Do quantum dots allow one access to pseudogap Kondo physics?
Extracted source text · arXiv cond-mat/0407583
%----------------------------------------------------------------------
% Template article for SCES '04 conference:
% document class 'phb-proc4-auth'
%
% This template should be used for all manuscripts submitted to the
% SCES '04 - The International Conference on Strongly Correlated Electron Systems
% Universität Karlsruhe (TH) July 26-30, 2004
%
% Send any comments to the SCES '04 Administration <sces04@physik.uni-karlsruhe.de>
%----------------------------------------------------------------------
% The template contains a number of LaTeX commands of the form :
%
% \command{}
%
% In order to complete this abstract, fill in the blank fields between
% the curly braces with the requested information.
%
% e.g. in order to add an abstract title, replace
% \title{} with \title{Evidence of Solitons in Tedium Diboride}
%
% The \author and \address commands can take an optional label
% in square brackets of the form :
%
% \command[label]{}
%
% The text of the abstract should be inserted between the two commands
% \begin{abstract} and \end{abstract}.
%
% Please leave all commands in place even if you don't fill them in.
% Do not alter commands which have non-blank arguments unless you are
% familiar with them.
%
% N.B. In LaTeX, all text following the % sign is treated as a comment.
% Some commands, e.g. \thanks{}, have been "commented
% out". Remove the % sign if you wish to use it.
% Blank lines are ignored (except in the manuscript text where
% they will be treated as a paragraph break).
%
%----------------------------------------------------------------------
% Invoke the LaTeX document class for these conference proceedings
\documentclass{phb-proc4-auth}
%----------------------------------------------------------------------
% Install LaTeX packages to allow enhanced math formatting and
% graphics installation. These are the recommended packages. If you
% are experienced in using LaTeX, you may choose a different list.
\usepackage{graphicx}
\usepackage{amssymb}
%----------------------------------------------------------------------
% Start processing the document and front page
\begin{document}
\begin{frontmatter}
%----------------------------------------------------------------------
% Specify destination and version number of the manuscript
\journal{SCES '04}
%----------------------------------------------------------------------
% Title of manuscript
\title{Do quantum dots allow one access to pseudogap Kondo physics?}
%----------------------------------------------------------------------
% List of authors
%
% List each author using a separate \author{} command
%
% If there is more than one author address, add a label to each author
% of the form \author[label]{name}. This label should be identical to
% the corresponding label provided with the \address command.
%
% e.g. if there are three authors from two institutions in USA and
% France, you can link them to their respective addresses, using
%
% \author[US]{John Doe}
% \author[US,FR]{Jane Doe}
% \author[FR]{Jean Dupont}
% \address[US]{University of Life, Somewhere, USA}
% \address[FR]{Universite de la Vie, Quelque Part, France}
%
% N.B. Unlike the document class used for abstract submissions, it is
% possible to have the author associated with more than one address,
% as shown in the example above.
%
\author{{\corauthref{1}}John Hopkinson}
\author{\hskip-0.21pc, Karyn Le Hur,}
\author{and {\'{E}}milie Dupont}
%----------------------------------------------------------------------
% List of addresses
%
% If there is more than one address, list each using a separate
% \address command using a label to link it to the respective author
% as described above
\address{D{\'e}partement de Physique, Universit{\'e} de Sherbrooke, Sherbrooke, Qu{\'e}bec, Canada}
%----------------------------------------------------------------------
% Title page footnotes
%
% If you need to add qualifying information to any of the authors,
% use the \thanksref{} command within the \author command. The
% argument is the label of a corresponding \thanks[label]{text}
% command which contains the footnote text
%
% e.g. you can acknowledge a funding authority for John Doe, using
%
% \author{John Doe\thanksref{ABC}}
% \thanks[ABC]{This work was supported by Institute of Unphysical
% Phenomena under contract no. ABC-123}
%
\thanks{This work was supported by NSERC, FQRNT and CIAR.}
%----------------------------------------------------------------------
% Contact Information
%
% Add the complete postal address, telephone number, fax number, and
% email address of the corresponding author as a special footnote using
% the \corauth[]{} command. This works in a similar way to the \thanks
% command. Add the \corauthref{} command within the \author command.
% The argument is the label of a corresponding \corauth[label]{text}
% command which contains the contact information. Prefix the text with
% Corresponding Author:
%
% e.g. if the contact author is John Doe,
%
% \author{John Doe\corauthref{1}}
% \corauth[1]{Corresponding Author: University of Life, 123 Some St.,
% Somewhere, MI 12345, USA. Phone: (555) 555-5555
% Fax: (555) 555-7777, Email: JDoe@uol.edu}
%
\corauth[1]{Corresponding Author: Universit{\'e} de Sherbrooke, QC, CA Phone: (819) 821-8000 ext. 3043 Fax: (819) 821-8046}
%----------------------------------------------------------------------
% Text of abstract
\begin{abstract}
For the last decade{\cite{revival}}, tunable quantum dot systems have allowed the investigation of Kondo physics wherein the quenching of a single spin on an artificial atom affects the conductance. The pseudogap Kondo model (pKm) featuring a density of states $\rho(\epsilon)$ = C$|\epsilon|^{r}$, introduced by Withoff and Fradkin{\cite{withoff}} in 1990 was predicted to exhibit Kondo-like physics above a critical value of the Kondo coupling, J$_c$, which several groups have shown by numerical renormalization group (RG) is finite for r$< \frac{1}{2}$. Gonzalez-Buxton {\it{et al}} {\cite{gonzalez}} showed that the strong coupling limit of the particle-hole symmetric model leads to a non-trivial $\frac{\pi (1-r)}{2}$ phase shift at low temperatures indicating incomplete screening of the local moment, while away from particle-hole symmetry (p-hs) one generically flows towards a ground state with $\delta \sim \pi$. We examine the implications of this model for quantum dots whose leads are Fermi-liquid-like, yet possess a tunneling density of states (TDOS) which is suppressed at the Fermi energy as a power law.
\end{abstract}
%----------------------------------------------------------------------
% Manuscript keywords
%
% Please give two or three keywords in the form: keyword \sep keyword
% e.g. NMR \sep superconductivity
%
% NB The syntax is different from the abstract document class
\begin{keyword}
pseudogap Kondo model \sep quantum dots \sep quantum critical point
\end{keyword}
%----------------------------------------------------------------------
% End of front page
\end{frontmatter}
%----------------------------------------------------------------------
% Manuscript text
%
% Fill in the following space with the manuscript text.
%4
% A number of LaTeX commands may be invoked in this space, e.g.
%
% \section{} : to insert a new section title
% \label{} : to label the numbered section for use in \ref{}
% \cite{} : to add a reference using the label in \bibitem{}
%
% A number of LaTeX environments may be used, e.g.
% \begin{equation}
% An equation inserted here will be automatically numbered
% \end{equation}
%
% Please refer to other LaTeX documentation for help on using these
% environments.
Many of the unusual features arising in strongly correlated electron systems have been attributed to the proximity to quantum critical points. In particular, heavy fermion systems are thought to feature a competition between RKKY and Kondo interactions, within the context of a Kondo lattice model, leading to (an increasingly heavy) metallic antiferromagnetic phase which is suppressed as a function of pressure, chemical substitution or magnetic field to produce a cone of non-Fermi liquid physics preceeding a (decreasingly heavy) metallic state. Unfortunately, despite decades of intensive research on the Kondo lattice model, the detailed nature of its solution remains a subject of some debate. A simpler model exhibiting a quantum phase transition between a state with local moments and spin-screened state, the pKm, was introduced in Ref. 2 and studied extensively by subsequent groups{\cite{gonzalez,chen,ingersent1}}. This paper explores the possibility that this model could be experimentally realizable in a quantum dot system.
%The central question posed us by LiV$_2$O$_4$ is: how can a mixture of S = 1 and S = $\frac{1}{2}$ quantum spins show behaviour characteristic of localized S = $\frac{1}{2}$ spins over a wide temperature range before condensing to a heavy fermi liquid state? There have been several attempts to answer this. Anisimov et al{\cite{anisimov}} have postulated the existence of two species of electrons and invoked the exhaustion phenomenon to explain the low temperature fermi liquid phase. Varma{\cite{varma}} proposed that the spin $\frac{1}{2}$ state arises due to mixed valence but it is not immediately clear how this can give rise to a second stage spin quenching process. Kusunose et al{\cite{kusunose}} have considered a two-band Hubbard model in the limit of small Hund's coupling. Fulde et al {\cite{fulde}} have considered each magnetic tetrahedron to be occupied by two S = 1 and two S = $\frac{1}{2}$ spins, the latter forming gapless Heisenberg chains similar to a Luttinger liquid at low temperatures. And recently, Burdin et al {\cite{burdin}} have proposed a single-band Kondo lattice model on a frustrated lattice. We present here a provisional description of a simple two-band model which we believe captures the essential physics behind these two stages.
%our model--the physics behind it (the gist).
{\it{Resonant level limit}} We consider the small dot, weak tunneling limit in which the dot is a collection of discrete levels of average spacing $\delta$E, an order of magnitude smaller than the charging energy, E$_c$ (${\mathcal{O}}$($\frac{e^2}{2C}$)), of the dot (e: electron charge; C: capacitance of dot-lead system). The energy to add or subtract an electron from the dot can be estimated as $\delta$E$^{\pm}_c$ = E$_c$(Q$_G$,n+1) - E$_c$(Q$_G$, n) = (n+$\frac{1}{2}$ - C$_G$V$_G$)$\frac{e^2}{C}$, where n is the initial(final) number of electrons on the dot for +(-), V$_G$ is an external gate voltage coupled capacitively (C$_G$) to the dot. A resonant condition periodic in the number of electrons on the dot corresponds to $\delta$E$_c^{+(-)}$ = 0, $\delta$E$_c^{-(+)}$ = $\frac{e^2}{C}$. For temperatures satisfying $\delta$E$<$T$<$E$_c$, the physics of the dot is dominated by Coulomb blockade. For temperatures below $\delta$E, one can write the current through the dot in the single level Landauer-B{\"u}ttiker (elastic scattering) form,
\begin{equation}
I = \frac{2e}{h}\int_{-D}^{D} \frac{4(f_L(\epsilon)-f_R(\epsilon))\Gamma_L\Gamma_R|\frac{\epsilon}{D}|^{2r}d\epsilon}{(\epsilon-\epsilon_d)^2 + (\Gamma_L + \Gamma_R)^2|\frac{\epsilon}{D}|^{2r}}.
\end{equation}
$f_{L(R)}(\epsilon)$ is the Fermi function of the left(right) lead, we have assumed that we are at sufficiently low T that the intervals $\{\pm D,\pm\delta E\}$ contribute weakly so we can neglect multiple states and have chosen a power-law dot-lead hybridization $ \Gamma(\epsilon) = \Gamma_0|\frac{\epsilon}{D}|^r$ where $\Gamma_L = \Gamma_R = \Gamma_0 = \pi(r+1)t_kt_{k'}/(2D)$ ($t_k$: dot-lead hopping, $|\epsilon|<D$). Implicit is the assumption of equivalent leads and dot-lead couplings, and that the effect of electron-electron interactions within the leads is completely captured by a renormalization of the TDOS of the conduction electrons, (in the spirit of the one-dimensional (1 D) wire RG of Nazarov and Glazman{\cite{nazarov}}). Using (1) we can calculate the bare conductance (G$_0$ = $\frac{dI}{dV}|_{V\rightarrow 0}$). Notice that at T=0 the manifestation of Coulomb blockade is an infinitely sharp transmission amplitude at resonance ($\epsilon_d = 0$).
{\it{Kondo limit}} For T$<\delta E$, directly between two resonant points, with an odd number of spins on the dot, the symmetric condition $\delta$E$_c^{(+)}$= $\delta$E$_c^{(-)}$ = $\frac{e^2}{2C}$ allows one to write $H = \sum_{k\alpha\sigma}(\epsilon_k c_{k\alpha\sigma}^{\dagger} c_{k\alpha\sigma} + (t_{k\alpha}c_{k\alpha\sigma}^{\dagger}d_{\sigma} + h.c.)) + \epsilon_dn_d+ Un_{d\downarrow}n_{d\uparrow}$,
%\begin{eqnarray}
%H&=&\sum_{k\alpha\sigma}(\epsilon_k c_{k\alpha\sigma}^{\dagger} c_{k\alpha\sigma} + (t_{k\alpha}c_{k\alpha\sigma}^{\dagger}d_{\sigma} + h.c.)) + \epsilon_dn_d \nonumber \\& &+ Un_{d\downarrow}n_{d\uparrow},
%\end{eqnarray}
where c$_{k\alpha\sigma}$ destroys a conduction electron of momentum k, spin $\sigma$, from lead $\alpha$ and we have defined an effective on-site Hubbard repulsion U = -2$\epsilon_d$, where $\epsilon_d$ is the energy of the highest occupied level on the dot. This model is the pseudogap Anderson model (pAm) studied in Ref. 3 with a reduced r{\'e}gime of validity (D$\rightarrow$E$_c$ in the energy summation). In the limit $-\epsilon_d$ (n odd dot energy), U + $\epsilon_d$ (energy of 1st excitation) $>>$ $\Gamma_0$, T the Schrieffer-W{\"o}lff transformation leads to{\cite{gonzalez}} $H_K = \sum_{k,k'}(J_{k,k'}c_{k\alpha}^{\dagger}\frac{\sigma_{\alpha\beta}}{2}c_{k'\beta}S + V_{k,k'}c_{k\sigma}^{\dagger}c_{k'\sigma})$,
%\begin{equation}
%H_K = \sum_{k,k'}(J_{k,k'}c_{k\alpha}^{\dagger}\frac{\sigma_{\alpha\beta}}{2}c_{k'\beta}S + V_{k,k'}c_{k\sigma}^{\dagger}c_{k'\sigma}),
%\end{equation}
where the Kondo coupling, J$_{k,k'}$ = 2($\frac{1}{|\epsilon_d|} + \frac{1}{|U+\epsilon_d|}$)t$_k$t$_{k'}$, and the potential scattering V$_{k,k'}$ = $\frac{1}{2} (\frac{1}{|\epsilon_d|}- \frac{1}{|U+\epsilon_d|}$)t$_k$t$_{k'}$ vanishes at the symmetric point U=-2$\epsilon_d$.
%The essence of our model is a high-temperature mixed valent spin-quenching which allows a symmetric spin channel to become heavy free electrons. At low temperatures, these free electrons strongly couple in an orthogonal channel{\cite{coleman&tsvelik}} via a Kondo-like term that arises from antiferromagnetic interactions between the vanadium atoms--which is not permitted to win energetically due to the underlying frustration of the lattice, hence a very heavy fermi liquid of holes forms at low temperatures.
%our model--explained
% However, if the Hamiltonian takes a form similar to Eq. 1 with U$\sim min(E_c^{(+)},E_c^{(-)})$ at low temperatures, the conduction electrons can be scattered at $\epsilon_F$ when r$\ne 0$. This leads one to expect that the bare conductance should be modified by the phase shift resulting from the power-law density of states. At resonance, Eq. 1 becomes the non-interacting Anderson model whose phase shift is independent of temperature satisfying,
%\begin{equation}
%\delta_0(\epsilon) = \frac{\pi}{2} - \tan^{-1}(\frac{\epsilon_d - \epsilon - \frac{\Gamma_0}{\pi}|\frac{\epsilon}{D}|^r sgn(\epsilon)\tan(\frac{\pi r}{2}) - \frac{2\Gamma_0 sgn(\epsilon)}{\pi(r-1)}|\frac{\epsilon}{D}| _2F_1(1,\frac{1-r}{2};\frac{3-r}{2};(\frac{\epsilon}{D})^2)}{\Gamma_0 |\frac{\epsilon}{D}|^r})),
%\end{equation}
Following Glazman and Ra\u{\i}kh{\cite{raikh}},
%we see that even for a resonant level,
the effects of correlation in the leads (r=0 for free electrons) is to decrease the effective conductance relative to unitarity at the Kondo limit as G = G$_0$sin$^2$($\delta$), a result one might have anticipated from Kane and Fisher's{\cite{kanef}} repulsive interactions in Luttinger liquids (there G$_0$ $\rightarrow$ gG$_0$, g: Luttinger parameter). In the Kondo limit, Ref. 3 found that at p-hs, the pKm flows to a strong coupling fixed point corresponding to the U=0 limit of the pAm provided $\Gamma$ exceeds a critical value which increases monotonically with r to diverge at $\frac{1}{2}$. The phase shift of this model at the Fermi energy is $\delta = \frac{\pi}{2} (1-r$sgn$(\epsilon))${\cite{gonzalez,chen}}. However, this strong coupling fixed point is unstable with respect to the potential scattering term V. At the symmetric point, V=0, one should{\cite{ingersent1}} enter the particle-hole symmetric fixed point below the Kondo scale T$_{\chi} \sim \delta E(\frac{J-J_c}{J})^{\frac{1}{r}}$. If V$\ne$0, for 0$<$r$<$r$^*$=0.375, the initial flow is towards the p-hs fixed point. Below a temperature scale T$_{\chi'} = |\frac{V}{\delta E}|^{\frac{1}{r}}$T$_{\chi}$ the problem flows to an asymmetric strong coupling fixed point{\cite{ingersent1}} whose phase shift{\cite{gonzalez}} about the Fermi energy is $\pi$. While such a phase shift would be opaque to conductance measurements, it is interesting to note that in Ahanorov-Bohm experiments{\cite{yang}} a flux corresponding to a phase-shift of $\pi$ has been apparently observed, and remains unexplained.
In Fig. 1 the effect of p-hs breaking is sketched. While one expects to see evidence of Kondo physics (r$<$r$^*$) at low temperatures of height comparable to the resonant peak, the width of the peak (in V$_G$) will similarly sharpen at T$\rightarrow$0 to approach a $\delta$-function at T=0, unlike at r=0. While single-walled metallic carbon nanotubes can be created with a sufficiently small power law TDOS, their 1 D nature introduces new complications. %Let us briefly comment on systems which might exhibit such a small power law in their tunneling density of states. For metallic carbon nanotubes it is believed that{\cite{egger}} the tunneling density of states from the end/bulk of the tube follows the power law $\alpha_{end} = \frac{\frac{1}{g}-1}{4}, \alpha_{bulk} = \frac{\frac{1}{g} + g -2}{8}$ from which we see that the (pKm) requirement 0$<$r$<$ r$^*$ imposes the constraints 1$>$g$_{end}>$0.4 or 1$>$g$_{bulk}> \frac{5-\sqrt{21}}{2}$ on the electron-electron interactions. An additional issue to consider{\cite{kanef}} at low temperatures is that any such system must be coupled to Fermi liquid leads such that at temperatures low with respect to the thermal coherence length L$_T$ = $\frac{\hbar v_{F}}{k_B T}$ one would expect see a resumption of regular Kondo physics.
The relation of these results with those of 1 D treatments{\cite{1dstuff}} where to our knowledge no critical Kondo coupling has yet been found will be discussed elsewhere{\cite{us}}. However, a small power-law TDOS (due to interactions between electrons) could be obtained with 2 D leads with a non-negligible resistance{\cite{slot}}.
%----------------------------------------------------------------------
% Reference section
%
% List each reference with a separate \bibitem{} command. The
% argument contains the label that is used in the \cite{} command
% in the main text
%
% e.g.
%
% This follows our pioneering work on TdB2\cite{TdB2}.
%
% \bibitem{TdB2}
% J. Doe, J. Doe, and J. Dupont, J. Irrep. Res. 10 (2000) 1000.
\begin{thebibliography}{00}
\bibitem{revival} For a review see: L. Kouwenhoven and L. Glazman, Physics World, Vol. 14, No. 1, 33 (2001).
\bibitem{withoff} D. Withoff {\it{et al.}}, Phys. Rev. Lett. {\bf{64}}, 1835 (1990). %Kondo, Johnston and Miller, PRB {\bf{59}}, (1999), 2609.
%\bibitem{others}
\bibitem{gonzalez}C. Gonzalez-Buxton {\it{et al.}}, Phys. Rev. B {\bf{57}}, 14254 (1998).%Urano, Nohara, Kondo, Sakai, H.Takagi, Shiraki and Okubo, PRL {\bf{85}}, (2000), 1052.
%\bibitem{RKKY}M. A. Ruderman and C. Kittel, Phys. Rev. {\bf{96}}, 99 (1954); T. Kasuya, Prog. Theoret. Phys. {\bf{16}}, 45 (1956); K. Yoshida, Phys. Rev. {\bf{106}}, 893 (1977).
%Anisimov, Korotin, Zolfl, Pruschke, LeHur and Rice, PRL {\bf{83}}, (1999), 364.
\bibitem{chen} K. Chen {\it{et al.}}, J. Phys. Cond. Matt. {\bf{7}}, L491 (1995).\bibitem{ingersent1} K. Ingersent, Phys. Rev. B {\bf{54}}, 11936 (1996).%Varma, PRB {\bf{60}}, no.10, (1999), R6973.
\bibitem{nazarov} Yu. V. Nazarov {\it{et al.}}, Phys. Rev. Lett. {\bf{91}}, 126804 (2003).%Kusunose, Yotsuhasi and Miyake, PRB {\bf{62}}, (2000), 4403.
\bibitem{raikh} L. I. Glazman and M. {\'E}. Ra\u{\i}kh, JETP Lett. {\bf{47}}, 452 (1988); T. K. Ng and P. A. Lee, Phys. Rev. Lett. {\bf{61}}, 1768 (1988).%Fulde, Yaresko, Zvyagin and Grin, cond-mat/0101455 (2001).
\bibitem{kanef} C. L. Kane {\it{et al.}}, Phys. Rev. B {\bf{46}}, 15233 (1992).
%Burdin, Grempel and Georges, cond-mat/0107288 (2001).
\bibitem{yang}Y. Ji {\it{et al.}}, Science {\bf{290}} 779 (2000).% }Coleman and Tsvelik, PRB {\bf{59}}, (1998), 12757.
%\bibitem{egger} R. Egger, A. Bachtold, M. S. Fuhrer, M. Bockrath, D. H. Cobden and P. L. McEuen, cond-mat/0008008 (Interacting Electrons in Nanostructures, ed. R. Haug and H. Schoeller (Springer)).
\bibitem{1dstuff}K. Le Hur, Phys. Rev. B {\bf{59}}, R11637 (1999); P. Fr\"{o}jdh and H. Johannesson, Phys. Rev. B {\bf{53}}, 3211 (1996); A. Furusaki and N. Nagaosa, Phys. Rev. Lett. {\bf{72}}, 892 (1994); E. H. Kim, cond-mat/0106575.
\bibitem{us}J. Hopkinson {\it{et al.}}, cond-mat/0407165.
\bibitem{slot}E. Slot {\it{et al.}} cond-mat/0403385; A. Bachtold {\it{et al.}}, Phys. Rev. Lett. {\bf{87}}, 166801 (2001); A. N. Cleland {\it{et al.}}, ibid. {64} 1565 (1990); G. L. Ingold {\it{et al.}} in {\it{Single charge tunneling}}, ed. H. Grabert {\it{et al.}}, Plenum Press 1992.
\end{thebibliography}
\begin{figure}
\includegraphics[scale=0.45]{hopkinson-0581-f1.eps}
\caption{\label{figure}A schematic sketch for r=0.2 of the conductance of the Coulomb blockade valley (a) at p-hs U=-2$\epsilon_d$; (b) for V$\ne$0.}
% consistent with $S = \frac{1}{2}$ from 10K to room temperature, and decreasing slope at higher temperatures.[12],[1],[2],[13]}
%{\cite{fujiwara,kondo,urano,muhtar}}}
\end{figure}
%----------------------------------------------------------------------
% Figures and Tables
%
% Insert figures and tables at the end of the document unless you
% are familiar with the LaTeX positional options.
%
% \begin{figure}
% \centering
% \includegraphics{filename.eps}
% \caption{Insert figure caption here}
% \end{figure}
%
% \begin{table}
% \centering
% \begin{tabular}
% Insert table here
% \end{tabular}
% \caption{Insert table caption here}
% \end{table}
%
% Please refer to other LaTeX documentation for help on using these
% environments.
%----------------------------------------------------------------------
% Terminate document
\end{document}
%----------------------------------------------------------------------