1: \documentclass[twocolumn,showpacs,preprintnumbers,amsmath,amssymb,prb]{revtex4}
2: \usepackage[dvips]{color,graphics,epsfig,rotating}
3: \usepackage{graphicx}% Include figure files
4: \usepackage{dcolumn}% Align table columns on decimal point
5: \usepackage{bm}% bold math
6:
7: \begin{document}
8:
9: \title{Bonding of H in O vacancies of ZnO}
10:
11: \author{H. Takenaka and D.J. Singh}
12: \affiliation{Materials Science and Technology Division and
13: Center for Radiation Detection Materials and Systems, Oak Ridge National
14: Laboratory, Oak Ridge, Tennessee 37831-6032}
15:
16: \date{\today}
17:
18: \begin{abstract}
19: We investigate the bonding of H in O vacancies in ZnO using
20: density functional calculations. We find that H is anionic
21: and does not form multicenter bonds with Zn in this compound.
22: \end{abstract}
23:
24: \pacs{71.20.Ps,71.55.Gs}
25:
26: \maketitle
27:
28:
29: ZnO is of importance as an extremely
30: fast inorganic scintillator material when
31: doped with Ga or In. It is useful in alpha particle detection, e.g. for
32: devices such as deuterium-tritium neutron generators used in
33: radiography.
34: \cite{lehmann,luckey,batsch,derenzo,neal,bourret}
35: In this application, H treatment has been shown to improve properties.
36: ZnO has also attracted much recent attention
37: motivated by potential applications as an oxide electronic
38: material,
39: \cite{hoffman,nomura,suzuki,look-02}
40: and in optoelectronic and lighting applications.
41: \cite{nicoll,bagnall,look-99,look,huang}
42: H has been implicated as playing an important role in the electronic
43: properties for ZnO for those applications as well.
44: \cite{vandewalle,cox,hoffmann}
45: From a fundamental point of view,
46: the behavior, and especially bonding of H, is of great interest;
47: H plays an exceptionally important role in chemistry, and
48: shows unique bonding characteristics. For example, it readily forms
49: compounds where it behaves as a halogen ion and forms structures similar to
50: fluorides, such as rutile,
51: perovskite, rocksalt, etc.,
52: \cite{bertheville,gingle,yvon}
53: and at the same time readily occurs
54: a cation in other chemical
55: environments. Polar covalent bonds involving H and hydrogen bonds
56: are central to
57: much of organic chemistry as well the properties of important
58: substances such as water.
59: \cite{pauling-book}
60: Thus the recent report by Janotti and Van de Walle (JV) that H forms
61: a new type of strong
62: multicenter bond in O vacancies in ZnO is of wide ranging interest.
63: \cite{janotti}
64:
65: In this paper, we present standard local density approximation (LDA)
66: calculations of the
67: electronic properties and structure of H containing O vacancies in ZnO.
68: We do not find the multicenter covalent bonds claimed by JV,
69: and instead characterize the behavior of H as quite conventional in that
70: it occurs as an anion on the anion site in a polar crystalline environment.
71:
72: Our calculations were done within the standard local density approximation
73: using the general potential linearized augmented planewave (LAPW) method,
74: including local orbitals. \cite{singh-book,singh-lo} Specifically,
75: we constructed a 72 atom 3x3x2 wurtzite supercells of ZnO, with one O
76: atom removed and replaced by H. The calculations were done using the
77: bulk lattice parameters of ZnO, but the internal coordinates of all
78: atoms in the supercell were fully relaxed.
79: No symmetry was assumed in the relaxations.
80: The LAPW method is an all electron method that makes no shape approximations
81: to either the potential or charge density.
82: It divides space into non-overlapping atom centered spheres and
83: an interstitial region.
84: The method then employs accurate basis sets appropriate
85: for each region. \cite{singh-book}
86: In the present calculations,
87: LAPW sphere radii of
88: 2.0 $a_0$, 1.6 $a_0$ and 1.2 $a_0$ were used for Zn, O, and H
89: respectively, along with a basis set consisting of more than
90: 8500 LAPW functions and local orbitals. Convergence tests were done
91: with a larger basis set of approximately 12000 functions, but no
92: significant changes were found. The relaxations were done without
93: any imposed symmetry, with a 2x2x2 special ${\bf k}$-point zone
94: sampling. A sampling using only the $\Gamma$ point was found to
95: yield slightly different quantitative results, due to the limited size of
96: our supercell, but would lead to the same conclusions.
97: The calculated value of the internal parameter is $u$=0.119, which
98: agrees almost exactly with the experimental value.
99: The densities of states used to analyze the electronic
100: properties were obtained using the linear tetrahedron method based
101: on eigenvalues and wavefunctions at 36 ${\bf k}$-points in the half
102: zone ({\bf k} and -{\bf k} are connected by time reversal).
103:
104: In our relaxed structure for a neutral cell, we find that H occurs
105: in a slightly asymmetric position, with three Zn neighbors at 2.03 \AA,
106: and one Zn neighbor (the one along the $c$-axis direction) at 2.17 \AA.
107: For the singly charged cell, we obtain a very similar result,
108: specifically three Zn neighbors at 2.02 \AA, and the apical Zn at
109: 2.21 \AA. In the following, we focus on the neutral cell except as
110: noted.
111:
112: \begin{figure}[tbp]
113: \vspace{0.5cm}
114: \includegraphics[width=3.4in,angle=0]{figure-1.eps}
115: \caption{Projection of the electronic density of states
116: onto the $s$ and $p$ components inside the H LAPW sphere, radius
117: 1.2 $a_0$, for
118: the 72 atom neutral cell.
119: The two peaks identified by JV as bonding and antibonding
120: combinations are indicated by ``B" and ``A", respectively. The
121: Fermi level lies in the conduction bands. The position of the
122: valence band maximum is denoted by ``VBM" (note that the LDA
123: strongly underestimates the 3.3 eV band gap of ZnO).
124: }
125: \label{dos}
126: \end{figure}
127:
128: Fig. \ref{dos} shows the projection of the electronic density
129: of states onto the H LAPW sphere, of radius 1.2 $a_0$.
130: The Fermi energy for our neutral cell lies at a position one electron per
131: cell into the conduction bands, corresponding to the valence difference of
132: one between O and H.
133: As may be seen, there are two prominent peaks in the H component of the
134: density of states, one, denoted
135: ``B", at $\sim$ -8 eV with respect to the Fermi level
136: (-6 eV with respect to the valence band maximum), and the other,
137: denoted ``A", high in
138: the conduction bands at $\sim$ 6 eV.
139: JV identified these peaks, ``B"
140: and ``A", respectively, as the bonding and antibonding combinations of
141: metal and H orbitals giving rise to the multicenter bond.
142: In addition, there is significant H $s$ character distributed over
143: the valence bands, especially near the valence band maximum.
144: We note that the very large bonding-antibonding splitting of 14 eV
145: implied by the
146: assignment of JV
147: indicates
148: extremely strong covalent bonds, which is somewhat surprising
149: considering the Zn-H distances.
150: In any case, such a large covalent gap would imply that the bonding
151: and antibonding states should have mixed character. In other words, the
152: bonding state should be of roughly half H $s$ character, while
153: the remaining H 1$s$ character should occur in the unoccupied
154: antibonding level, so that the occupancy of the H 1$s$ orbital should
155: be roughly 1 e, and certainly significantly less than 2 e.
156:
157: To analyze the bonding further it is convenient to compare the
158: charge density with an ionic model, as was done for some alanates.
159: \cite{aguayo,singh-alanate}
160: As mentioned, H is known to enter
161: some solids as an anion, including tetragonal MgH$_2$. \cite{yu}
162: ZnH$_2$ also exists though it is not as
163: well characterized. \cite{wiberg}
164: Furthermore, the simplest
165: hydride, LiH, is of this ionically
166: bonded type and includes H$^-$ anions coordinated by six
167: metal atoms. \cite{kunz,dovesi}
168: In these hydrogen
169: anion based materials, the negative H ion is stabilized by the Ewald
170: field.
171: In fact, the importance
172: of the Ewald field is one of the essential differences between
173: chemistry in solid state and the chemistry of molecules. The
174: long range Coulomb interaction stabilizes ionic bonding for species
175: that would generally be largely covalent in small molecules, and
176: in particular stabilizes anions such as O$^{2-}$
177: and H$^-$, which are common in solid state chemistry but much less
178: so in gas phase molecules.
179: This stabilization by the Ewald field is reflected in the
180: variability of the effective size of H in
181: crystal structure data for anionic hydrides. \cite{morris,shannon,pauling}
182: % In fact, in some anionic hydrides, the effective H$^-$ radius more similar
183: % to Br$^-$ or I$^-$ than to F$^-$.
184: In view of the common occurrence of H as an anion in many metal hydrides,
185: it would not
186: be surprising if H$^{-}$ were stabilized by the
187: Ewald field of an anion vacancy in a polar
188: crystal such as ZnO.
189: Thus we consider an ionic model, based on the charge density of a
190: H$^-$ ion stabilized by the Ewald field, as simulated by a Watson
191: sphere, \cite{watson}
192: as in Ref. \onlinecite{singh-alanate}.
193: For such a H$^-$ ion, 0.525 e out of 2.0 e, i.e. $\sim$ 26\% of the charge,
194: is inside a radius of 1.2 $a_0$, so the majority of the charge is outside.
195: Because of the small sphere radius used for H in our calculations
196: the amount of charge inside the sphere
197: is only weakly dependent on the
198: Watson sphere radius, which reflects the environment.
199: For a non-spin polarized neutral H in free space as described in the LDA,
200: 0.378 e (38\%) would be inside a radius of 1.2 $a_0$, showing that there is a
201: strong dependence on the charge state, though not precise proportionality.
202: Fig. \ref{int} shows the integral of the H $s$ character as
203: a function of energy normalized by the fraction of the H$^-$ charge inside
204: a 1.2 $a_0$ sphere (0.525/2). Over the valence band region the $p$ contribution
205: is less than 2\%, and the $d$ contribution is less than 0.2\%.
206: The conduction bands, which are more Zn $sp$
207: derived, show a larger proportion of H $p$ character, as may be seen
208: in Fig. \ref{dos}.
209: Thus the charge inside the sphere, which comes
210: from the occupied valence bands, is mainly
211: due to H $s$ states, and not from orbitals on neighboring atoms.
212:
213: \begin{figure}[tbp]
214: \vspace{0.5cm}
215: \includegraphics[width=3.4in,angle=0]{figure-2.eps}
216: \caption{Integration of the projection of the H $s$ projected density
217: of states as in Fig. \ref{dos} normalized according to the
218: fraction of charge inside a 1.2 $a_0$ sphere for a Watson sphere
219: stabilized H$^-$ anion (see text).
220: }
221: \label{int}
222: \end{figure}
223:
224: Using the ionic model for H$^-$, i.e. incorporating the factor of 0.525/2
225: as the fraction of charge inside the H LAPW sphere, and integrating,
226: one finds that the peak ``B" contains $\sim$ 0.8 H $s$ electrons. Integrating
227: over the remaining valence bands brings the H $s$ count to 2.0 electrons,
228: i.e. what is expected for H$^-$.
229: This leads to an interpretation of the electronic structure, where the
230: peak ``B" comes from the H 1$s$ state. This hybridizes
231: with valence band states, which have mixed Zn $d$ and O $p$ character.
232: The second peak ``A", 14 eV higher, is then the H $s$ resonance.
233: This is a very reasonable position for the resonance
234: of H$^-$. In particular,
235: the H$^{--}$ resonance of atomic H$^-$ is at $\sim$ 14.5 eV.
236: \cite{walton,taylor,schulz}
237: JV emphasized the shape of the charge density associated with the states
238: in the peak ``B" and argued for bonding based partly on real space images
239: of this charge. As mentioned,
240: in our projected density of states we find that this peak
241: contains 0.8 $s$ electrons (i.e. 40\%
242: H $s$ character), which would
243: be consistent with a bonding orbital. However, the hybridization is with
244: other occupied states, and when the integration is done over all the valence
245: bands, we find 2 $s$ electrons, consistent with H$^-$. We emphasize that
246: mixing of occupied states does not contribute to the energy, and that
247: such hybridizations do not constitute bonds.
248: Our calculated binding energy relative H$_2$
249: and a relaxed neutral supercell with an O vacancy is 87 kJ/mol H. \cite{lda-h2}
250: This may be an overestimate due to LDA errors, \cite{lda}
251: but in any case is much
252: smaller than the binding that would be suggested by a 14 eV bonding-antibonding
253: splitting.
254:
255: We also calculated the positron wavefunction and lifetimes
256: for ZnO with an O vacancy and with the H containing O vacancy.
257: This was done using the LAPW method in the full inverted self-consistent
258: Coulomb potential plus
259: the
260: correlation and enhancement factors of Boronski and Nieminen
261: \cite{boronski}
262: as calculated from the full charge density.
263: We obtain a bulk positron lifetime for
264: ZnO of 144 ps, which is
265: at the lower end of the experimental range.
266: Reported experimental values are
267: 151 ps (Ref. \onlinecite{bauer}),
268: 170 ps (Ref. \onlinecite{tuomisto}),
269: 141 ps - 155 ps (Ref. \onlinecite{dutta}), and
270: 182 ps (Ref. \onlinecite{chen}).
271: Significantly, positrons,
272: which are positively charged, tend to localize in voids and in sites
273: that are favorable for cations, and localize weakly if at all in anion
274: sites, due to the unfavorable Coulomb potential. We do not find positron
275: localization at the O vacancy in our ZnO supercell, indicating that the
276: O is indeed an anion as expected, nor do we find positron localization or a
277: significant lifetime increase in the cell with a H containing O vacancy.
278: We also find no significant change in lifetime for H in an
279: O vacancy within a charged supercell
280: with one electron removed.
281: In contrast, we obtain a bound positron state for Zn vacancies,
282: both with and without H, reflecting
283: the fact that Zn is on a cation site.
284: The calculated lifetime in a supercell with a Zn
285: vacancy is 212 ps, while with a H filled Zn vacancy we obtain
286: 175 ps (in this case H bonds to a single adjacent O to form a
287: hydroxyl like unit with H-O bond length of 1.01 \AA).
288: \cite{rmt-note}
289:
290: To summarize, we have performed density
291: functional calculations for ZnO supercells with both
292: empty and H filled O vacancies. Based on an analysis of the electronic
293: structure we do not
294: find any evidence for hydrogen multicenter bonds, but rather
295: find that H occurs as H$^-$.
296:
297: We are grateful for helpful discussions with
298: L.A. Boatner, J.S. Neal, and L.E. Halliburton.
299: This work was supported by the Department of Energy,
300: Office of Nonproliferation Research and Development, NA22.
301:
302: \begin{references}
303:
304: \bibitem{lehmann}
305: W. Lehmann, Solid State Electronics {\bf 9}, 1107 (1966).
306:
307: \bibitem{luckey}
308: D. Luckey, Nucl. Instr. and Meth. {\bf 62}, 119 (1968).
309:
310: \bibitem{batsch}
311: T. Batsch, B. Bengtson, and M. Moszynski,
312: Nucl. Instr. and Meth. {\bf 125}, 443 (1975).
313:
314: \bibitem{derenzo}
315: S.E. Derenzo, E. Bourret-Courchesne, M.J. Weber, and M.K. Klintenberg,
316: Nucl. Instr. Meth. Phys. Res. A {\bf 537}, 261 (2005).
317:
318: \bibitem{neal}
319: J.S. Neal, L.A. Boatner, N.C. Giles, L.E. Halliburton, S.E. Derenzo,
320: and E.D. Bourret-Courchesne,
321: Nucl. Inst. Meth. Phys. Res. A {\bf 568}, 803 (2006).
322:
323: \bibitem{bourret}
324: E.D. Bourret-Courchesne, and S.E. Derenzo,
325: 2006 IEEE Nuclear Science Symposium Conference Record N40-5, 1541
326: (2006).
327:
328: \bibitem{nomura}
329: K. Nomura, H. Ohta, K. Ueda, T. Kamiya, M. Hirano, and H. Hosono,
330: Science {\bf 300}, 5623 (2003).
331:
332: \bibitem{hoffman}
333: R.L. Hoffman, B.J. Norris, and J.F. Wager,
334: Appl. Phys. Lett. {\bf 82}, 733 (2003).
335:
336: \bibitem{suzuki}
337: A. Suzuki, T. Matsushita, T. Aoki, Y. Yoneyama, and M. Okuda,
338: Jpn. J. Appl. Phys., Part 2 {\bf 38}, L71 (1999).
339:
340: \bibitem{look-02}
341: D.C. Look, D.C. Reynolds, C.W. Litton, R.L. Jones, D.B. Eason, and G. Cantwell,
342: Appl. Phys. Lett. {\bf 81}, 1830 (2002).
343:
344: \bibitem{nicoll}
345: F.H. Nicoll, Appl. Phys. Lett. {\bf 9}, 13 (1966).
346:
347: \bibitem{bagnall}
348: D.M. Bagnall, Y.F. Chen, Z. Zhu, T. Yao, S. Koyama, M.Y. Shen, and T. Goto,
349: Appl. Phys. Lett. {\bf 70}, 2230 (1997).
350:
351: \bibitem{look-99}
352: D.C. Look, J.W. Hemsky, and J.R. Sizelove,
353: Phys. Rev. Lett. {\bf 82}, 2552 (1999).
354:
355: \bibitem{look}
356: D.C. Look, Mater. Sci. Eng. B {\bf 80}, 383 (2001).
357:
358: \bibitem{huang}
359: M.H. Huang, S. Mao, H. Feick, H.Q. Yan, Y.Y. Wu, H. Kind, E. Weber,
360: R. Russo, and P. Yang, Science {\bf 292}, 1897 (2001).
361:
362: \bibitem{vandewalle}
363: C.G. Van de Walle, Phys. Rev. Lett. {\bf 85}, 1012 (2000).
364:
365: \bibitem{cox}
366: S.F.J. Cox, E.A. Davis, S.P. Cottrell, P.J.C. King, J.S. Lord,
367: J.M. Gil, H.V. Alberto, R.C. Vilao, J. Pironto Duarte,
368: N. Ayres de Campos, A. Weidinger, R.L. Lichti, and S.J.C. Irvine,
369: Phys. Rev. Lett. {\bf 86}, 2601 (2001).
370:
371: \bibitem{hoffmann}
372: D.M. Hoffmann, A. Hofstaetter, F. Leiter, H. Zhou, F. Henecker,
373: B.K. Meyer, S.B. Orlinskii, J. Schmidt, and P.G. Baranov,
374: Phys. Rev. Lett. {\bf 88}, 045504 (2002).
375:
376: \bibitem{bertheville}
377: B. Bertheville, T. Herrmannsdorfer, and K. Yvon,
378: J. Alloys Compd. {\bf 325}, L13 (2001).
379:
380: \bibitem{gingle}
381: F. Gingle, T. Vogt, E. Akiba, and K. Yvon,
382: J. Alloys Compd. {\bf 282}, 125 (1999).
383:
384: \bibitem{yvon}
385: K. Yvon, Chimia {\bf 52}, 613 (1998).
386:
387: \bibitem{pauling-book}
388: L. Pauling, {\em Nature of the Chemical Bond} (Cornell University Press,
389: Ithaca, 1960).
390:
391: \bibitem{janotti}
392: A. Janotti, and C.G. Van de Walle, Nature Materials {\bf 6}, 44 (2007).
393:
394: \bibitem{singh-book}
395: D.J. Singh and L. Nordstrom, {\em Planewaves, Pseudopotentials and the
396: LAPW Method, 2nd. Ed.} (Springer, Berlin, 2006).
397:
398: \bibitem{singh-lo}
399: D. Singh, Phys. Rev. B {\bf 43}, 6388 (1991).
400:
401: \bibitem{yu}
402: R. Yu and P.K. Lam, Phys. Rev. B {\bf 15}, 8730 (1988).
403:
404: \bibitem{morris}
405: D.F.C. Morris and G.L. Reed,
406: J. Inorg. Nucl. Chem. {\bf 27}, 1715 (1965).
407:
408: \bibitem{shannon}
409: R.D. Shannon, Acta Cryst.{\bf A32}, 751 (1976).
410:
411: \bibitem{pauling}
412: L. Pauling, Acta Cryst., Sect. B {\bf 34}, 746 (1978).
413:
414: \bibitem{aguayo}
415: A. Aguayo and D.J. Singh, Phys. Rev. B {\bf 69}, 155103 (2004).
416:
417: \bibitem{singh-alanate}
418: D.J. Singh, Phys. Rev. B {\bf 71}, 216101 (2005).
419:
420: \bibitem{wiberg}
421: E. Wiberg, W. Henle, and R. Bauer,
422: Z. Naturforsch. B {\bf 6}, 393 (1951).
423:
424: \bibitem{kunz}
425: A.B. Kunz and D.J. Mickish, Phys. Rev. B {\bf 11}, 1700 (1975).
426:
427: \bibitem{dovesi}
428: R. Dovesi, C. Ermond, E. Ferrero, C. Pisani, and C. Roetti,
429: Phys. Rev. B {\bf 29}, 3591 (1984).
430:
431: \bibitem{watson}
432: R.E. Watson, Phys. Rev. {\bf 111}, 1108 (1958); we used a H anion stabilized
433: by a sphere of radius 1.62 \AA.
434:
435: \bibitem{walton}
436: D.S. Walton, B. Peart, and K. Dolder, J. Phys. B {\bf 3}, L148 (1970).
437:
438: \bibitem{taylor}
439: H.S. Taylor and L.D. Thomas, Phys. Rev. Lett. {\bf 28}, 1091 (1972).
440:
441: \bibitem{schulz}
442: G.J. Schulz, Rev. Mod. Phys. {\bf 45}, 378 (1973).
443:
444: \bibitem{lda-h2}
445: The value used for the energy of H$_2$ is -2.294 Ry;
446: D.J. Singh, M. Gupta, and R. Gupta, Phys. Rev. B {\bf 75}, 035103 (2007).
447:
448: \bibitem{lda}
449: The LDA generally overbinds solids, and this leads to overestimates
450: of the binding of H in solids, typically in the range of 0 to 20 kJ/mol H;
451: H. Smithson, C.A. Marianetti, D. Morgan, A. Van der Ven, A. Predith,
452: and G. Ceder, Phys. Rev. B {\bf 66}, 144107 (2002);
453: S.V. Halilov, D.J. Singh, M. Gupta, and R. Gupta, Phys. Rev. B {\bf 70},
454: 195117 (2004);
455: K. Miwa and A. Fukumoto, Phys. Rev. B {\bf 65}, 155114 (2002).
456:
457: \bibitem{boronski}
458: E. Boronski and R.M. Nieminen, Phys. Rev. B {\bf 34}, 3820 (1986);
459: see also M.J. Puska and R.M. Nieminen, Rev. Mod. Phys. {\bf 66}, 841 (1994);
460: P. Schultz and K.G. Lynn, Rev. Mod. Phys. {\bf 60}, 701 (1988).
461:
462: \bibitem{bauer}
463: G. Bauer, W. Anwand, W. Skorupa, J. Kuriplach, O. Melikhova, C. Moisson,
464: W. von Wenckstern, H. Schmidt, M. Lorenz, and M. Grundmann,
465: Phys. Rev. B {\bf 74}, 045208 (2006).
466:
467: \bibitem{tuomisto}
468: F. Tuomisto, V. Ranki, K. Saarinen and D.C. Look,
469: Phys. Rev. Lett. {\bf 91}, 205502 (2003).
470:
471: \bibitem{dutta}
472: S. Dutta, M. Chakrabarti, S. Chattopadhyay, D. Jana, D. Sanyal, and A. Sarkar,
473: J. Appl. Phys. {\bf 98}, 053513 (2005).
474:
475: \bibitem{chen}
476: Z.Q. Chen, S. Yamamoto, M. Maekawa, A. Kawasuso, X.L. Yuan, and
477: T. Sekiguchi, J. Appl. Phys. {\bf 94}, 4807 (2003).
478:
479: \bibitem{rmt-note}
480: The calculations for H in a Zn vacancy were done using a different
481: set of LAPW sphere radii, as was necessary due to the short H-O
482: bond length.
483:
484: \end{references}
485:
486: \end{document}
487:
488:
489: