1: %\documentclass[twocolumn,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}

2: %\documentclass[preprint,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}

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

4: %\documentclass[preprint,aps,draft]{revtex4}

5: \documentclass[twocolumn,prb,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}

6: \usepackage{graphicx}

7: \usepackage{dcolumn}

8: \usepackage{bm}

9: 

10: %\nofiles

11: 

12: \begin{document}

13: 

14: %\preprint{APS/123-QED}

15: 

16: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

17: 

18: \title{Temperature and magnetization dependent band-gap

19:   renormalization and optical many-body effects in diluted magnetic

20:   semiconductors}

21: 

22: \author{Ying Zhang} 

23: 

24: \author{S. Das Sarma} 

25: 

26: \affiliation{Condensed Matter Theory Center, Department of Physics,

27:   University of Maryland, College Park, MD 20742-4111}

28: 

29: \date{\today}

30: 

31: \begin{abstract}

32: We calculate the Coulomb interaction induced density, temperature and

33: magnetization dependent many-body band-gap renormalization in a

34: typical diluted magnetic semiconductor ${\rm Ga}_{1-x}{\rm Mn}_x{\rm

35:   As}$ in the optimally-doped metallic regime as a function of carrier

36: density and temperature. We find a large ($\sim 0.1 eV$) band gap

37: renormalization which is enhanced by the ferromagnetic transition.  We

38: also calculate the impurity scattering effect on the gap narrowing. We

39: suggest that the temperature, magnetization, and density dependent

40: band gap renormalization could be used as an experimental probe to

41: determine the valence band or the impurity band nature of carrier

42: ferromagnetism.

43: \end{abstract}

44: 

45: \pacs{78.30.Fs;75.10.-b;75.10.Lp;71.45.Gm}

46: 

47: \maketitle

48: 

49: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

50: \section{Introduction}

51: 

52: A central open question~\cite{DMS} in the physics of carrier-mediated

53: ferromagnetism in diluted magnetic semiconductors (DMS) is the nature

54: of the semiconductor carriers mediating the ferromagnetic interaction,

55: in particular, whether the carriers (i.e. the holes in the

56: well-studied Ga$_{1-x}$Mn$_x$As system) are itinerant band carriers

57: (i.e. holes in the GaAs valence band for Ga$_{1-x}$Mn$_x$As) residing

58: in the (valence) band of the parent semiconductor or impurity band

59: carriers residing in a narrow impurity band within the band gap of the

60: semiconductor. This issue is of fundamental

61: significance~\cite{chattopadhyay} since the valence band~\cite{dietl}

62: or impurity band~\cite{berciu} competing pictures respectively imply

63: competing RKKY-Zener~\cite{dietl2} or double exchange~\cite{akai}

64: mechanisms for DMS ferromagnetism.  In spite of intense experimental

65: and theoretical activity in Ga$_{1-x}$Mn$_x$As over the last ten

66: years, the key issue of valence band versus impurity band carriers

67: mediating the DMS ferromagnetism has remained controversial even in

68: the optimally doped, ($x \sim 0.05$) nominally metallic, high-$T_c$

69: ($T_c \sim 150 - 200K$, the ferromagnetic transition temperature or

70: the Curie temperature) Ga$_{1-x}$Mn$_x$As material. For example, {\it

71:   ab initio} first principles band structure

72: calculations~\cite{sanvito} typically indicate a strong-coupling

73: narrow impurity band behavior whereas the extensively used

74: phenomenological mean-field description, parameterized by a single

75: effective impurity moment-carrier spin (``pd'') exchange

76: interaction~\cite{DMS, dietl, berciu, dassarma2}, leads to reasonable

77: quantitative agreement with experimental results in the metallic ($x

78: \sim 0.05$) Ga$_{1-x}$Mn$_x$As requiring a relatively weak exchange

79: coupling (and therefore, a weak perturbation of the GaAs valence band)

80: between the Mn moments and the valence band holes. Similarly, optical

81: absorption spectroscopic data in GaMnAs were first

82: interpreted~\cite{hirakawa} using an impurity band theoretic

83: description~\cite{hwang}, but later it was shown~\cite{yang} that the

84: same data could also be explained as arising from the valence band

85: picture. Given the great complexity of many competing interaction and

86: disorder effects in the DMS Hamiltonian, it is increasingly clear that

87: this important question cannot be settled purely by theoretical means,

88: and an unambiguous experimental test is warranted.

89: 

90: In this paper we propose an experimental (optical) measurement of the

91: carrier-induced many-body band gap renormalization (BGR) in

92: Ga$_{1-x}$Mn$_x$As for the definitive resolution of this controversy;

93: in particular, we establish theoretically that the BGR in

94: Ga$_{1-x}$Mn$_x$As should be {\em extremely} large ($\sim 0.1 eV$) and

95: a strong function of hole density if the carriers are indeed GaAs

96: valence band holes, allowing for a clear distinction between the

97: (valence band) RKKY-Zener and the (impurity band) double exchange

98: mechanisms for DMS ferromagnetism. In addition, we calculate the

99: temperature (as well as density) and hole spin-polarization dependent

100: BGR in Ga$_{1-x}$Mn$_x$As, finding a strong quantitative dependence of

101: the BGR on the magnetic properties -- in particular, the calculated

102: BGR depends strongly on whether the system is ferromagnetic or

103: paramagnetic with the ferromagnetic Ga$_{1-x}$Mn$_x$As typically

104: having a factor of $1.5$ to $2$ higher BGR than the corresponding

105: paramagnetic system at the same density (and temperature). Our

106: predicted density, temperature, and spin-polarization (i.e.

107: ferromagnetic being fully spin-polarized and paramagnetic being

108: spin-unpolarized) dependence of BGR in Ga$_{1-x}$Mn$_x$As (provided

109: the carriers are the usual GaAs valence band holes) should enable a

110: clear distinction between the valence band and the impurity band

111: picture of DMS ferromagnetism through a careful analysis of

112: experimental optical absorption data~\cite{singley}. We want to

113: emphasize that our work is not only about studying BGR in the

114: Ga$_{1-x}$Mn$_x$As system. Our main purpose is to propose a technique

115: for distinguishing whether the GaMnAs carriers are in the valence band

116: or the impurity band, which is a subject of considerable importance

117: and controversy in the DMS community as exemplified by many published

118: works on the topic. On the other hand, to our knowledge the

119: temperature and magnetization dependence of the BGR are never studied

120: before, and they are crucial to the experimental observation of BGR.

121: 

122: Free carriers (e.g. holes in Ga$_{1-x}$Mn$_x$As) affect the band gap

123: of semiconductors in two essential ways. First, there is the trivial

124: single-particle effect of band filling (sometimes called the

125: Moss-Burstein shift) arising from a finite Fermi level $E_F$ ($\propto

126: n^{2/3}$, where $n$ is the free carrier density) in the valence or the

127: conduction band of the semiconductor as the band is filled with a

128: finite carrier density. The band filling effect obviously increases

129: the apparent band gap by an amount $E_F$ which should be subtracted

130: out from the measured band gap energy. In our results presented in

131: this paper we subtract out the trivial band filling Moss-Burstein

132: shift in order to avoid any confusion. The second effect, which is the

133: main subject of this work, is the self-energy correction of the

134: semiconductor band edge due to the many-body exchange-correlation

135: effect arising from finite carrier density. The band gap

136: renormalization~\cite{dassarma3, trankle} is a true many-body effect

137: arising from the hole-hole Coulomb interaction in the

138: Ga$_{1-x}$Mn$_x$As valence band, with the intrinsic (i.e. the zero

139: carrier density limit) band gap being increasingly reduced (i.e.

140: renormalized) by the carrier-induced self-energy correction as the

141: carrier density increases. This density dependent many-body

142: self-energy correction induced reduction or renormalization of the

143: semiconductor band gap, the BGR, in Ga$_{1-x}$Mn$_x$As is the subject

144: matter of our work.

145: 

146: The rest of the paper is structured as following: In Section

147: \ref{sec:form} we present the formalism of our BGR calculation as a

148: function of temperature, carrier density and spin degeneracy. In

149: Section \ref{sec:results} we describe the result of BGR for an

150: optimally-doped metallic Ga$_{1-x}$Mn$_x$As DMS system. In Section

151: \ref{sec:diss} we discuss the implication of our BGR results how they

152: might serve as a probe to determine the valence band or impurity band

153: nature of carrier ferromagnetism, which is the purpose of this paper.

154: Especially, we discuss the effect of impurity scattering on the

155: Ga$_{1-x}$Mn$_x$As optical properties, and why in spite of this

156: complication, the many-body BGR effect can still be experimentally

157: observed.

158: 

159: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

160: \section{Formalism}

161: \label{sec:form}

162: 

163: \begin{figure}[htbp]

164: \centering \includegraphics[width=2in]{fig1.eps}

165:   \caption{Feynman diagrams for calculating the quasihole

166:     self-energy. The solid line denotes the hole propagator. (a)

167:     one-loop quasihole self-energy due to Coulomb interaction, with

168:     the wiggly line denoting long-range Coulomb interaction; (b)

169:     quasihole self-energy due to impurity scattering in the

170:     single-site approximation, with the dashed line denoting

171:     short-range impurity interaction.}

172:   \label{fig:feynman}

173: \end{figure}

174: 

175: We calculate the BGR in the single-loop (Fig.~\ref{fig:feynman})

176: self-energy theory which is the leading-order approximation in the

177: dynamically screened Coulomb interaction. This approximation, often

178: referred to as the ``GW approximation''~\cite{hedin} in the

179: semiconductor literature~\cite{louie}, is known to be exact in the

180: high carrier density limit which happens to be the situation of

181: interest to us since Ga$_{1-x}$Mn$_x$As is an extremely heavily

182: hole-doped system with the hole density $n \approx 10^{20} cm^{-3}$.

183: In Ga$_{1-x}$Mn$_x$As the free carriers are holes, so we only consider

184: the self-energy correction to the band gap arising from the quasihole

185: self-energy correction to the valence band edge within the

186: GW-approximation (Fig.~\ref{fig:feynman}), obtaining for the

187: finite-temperature Matsubara self-energy ($\hbar = k_B = 1$

188: throughout):

189: 

190: \begin{eqnarray}

191:   \label{eq:self}

192:   \Sigma({\bf k}, i \nu_l) = - \int \frac{d^3 q}{(2 \pi)^3}

193:     T \sum_{\omega_n} \frac{v_q}{\epsilon({\bf q}, i \omega_n)}

194:       \nonumber \\

195:       \times \frac{1}{i \nu_l + i \omega_n - \xi_{{\bf q} - {\bf k}}},

196: \end{eqnarray}

197: where $v_q = 4 \pi e^2 / \kappa q^2$ is the 3D bare Coulomb

198: interaction with $\kappa$ the lattice dielectric constant, $i \nu_l =

199: i (2l + 1) \pi T$ and $i \omega_n = i 2 n \pi T$ are the usual

200: fermion/boson odd/even Matsubara frequencies ($l$, $n$ integers),

201: $\xi_{\bf k} = k^2 /(2 m) - \mu$ with $\mu$ the chemical potential and

202: $m$ the bare band mass, and $\epsilon({\bf k}, i \omega_n)$ is the

203: dynamical dielectric function, given by the infinite sum of the

204: polarization bubble diagrams:

205: \begin{equation}

206:   \label{eq:epsilon}

207:   \epsilon({\bf k}, i \omega_n) = 1 - v_q \Pi({\bf k}, i \omega_n),

208: \end{equation}

209: with $\Pi({\bf k}, i \omega_n)$ the 3D hole polarizability:

210: \begin{equation}

211:   \label{eq:Pi}

212:   \Pi({\bf k}, i \omega_n) = g \int \frac{d^2 q}{(2 \pi)^2}

213:   \frac{n_F(\xi_{\bf q}) - n_F(\xi_{{\bf q} - {\bf k}})}

214:   {\xi_{\bf q} - \xi_{{\bf q} - {\bf k}} + i \omega_n},

215: \end{equation}

216: where $n_F(x) = 1/(e^{x/T} + 1)$ is the Fermi distribution function

217: and $g$ the hole spin degeneracy factor.

218: 

219: In Ga$_{1-x}$Mn$_x$As the valence band structure is complicated with

220: spin-split light and heavy holes both being important. Fortunately,

221: our calculated results for BGR do not depend strongly on the carrier

222: effective mass, and change little if the band mass is changed between

223: $0.3 - 0.6 m_e$. We choose, consistent with experiment, $m = 0.5 m_e$

224: as the average band mass in all of our calculations. The dielectric

225: constant $\kappa$ in the system is $10.9$. The spin degeneracy $g$

226: varies from $1$ to $2$ according to the magnetization of the DMS

227: system. ($g=2(1)$ for the paramagnetic (ferromagnetic) case.)

228: Spin-orbit coupling effects, which are known to be important for GaAs

229: valence band, are neglected in our theory -- spin-orbit coupling is

230: expected to lead to small quantitative corrections of our calculated

231: BGR.

232: 

233: 

234: To calculate the retarded self-energy $\Sigma ({\bf k}, i \nu_l \to

235: \omega + i 0^+) \equiv \Sigma({\bf k}, \omega)$, we perform a certain

236: contour distortion to transfer the real frequency integration into

237: summations over imaginary frequencies using the analytic properties of

238: the dielectric function. This technique is described in detail in

239: Ref.~\onlinecite{massT}. To obtain quasihole self-energy with

240: wavevector ${\bf k}$, we simply put $\omega = \xi_{\bf k}$ instead of

241: solving the Dyson equation $\omega = \xi_{\bf k} +

242: \mbox{Re}\Sigma({\bf k}, \omega)$. This is known~\cite{rice} to be the

243: correct approximation consistent with the Fermi liquid theory, within

244: the single-loop self-energy approximation of Eq.~(\ref{eq:self}).  The

245: quasihole self-energy then becomes a function of wavevector only:

246: \begin{eqnarray}

247:   \label{eq:Esum}

248:   \Sigma({\bf k}) =&-& \int \frac{d^3 q}{(2 \pi)^3}

249:   v_q n_F(\xi_{{\bf q} - {\bf k}}) \nonumber \\

250:   &-& \int \frac{d^3 q}{(2 \pi)^3} v_q \left[

251:     \frac{1}{\epsilon(q, \xi_{{\bf q} - {\bf k}} - \xi_{\bf k})} - 1 \right]

252:   \nonumber \\

253:   &&~~~~~\cdot \left[n_B(\xi_{{\bf q} - {\bf k}} - \xi_{\bf k})

254:     +n_F(\xi_{{\bf q} - {\bf k}}) \right] \nonumber \\

255:   &-& \int \frac{d^3 q}{(2 \pi)^3} T \sum_{\omega_n}

256:   v_q \left[ \frac{1}{\epsilon(q, i \omega_n)} - 1 \right]

257:   \nonumber \\

258:   && ~~~~~~~ \times \frac{1}{i \omega_n - (\xi_{{\bf q} - {\bf k}} -

259:   \xi_{\bf k})}.

260: \end{eqnarray}

261: We directly calculate BGR from Eq.~(\ref{eq:Esum}) for various hole

262: densities, temperatures and spin polarizations by writing $BGR \equiv

263: |{\rm Re}~\Sigma(k = 0)|$, the magnitude of the real part of band edge

264: self-energy.

265: 

266: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

267: \section{Results}

268: \label{sec:results}

269: 

270: \begin{figure}[htbp]

271: \centering \includegraphics[width=3in]{fig2.eps}

272:   \caption{(Color online.) Calculated band-gap renormalization

273:     (narrowing) as a function of hole-density at different

274:     temperatures in a ${\rm Ga}_{1-x}{\rm Mn}_x{\rm As}$ system with

275:     spin degeneracy $g = 1$ and $2$. The upper (lower) 7 curves

276:     correspond to $g=1 (2)$ case. For each $g$ value, results for 7

277:     different temperatures are shown: $0K, 1K, 5K, 10K, 25K, 100K,

278:     300K$. Note that BGR results for $T= 100K$ and $300K$ are almost

279:     identical, indicating a saturation temperature effect. $E_F =

280:     (n/g)^{2/3} \times 249.41 meV$ where $n$ is in the unit of

281:     $10^{20} cm^{-3}$.}

282:   \label{fig:BGR}

283: \end{figure}

284: 

285: In Fig.~\ref{fig:BGR} we present the calculated band-gap

286: renormalization as a function of hole-density $n$ at different

287: temperatures. We note from Fig.~\ref{fig:BGR} that the BGR reduces the

288: band gap approximately (but not precisely) as an $n^{1/3}$ functional

289: dependence on hole density, and the effect can be as large as a few

290: hundred meVs.  Although BGR itself increases in magnitude with hole

291: density, we note that the scaled BGR, i.e. $BGR/E_F$, decreases in

292: magnitude with increasing hole density since $E_F \sim n^{2/3}$.  This

293: is consistent with the fact that the relative importance of

294: interaction effects goes down in a quantum Coulomb system with

295: increasing density. Fig.~\ref{fig:BGR} shows that an increasing

296: temperature also enhances BGR, and this enhancement has a tendency to

297: saturate at high temperatures. The increasing BGR with temperature

298: arises from the weakening of screening with increasing temperature.

299: The temperature dependence we obtain is moderate but observable. Even

300: though the percentage correction to BGR due to finite temperature is

301: larger at lower densities, the absolute finite temperature correction

302: is approximately the same for all densities within the range $10^{19}

303: - 10^{21} cm^{-3}$.

304: 

305: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

306: \section{Discussion}

307: \label{sec:diss}

308: 

309: One of the important features of Fig.~\ref{fig:BGR} is that the BGR

310: depends strongly on the magnetic properties of Ga$_{1-x}$Mn$_x$As with

311: the ferromagnetic spin-polarized hole $g=1$ situation having larger,

312: by a factor of $1.5$ to $2$ BGR, than the corresponding

313: spin-unpolarized ($g=2$) paramagnetic case. The strong spin degeneracy

314: dependence of BGR apparent in Fig.~\ref{fig:BGR} is understandable on

315: the basis of the fully spin-polarized system having weaker screening

316: since the density of states is lower in magnitude in the polarized

317: system (i.e. $g=1$ is the spin-polarized system versus $g=2$ in the

318: unpolarized system in Eq.~(\ref{eq:Pi})). Also, BGR depends indirectly

319: on the Fermi energy, and since the polarized system has a higher $E_F$

320: than the unpolarized state at the same density, the BGR is higher in

321: the fully spin-polarized ferromagnetic system. Such a strong

322: dependence of BGR on the hole spin-polarization (by $50 - 100\%$)

323: should be reasonably easy to detect experimentally by measuring BGR in

324: Ga$_{1-x}$Mn$_x$As well below and well above the Curie temperature.

325: 

326: We have also carried out calculations for the quasiparticle effective

327: mass and spin susceptibility (or equivalently the Landau $g$-factor)

328: renormalization for the Ga$_{1-x}$Mn$_x$As valence band holes induced

329: by the hole-hole many-body Coulomb interaction. Our results (not

330: shown) indicate that many-body hole effective mass and spin

331: susceptibility renormalizations are rather small ($10-20\%$) at the

332: hole densities of interest in Ga$_{1-x}$Mn$_x$As. This rather small

333: quasiparticle Fermi liquid renormalization in Ga$_{1-x}$Mn$_x$As is

334: consistent with the weakly interacting high-density nature of

335: Ga$_{1-x}$Mn$_x$As. We have also calculated the hole self-energy

336: correction $\Sigma({\bf k})$ in Eq.~(\ref{eq:Esum}) as a function of

337: wavevector to check for the nonparabolicity introduced by hole-hole

338: interaction. We find that ${\rm Re}~\Sigma(k_F)$ to be slightly ($\sim

339: 10 - 20\%$) larger in magnitude than ${\rm Re}~\Sigma(k = 0)$ at the

340: hole densities of interest, indicating the correlation-induced

341: many-body modification of the hole energy dispersion to be rather

342: small. We have also calculated the imaginary part of the hole

343: self-energy, ${\rm Im}~\Sigma(k=0)$, to ensure that the quasiparticle

344: picture does not completely break down at the band edge ($k=0$). We

345: find $|{\rm Im}~\Sigma(k=0)|/E_F$ to be in the $0.2$ to $0.05$ range

346: for $n = 10^{19} - 10^{21} cm^{-3}$ hole density range of interest,

347: indicating that the quasiparticle band description remains

348: experimentally valid in Ga$_{1-x}$Mn$_x$As in the density range of our

349: interest.

350: 

351: Now we discuss the effect of impurity scattering on the

352: Ga$_{1-x}$Mn$_x$As optical properties, which is likely to complicate

353: the experimental observation of the many-body BGR correction predicted

354: in this paper, particularly since Ga$_{1-x}$Mn$_x$As samples, even in

355: the optimally ($x \sim 0.05$) doped metallic system, tend to have

356: rather large resistivity indicating strong impurity scattering.

357: Impurity scattering destroys momentum conservation and consequently

358: may strongly affect inter-band optical absorption experiments which

359: depend on wavevector conservation. In particular, strong impurity

360: scattering would lead to two distinct effects on the optical

361: absorption in heavily doped Ga$_{1-x}$Mn$_x$As. First, there will be

362: an upward shift in the valence band edge (i.e. a band gap narrowing)

363: arising from the real part of the self-energy $\Sigma_{\rm i}$

364: associated with the hole-impurity interaction. This impurity-induced

365: band-gap narrowing effect has the same practical effect as BGR on the

366: experimentally measured optical absorption gap, and the net band gap

367: narrowing will be a sum of the hole-hole self-energy (i.e. BGR) and

368: the hole-impurity self energy. The hole-impurity self-energy is

369: therefore significant for the optical absorption experiments of

370: interest to us. On the other hand the imaginary part of the

371: hole-impurity self-energy, ${\rm Im}~\Sigma_{\rm i}$, leads to a

372: broadening of the momentum eigenstates and is therefore a measure of

373: the level broadening in the optical absorption spectra. This impurity

374: induced level broadening is, therefore, also an important

375: consideration for estimating BGR from the optical absorption spectrum

376: since this will control the broadening of the absorption spectra.

377: 

378: \begin{figure}[htbp]

379: \centering \includegraphics[width=3in]{fig3.eps}

380:   \caption{(Color online.) The contribution of impurity scattering 

381:     to the real and imaginary quasihole self-energy as a function of

382:     hole-density at $k = 0, k_F$ and $g = 1, 2$ in a ${\rm

383:       Ga}_{1-x}{\rm Mn}_x{\rm As}$ system. We assume $n_{\rm i} = n$.}

384:   \label{fig:I}

385: \end{figure}

386: 

387: We have obtained the impurity scattering effect on the hole states of

388: Ga$_{1-x}$Mn$_x$As by calculating the hole-impurity self-energy

389: $\Sigma_{\rm i} ({\bf k})$ in the self-consistent single-site

390: approximation shown in Fig.~\ref{fig:feynman}(b). The self-consistent

391: single-site approximation (Fig.~\ref{fig:feynman}(b)), which is a

392: substantial improvement on the leading-oder Born approximation, should

393: be qualitatively valid in the metallic regime of Ga$_{1-x}$Mn$_x$As

394: (as long as the carrier are indeed valence band hole states). The

395: integral equation represented by the self-consistent hole-impurity

396: scattering diagrams of Fig.~\ref{fig:feynman}(b) can be exactly solved

397: for all ${\bf k}$ in the strong impurity screening limit (assuming the

398: impurities to be random charge center of density $n_{\rm i}$ in

399: Ga$_{1-x}$Mn$_x$As), and the result for $k=k_F$ and $k=0$ are

400: \begin{eqnarray}

401: \label{eq:SigmaI}

402: \Sigma_{\rm i}(0) &=& V_0 n_{\rm i} (1 + \alpha \sqrt{r_s})^{-1};

403: \nonumber \\

404: \Sigma_{\rm i}(k_F) &=& V_0 n_{\rm i} [1 + \alpha \sqrt{r_s} f(\beta

405: \sqrt{r_s}) + i \gamma],

406: \end{eqnarray}

407: where 

408: \begin{equation}

409: \label{eq:fx}

410: f(x) = {1 \over 2} + ({x \over 8} - {1 \over 2 x}) \ln \left| {x + 2

411:   \over x - 2} \right|,

412: \end{equation}

413: and $\alpha = (6g/\pi)^{1/3}$; $\beta = 2^{2/3} 3^{-1/3} \pi^{-2/3}

414: g^{2/3}$; $\gamma = \pi/2$; $r_s = (9 \pi / 2 g)(a_B k_F)^{-1}$ with

415: $a_B$ the effective hole Bohr radius in GaAs; $V_0 = 4 \pi e^2/\kappa

416: q_{TF}^2$ where $q_{TF}$ is the GaAs hole Thomas-Fermi screening

417: wavevector (and $\kappa$ the GaAs lattice dielectric constant). In

418: Fig.~\ref{fig:I} we show our calculated real and imaginary parts of

419: the hole-impurity self-energy assuming an optimal Ga$_{1-x}$Mn$_x$As

420: metallic system, of $n_{\rm i} \equiv n$. It is important to point out

421: that our calculated ${\rm Im}~\Sigma$ in Fig.~\ref{fig:I} implies a

422: level broadening which is consistent with the measured resistivity of

423: Ga$_{1-x}$Mn$_x$As, providing support for our theoretical

424: approximation. (The actual $\Sigma_{\rm i}$ may be somewhat smaller

425: than that given in Fig.~\ref{fig:I} due to impurity clustering effect

426: ignored in our theory.)  We note that the impurity induced level

427: broadening, while being somewhat less in magnitude, is of the same

428: order as the BGR, which may complicate the interpretation of the

429: optical absorption data. (The net band gap shrinkage is given by the

430: sum of BGR (Fig.~\ref{fig:BGR}) and the real part of impurity-hole

431: self-energy given in Fig.~\ref{fig:I}.) But, we believe that it should

432: still be possible to analyze the optical absorption data to check

433: whether the density dependent optical absorption spectra are

434: consistent with a density, temperature, and spin polarization

435: dependent BGR predicted in our Fig.~\ref{fig:BGR}. Especially, the

436: temperature dependence of the impurity scattering effect is very

437: small, and this enables us to identity whether the band shrinking is

438: indeed a many-body BGR effect instead of impurity scattering by

439: examining whether this gap shrinking (if observed) possesses an

440: obvious temperature dependence.

441: 

442: If such a many-body BGR is observed in the experimental data, then

443: that would be strong evidence supporting a valence band RKKY-Zener

444: mechanism for DMS ferromagnetism in Ga$_{1-x}$Mn$_x$As.  This is

445: because if the holes are located in the impurity band, the BGR effect

446: should be very small because of the large band mass associated with

447: the impurity band. (Note the impurity band BGR effect just mentioned

448: should not be confused with impurity scattering effect mentioned in

449: the last paragraph.) We note that in addition to optical data, STM

450: measurements~\cite{STM} may also be helpful in the observation of BGR.

451: We emphasize that because of the large defect and impurity content in

452: GaMnAs, invariably present~\cite{STM} in the low-temperature MBE

453: growth of DMS materials, the observation of BGR will be complicated,

454: but our calculated density, temperature, and magnetization dependence

455: results should enable such a BGR observation of it is present.  We

456: also note that at finite temperature electron-phonon

457: interaction~\cite{cardona} would also contribute to the BGR, but the

458: phonon effect is smaller in magnitude than the exchange-correlation

459: correction in the high hole density GaMnAs of interest to us. Also,

460: the phonon correction does not exhibit a strong density or

461: magnetization dependence.

462: 

463: In summary, we have developed a theory for hole-hole (and

464: hole-impurity) free carrier interaction induced many-body effects on

465: the optical absorption spectra of Ga$_{1-x}$Mn$_x$As, finding (large)

466: density and magnetization dependent and (moderate) temperature

467: dependent many-body band gap renormalization corrections, which should

468: be observable experimentally provided the holes in Ga$_{1-x}$Mn$_x$As

469: indeed reside in the GaAs valence band, and {\em not} in the impurity

470: band.

471: 

472: This work is supported by ONR and ARDA.

473: 

474: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

475: 

476: \bibliography{BGR}

477: \end{document}

478: