1: \documentstyle[aps,epsfig,multicol]{revtex}
2: %\documentclass[prl,aps,twocolumn,floats,psfig]{revtex}
3: %\documentclass[prl,aps,psfig,multicolumn,floats]{revtex}
4:
5: \renewcommand{\vec}[1]{{\bf #1}}
6: \newcommand{\nin}{\noindent}
7: \newcommand{\be}{\begin{equation}}
8: \newcommand{\ee}{\end{equation}}
9: \newcommand{\bea}{\begin{eqnarray}}
10: \newcommand{\eea}{\end{eqnarray}}
11: \newcommand{\lb}{\left[}
12: \newcommand{\rb}{\right]}
13: \newcommand{\lp}{\left(}
14: \newcommand{\rp}{\right)}
15: \newcommand{\lf}{\left\{}
16: \newcommand{\rf}{\right\}}
17: \newcommand{\ftn}{\footnote}
18: \renewcommand{\H}{{\cal H}}
19: \newcommand{\LR}{\langle l|r\rangle}
20: \newcommand{\lra}{\leftrightarrow}
21:
22: \begin{document}
23: %%%%%%%%% from BDS (begin)
24: \def\breakon{\end{multicols}\widetext\vspace{-.2cm}
25: \noindent\rule{.48\linewidth}{.3mm}\rule{.3mm}{.3cm}\vspace{.0cm}}
26:
27: \def\breakoff{\vspace{-.2cm}
28: \noindent
29: \rule{.52\linewidth}{.0mm}\rule[-.27cm]{.3mm}{.3cm}\rule{.48\linewidth}{.3mm}
30: \vspace{-.3cm}
31: \begin{multicols}{2}
32: \narrowtext}
33: %%%%%%%%% from BDS (end)
34:
35: \draft
36:
37: %\twocolumn[\hsize\textwidth\columnwidth\hsize\csname @twocolumnfalse\endcsname
38: \title{Electron properties of Carbon nanotubes in the field effect regime}
39:
40: \author{D. S. Novikov and L. S. Levitov}
41: \address{
42: \mbox{Physics Department, Center for Materials Sciences \& Engineering,
43: Massachusetts Institute of Technology,
44: Cambridge, MA 02139}
45: }
46: %\date{\today}
47:
48: \maketitle
49:
50: \begin{abstract}
51: Electron properties of Carbon nanotubes can change qualitatively
52: in a transverse electric field.
53: In metallic tubes the sign of Fermi velocity
54: can be reversed in a sufficiently strong field,
55: while in semiconducting tubes the effective mass can change sign.
56: These changes in the spectrum manifest themselves in a breakup
57: of the Fermi surface and in the energy gap suppression, respectively.
58: The effect is controlled by the field inside the tube which is screened
59: due to the polarization induced on the tube.
60: The theory of screening links it with the chiral anomaly for $1D$ fermions
61: and obtains a universal screening function determined
62: solely by the Carbon $\pi$ electron band.
63: \vskip2mm
64: \end{abstract}
65:
66: \bigskip
67:
68: \begin{multicols}{2}
69: \narrowtext
70:
71: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
72:
73: The possibility to change electron spectrum of Carbon nanotubes
74: by external field is of interest for basic research as well as
75: for nanoscale device engineering. Carbon nanotube (NT) is a $1D$ metal
76: or semiconductor depending on the chiral angle\cite{Dresselhaus}.
77: Metallic behavior can be suppressed
78: by parallel magnetic field that induces a minigap
79: at the band crossing\cite{B-paral}.
80: Similar minigaps appear in nominally metallic chiral NTs
81: due to the intrinsic curvature\cite{KaneMele97,exp-curvature,curv-gaps-exp}.
82: Novel properties are predicted for the BN tubes
83: having no inversion symmetry\cite{Mele01,Kral00}.
84:
85: Here we examine the changes induced in the NT electron spectrum
86: by transverse electric field ${\cal E}$
87: strong enough to mix different NT subbands:
88: %
89: \be\label{eq:Erequired}
90: e{\cal E}R \simeq \Delta_0\equiv \hbar v/R
91: ,\quad
92: {\cal E}\,[{\rm MV/cm}] \simeq 5.26/R^2\,[{\rm nm}^2],
93: \ee
94: %
95: where $R$ is the tube radius and $v$ is electron velocity.
96: In such a field the effect on electron spectrum is dramatic:
97: in metallic tubes the electron velocity $v=d\epsilon/dp$
98: can be reduced and even reverse the
99: sign, causing Fermi surface breakup, while
100: in semiconducting tubes the effective mass sign can change,
101: which is accompanied by strong suppression of the excitation gap
102: (Fig.\ref{fig:MNT,SNT}).
103:
104: The NT electron system in this regime can be a host of intriguing
105: many-body phenomena. The reduction of electron velocity in metallic tubes
106: leads to an increase of the dimensionless interaction $e^2/\hbar v$
107: that controls the Luttinger liquid properties\cite{all-Luttinger}.
108: One expects this to enhance the intrinsic, interaction-induced energy gap
109: predicted to be small
110: in pristine NTs\cite{all-Luttinger}.
111:
112: Even more peculiar is the
113: negative $v$ state with intertwining electron and hole Fermi surfaces
114: (Fig.\ref{fig:MNT,SNT} top). This system provides a
115: realization of a metallic state unstable with respect
116: to electron-hole pairing into excitons. Such an instability,
117: long-envisioned \cite{excitonic} by Mott, Keldysh and Kopaev, and others,
118: is especially interesting for the
119: mirror symmetric electron and hole
120: bands, described by an analog of the BCS theory.
121: The chiral gauge symmetry (\ref{eq:chiralGtrans}) of the NT electron
122: Hamiltonian discussed below
123: eliminates the interband matrix elements of particle density.
124: This makes the phase of the excitonic order parameter
125: a gapless Goldstone field, similar to the BCS order parameter.
126: Despite its simplicity, no system with such properties has been
127: unambiguously identified so far, and it is thus possible that nanotubes
128: in a transverse field present a unique opportunity to study this
129: phenomenon.
130:
131: The field (\ref{eq:Erequired}) required to create this state is to be
132: achieved within the tube where the external field is partially
133: screened\cite{Benedict95}.
134: However, because of the discrete bands with relatively large
135: interband separation
136: $\Delta_0\,[{\rm eV}]\equiv \hbar v/R=0.53/R\,[{\rm nm}]$,
137: the transverse field penetrates in the NT fairly well.
138: We find the screening factor to be $R$-independent and close to $5$
139: (in accord with \cite{Benedict95}),
140: for both metallic and semiconducting tubes.
141: With the screening taken into account the numbers
142: for the required field remain feasible.
143: (Fields up to $50\,{\rm MV/cm}$ have been
144: demonstrated recently in $2D$ gated structures\cite{Battlog}.)
145:
146: Unexpectedly, the problem of transverse field screening has
147: a relation with the chiral anomaly.
148: A significant part of electron energy in an external field, in the
149: massless Dirac model (standard for NT),
150: arises from the
151: effects at the Fermi sea bottom, where a regularization of this
152: model is required.
153: However, the anomaly links the regularized energy
154: with the properties near the Fermi level (the number of fermion
155: species and their velocity),
156: and thereby generates
157: a universal screening function determined only
158: by the Carbon $\pi$ electron band.
159:
160: We consider the NT at half-filling ignoring charge accumulation
161: due to gating. Gating in itself
162: will not modify the transverse field within the tube, since
163: a uniformly charged cylinder
164: is equipotential. Charging may affect
165: the inner NT field indirectly via changing screening,
166: but this effect should not be significant at moderate gating.
167:
168: The $\pi$ electron Carbon band is described by a nearest-neighbor
169: tight-binding problem
170: %% on a honeycomb lattice,
171: $\epsilon \psi_r=t\sum_{r'}\psi_{r'}$
172: with the hopping amplitude $t\simeq 3\,{\rm eV}$.
173: The states
174: with small $|\epsilon|\ll t$
175: are described,
176: separately at each of the $K$ and $K'$ points, by a massless Dirac
177: Hamiltonian
178: %
179: \be\label{eq:DiracKK'}
180: {\cal H}_0=-i\hbar v(\sigma_y \partial_x-\sigma_x \partial_y)
181: ,\quad v=3ta_{\rm c-c}/2\hbar
182: \ee
183: %
184: For a NT in the presence
185: of a transverse electric field ${\cal E}$,
186: %
187: \be\label{eq:Hdirac}
188: {\cal H}_0=\hbar v \lp i\sigma_x \partial_y + \sigma_y k \rp
189: -e{\cal E}R \cos (y/R)
190: \ee
191: %
192: with $k$ the longitudinal momentum and $R$ the NT radius. The
193: boundary conditions are quasiperiodic:
194: %
195: \be\label{eq:q-periodicity}
196: \psi(y+2\pi R)=e^{2\pi i\delta}\psi(y)
197: \ ,\quad
198: \delta=\cases{0\ ,& metallic\cr
199: \pm\frac13\ ,& semic.}
200: \ee
201: %
202: The effects of NT curvature\cite{KaneMele97} as well as of
203: a parallel magnetic field\cite{B-paral} can in be included
204: by slightly shifting $\delta$ away from the ideal values
205: (\ref{eq:q-periodicity}).
206:
207: We employ a chiral gauge transformation
208: %
209: \be\label{eq:chiralGtrans}
210: \psi(y)=e^{-i\sigma_x\phi(y)}\widetilde\psi(y)
211: \ ,\quad
212: \phi(y)=\frac{e{\cal E}R^2}{\hbar v}\sin (y/R)
213: \ee
214: %
215: which preserves the condition (\ref{eq:q-periodicity}) and
216: turns Eq.(\ref{eq:Hdirac}) into
217: %
218: \be\label{eq:Htransformed}
219: \widetilde{\cal H}_0=\hbar v \lp i\sigma_x \partial_y + k e^{2i\sigma_x\phi(y)}\sigma_y \rp
220: \ee
221: %
222: The transformed Hamiltonian reveals that,
223: in particular, the spectrum at $k=0$ is totally independent of
224: the transverse field (Fig.\ref{fig:MNT,SNT}).
225:
226:
227: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
228: \begin{figure}[t]
229: \centerline{\psfig{file=fig1.eps,width=3.5in,height=2.5in}}
230: \vspace{-3mm}
231: \centerline{
232: \begin{minipage}[t]{3.5in}
233: \vspace{0pt}
234: \centering
235: \includegraphics[width=3.5in]{fig2.eps}
236: \end{minipage}
237: \hspace{-1.5in}
238: \begin{minipage}[t]{1.4in}
239: \vspace{0pt}
240: \centering
241: \includegraphics[width=1.4in]{fig3.eps}
242: \end{minipage}
243: }
244: \vspace{0.5cm}
245: \caption[]{
246: Electron bands transformation:
247: velocity reversal in metallic NT (top);
248: effective mass sign change in semiconducting NT (bottom).
249: The bands are shown for the dimensionless transverse field
250: $u=e{\cal E}R/\Delta_0$
251: below and above critical.
252: {\it Inset:} Energy gap suppression in a semiconducting NT.
253: }
254: \label{fig:MNT,SNT}
255: \end{figure}
256: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
257:
258: The spectrum of a metallic NT
259: near the band crossing at $kR\ll1$ can be found by employing the
260: degenerate perturbation theory. We project the Hamiltonian
261: (\ref{eq:Htransformed})
262: on the two states $|\uparrow\rangle$,
263: $|\downarrow\rangle$ degenerate at $k=0$, and obtain
264: %
265: \be\label{eq:J0,u}
266: \epsilon(k)=\pm \lb \hbar v J_0(2u)\rb\, k
267: \ ,\quad
268: u\equiv e{\cal E}R/\Delta_0 = e{\cal E}R^2/\hbar v
269: \ee
270: %
271: Electron velocity changes sign at the roots of
272: the Bessel function $J_0$, first at $2u=\mu_1\approx 2.405$
273: (Fig.\ref{fig:MNT,SNT}). At $u$ above critical the Fermi surface
274: fractures: an additional small pocket appears for each spectral
275: branch.
276:
277: The level shifts in semiconducting NT at small $k$ are given by
278: the second order perturbation theory in the $k$-term of
279: the transformed Hamiltonian (\ref{eq:Htransformed}):
280: %
281: \bea
282: && \epsilon_n^\pm(k)=\pm\Delta_0 \lp (n+\delta)+A_n (kR)^2 \rp
283: \\
284: && A_n=
285: \sum_{m=-\infty}^\infty J_m^2(2u)\frac{2(n+\delta)}{4(n+\delta)^2-m^2}
286: \eea
287: %
288: For $\delta=1/3$ the curvature of the lowest band $A_0$ changes sign at
289: $u_c\approx 0.6215$ (Fig.\ref{fig:MNT,SNT}).
290: This leads to a singular behavior of the excitation gap
291: which is constant at $u<u_c$ and sharply decreases at $u>u_c$
292: (Fig.\ref{fig:MNT,SNT} inset). This occurs because of
293: the lowest excitation energy shifting from $k=0$ at $u<u_c$
294: to $k\ne 0$ at $u>u_c$. The threshold-like suppression of the gap
295: can be detected by a transport measurement in a thermally activated regime.
296:
297: The chiral gauge symmetry (\ref{eq:chiralGtrans}) that protects
298: the spectrum at $k=0$ is a distinct feature of
299: the Dirac model (\ref{eq:DiracKK'}), (\ref{eq:Hdirac}).
300: The $\pi$ electron tight-binding
301: problem, in the next-lowest gradient order,
302: generates a correction to the Hamiltonian (\ref{eq:DiracKK'})
303: violating the symmetry (\ref{eq:chiralGtrans}):
304: %
305: \be\label{eq:H1}
306: {\cal H}={\cal H}_0+\lambda
307: e^{-\frac{3i}2\theta \sigma_z}\!\! \left[
308: \sigma_x\left(\partial_x^2\!-\!\partial_y^2\right)
309: \! -\!2\sigma_y\partial_x\partial_y
310: \right] e^{\frac{3i}2\theta \sigma_z}
311: \ee
312: %
313: with $\lambda\!=\!\frac14 a_{\rm c-c}\hbar v$ and
314: $\theta$ the NT chiral
315: angle\cite{Dresselhaus}.
316: The transformation (\ref{eq:chiralGtrans}) applied to (\ref{eq:H1})
317: gives a minigap\cite{E-perp}
318: %
319: \be\label{eq:gap-perp}
320: \Delta_{\cal E} = |\sin 3\theta|\, (a_{\rm c-c}/8\hbar v)\,(e{\cal E}R)^2
321: \ee
322: %
323: Since $8R \gg a_{\rm c-c}$,
324: the minigap (\ref{eq:gap-perp})
325: is too small to alter the behavior at the energies of
326: interest, $\epsilon\simeq\Delta_0$.
327:
328: The main effect of electron interaction is screening of the inner field that
329: couples to the electron motion. Here we derive the relation between
330: the inner and outer fields.
331: We first show how the screening problem is reduced to the calculation
332: of the NT electron energy in the presence of an external field.
333: Hereafter we measure all energies in the units of $\Delta_0=\hbar v/R$
334: and use dimensionless field $u=e{\cal E}R/\Delta_0$.
335: From Gauss' law, the fields inside and outside the tube are related
336: with the induced surface charge density per one fermion species
337: (spin and valley) by
338: %
339: \be\label{eq:Gauss}
340: {\cal E}_{\rm ext}={\cal E}+{\textstyle \frac12} \cdot 4\pi \cdot 4\sigma
341: \ee
342: %
343: where the factor $1/2$ accounts for depolarization in the cylindrical geometry.
344: In Eq.(\ref{eq:Gauss}) we projected the actual charge density on
345: the $\cos\varphi$ harmonic as $\sigma(\varphi) \to 4\sigma \cos\varphi$,
346: ignoring the higher order harmonics. (Here $\varphi\equiv y/R$.)
347:
348: To obtain the $\cos\varphi$ harmonic of the induced charge,
349: we evaluate the dipole moment per unit length
350: as $P=- dW({\cal E})/d{\cal E}$, where $W({\cal E})$ is
351: the energy of one fermion species
352: as a function
353: of the inner field. Combining this with the relation
354: $\sigma=P/(\pi R^2)$
355: and with the Gauss' law (\ref{eq:Gauss}),
356: and passing to dimensionless $u_{\rm ext}$, $u$,
357: we obtain
358: %
359: \be\label{eq:UU'W}
360: u_{\rm ext}=u+8\frac{e^2}{\hbar v}P(u)
361: \ee
362: %
363: After the dipole moment $P(u)$ is known
364: Eq.(\ref{eq:UU'W}) can be solved for the inner field $u$ in terms
365: of the outer field $u_{\rm ext}$.
366:
367: We consider the general problem of electron energy
368: in a transverse field in a free particle model.
369: The electron levels $\epsilon_{n,k}$ perturbed by the field
370: can be easily found numerically
371: at each value of the longitudinal momentum
372: $k$
373: by using a transfer matrix for Eq.(\ref{eq:Htransformed}).
374: The level shifts
375: $\delta \epsilon_{n,k}=\epsilon_{n,k}(u)-\epsilon_{n,k}^{(0)}$ decrease at large $|n|$, and
376: the series
377: the total change of the occupied states energy
378: %
379: \be\label{E0}
380: E_0(k)=\sum\limits_{n=-\infty}^{+\infty} \delta \epsilon_{n,k}
381: \quad ({\rm such\ that\ } \epsilon_{n,k}(u)<0)
382: \ee
383: %
384: rapidly converge at $n\to\pm\infty$.
385:
386: There are two basic problems with Eq.(\ref{E0}):
387: 1) Due to an upward shift of the filled levels (Fig.1),
388: $E_0$ is positive and also has positive derivative $dE_0/du$.
389: Hence Eq.(\ref{E0}) leads to the dipole moment $P_0=-dE_0/du$
390: opposite to the field, i.e. to an
391: unphysical ``diamagnetic'' polarization sign instead of
392: the expected ``paramagnetic'' effect.
393: 2) The dependence of the energy $E_0$ on the longitudinal wavevector $k$
394: leads to an ultraviolet divergence in the integral
395: $P=\int P(k)dk$, because $E_0(k)$ increases with $|k|$, saturating
396: at $|k|R\gg1$ at an asymptotic value $\frac12 u^2$.
397:
398: Both difficulties are resolved by taking into account
399: a fundamentally important contribution to the energy
400: that arises due to the effects at the {\it Fermi sea bottom}.
401: Physically, the finite electron band width
402: %% ($\simeq 12\,{\rm eV}$ in Carbon)
403: invalidates the massless Dirac
404: approximation at large negative energies.
405: This contribution, however,
406: depends solely on the number of Dirac fermion species and their velocity $v$,
407: and is totally insensitive to any other details including
408: the longitudinal momentum $k$ value. We find that
409: %
410: \be\label{eq:Eanomaly}
411: E_{\rm anom}=-{\textstyle\frac12}u^2
412: \ee
413: %
414: for each fermion species.
415: Remarkably, Eq.(\ref{eq:Eanomaly}) can be obtained without detailed discussion
416: of the behavior at the interatomic length scales
417: --- the universality of
418: Eq.(\ref{eq:Eanomaly}) is rooted in the physics of {\it the chiral anomaly}
419: in the $1+1$ fermion problem.
420: The resulting total energy integral
421: %
422: \be\label{eq:Wintegral}
423: W = \int_{-\infty}^\infty \lp E_0(k)+E_{\rm anom}\rp {\textstyle \frac{dk}{2\pi}}
424: \ee
425: %
426: converges after $E_0(k)$ is offset by $E_{\rm anom}$ (Fig.\ref{fig:P(k)}).
427:
428: We evaluate the energy (\ref{E0}) and derive the anomaly (\ref{eq:Eanomaly})
429: for a weak field $u\ll 1$. The NT bands at $u=0$ are
430: %
431: \be
432: \epsilon_n^\pm(k)=\pm\sqrt{(n+\delta)^2+k^2}\ ,\quad -\infty<n<+\infty
433: \ee
434: %
435: In a half-filled system with just the $\epsilon_n^-(k)\!\!<\!\!0$ bands
436: filled, the external field $u$ changes the
437: Fermi sea energy by
438: %
439: \be
440: W=- \int {\sum_n}'
441: %^{\rm (reg)}
442: \delta \epsilon_n^-(k) \frac{dk}{2\pi}
443: \ee
444: %
445: Here the superscript in $\sum'$ indicates
446: regularization by truncating the interaction with
447: the external field at a certain large negative energy.
448: We check that this contribution
449: to the energy is independent of
450: the details of truncation and obtain the anomaly (\ref{eq:Eanomaly})
451: by choosing a convenient truncation scheme.
452:
453: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
454: \begin{figure}
455: \centerline{\psfig{file=fig4.eps,width=3.5in,height=2.5in}}
456: \vspace{0.1cm}
457: % \centerline{\psfig{file=spectrum-saw2.eps,width=3.5in,height=2.5in}}
458: % \vspace{0.5cm}
459: \caption[]{
460: Dipole moment $P(k)=-d(E_0(k)+E_{\rm anom})/du$
461: per one fermion species in a semiconducting NT
462: as a function of $k$. Note that the energy anomaly (\ref{eq:Eanomaly})
463: cancels with $E_0(k)$
464: at $kR\gg1$, assuring convergence of $P_{\rm total}=\int P(k)dk/2\pi$.
465: Note also that $P(k\to 0)$ is dominated by the anomaly, since
466: $E_0=0$ at $k=0$ due to the chiral gauge invariance
467: (\ref{eq:chiralGtrans}).
468: }
469: \label{fig:P(k)}
470: \end{figure}
471: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
472:
473: The level shifts $\delta \epsilon_n^-(k)$, in the second order of
474: the perturbation theory in the external field
475: $\hat V\!\!=\!\!-e{\cal E}R\cos\varphi$, are
476: %
477: \be\label{eq:sum-perturb}
478: \delta \epsilon_n^-=
479: \sum_{m}\frac{|\langle m^+|\hat V|n^-\rangle|^2}{\epsilon_n^- -\epsilon_m^+}
480: + \sum_{m}
481: \frac{|\langle m^-|\hat V|n^-\rangle|^2}{\epsilon_n^- -\epsilon_m^-}
482: \ee
483: %
484: where the superscript $\pm$ indicates the electron and hole branches
485: and the $k$ dependence is suppressed. Due to the integration over $\varphi$
486: with $\hat V\propto\cos\varphi$ in the matrix elements the only nonzero
487: terms in (\ref{eq:sum-perturb}) are those with $m=n\pm 1$.
488:
489: We now show that the sums over $\epsilon_m^+$ and $\epsilon_m^-$ in
490: (\ref{eq:sum-perturb}), respectively, give the regular and the anomalous
491: contributions to the total energy $W=\sum_n'\delta \epsilon_n^-$.
492: Different behavior of the two sums under regularization stems
493: from their different convergence type. Individual terms in the sum over
494: $\epsilon_m^+$ decrease rapidly at large $m$, so that
495: the series for $W$
496: is absolutely convergent. On the other hand, in the sum
497: over $\epsilon_m^-$ the terms do not change at large $m$ and thus the
498: corresponding contribution to $W$ is given by poorly
499: convergent and regularization-sensitive series.
500:
501: Taking from (\ref{eq:sum-perturb}) just the terms with $\epsilon_m^+$,
502: evaluating the matrix elements $\langle m^+|\hat V|n^-\rangle$ and
503: summing over $n$ yields
504: %
505: \be\label{eq:Ereg-perurb}
506: E_0(k) =
507: \frac{u^2}4 \sum_n \frac{\epsilon_n^+(k)\epsilon_{n'}^+(k)\!-\!(n\!+\!\delta)(n'\!+\!\delta)\!-\!k^2}{\epsilon_n^+(k)\epsilon_{n'}^+(k)(\epsilon_n^-(k) - \epsilon_{n'}^+(k))}
508: \ee
509: %
510: with $n'=n+1$. The sum (\ref{eq:Ereg-perurb}) rapidly converges at large $n\to\pm\infty$
511: and can be easily evaluated numerically.
512:
513: Now we consider the sum of the level shifts $W={\sum}_n'\delta \epsilon_n^-$
514: taking into account only the second term in
515: (\ref{eq:sum-perturb}). At the first sight this sum is identically zero.
516: Indeed, due to the symmetry
517: $\langle m^-|\hat V|n^-\rangle = \langle n^-|\hat V|m^-\rangle$,
518: in the sum over $n$ with $m=n\pm1$ all the terms cancel in pairs.
519: However, truncation of the interaction at a large negative energy
520: compromises the cancellation and yields a finite result. If one sets
521: $\langle m^-|\hat V|n^-\rangle =0$ for all
522: $|m|$ or $|n|$ exceeding a large number $N$, there will be just two terms
523: in the sum over $n$ that do not cancel:
524: %
525: \be\label{eq:Nterms}
526: E_{\rm anom}=
527: \frac{|\langle N'^-|\hat V|N^-\rangle|^2}{\epsilon_N^- -\epsilon_{N'}^-}
528: +
529: \frac{|\langle \!-\!N'^-|\hat V|-\!N^-\rangle|^2}{\epsilon_{-N}^- -\epsilon_{-N'}^-}
530: \ee
531: %
532: with $N'=N-1$.
533: Evaluating the matrix elements and energy levels is straightforward because
534: at large $N\gg |k|$ one can set $k=0$.
535: The result, coinciding with (\ref{eq:Eanomaly}), is robust
536: under a change of the regularization.
537:
538: The expression (\ref{eq:Eanomaly}) for the energy anomaly,
539: derived above for the weak field, is in fact more general. To illustrate
540: this we consider a special case of zero longitudinal momentum $k=0$
541: and derive (\ref{eq:Eanomaly}) from bosonization, without using
542: perturbation theory in $u\ll 1$. After the problem (\ref{eq:Hdirac})
543: is bosonized in the standard way\cite{Stone},
544: using $\psi_{L,R}\propto e^{i\phi_{L}}, e^{-i\phi_{R}}$,
545: we obtain
546: a quadratic Hamiltonian
547: \be\label{H-bos}
548: {\cal H} = \int_0^{2\pi R}
549: \sum\limits_{j=L,R}\lb
550: {\textstyle \frac{\hbar v}{4\pi}} (\partial_{x}\phi_{j})^2
551: +{\textstyle \frac{e}{2\pi}}\partial_{x}\phi_{j}U(y)\rb d y
552: \ee
553: The second term in (\ref{H-bos}) representing interaction
554: with the external field $U(y)$ can be decoupled by a shift
555: $\phi_j \to \phi'_j - \frac1{\hbar v_F}\int_0^y U(y')dy'$.
556: The Hamiltonian for $\phi'_j$ takes the
557: form (\ref{H-bos}) with $U=0$, while the ground state energy
558: %
559: \be
560: \delta E = - \frac{e^2}{2\pi \hbar v_F} \int_0^{2\pi R}\!\!\! U^2(y)dy
561: \ee
562: %
563: is nothing but the anomaly (\ref{eq:Eanomaly}) scaled by $\Delta_0=\hbar v/R$.
564:
565: After adding the energies (\ref{eq:Ereg-perurb}) and (\ref{eq:Eanomaly}),
566: and integrating in (\ref{eq:Wintegral}) over $k$ numerically, one obtains
567: %
568: \be\label{Wlinearized}
569: W=-\frac{\alpha}2u^2,\quad
570: \alpha=\cases{0.196... & \ for $\delta=1/3$\cr
571: 0.179... & \ for $\delta=0$}
572: \ee
573: %
574: Eq.(\ref{eq:Gauss}) with the dipole moment $P = - dW/d{\cal E}$ and
575: the charge density $\sigma = P/\pi R^2$ yield the screening function
576: %
577: \be\label{u-ratio}
578: {\cal E}_{\rm ext} = \lp 1+8\alpha \, {\textstyle \frac{e^2}{\hbar v}} \rp {\cal E}
579: \ee
580: %
581: With $e^2/\hbar v=2.7$
582: this gives ${\cal E}_{\rm ext}/{\cal E} = 5.24$ for $\delta=1/3$, and
583: ${\cal E}_{\rm ext}/{\cal E} = 4.87$ for $\delta=0$.
584: The outer-to-inner field ratio $\simeq 5$ (see \cite{Benedict95} for another
585: derivation) renders
586: the required fields (\ref{eq:Erequired}) feasible\cite{tobepublished}.
587: Interestingly, the screening (\ref{u-ratio})
588: is independent of the tube radius $R$ and
589: is almost the same in the metallic and semiconducting NTs.
590: The latter is not surprising, since the screening
591: is absent in a single 1D mode approximation: the polarizability
592: is related with dipolar transitions between {\it different} subbands.
593:
594: The radius-independence of (\ref{u-ratio}) resembles an effect
595: of a dielectric constant. We note, however, that the change of the inner field
596: due to individual Carbon atoms polarizability is small
597: in $a_{\rm c-c}/2\pi R \ll 1$. The result (\ref{u-ratio})
598: reflects the semimetallic character of the $\pi$ electron
599: band with the density of states vanishing at the band center.
600:
601: In summary, nanotube electron states undergo interesting transformations
602: in the field effect regime, leading to novel phenomena in both
603: the single particle and many-body properties. The analysis of
604: screening, performed using a relation with the theory of chiral anomaly,
605: indicates that the fields required for the observation
606: of the proposed effects are in the experimentally feasible range.
607:
608: This work was supported by the MRSEC Program
609: of the National Science Foundation under Grant No. DMR 98-08941.
610:
611: %% \vskip-0.5cm
612: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
613:
614: \begin{thebibliography}{99}
615:
616: \bibitem{Dresselhaus} R. Saito, G. Dresselhaus, and M. S. Dresselhaus,
617: {\it Physical Properties of Carbon Nanotubes}
618: (Imperial College Press, London, 1998).
619:
620: \bibitem{B-paral}
621: H. Ajiki and T. Ando,
622: J. Phys. Soc. Jpn. {\bf 65}, 505 (1996);
623: %%
624: % Parallel B-field, experiment:
625: %% \bibitem{exp-B-parallel}
626: J. -O. Lee,
627: %%% {\em et al.},
628: J. R. Kim, J. J. Kim, J. Kim, N. Kim, J. W. Park, and K. H. Yoo,
629: %% Observation of magnetic-field-modulated energy gap in Carbon nanotubes,
630: Sol. Stat. Comm. {\bf 115}, 467 (2000).
631:
632: \bibitem{KaneMele97}
633: C. L. Kane, E. J. Mele,
634: %% Size, Shape, and Low Energy Electronic Structure of Carbon Nanotubes,
635: Phys. Rev. Lett. {\bf 78}, 1932 (1997).
636:
637: % Curvature effect in zig-zag NT, experiment:
638: \bibitem{exp-curvature}
639: C. Zhou, J. Kong, H. Dai,
640: %% Intrinsic electrical properties of individual single-walled Carbon
641: %% nanotubes with small band gaps,
642: Phys. Rev. Lett. {\bf 84}, 5604 (2000).
643:
644: \bibitem{curv-gaps-exp}
645: M. Ouyang, J. L. Huang, C. L. Cheung, C.M. Lieber,
646: %% Energy gaps in "metallic" single-walled carbon nanotubes,
647: Science {\bf 292}, 702 (2001)
648:
649: \bibitem{Mele01}
650: E. J. Mele, P. Kral,
651: Phys. Rev. Lett. 88, 056803 (2002);
652: %% Electric Polarization of Heteropolar Nanotubes as a Geometric Phase,
653: %% cond-mat/0111020;
654:
655: \bibitem{Kral00}
656: P. Kral, E. J. Mele, D. Tomanek,
657: Phys. Rev. Lett. 85, 1512 (2000)
658: %% Photogalvanic Effects in Heteropolar Nanotubes,
659: %% cond-mat/0004116
660:
661: \bibitem{all-Luttinger}
662: C. Kane, L. Balents, M. P. A. Fisher,
663: Phys. Rev. Lett. 79, 5086 (1997);
664: R. Egger, A. O. Gogolin, Phys. Rev. Lett. 79, 5082 (1997);
665: L. Balents, M. P. A. Fisher, Phys. Rev. B 55, R11 973 (1997);
666: Yu. A. Krotov, D.-H. Lee, S.~G.~Louie, Phys. Rev. Lett. 78, 4245 (1997).
667:
668: \bibitem{excitonic}
669: N. F. Mott, Phil. Mag. 6, 287 (1961);
670: L. V. Keldysh, Yu. V. Kopaev, Fiz. Tverd. Tela 6, 2791 (1964); [Soviet Phys.-Solid State 6, 2219 (1965)];
671: J. J. des Cloizeaux, Phys. Chem. Solids 26, 259-266 (1965);
672: B. I. Halperin, T. M. Rice, in Solid State Physics 21, 115-192 (1968).
673:
674: \bibitem{Benedict95}
675: L. X. Benedict, S. G. Louie, and M. L. Cohen,
676: %% Static polarizabilities of single-wall carbon nanotubes,
677: Phys. Rev. B{\bf 52}, 8541 (1995)
678:
679: \bibitem{Battlog}
680: J. H. Sch\"on, S. Berg, Ch. Kloc, B. Battlog,
681: %% Ambipolar pentacene field-effect transistors and inverters,
682: Science {\bf 287}, 1022 (2000);
683: J. H. Sch\"on, A. Dodabalapur, Z. Bao, Ch. Kloc, O. Schenker, B. Battlog,
684: %% Gate-induced superconductivity in a solution-processed organic polymer film,
685: Nature {\bf 410}, 189 (2001)
686:
687: \bibitem{Stone} M. Stone (ed.), ``Bosonization'' (World Scientific, 1994)
688:
689: \bibitem{tobepublished}
690: For the field $u\simeq 1$ the screening factor
691: is close to that found in the weak field $u\ll1$ (to be published).
692:
693: %% \bibitem{Tans}
694: %% S. J. Tans,
695: %%% {\em et al.},
696: %% M. H. Devoret, H. Dai, A. Thess, R. E. Smalley, L. J. Geerligs, and
697: %% C. Dekker,
698: %%% Individual single-wall carbon nanotubes as quantum wires,
699: %% Nature, {\bf 386} 474 (1997).
700:
701: %% \bibitem{Bockrath'97}
702: %% M. Bockrath, D. H. Cobden, P. L. McEuen, N. G. Chopra, A. Zettl, A. Thess, R. E. Smalley,
703: %%% Single-electron transport in ropes of carbon nanotubes,
704: %% Science {\bf 275}, 1922 (1997)
705:
706: %% \bibitem{McEuen'99}
707: %% P. L. McEuen,
708: %%% {\em et al.},
709: %% M. Bockrath, D. H. Cobden, Y.-G. Yoon, and S. G. Louie,
710: %%% Disorder, Pseudospins, and Backscattering in Carbon Nanotubes,
711: %% Phys. Rev. Lett. {\bf 83}, 5098 (1999).
712: %cond-mat/9906055
713:
714: %% \bibitem{B-perp}
715: %% S. Roche and R. Saito,
716: %% Effects of magnetic field and disorder on the electronic properties of
717: %% carbon nanotubes,
718: %% Phys. Rev. B {\bf 59}, 5242 (1999).
719:
720: \bibitem{E-perp}
721: The minigap (\ref{eq:gap-perp}) is $\simeq 8R/a_{\rm c-c}$ times
722: smaller than that found in
723: X. Zhou,
724: %% {\em et al.},
725: Hu Chen, Ou-Yang Zhong-can,
726: %% Electric Field-Induced Energy Gaps In Carbon Nanotubes: Prediction Of
727: %% Controllable Nanoscale Switching Devices
728: J. Phys.-Cond. Mat. {\bf 13} (27), L635 (2001)
729: %% cond-mat/0010124;
730: %% The Polarizability and Electric Field-Induced Energy Gaps In Carbon
731: %% Nanotubes,
732: %% cond-mat/0102094
733:
734:
735: % \bibitem{}
736:
737: \end{thebibliography}
738:
739: \end{multicols}
740: \end{document}
741: