A possible scenario of metallization in boron doped diamond CB$_x$
Extracted source text ยท arXiv cond-mat/0405040
% ****** Start of file apssamp.tex ******
%
% This file is part of the APS files in the REVTeX 4 distribution.
% Version 4.0 of REVTeX, August 2001
%
% Copyright (c) 2001 The American Physical Society.
%
% See the REVTeX 4 README file for restrictions and more information.
%
% TeX'ing this file requires that you have AMS-LaTeX 2.0 installed
% as well as the rest of the prerequisites for REVTeX 4.0
%
% See the REVTeX 4 README file
% It also requires running BibTeX. The commands are as follows:
%
% 1) latex apssamp.tex
% 2) bibtex apssamp
% 3) latex apssamp.tex
% 4) latex apssamp.tex
%
\documentclass[twocolumn,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
%\documentclass[preprint,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
% Some other (several out of many) possibilities
%\documentclass[preprint,aps]{revtex4}
%\documentclass[preprint,aps,draft]{revtex4}
%\documentclass[prb]{revtex4}% Physical Review B
\usepackage{graphicx}% Include figure files
\usepackage{dcolumn}% Align table columns on decimal point
\usepackage{bm}% bold math
%\nofiles
\begin{document}
%\preprint{APS/123-QED}
\title{A possible scenario of metallization in boron doped diamond CB$_x$}
% Force line breaks with \\
\author{Yu.G. Pogorelov}
\altaffiliation{Electronic address: ypogorel@fc.up.pt}
\affiliation{CFP and Departamento de Fisica, Faculdade de
Ciencias, Universidade do Porto, Rua do Campo Alegre, 687,
4169-007 Porto, Portugal}%
\author{V.M. Loktev}
\altaffiliation{Electronic address: vloktev@bitp.kiev.ua}
\affiliation{Bogolyubov Institute for Theoretical
Physics, Metrologichna str. 14-b, Kiev, 03143 Ukraine}%
\date{\today}
% It is always \today, today,
% but any date may be explicitly specified
\begin{abstract}
Possibility for collectivization of acceptor states in a semiconductor,
converting it to metal, is discussed within the scope of Anderson s-d
hybride model. This model is generalized for multicomponent band structure
and composite acceptor states, localized on pairs of neighbor dopants
(impurity ``dumbbells''), in order to describe the boron doped diamond CB$_x$.
The resulting parameters of band structure, in particular, position of the
Fermi level, are compared to the recent experimental data on metallized and
superconducting CB$_x$.
\end{abstract}
\pacs{71.55.-i, 74.20.-z, 74.20.Fg, 74.62.Dh, 74.72.-h}
% 74.20.Rp - Pairing symmetries (other than s-wave)
% 74.25.Dw - Superconductivity phase diagram
% 74.62.Dh - Effects of crystal defects, doping and substitution
% 74.72.-h - High-T_c compounds
\maketitle
\noindent Recent discovery of superconducting transition in boron doped
diamond CB$_x$ \cite{ekimov} brought a new attention to the problem of
impurity induced metallization in semiconductors. This topic is already at
the center of discussion on high-$T_{c}$ metal-oxide perovskites (HTSC),
where shallow acceptors by, e.g., divalent alcali earth substitutes A for
trivalent lanthanum in La$_{2-x}$A$_x$CuO$_{4}$, give rise to metallic and
superconducting state in basal CuO$_{2}$ planes at doping $x$ above some
critical level $\sim 5\%$. A description of this process on the basis of
Lifshitz model of impurity states was proposed by the authors \cite{lp}.
However the situation in CB$_x$ differs from HTSC in essential aspects:
diamond has an exemplary 3D isotropic (cubic) lattice structure \cite{herman}
and B atoms at relevant doping levels ($x\sim 4\%$) mostly occupy interstitial
positions in it \cite{gild},\cite{prins},\cite{chen},\cite{thonke} where they
nominally should stay neutral. Even at lower doping ($x\lesssim 0.5\%$),
when they are mostly acceptor substitutes \cite{werner}, it was already
recognized that common effective mass approach does not apply for the deep
0.37 eV level by isolated acceptor \cite{glez} whereas the conductivity
may result from the interplay between that and some ``additional'', much
more shallow levels ($\sim 0.06$ eV) by acceptor clusters \cite{mamin}.
This clustering effect should be even more important at higher doping, as seen
from statistical weights of configurations around an interstitial impurity
(Fig.\ref{cap:1}) at $x = 4\%$ ($8\%$ per diamond unit cell), reaching maximum
$\sim 38\%$ for its clusters with another impurity in one of 12 nearest
neighbor interstitices. The facts that carrier density still grows with
doping and that energy would be gained when doped holes are spread around the
pair of impurities by B-C covalency effects suggest that such clustered B
interstitials (impurity ``dumbbells'') also give rise to shallow acceptor
levels, supposedly not described by effective mass, like those from clusters
of substitutional impurities.
When choosing an adequate model for such perturbation of electronic spectrum,
one have to rule out either the Mott metallization (since the effective
Bohr radius is too small, $\sim 3$ \AA \cite{baskaran}) and the Lifshitz
impurity model (which does not provide metallization in 3D doped systems
\cite{iv}). Then the most natural choice is perhaps the Anderson s-d hybride
model \cite{and}, where the possibility for metallization was studied long
ago \cite{ivanov}. Here we try to adapt the latter results for the intriguing
physics of doped diamond, leaving aside the issue of pairing mechanism in
the metallized CB$_x$ (discussed recently in RVB \cite{baskaran} or
bond-stretching \cite{lee} scenarios).
\begin{figure}
\includegraphics[scale=0.5]{CB1.eps}
\caption{\label{cap:1} Unit cell of diamond (bigger circles for carbons)
where boron dopants (dark circles) occupy an interstitial site at the center
and one of 12 nearest neighbor interstices (dashed circles). The resulting
impurity ``dumbbell'' has two acceptor levels, one of them, shallow
antibonding, being of principal importance for metallization.}
\end{figure}
Referring to the well established theoretical band structure of pure diamond
\cite{herman},\cite{cohen} and recent ARPES measurement data \cite{jimenez},
the analysis should concentrate around the top of the valence band (chosen
as the energy reference). Here the three valence subbands have almost
isotropic dispersion $\varepsilon _{j,{\bf k}}=-\hbar^2 k^2/2m_j$ with
effective masses $m_{1,2,3}\approx 2.12, 1.06, 0.7 m_e$ (neglecting a small
spin-orbit splitting down by $\approx 6$ meV for $j=2$). Considering the most
relevant acceptor level $\varepsilon_i$ by impurity cluster, closely atop
the valence band, and neglecting all other (much higher or/and much fewer)
levels by clustered or isolated impurities, the corresponding generalization
of hybride model can be presented by the Hamiltonian
\begin{eqnarray}
H=\sum_{j,{\bf k}}&&\varepsilon _{j,{\bf k}}a_{j,{\bf k}}^{\dagger }
a_{j,{\bf k}}+\sum_{{\bf p}}\varepsilon _{0}b_{{\bf p}}^{\dagger }
b_{{\bf p}}+\notag\\&&+\frac 1{\sqrt{N}}\sum_{j,{\bf k,p}}\left( \gamma_j
{\rm e}^{i{\bf k}\cdot{\bf p}}a_{j,{\bf k}}^{\dagger }b_{{\bf p}}+
h.c.\right) , \label{eq1}
\end{eqnarray}
where $N$ is the number of unit cells in crystal, $a_{j,{\bf k}}$ and
$b_{{\bf p}}$ are the Fermi operators (spin indices suppressed) for
excitations of $j$th subband and of impurity clusters (called simply
impurities in what follows) randomly located at ${\bf p}$ with concentration
$c \approx x/2$ (per unit cell), and $\gamma_j$ is the constant of
hybridization between them \cite{note}.
The resulting spectrum generally includes either band-like and localized
states and, since each of $cN$\ impurities provides one (hole) carrier to
the system, the issue of metallization depends on whether the Fermi level
$\varepsilon _{{\rm F}}$ from the equation
\begin{equation}
c=\int_{\varepsilon _{{\rm F}}}^{\infty }\rho \left( \varepsilon \right)
d\varepsilon
\label{eq2}
\end{equation}
lies within the domain of band states. The density of states (DOS, per unit
cell) $\rho\left(\varepsilon\right) =N^{-1}{\rm{Im Tr}}\left(\varepsilon -
H\right)^{-1}$ in the relevant basis of states for Eq.\ref{eq1}:
$\left\{\left|j,{\bf k}\right\rangle ,\left|{\bf p}\right\rangle\right\}$
(ignoring the conduction band), is calculated through the respective diagonal
Green functions.
Let's begin supposing that only single (say, 1st) subband is hybridized with
impurity states: $\gamma_{2,3}=0$, $\gamma_1=\gamma\neq 0$ (comparison with an
alternative choice $\gamma_1=\gamma_2=\gamma_3=\gamma$ will be done at the end),
and obtain DOS as a sum of contributions from host (h) and impurity (i)
subsystems:
\begin{eqnarray*}
\rho\left(\varepsilon\right)&=&\rho_{h}\left(\varepsilon\right)
+\rho_{i}\left(\varepsilon\right) , \\
\rho _{h}\left( \varepsilon \right)&=&\sqrt{-\varepsilon}\left(W_2^{-3/2}
+ W_3^{-3/2}\right) + \\&&+\frac 1{\pi N}{\rm {Im}}\sum_{\bf k}\frac 1 {\varepsilon -\varepsilon _{1,{\bf k}}- \Sigma_{{\bf k}}}, \\
\rho_{i}\left( \varepsilon \right) &=&\frac c{\pi}{\rm {Im}}
\frac 1{\varepsilon -\varepsilon _{0}-\Sigma _{i}}.
\end{eqnarray*}
Here $W_j = \pi^{4/3}\hbar^2/(m_j v^{2/3})$ ($v$ being the unit cell volume)
is of the order of width of respective subband, whereas the self-energies
$\Sigma_{\bf k}$ and $\Sigma_i$ can be presented as expansions in groups of
interacting impurities \cite{ivanov}:
\begin{equation*}
\Sigma _{{\bf k}}=\frac{c\gamma ^{2}}{\varepsilon-\varepsilon _{0}-
\Sigma _{i}}\left( 1+cB_{{\bf k}}+\ldots\right)
\end{equation*}
and
\begin{equation*}
\Sigma _{i}=\frac{\gamma ^{2}}{N}\sum_{{\bf k}}\frac 1 {\varepsilon
-\varepsilon _{1,{\bf k}}-\Sigma _{{\bf k}}}\left(
1+cB_{i}+\ldots \right).
\end{equation*}
The relevant impurity level is defined by the equation $\varepsilon_i =
\varepsilon_0 + {\rm Re} \Sigma_i(\varepsilon_i)$ in the limit $c \to 0$,
whereas the next terms after unity in the group expansions:
\begin{eqnarray*}
B_{{\bf k}}&=&-A_{\bf 00}-A_{\bf 00}^{2}+\sum_{{\bf n}\neq {\bf 0}}
\frac{A_{\bf 0n}^3 {\rm e}^{-i{\bf k} \cdot {\bf n}}+A_{\bf 0n}^4}
{1-A_{\bf 0n}^2}, \\
B_{i}&=&-A_{\bf 00}^{2}+\sum_{{\bf n}\neq {\bf 0}}\frac{A_{\bf 0n}^{4}}{1-
A_{\bf 0n}^{2}}, \\
A_{\bf 0n}&=&\frac{\gamma^2}{N\left(\varepsilon -\varepsilon _{0}-
\Sigma _{i}\right)}\sum_{{\bf k}}\frac{{\rm e}^{i{\bf k}\cdot{\bf n}}}
{\varepsilon -\varepsilon _{1,{\bf k}}-\Sigma _{{\bf k}}}
\end{eqnarray*}
describe all indirect interactions between two impurities (in fact,
two ``dumbbels'') located at cells ${\bf 0}$ and ${\bf n}$, via exchange
of virtual excitations from 1st subband. The respective contribution to
DOS from the states localized at such pairs of ``dumbbells'' follows from
the formula: ${\rm Im} \sum_{\bf n} f_{\bf n}/(1 - A_{\bf 0n}^{2}) \approx
(\pi/v)\int d{\bf r} f_{\bf r}\delta(1 - {\rm Re}A_{\bf 0r}^{2})$
(for any real function $f_{\bf n}$), and the concentration broadening
$\Gamma_i$ of the impurity level $\varepsilon_i$ from the condition:
$|{\rm Re}A_{{\bf 0}\overline{\bf r}}(\varepsilon_i \pm \Gamma_i)| = 1$,
where $\overline{r}\sim (v/c)^{1/3}$ is the average distance between
impurities.
We construct the solution by analogy with the case of donor impurity
level $\varepsilon _i < 0$ hybridized to a single parabolic band
$\varepsilon _{{\bf k}}=\hbar ^{2}k^{2}/2m$ \cite{ivanov}. There a
qualitative restructuring of spectrum happens when the concentration
$c$ surpasses the characteristic (supposedly small) value
$c_{0}\sim \left( \varepsilon_{i}/W\right)^{3/2}/4\pi$, where
$W = \pi^{4/3}\hbar^2/(m v^{2/3})$. In other words, then $\overline{r}$
turns smaller of the localization radius $r_{0}\sim v^{1/3}\sqrt{W/
\varepsilon _i}$ and localized impurity states effectively overlap.
If $c$ is also greater than other characteristic value $c_{cr}\sim
\left(\gamma /W\right)^6$, this overlapping leads to formation of two
separate bands of coherent extended states, whose structure depends on
a specific relation between $c_0$ and $c_{cr}$. In particular, if the level
$\varepsilon_i$ is so shallow that $c_0 < c_{cr}$, the dispersion
of two split bands is
\begin{equation}
\varepsilon _{\pm}\left({\bf k}\right) \approx \frac{\varepsilon _{\bf k}\pm
\sqrt{\varepsilon _{\bf k}^2+4c\gamma^2}}2,
\label{eq3}
\end{equation}
for wave numbers $k$ restricted to $k \gtrsim k_{min}\sim (c_{cr}/v)^{1/3}$
for the upper band, $\varepsilon _{+}\left({\bf k}\right)$, and to
$k_{min}\lesssim k \lesssim k_{max} \sim 1/\overline{r}$ for the lower band,
$\varepsilon _{-} \left({\bf k}\right)$. These restrictions give estimates for
the Mott mobility edges separating band and localized states: $\varepsilon_c
\sim \varepsilon_{+}(k_{min})$, $\varepsilon_c^\prime \sim \varepsilon_-(k_{max})$,
$\varepsilon_c^{\prime\prime}\sim\varepsilon_-({k_{min}})$ (Fig.\ref{cap:2}). Then
it can be shown that the lower band only adopts $\sim Nc^{3/4}/c_{cr}^{1/4}$
states, that is less than the total of $Nc$ carriers. Hence the rest of
carriers should occupy the ''tail'' states formed by interacting impurity pairs
whose density is defined as
\begin{equation}
\rho _{i}\left(\varepsilon \right) \sim \frac {c^2 \gamma^2}{\varepsilon
^2}{\rm Im} B_{i}\sim \frac {c^2 \gamma^8}{W^3\varepsilon^5}
\label{eq4}
\end{equation}
(for $\varepsilon - \varepsilon_i \gg \Gamma_i \sim c^{1/3}\gamma^2/W$) and
extends to the edge of the upper band. At $\varepsilon > \varepsilon_+(0)$,
the DOS in this band $\rho_h(\varepsilon)$ is much higher than the ``tail''
function $\rho _{i}\left(\varepsilon \right)$, Eq.\ref{eq4}, so that the number
of empty tail states above the Fermi level approximately equals the number of
hole carriers within the upper band:
\begin{equation}
\int_{\varepsilon _{{\rm F}}}^{\infty }\rho
_{i}\left( \varepsilon \right) d\varepsilon \approx \int_{\varepsilon_+(0)}
^{\varepsilon _{\rm F}}\rho_{h}\left(\varepsilon\right) d\varepsilon.
\end{equation}
This gives the distance from the upper band edge to the Fermi energy:
$\varepsilon _{{\rm F}}- \varepsilon_+(0) \sim \sqrt{c}\gamma$, which is
$\sqrt{c/c_{cr}}$ times bigger than the distance from this band edge to
the mobility edge $\sim c_{cr}^{2/3}W$. Therefore the system turns to be
metal with the band structure as shown in Fig.\ref{cap:2}.
\begin{figure}
\includegraphics[scale=0.5]{CB2.eps}
\caption{\label{cap:2} Restructuring of the conduction band at high
enough doping ($c \gg c_{cr}$, see in text) by shallow donors within
the hybride model \cite{ivanov}. There are two separate regions of band
states near the bottom of conduction band and the system turns metal.}
\end{figure}
In the multicomponent valence band case defined by Eq.\ref{eq1} with
$\gamma_1=\gamma$ and $\gamma_{2,3}\approx 0$, we obtain the same splitting
as by Eq.\ref{eq3} for the $j=1$ valence subband (within to obvious inversion
$\varepsilon_+\leftrightarrows\varepsilon_-$), but the essential difference
from the situation in Fig.\ref{cap:2} is that here the excess carriers can
fill the two subbands, \emph{non-hybridized to impurities}, resulting in the
metallic state with the Fermi level below zero energy: $-\varepsilon_{\rm F}
\sim (c-c^{3/4}c_{cr}^{1/4})^{2/3}W$, and two close regions of high occupation,
as shown in Fig.\ref{cap:3}a. These two regions can play the same role for
the non-monotonic temperature dependence of conductivity as two activation
energies, considered in the low doping regime \cite{mamin}. With tentative
values $\gamma \approx 1$ eV, $\varepsilon_i \approx 0.05$ eV, and
$W_1\approx 3.5$ eV, we obtain from the above formulas: $c_{0}\sim 0.02\%$,
$c_{cr}\sim 0.15\%$, and for $c\approx 2\%$ arrive at $|\varepsilon _{{\rm F}}|
\sim 150$ meV. This value turns comparable to the splitting $\sqrt{c}\gamma$
of the 1st subband and quite sufficient to accomodate possible superconducting
gap $2\Delta \sim 3.52T_{c}\sim 0.6$ meV (for the reported $T_{c}=2.3$ K in
CB$_x$ \cite{ekimov}).
An alternative hybridization scheme, with $\gamma_{1,2,3}=\gamma$, will
produce similar splitting of all three valence subbands, so that the total
capacity of upper split bands (about triple of that for the single
$\varepsilon_-({\bf k})$ band in Eq.\ref{eq3}) will be most probably
sufficient to accomodate the Fermi level at positive energies, as shown
in Fig.\ref{cap:3}b. The value of $\varepsilon_+(0)-\varepsilon _{{\rm F}}$
in the resulting conduction band should be comparable to the above estimate for
$|\varepsilon _{{\rm F}}|$, but this single region of high occupation looks
unlike for the above referred non-monotonic conductivity vs temperature,
seen also in the metallized CB$_x$ \cite{ekimov}. This favors to the former
version of impurity preferential hybridization to a single valence subband.
\begin{figure}
\includegraphics[scale=0.5]{CB3.eps}
\caption{\label{cap:3} Restructuring of the multicomponent valence band
at doping by shallow acceptors. a) Exclusive hybridization to the 1st subband.
b) Equal hybridization to all three subbands.}
\end{figure}
In conclusion, we analyzed the specifics of impurity induced metallization
in a version of Anderson s-d hybride model for multiband semiconductor and
showed its difference compared to the simple single band case. The resulting
split band structure with a relatively small Fermi surface of radius
$k_{{\rm F}}\sim \left(c/v\right) ^{1/3}$ near the Brillouin zone
center also essentially differs from the non-split structures obtained in the
simplest appoximations of effective mass or virtual crystal \cite{lee}. This
difference can be verified experimentally, e.g., in the optical response at
about 8 $\mu$m. At least, the proposed metallic state in CB$_x$ can be further
used to describe the superconducting transition in such doped system, which,
regardless of the pairing mechanism (perhaps phonon-mediated), can be
effectively enhanced due to multiband spectrum structure \cite{mosk}.
\begin{thebibliography}{99}
\bibitem{ekimov} E.A. Ekimov, V.A. Sidorov, E.D. Bauer, N.N. Mel'nik, N.J.
Curro, J.D. Thompson, and S.M. Stishov, Nature{\bf \ 428}, 542 (2004).
\bibitem{lp} V.M. Loktev and Yu.G. Pogorelov, Physica C (1996); Low Temp.
Phys. {\bf 27}, 1039 (2001); Europhys. Lett. {\bf 58}, 549 (2002).
\bibitem{herman} F. Herman, Phys. Rev. {\bf 93}, 1214 (1954).
\bibitem{gild} G.Sh. Gildenblat, S.A. Grot, and A. Badzian, Proc. IEEE {\bf %
79}, 647 (1991).
\bibitem{prins} J.F. Prins, Thin Solid Films {\bf 212}, 11 (1992).
\bibitem{chen} Y.-H. Chen, C.-T. Hu, and I.-N. Lin, Appl. Phys. Lett.
{\bf 18}, 2857 (1999).
\bibitem{thonke} K. Thonke, Semicond. Sci. Technol. {\bf 18}, S20 (2003).
\bibitem{werner} M. Werner, R. Job, A. Zaitzev, W.R. Fahrer, W. Seifert,
C. Johnston, and P.R. Chalker, phys. stat. sol. (a) {\bf 154}, 385 (1996).
\bibitem{glez} J.W. Glezener, Appl. Phys. Lett {\bf 64}, 217 (1994).
\bibitem{mamin} R.F. Mamin, T. Inushima, Phys. Rev. B {\bf 63}, 033201 (2001).
\bibitem{baskaran} G. Baskaran, cond-mat/0404286.
\bibitem{iv} M.A. Ivanov, Yu.G. Pogorelov, Sov. Phys. JETP {\bf 45}, 1155
(1977).
\bibitem{and} P.W. Anderson, Phys. Rev. {\bf 124}, 41 (1961).
\bibitem{ivanov} M.A. Ivanov and Yu.G. Pogorelov, Sov. Phys. JETP {\bf 49},
510 (1979); ibid {\bf 61}, 1033 (1985).
\bibitem{lee} K.-W. Lee, W.E. Pickett, cond-mat/0404547.
\bibitem{cohen} W. Saslow, T.K. Bergstresser, and M.L. Cohen, Phys. Rev.
Lett. {\bf 16}, 354 (1966).
\bibitem{jimenez} I. Jim\'{e}nez, L.J. Terminello, D.G.J. Sutherland, J.A.
Carlisle, E.L. Shirley, and F.J. Himpsel, Phys. Rev. B {\bf 56}, 7215 (1997).
\bibitem{note} It should be noted that, strictly speaking, there are 3
different types of the considered impurity dumbbels, related to their
orientations along $x$, $y$, or $z$ crystalline axes. However, we do not
expect any specific asymmetry between them in the cubic crystal and attribute
the same $\varepsilon_0$ and $\gamma_j$ values to all of them.
\bibitem{mosk} H. Suhl, B.T. Matthias, L.R. Walker, Phys. Rev. Lett. {\bf 3},
552 (1959); V.A. Moskalenko, Fiz. Metallov i Metallovedenie {\bf 4}, 503 (1959).
\end{thebibliography}
\end{document}
%
% ****** End of file apssamp.tex ******