188: After the
189: deposition of the top electrode, the device was sealed off
190: from the ambient using a 120 nm
191: silicon oxide passivation layer.
192: This passivation layer is very critical to obtaining
193: reproducible results.
194: %
195: %
196: \begin {figure}
197: \includegraphics*[width=7.0cm, height=8.0cm]{fig1.ps}
198: \caption
199: { Top: Schematic of the molecular device, and voltage
200: profile in the device for poled AlQ. The ''bottom'' Al-electrode
201: is at the right end.
202: Lower panel: $I-V$ characteristics of the device without
203: electric poling and with poling where NDR structures arise.
204: }
205: \label{fz}
206: \end{figure}
207: %
208: %
209:
210: {\it Measurements and results.--}
211: The current-voltage ($I-V$) measurements were
212: conducted using a HP4140B meter with a Materials Development probe
213: station. Figure 1 (lower panel) shows $I-V$ characteristics of one typical
214: device before poling (dashed line) at 298 K and after poling (solid
215: line) at 255 K. The electric poling was made by sweeping the applied
216: voltage from the starting poling voltage $V_p$ onwards
217: with a dwell time of 1 second per voltage point,
218: in the same mode of $I-V$ measurement. The starting $V_p$ is
219: $\sim 13$ V at 298K
220: and $\sim 10$ V at 255K. The maximum poling voltage is $\sim 18$ V. It is quite
221: clear that after electric poling, sharp NDR peaks near -3 V and +3V were
222: developed. The peak structures are very similar to those observed on a
223: single molecule or self-assembled molecular monolayer\cite{datta,read}.
224: We also
225: found that the NDR peak can be erased by passing a high current at a
226: higher positive bias-voltage, and the erased peak can be re-generated by
227: poling. It can be repeated many times on some devices. Because there are
228: no strong primary bonds between molecules, the molecular alignment may
229: be readily destroyed by thermal effects. Nevertheless we found that over
230: 90\% of the device showed similar characteristics.
231: Similar, sporadic and unexplained observations of NDR from AlQ has
232: been reported in the literature\cite{kim}.
233: Here we present controlled
234: observations whose origin is quantitatively explained.
235:
236:
237: {\it Theoretical considerations.--}
238: The electrode (see Fig. 1) chemical potentials are
239: $\mu_L$, and $\mu_R$, with
240: $-|e|V=\mu_L-\mu_R$, where $V=V_R-V_L$ is the applied potential.
241: The
242: electron flow involves (i) electron injection
243: into the molecular layer at the right, denoted by $R\to M$,
244: and described by a transmission function $T_{RM}$ which is essentially
245: the joint density of states of the electrode surface and the molecular
246: layer, weighted by a matrix element.
247: (ii) Transfer of carriers from
248: molecule to molecule, $M\to M$, described by $T_{MM}$ and finally,
249: (iii) transfer from the
250: the molecular layer to the left electrode, $M \to L$, described by $T_{ML}$.
251:
252: We begin by a qualitative discussion.
253:
254: The interaction between $Al$ surfaces and organic molecules like
255: $AlQ$ has been studied extensively\cite{mason,gennip}. The $AlQ$
256: molecules placed directly in contact with the $Al$ surface is known
257: to undergo reactions (e.g.,between $Al$ and the
258: quinolate-$O$). Clearly, an important role of the $LiF$ is
259: in shielding the molecules against such reactions.
260: Various indirect interfaces, e.g, $Al/X/AlQ$, where X may be,
261: say, $LiF$, $CaF_2$, $SiO_2$ etc., have been examined. It has been found
262: that a few atomic layers of $LiF$ separating the $AlQ$ from the
263: $Al$ electrode produces a striking improvement in the $I-V$
264: characteristics and lowers the threshold for
265: electron injection. Although this system has been extensively investigated,
266: a clear explanation of how $LiF$ helps has not been forthcoming.
267:
268: We show that the $Al/LiF$-electrode DOS is like a $\delta$-function
269: sitting on the slowly varying DOS of the $Al$ substrate. The DOS of the
270: $AlQ$ molecule is also found to have a strong sharp feature (SF) in the
271: unoccupied manifold of states (UMS). At the appropriate bias, electrons
272: are injected into the SF of the UMS. Unlike injection into the
273: lowest unoccupied state (``LUMO'') which is affected by the Coulomb blockade
274: and the need for molecular rearrangement, injection to a higher energy
275: UMS couples only weakly to the ground-state electron distribution.
276: The injected electrons rapidly migrate in the poled molecular film by
277: resonant tunneling. Hence the inter-molecular transmission function
278: $T_{MM}$ is essentially
279: unity. Finally, electrons arriving at the drain electrode
280: are in an excited state above the Fermi energy of the $Al$ electrode,
281: and can resonantly tunnel into the slowly varying
282: unoccupied DOS of the $Al$ substrate. This resonant process
283: also involves a negligible potential drop if the intervening $LiF$ layer
284: is atomically thin. Thus essentially all the potential drop occurs
285: at the injection electrode (see Fig.1, top panel) where there is also
286: an ``empty'' region (not shown in the figure)
287: between the $LiF$ surface and the first
288: $AlQ$ layer. All these lead to an effective tunneling length $s$ which
289: will appear in the matrix element connecting the electrode states
290: and the $AlQ$ states.
291: Thus the potential profile in
292: Fig. 1 is rather schematic and does not show the $LiF-AlQ$ tunneling gap
293: which is found from our calculations (see below) to be about $\sim 0.1$ nm,
294: while the $LiF$ layer is about 0.2-0.3 nm.
295: We have, to
296: leading order,
297: \begin{equation}
298: T_{RM}(E,V)=(2\pi)^2|A_{RM}(V)|^2\rho_R(E-eV)\rho_M(E)
299: \end{equation}
300: Here $A_{RM}$ is the matrix element connecting the molecule
301: and the electrode, and $\rho_R,\, \rho_M$ are the DOS of the
302: right electrode (R) and the molecule (M) respectively.
303: Given our previous conclusion that
304: $T_{MM}$, and $T_{ML}$ are resonant processes of the
305: order of unity, the current is effectively determined by
306: $T_{RM}$. The bias dependence of $A_{RM}(V)$ can be neglected
307: only if two sharp features in the two DOS functions are involved.
308: However, the bias dependence of $A_{RM}(V)$ is
309: important, and this can be included by evaluating the matrix element
310: as in Lang\cite{lang}.
311:
312:
313: A standard Fermi Golden-rule analysis, or a
314: Landauer-B\u{u}ttiker approach may be used to write an expression for the
315: current $I$ in terms of the transmission function $T(E,V)$.
316: \begin{equation}
317: I=2\sigma_0\int_{-\infty}^{\infty}dET(E,V)[f(E-\mu_R)-f(E-\mu_L)]\\
318: \end{equation}
319: Here $\sigma_0$ is the quantum of conductance $e^2/h=1$
320: in atomic units, which corresponds to a resistance of 12.9 $K\Omega$
321: in conventional units.
322: The Fermi factors
323: are effectively step functions even at room temperature.
324: Thus the
325: dominant contribution to the current is given by:
326: \begin{equation}
327: \label{i-v}
328: I(V)=A\int_{\mu_L}^{\mu_R}dE\,\rho_R(E-eV)\rho_M(E)
329: e^{-2s\{2(E-W)+eV\}^{1/2}}
330: \end{equation}
331: where $A$ is a numerical factor and $\mu_L=E_F$ is the Fermi energy,
332: while $\mu_R=E_F+eV$.
333: Here $s$ is an effective tunneling length
334: linking the molecule and the $Al/LiF$ electrode, and
335: $W$ is the workfunction\cite{lang}. These are
336: treated as parameters of the problem. The main effect of higher temperatures
337: comes in via a convolution of the $T_{MM}$ which involves
338: molecular Debye-Waller effects. We have neglected such effects.
339:
340: The left electrode was made by depositing $LiF$ and then $Al$
341: on $AlQ$, and hence follows the reverse of the fabrication sequence
342: used for the right electrode. However, the electronic processes
343: in the device are essentially symmetric, with slightly
344: different values for $s$, $W$ etc. When the bias is
345: reversed, the carrier processes reverse direction.
346: This is different to the situation
347: in $Au-thiol\,$-STM devices where the thiol is chemically attached
348: to the $Au$ electrode. Then
349: NDR is obtained by having two potential drops at the
350: two electrodes\cite{datta}. Also, unlike in the $Au-thiol\,$-STM
351: system, the $AlQ$ molecules are shielded from the $Al$ electrode by
352: the $LiF$ overlayers, and the main potential drop
353: occurs at the injector $LiF$ dielectric layer, at the empty region
354: between the $LiF$ surface and the the first $AlQ$ moleculer layer.
355: This is estimated to be about $\sim 1.2$ Angstroms and
356: plays the role of a tunneling length.
357:
358:
359: {\it Density of states of the Al/LIF electrode.--}
360: Bulk $LiF$ and bulk $Al$ both have FCC structures and are very nearly lattice
361: matched. While $Al$ is a very good metal, $LiF$ is a large-bandgap insulating
362: ionic crystal. We have carried out first-principles density functional
363: calculations to simulate $Al/(LiF)_n$ where the number of atomic planes
364: $n$ was varied from 1 to 4 to give LiF-overlayers. The crystallographic
365: details of the experimental $Al$ surface is unknown. We have carried out
366: calculations with the $LiF$ layers arranged along [001] as well
367: as along [111] directions with very
368: similar results.
369: The $Al$ substrate was modeled with 6 planer layers of $Al$ in the growth
370: direction, and periodically continued in the $x-y$ directions.
371: The last layer of $Al$ and the $LiF$ layer were geometry optimized using
372: standard
373: energy minimization methods available with the VASP plane-wave
374: code\cite{vasp}. The simulation cell (SC) included five vacuum layers
375: to separate the $Al$ substrate from the $LiF$.
376: DOS calculations were done using an
377: $11\times 11\times 11$ Monkhorst-Pack $k$-sampling scheme.
378: The $Al$ atomic layer
379: adjacent to the $LiF$ suffers virtually no reconstruction,
380: while the $F$ and $Li$ atoms
381: which were initially coplaner in the (001) layers,
382: reconstruct in opposite directions, with the $F^-$
383: moving inwards, towards the $Al$ layer, while the $Li^+$ ions move outwards,
384: along the growth axis.
385: This reconstruction is very small ($\sim 0.1 A$).
386: Similar effects, as well as
387: metal-induced surface states, have been reported in $Cu/LiCl$ and
388: related systems recently\cite{arita}.
389: A crucial effect of the $LiF$
390: layer on $Al$ is seen in Fig.2. The DOS of $Al$ by itself,
391: $LiF$ by itself, and in the combined structure
392: $Al/(LiF)_n$ are shown in Fig.~\ref{figdos}, all referred to the
393: same Fermi level
394: set to zero for clarity. It is clear that a sharp structure (similar to that
395: in the DOS of an STM tip) is produced in the $Al/(LiF)_n$ system
396: if $n$ is small. Thus, if an $Al$ electrode has an {\it atomically thin}
397: overlayer of $LiF$, we have an injection DOS similar to those
398: created using
399: mechanical break junctions or STM-tip molecular contacts.
400: %
401: %
402: \begin{figure}
403: \includegraphics*[width=9.0cm, height=10.0cm]{fig2.ps}
404: \caption
405: {The top panel contains the occupied-state DOS of bulk-Al and bulk-LiF,
406: with both Fermi levels set to zero. The lower two panels show the
407: DOS with a single monoatomic layer of LiF, and 2 atomic layers of LiF,
408: deposited on Al and structure optimized by total-energy minimzation.
409: }
410: \label{figdos}
411: %fig 1
412: \end{figure}
413:
414: The calculation of the transmission function $T(E,V)$ requires the DOS
415: of the $AlQ$ molecule as well. Here a cluster-type calculation is
416: appropriate, while it would {\it not} be appropriate for the $Al$ metallic
417: layers.
418: The electronic structure of the $AlQ$ molecule and the $AlQ^-$
419: negative ion were calculated using the Gausssian-98 code at the
420: B3LYP/3-21G* level\cite{acro,G98}. These calculations included geometry
421: optimization together with total energy minimization and state of the
422: art gradient corrected exchange-correlation functionals. Density of states
423: (DOS) curves were
424: constructed using a 0.25 eV Lorentz broadening of the
425: one-particle eigenenergies.
426: These calculations reveal a highest occupied
427: molecular orbital (HOMO) at -5.02 eV and a lowest unoccupied molecular
428: orbital (LUMO) at -1.79 eV, while the optical gap was previously
429: calculated by us~\cite{marekgap} to be 2.7 eV.
430:
431: {\it Comparison of calculated and observed $I-V$ characteristics.--}
432: The analysis leading to Eq.~\ref{i-v} showed that the $I-V$ characteristic
433: would depend on the main features of the DOS of the molecular film, i.e.,
434: the molecule itself as we have a simple chemically uncoupled array of
435: molecules.
436: In fig.~\ref{expth}, top panel, we present the ``tip-like''
437: DOS of the $Al/LiF$ layer,
438: with the
439: background $Al$ density of states subtracted off (e.g, bottom panel
440: of Fig.2 minus
441: the $Al$-slab DOS from the top panel of Fig.2). The DOS of the
442: $AlQ$ molecule is shown in the middle panel.
443: The bottom panel shows
444: the experimental $I-V$ data at 255 K
445: compared with the theoretical result obtained from Eq.~\ref{i-v}.
446: The experimental $I-V$ agreed with the theory for
447: a workfunction $W\sim 4.6-4.9 eV $, and
448: with the effective tunneling length $s\sim 0.8-1.3$
449: Angstroms.
450: The numerical factor $A$ was $\sim 0.001$.
451: %
452: %
453: \begin{figure}
454: \includegraphics*[width=7.0cm, height=8.0cm]{fig3.ps}
455: \caption
456: { Top:
457: the $Al/LiF$ ''tip'' like DOS with the $Al$ DOS subtracted out.
458: Middle panel shows the DOS of the neutral $AlQ$ molecule.
459: Lower panel: $I-V$ characteristics of the device
460: compared with theory.
461: }
462: \label{expth}
463: \end{figure}
464: %
465:
466: {\it Discussion.--}
467: In the usual picture ({\it not used by us}$\,$)
468: of electron transport across the A/AlQ/B structures
469: under bias, it is assumed that an electron is injected
470: from the source
471: into the LUMO of the $AlQ$ molecule,
472: converting it to a transient $AlQ^{-*}$ anion.
473: This involves Coulomb blockade,
474: rearrangement of bond lengths,
475: bond angles etc., to give the actual $AlQ^-$ anion.
476: The carrier then hops to a
477: neighboring $AlQ$ molecule and moves towards the drain electrode.
478: There is no resonant transfer possible not only because the $AlQ$
479: molecules are not properly oriented, but also because the
480: available LUMO states are not in resonance with the $AlQ^-$
481: eigenstate containing the carrier electron. The $I-V$ characteristics
482: show no NDR effects.
483:
484: In contrast, in the structure discussed here,
485: the molecules in the electric-field poled $AlQ$ layer form an oriented
486: array. The electron is injected to an unoccupied high-energy state
487: by the sharp DOS structure of the $Al/LiF$ electrode.
488: The electron reaches the drain electrode by inter-molecular resonant transfer.
489: Since only high-energy unoccupied states are invoked, this process
490: does not involve significant Coulomb blockade or
491: bond-rearrangement bottlenecks.
492: The device is robust, easy to make and works at ambient
493: temperatures.
494:
495: {\it Conclusion--} We have shown, experimentally and theoretically, that
496: a robust room-temperature molecular device, showing $I-V$ characteristics
497: similar to those obtained in STM-based molecular devices,
498: can be fabricated using $Al/LlF$ electrodes and a poled molecular film.
499: New insight into the role of $LiF$ overlayers in electrode processes is
500: also presented.
501:
502:
503:
504: \begin{references}
505: \bibitem{none}
506: $\,^*$ electronic address: chandre.dharma-wardana@nrc.ca \\
507: \bibitem{gimzew}
508: For references to early work,
509: see N. D. Lang, Phys. Rev. B, {\bf 52}, 5335 (1995)
510: \bibitem{datta}
511: S. Datta et al., Phys. Rev. Lett. 79, 2530 (1997);
512: M. Di Ventra et al., {\it Ibid} {\bf 84}, 979 (2000);
513: E. G. Emberley et al., Phys. Rev. B {\bf 58}, 10911 (1998);
514: H. Mehrez et al., {\it Ibid} {\bf 65},195419 (2002);
515: Yong Chen et al., Appl. Phys. Lett., {\bf 82}, 1610 (2003)
516: %
517: \bibitem{ratner}
518: A. Nitzan and M. A. Ratner, Science, {\bf 300}, 1384 (2003)
519: %
520: \bibitem{moore}
521: www.intel.com/research/silicon/mooreslaw.htm
522: %
523: \bibitem{surfacedis}
524: See, e.g., {\it Compound Semiconductor Surface Passivation
525: and Novel Device Processing}, edited by H. Hasegawa et al.,
526: (MRS, Warrendale, Pennsylvania, 1999).
527: %
528: \bibitem{read}
529: M. A. Reed et al., Science {\bf 278}, 252 (1997).
530: %J. Chen et al., Science 286, 1550
531: %(1999).
532: %
533: \bibitem{critique}
534: see R. F. Service, Science, {\bf 302}, 556 (2003)
535: \bibitem{turak}
536: A. Turak et al.,
537: Appl. Phys. Lett. 81, 766 (2002).
538: %
539: \bibitem{polling}
540: M.C. Petty et al., Eds. {\it An Introduction to
541: Molecular Electronics}, (Edward Arnold, London, 1995).
542: %
543: \bibitem{grosea}
544: D. Grozea et al., Appl. Phys. Lett. 81, 3173 (2002).
545: %
546: \bibitem{kim}
547: S. K. Kim et al., Mol. Cryst. Liq. Cryst. {\bf 377}, 133 (2002)
548: %
549: \bibitem{mason}
550: M. G. Mason et al, J. App. phys. 89, 2756 (2001) and references
551: there in.
552: %
553: \bibitem{gennip}
554: W. J. H. van Gennip et al, J. Chem. Phys. {\bf 117}, 5031 (2002)
555: %
556: \bibitem{lang}
557: N. D. Lang, Phys. Rev. B {\bf 34}, 5947 (1986)
558: %
559: \bibitem{vasp}
560: G. Kress, J. Furthmuller and J. Hafner,
561: see http://cms.mpi.univie.ac.at/vasp/
562: %
563: \bibitem{arita}
564: M. Kiguchi et al., Phys. Rev. Lett, {\bf 90}, 196803 (2003)
565: %
566: \bibitem{acro}
567: For the acronyms, basis sets, etc.,
568: see:A.D. Becke, J. Chem. Phys., 98,
569: 5648 (1993); C. Lee, W. Yang and R.G. Parr, Phys. Rev. B, 37,785 (1988).
570: %
571: \bibitem{G98}
572: {\it Gaussian 98}, Revision A.9, M. J. Frisch et al, Gaussian Inc.,
573: Pittsburgh, PA (1998).
574: %
575: \bibitem{marekgap}
576: M. W. C. Dharma-wardana and Marek Z. Zgierski, J. Opt. A: Pure
577: Appl.Opt. 4, 278 (2002).
578:
579: \end{references}
580: %
581: %
582: \end{document}
583:
584:
585:
586:
587:
588:
589: