1: %\documentclass[preprint,aps,prd]{revtex4}

2: 

3: \documentclass[twocolumn,prl]{revtex4}

4: \RequirePackage{xspace}

5: \usepackage{graphicx}

6: 

7: 

8: \def\be{\begin{equation}}

9: \def\ee{\end{equation}}

10: \def\bearr{\begin{eqnarray}}

11: \def\eearr{\end{eqnarray}}

12: \def\tc{$T_c~$}

13: \def\tcl{$T_c^{1*}~$}

14: \def\c2{ CuO$_2~$}

15: \def\ruo{ RuO$_2~$}

16: \def\lsco{LSCO~}

17: \def\bi{bI-2201~}

18: \def\tl{Tl-2201~}

19: \def\hg{Hg-1201~}

20: \def\sro{$Sr_2 Ru O_4$~}

21: \def\rc{$RuSr_2Gd Cu_2 O_8$~}

22: \def\mgb{$MgB_2$~}

23: \def\pz{$p_z$~}

24: \def\ppi{$p\pi$~}

25: \def\sqo{$S(q,\omega)$~}

26: \def\tperp{$t_{\perp}$~}

27: \def\he4{${\rm {}^4He}$~}

28: \def\ags{${\rm Ag_5 Pb_2O_6}$~}

29: \def\nxcob{$\rm{Na_x CoO_2.yH_2O}$~}

30: \def\lsco{$\rm{La_2CuO_4}$~}

31: \def\lbco{$\rm{La_{2-x}Ba_x CuO_4}$~} 

32: \def\half{$\frac{1}{2}$~}

33: \def\tst{${\rm T^*$~}}

34: \def\tch{${\rm T_{ch}$~}}

35: \def\jeff{${\rm J_{eff}$~}}

36: \def\nbc{${\rm LuNi_2B_2C}$~}

37: \begin{document}

38: \preprint{IMSc-2004/10/XX}

39: 

40: 

41: \title{Strongly Correlated Impurity Band Superconductivity in Diamond:\\

42: X-ray Spectroscopic evidence for upper Hubbard and mid-gap bands}  

43:   

44: \author{ G. Baskaran \\

45: Institute of Mathematical Sciences\\

46: C.I.T. Campus,

47: Chennai 600 113, India }

48: 

49: 

50: \begin{abstract}

51: In a recent X-ray absorption study in boron doped diamond, Nakamura et al.

52: have seen a well isolated narrow boron impurity band in non-superconducting 

53: samples and an additional narrow band at the chemical potential 

54: in a superconducting sample. We interpret the beautiful spectra as evidence 

55: for upper Hubbard band of a Mott insulating impurity band and an additional 

56: metallic `mid-gap band' of a conducting `self-doped' Mott insulator. This supports

57: the basic framework of a recent theory of the present author of strongly

58: correlated impurity band superconductivity (SCIBS) in a template of a wide-gap 

59: insulator, with no direct involvement of valence band states.

60: \end{abstract}

61: \maketitle

62: 

63: 

64: Recent discovery of superconductivity in boron doped diamond\cite{ekimov}

65: by Ekimov et al., is a pleasant surprise and superconductivity is seen in 

66: a place where one suspected it the least. This discovery has important 

67: implications in basic science and technology. These results have been 

68: confirmed\cite{takano,bustarret} at an elevated superconducting \tc in doped thin films synthesized 

69: by microwave plasma assisted chemical vapor deposition (MPCVD), a method which 

70: has advantages over the original high pressure-high temperature synthesis. 

71: This paves way for a series of new 

72: experiments\cite{bustarret,ekimov2,nakamura1,nakamura2} to unravel the mechanism 

73: of superconductivity in doped diamond as well as various 

74: anomalous properties that have been predicted theoretically\cite{dmndGB}. 

75: One such first measurements is a recent X-ray absorption spectroscopy 

76: (XAS) in the non-superconducting and superconducting dopings, by 

77: Nakamura et al\cite{nakamura1,nakamura2}.  

78: 

79: Unique among theoretical 

80: proposals\cite{dmndGB,boeri,lee,xiang,blase,russianTheory} for the mechanism 

81: of superconductivity

82: in this unusual system is a suggestion of the present author\cite{dmndGB}

83: that superconductivity

84: here is an impurity band phenomenon driven by strong electron correlation effects.

85: It has raised important questions on the origin and mechanism of superconductivity, 

86: by pointing out\cite{dmndGB} that the disordered superconducting state is 

87: in the vicinity of the Mott insulator-metal transition point\cite{shiomi,prinz}. 

88: 

89: In the XAS study in reference\cite{nakamura1,nakamura2}, a well isolated 

90: narrow boron impurity band is seen in non-superconducting samples and 

91: an additional narrow band at the chemical potential is seen in a 

92: superconducting sample. 

93: In this letter we interpret the spectra and provide evidence for an 

94: upper Hubbard band of an impurity band Mott insulator and an additional 

95: metallic `mid-gap band' of a conducting `self-doped' Mott insulator. 

96: This interpretation that is simple and natural supports the basic framework 

97: of a recent theory of the present author\cite{dmndGB} of superconductivity 

98: in a correlated impurity band in a template of a wide-gap insulator, 

99: with no direct involvement of valence band states, after the depletion of 

100: impurity states from the original broad valence band. We also discuss 

101: briefly at the end why an interpretation of the data in terms of certain 

102: lattice `relaxed' hole states may not be tenable.

103: 

104: On doping, a boron atom substitutes a carbon atom and bonds to 

105: neighboring carbon atoms through $sp^3$ hybridization. As boron has one less 

106: electron compared to carbon, a hole (an unpaired spin) is left behind, in an 

107: otherwise filled band. In the ground state this hole is bound to the boron 

108: atom in one of the three 

109: fold degenerate impurity state with a binding energy of 0.37 eV. As more and

110: more boron atoms are doped, the impurity wave functions overlap and there 

111: is an Anderson-Mott insulator to conductor transition at a critical 

112: boron concentration $n_c$, in this uncompensated semiconductor. The present 

113: author pointed out that since 

114: superconductivity observed in reference\cite{ekimov} is at a boron density, 

115: strikingly close to this critical density $n_c$\cite{shiomi,prinz,thonke}, 

116: superconductivity is likely to be 

117: an impurity band phenomenon, dominated by strong correlation physics within the

118: impurity band sub system. We will call this as the strongly correlated

119: impurity band 

120: superconductivity (SCIBS). We modeled SCIBS in terms of an effective 

121: repulsive Hubbard model and argued how resonating valence bond 

122: correlations\cite{pwascience} among spin-\half moments of neutral $B^0$ acceptor 

123: states and `self-doping', a spontaneous creation of a small and equal number of 

124: nominal $B^+$ and $B^-$ will lead to a Mott insulator to superconductor 

125: transition across a critical boron density $n_c$. In real systems there could 

126: be also compensation from donors such as nitrogen, phosphorus and

127: certain boron clusters.

128: 

129: In what follows we will interpret the recent XAS results as giving a strong

130: evidence for the presence of a correlated impurity band, an upper Hubbard   

131: band and creation of a `mid-gap band' by a process of self doping of the 

132: Mott insulator across the Mott insulator superconductor transition point.

133: 

134: In XAS, an electron from filled 1s state of either boron (B-K edge) or 

135: carbon (C-K edge)is removed and placed at an empty state above the 

136: chemical potential by

137: soft X-rays. We wish to probe the boron acceptor states close to the chemical 

138: potential. As boron has a strong $sp^3$ hybridization with carbon atoms, 

139: the impurity hole wave functions have a large radius of about 5 to 8 

140: Bohr radius. Thus it  has a large carbon wave function content. Further, 

141: in the range of concentration of interest acceptor wave functions start

142: overlapping so that XAS on C-K edge are good enough 

143: to bring out the nature of states close to the chemical potential. It is 

144: clear from the experiments that C-K edge spectra has an advantage 

145: over B-K edge, as it has less contamination from deep levels from 

146: an unavoidable density of interstitial boron atoms and boron clusters.

147: 

148: \begin{figure}

149: \includegraphics[width=9cm]{xasfig1.eps}

150: \caption{\label{Fig1}

151: The C-$K$ XAS spectra of Nakamura et al.\cite{nakamura1} is reproduced;

152: S-superconducting sample and N1,N2 non-superconducting samples.

153: The dotted line represents the energy of the valence band maximum using 

154: the band gap of 5.5 eV.  Two characteristic peaks, $H$ and $I$, are observed 

155: at about 282.6 and 284.1eV. We interpret peak $I$ as upper Hubbard band and 

156: $H$ as the conducting mid-gap band of our impurity band subsystem (see text). 

157: Chemical potential lies within the small peak $H$, the metallic mid-gap band.}

158: \end{figure}

159: Before we proceed, we would like to make two observations

160: on the remarkable X-ray absorption spectra of reference\cite{nakamura1}: 

161: i) for all three boron concentrations where experiments

162: have been done, two non superconducting samples (below $n_c$) and one

163: superconducting sample (above $n_c$), the impurity band retains its integrity

164: and its center does not move. ii) at the chemical potential a new narrow

165: band evolves and gets prominent in the superconducting state and the chemical

166: potential is, surprisingly, pinned at an energy below the top of the valence band. 

167: \begin{figure}

168: \includegraphics[width=9cm]{xasfig2.eps}

169: \caption{\label{Fig2}

170: Schematic local single particle spectral function of an electron in impurity

171: band for two cases: A) Boron impurity band Mott insulator exhibiting upper 

172: and lower Hubbard bands and B) self-doped case, obtained by increasing boron 

173: density beyond Mott transition point; here mid-gap band contains equal number

174: of holons and doublons, whose density depends on band parameters and long 

175: range coulomb interaction. Hubbard band splitting by U and chemical potential 

176: $\mu$ is also shown. In describing low energy physics, elimination of upper 

177: and lower Hubbards bands generates superexchange and we obtain a Heisenberg 

178: model for case A and a 2 species t-J model\cite{rvbmsGB} for case B.

179: A standard spectral sum rule keeps the total area under upper, lower 

180: and mid-gap bands the same in both cases.}

181: \end{figure}

182: 

183: In what follows we discuss a simple theory of the shape of X-ray absorption

184: and emission spectra of impurity band Mott insulator and conductor in the 

185: vicinity of the metal insulator transition point. 

186: 

187: 

188: In XAS, the absorption

189: cross section is determined, using Fermi golden rule and impulse approximation

190: as

191: \be

192: I_{\rm a}(\omega) \sim  \frac{2\pi}{\hbar} \sum_{f} 

193: | \langle i | \sum_{\ell} \frac{e}{c}{\bf p}_{\ell}\cdot {\bf A} | f \rangle  |^2 

194: \delta (E_f - E_i - \hbar \omega)

195: \ee

196: 

197: We are interested in a narrow range of energy just above the chemical

198: potential $\mu$ and ignore frequency dependence of  matrix elements  

199: to get 

200: \be

201: I_{\rm a}(\omega) ~~\propto~~ \rho_{\rm e}(\omega), ~~~ \hbar \omega~ >~ \mu

202: \ee

203: where $\rho_{\rm e}(\omega)$ is the local(atomic) one electron spectral 

204: function. The

205: X-ray fluorescence or emission (XES) spectra, where electrons close to 

206: the chemical potential fall into the empty 1s-core hole state and emit X-rays,

207: the emission intensity is given by

208: \be

209: I_{\rm e}(\omega) ~~\propto~~ \rho_{\rm h}(\omega), ~~~ \hbar \omega~ <~ \mu

210: \ee 

211: Here $\rho_{\rm e}(\omega)$ is the local one hole spectral function.

212: 

213: 

214: Now we discuss the profile of the electron and hole spectral functions.

215: At the energy scale of the impurity band 

216: width our discussion is rigorous and the spectral profile is qualitatively

217: correct. These spectral features of a Mott insulator and doped Mott insulator

218: are known from early Hubbard I to III approximations and recent 

219: infinite dimensional and dynamical mean field theory approaches\cite{infU}. 

220: 

221: In the atomic limit, $ U >> t_{ij}$, for a Hubbard model at half filling,

222: \be

223: \rho_{\rm e,h}(\omega) \sim~  \delta \left(\hbar \omega \pm 

224: \frac{U}{2}\right)

225: \ee

226: For our impurity Hubbard bands, disorder and finite $t_{ij}$ will broaden

227: the spectral function to give a shape shown in figure 2A. If we dope

228: the `half filled' impurity band with electron (doublons), for example by a 

229: small amount of compensation by an additional nitrogen doping, we introduce 

230: a band of mid-gap states, whose area is proportional to the

231: compensation (dopant) fraction $x$. Mid gap states of a paramagnetic state

232: are correlated many body energy levels and have no single particle description, 

233: for example in a Hartree-Fock theory. The low energy dynamics of the metallic 

234: band of mid-gap states (doublons) in this case of partial compensation by

235: nitrogen is governed by  t-J model. This is because, when we eliminate the 

236: surviving upper and lower Hubbard bands of a doped Mott insulator 

237: to obtain low energy effective Hamiltonian, superexchange gets generated. 

238: 

239: \begin{figure}

240: \includegraphics[width=9cm]{xasfig3.eps}

241: \caption{\label{Fig3} We have overlaid the correlated Hubbard and mid-gap 

242: impurity bands with the one particle valence and conduction band density of 

243: states. Location of the chemical potential $\mu$ has meaning only within 

244: the Hubbard and mid-gap bands. As far the valence band it has no holes. 

245: XAS measures the conduction band, the upper Hubbard band and the 

246: mid-gap bands shown as thick lines. }

247: \end{figure}

248: In a recent paper we suggested\cite{rvbmsGB}, based on experimental results 

249: and theoretical

250: considerations on Mott insulator to metal transition, that the conducting

251: side close to the first order Mott transition point should be viewed as 

252: a `self doped' Mott insulator. As the doping is internal, number of 

253: negative and positive (doublons and holons) carriers are the same. The carrier

254: density is determined by long range coulomb interactions and band parameters.

255: To describe the low energy dynamics of such a doped Mott insulator, which 

256: preserves superexchange in the conducting state, we introduced a 2 species

257: t-J (or 2t-J) model. The schematic spectral profile, figure 2B, is the same 

258: as regular t-J model except that the mid-gap metallic band contains equal 

259: number of positive (holon) and negative (doublon) carriers (with reference 

260: to  neutral Mott insulator). 

261: As mentioned earlier, in reality there can be a small amount

262: of compensation from impurities such as nitrogen, phosphorus and 

263: certain boron clusters. The resonating valence bond physics of 

264: superconductivity in our self-doped Mott insulator\cite{rvbmsGB} is partly 

265: contained in the t-J-U model\cite{tJU}, but with key differences.

266: 

267: X-ray absorption spectroscopy measures the upper Hubbard band and 

268: the conducting `mid-gap band'. X-ray emission spectroscopy measures

269: the lower Hubbard band the conducting `mid-gap band'. However, in the

270: case of acceptor impurity band, such as ours, emission from the

271: close by valence band will mask the small signal from the impurity bands,

272: making XES problematic. In the same vein, if we had a donor band

273: XAS, rather than XES will have problems.

274: 

275: 

276: Now we discuss the experimental result of reference\cite{nakamura1}, 

277: figure 1, in the

278: light of  the above discussion. The spectra are for three boron dopings, two 

279: non-superconducting and one superconducting cases. The conduction band edge 

280: and the impurity bands are clearly visible. In the non-superconducting and

281: superconducting samples the impurity band remains sharp, denoted by the

282: peak $I$. This is our upper Hubbard band. In the superconducting state we 

283: have the additional band, at the chemical potential, which we identify as the 

284: metallic mid-gap band of self doped carriers, denoted by peak $H$. 

285: From the experimental result, we see that the 

286: peak to peak distance of the upper Hubbard band and mid-gap band,

287: $\frac{U}{2} \approx$ 0.7 to 1 eV. This value of U $\sim$ 1 to 2 eV

288: for our impurity orbital is easily rationalized, as the impurity Bohr

289: radius of our deep level acceptor state is increased only by a factor of 

290: 5 to 8 from atomic Bohr radius.

291: 

292: Since U $\sim 1~eV$ is large compared to acceptor binding energy 

293: $E_B \approx 0.37~eV$, an important question in the conducting state about 

294: possible hole transfer to top of the valence band arises. That is, soon

295: after metallization increased charge fluctuations in the impurity band

296: will introduce the Hubbard U repulsion energy among the holes by an 

297: amount $\sim x~U$ prompting hole transfer to the valence band. 

298: Here $x$ density of self doped carriers. In addition to this energy loss 

299: on metallization, there are two sources of energy gain; i) the well known 

300: delocalization energy, proportional to the impurity band width and ii) 

301: a Madelung energy gain\cite{rvbmsGB} within the impurity band arising from 

302: long range 

303: coulomb interaction $E_{\rm Md}$. This energy is missed in a Hubbard model,

304: which ignores long range coulomb interaction; it is this which also 

305: determines the self doping fraction $x$.  When $ xU - E_{\rm Md} > E_B$, 

306: it will be energetically favorable to drain holes from the impurity band 

307: to the top of the valence band. 

308: 

309: In the experiments\cite{nakamura1}, the chemical potential neither falls on 

310: top of

311: the valence band nor inside the impurity band, but seems to be pinned to

312: the metallic mid-gap band. This means all low energy charge and spin actions 

313: are taking place within the impurity band. This is a good support for our 

314: impurity band modeling\cite{dmndGB} of the superconductivity phenomena. 

315: One may safely say that there are no hole fermi surface from a finite density 

316: of valence band holes, as is assumed in current electron-phonon 

317: theories\cite{boeri,lee,xiang,blase}.

318: 

319: At this point it is also important to make a qualitative difference

320: between a renormalized one particle band such as the conduction or 

321: valence band and a correlated many body bands such as upper, lower and

322: mid-gap Hubbard bands. A finite value of spectral functions in these many 

323: body bands are basically projected one particle density of states rather 

324: than one particle level densities. We emphasize this in figure 3, where we 

325: overlay the correlated Hubbard and mid-gap impurity bands on the 

326: one particle valence and conduction band density of states. Location

327: of the chemical potential $\mu$ has meaning only within the Hubbard and 

328: mid-gap bands. As far the valence band it has no holes. 

329: This figure also shows the mid-gap band, upper Hubbard band and the

330: conduction band that will be seen in the X-ray absorption spectra by a 

331: thick line. 

332:  

333: The authors of the XAS experiments\cite{nakamura1} suggest that what we call 

334: as the mid-gap

335: band at the chemical potential could arise from certain lattice relaxed hole

336: states in the absence of simple rigid band shifts on doping. Unfortunately

337: the energy of the hole is shifted in the opposite direction and it does not

338: look like a relaxed state but a `strained metastable state' which could 

339: decay down to the top of the valence band (energy of a hole is always 

340: positive and measured from the chemical potential). How, on metallization such

341: a meta stable hole states will collectivized and become stable ? That too in 

342: the background of an impurity band that continues to be present, as the XAS

343: spectra indicates. One also has to answer why relative density of such lattice 

344: relaxed hole states (area under mid-gap band relative to impurity band) should 

345: increase with boron doping.    

346: 

347: It is also not clear whether a phase separation of boron impurities will

348: explain the observed XAS spectra. Indeed, some careful analysis has been 

349: done in the experiments\cite{nakamura1} to eliminate the possibility of 

350: precipitation.

351: 

352: In all our discussion we did not discuss disorder explicitly. As emphasized

353: in our original paper, the uncompensated case under discussion has a 

354: commensurate filling of the impurity band, with an average of one hole per 

355: boron atom. In such a situation

356: Mott-Hubbard correlations overwhelm; disorder to a large extent is 

357: `irrelevant' in the renormalization group sense. That is, once Mott localization 

358: sets in, Anderson localization has lesser role to play. Further, spin charge 

359: decoupling, an inevitable consequence in the Mott transition region,

360: also protects current carrying holon and doublon states from Anderson localization 

361: effects. As we move far away from commensurate situation,

362: for example by partial compensation, Anderson localization effects and notions

363: such as mobility edge become important.

364: 

365: 

366: 

367: To summarize, it is indeed nice to have diamond, a relatively `simple' 

368: system as a template for the fascinating impurity band Mott phenomena and 

369: superconductivity. A large band gap and a relatively large acceptor binding 

370: energy of a simple acceptor boron, nicely isolates out and makes the impurity

371: Hubbard bands and mid-gap bands visible in X-ray absorption spectroscopy. 

372: It will be very nice to increase the energy resolution and explore some

373: of the beautiful many body issues related to our mechanism\cite{dmndGB} of 

374: SCIBS by X-ray absorption and emission spectroscopy, for a range of boron doping

375: and also partial compensation.

376: 

377: I thank Manas Sardar for useful discussions and bringing to my attention 

378: reference\cite{nakamura2}. 

379: 

380: \begin{references}

381: \bibitem{ekimov} E.A. Ekimov, V.A. Sidorov, E.D. Bauer, N.N. Mel'nik,

382: N.J. Curro, J.D. Thompson, S.M. Stishov,

383: Nature, {\bf 428}, 542(2004). 

384: \bibitem{takano} Y. Takano, M. Nagao, I. Sakaguchi, M. Tachiki, T. Hatano,

385: K. Kobayashi, H. Umezawa, H. Kawarada,

386: cond-mat/0406053., Appl. Phys. Lett. {\bf 85}, (2004) (in press).

387: \bibitem{bustarret} E. Bustarret, J. Kacmarcik, C. Marcentaut, 

388: E. Gheeraert, C. Cytermann, J. Marcus and T. Klein, cond-mat/0408517

389: \bibitem{ekimov2}V. A. Sidorov, E.A. Ekimov, S.M. Stishov, E.D. Bauer 

390: and J.D. Thompson, cond-mat/04

391: \bibitem{nakamura1} J. Nakamura, T. Oguchi, N. Yamada, K. Kuroki, K. Okada,

392: Y. Takano, M. Nagao, I. Sakaguchi, H. Kawarada, R.C.C. Perera and D.L. Ederer

393: cond-mat/04010144.

394: \bibitem{nakamura2} J. Nakamura, E. Kabasawa, N. Yamada, Y. Einaga, D. Saito,

395: H. Isshiki, S. Yugo, R.C.C. Perera,

396: cond-mat/0407438, accepted to Phys. Rev. B.

397: \bibitem{dmndGB} G. Baskaran, cond-mat/0404286.

398: \bibitem{boeri} L. Boeri, J. Kortus, O.K. Andersen, cond-mat/0404447.

399: \bibitem{lee} K.W. Lee and W.E. Pickett, cond-mat/0404547.

400: \bibitem{xiang} H.J. Xiang, Z. Li, J. Yang, J.G. Hou, Q. Zhu,

401: cond-mat/0406446.

402: \bibitem{blase} X. Blase, Ch. Adessi, D. Connetable, cond-mat/0407604.

403: \bibitem{russianTheory}Y.G. Pogorelov, V.M. Loktev, cond-mat/0405040 

404: \bibitem{shiomi} H. Shiomi, Y. Nishibayashi, N. Fujimori, Jap. J. App. Phys.,

405: {\bf 30} 1363 (1991)

406: \bibitem{prinz} T. Tshepe, J.F. Prinz and M.J.R. Hoch, Diamond Relat.

407: Mater., {\bf 8} 1508 (1999) 

408: \bibitem{thonke} K. Thonke, Semicond. Sci. Tech. {\bf 18}, S20 (2003) 

409: \bibitem{pwascience} P.W. Anderson, Science {\bf 235} 1196 (87);

410: G. Baskaran, Z. Zou and P.W. Anderson,  Sol.  State Commm. {\bf 63} 973 (87);

411: P. W. Anderson, G. Baskaran, T. Hsu and Z. Zou, Phys. Rev. {\bf B 58} 2790 (87);

412: G. Baskaran and P.W. Anderson, Phys. Rev. {\bf B37} 580 (88); 

413: G. Baskaran, Physica Scripta, {\bf T27} 53 (89).

414: \bibitem{infU} J. Hubbard, Proc. Roy. Soc. (London), {\bf A276} 283 (1963);

415: {\bf A277} 238 (1964) and W. Metzner and D. Vollhardt, Phys. Rev. Lett.,

416: A. Georges, G. Kotliar, W. Krauth, M. Rozenberg, Rev. Mod. Phys., {\bf 68}

417: 13  (1996)

418: \bibitem{rvbmsGB} G. Baskaran, Phys. Rev. Lett., {\bf 90}~197007 (2003)

419: \bibitem{tJU} F.C. Zhang, Phys. Rev. Lett., {\bf 90}~207002 (2003)

420: 

421: 

422: \end{references}

423: \end{document}

424: ~

425: 

426: