1: %\documentclass{epl}
2: \documentclass[doublecol]{epl2}
3:
4: \usepackage{bm}% bold math
5: \usepackage{amsmath}
6:
7: \title{Vortex-boson duality in four space-time dimensions}
8: \author{M. Franz}
9:
10: \institute{
11: Department of Physics and Astronomy,
12: University of British Columbia, Vancouver, BC, Canada V6T 1Z1
13: }
14: \pacs{74.72.-h}{First PACS description}
15: \pacs{74.20.-z}{Second PACS description}
16: \pacs{11.25.Tq}{Third PACS description}
17:
18: \abstract{A continuum version of the vortex-boson duality in (3+1) dimensions is
19: formulated and its implications studied in the context of a pair
20: Wigner crystal in underdoped cuprate superconductors.
21: The dual theory to a phase fluctuating superconductor (or superfluid)
22: is shown to be a theory of bosonic strings interacting through a Kalb-Ramond
23: rank-2 tensorial gauge field. String condensation produces Higgs mass for the
24: gauge field and the expected Wigner crystal emerges as an interesting
25: space-time analog of the Abrikosov lattice.}
26:
27:
28: \begin{document}
29: \newcommand{\bk}{{\bf k}}
30: \newcommand{\bp}{{\bf p}}
31: \newcommand{\bv}{{\bf v}}
32: \newcommand{\bx}{{\bf x}}
33: \newcommand{\bX}{{\bf X}}
34: \newcommand{\tbq}{\tilde{\bf q}}
35: \newcommand{\tq}{\tilde{q}}
36: \newcommand{\bQ}{{\bf Q}}
37: \newcommand{\br}{{\bf r}}
38: \newcommand{\bR}{{\bf R}}
39: \newcommand{\bB}{{\bf B}}
40: \newcommand{\bA}{{\bf A}}
41: \newcommand{\bK}{{\bf K}}
42: \newcommand{\Bomega}{{\bf \Omega}}
43: \newcommand{\cS}{{\cal S}}
44:
45:
46: \maketitle
47:
48: When particles in quantum many body
49: systems interact very strongly standard perturbation
50: techniques break down and, in dimensions greater than 1, {\em dualities}
51: often provide the only insights into the physics of such systems.
52: Duality transformations, in general, map the strongly coupled sector of one
53: theory onto the weakly coupled sector of another. The original Kramers-Wanier
54: duality \cite{kramers1} for the Ising ferromagnet represents a prime example of
55: such a mapping. Dualities permeate modern statistical, condensed matter
56: and particle physics, and have emerged recently as a key tool in string
57: theory.
58:
59: In condensed matter physics perhaps the most useful and influential
60: duality is the one connecting vortices and bosons in two spatial dimensions
61: \cite{dasgupta,nelson,lee_fisher}. This duality, hereafter referred to as
62: Lee-Fisher duality, maps the system of interacting
63: bosons in (2+1)D onto a fictitious superconductor in an external magnetic field
64: whose flux in the temporal direction is proportional to the density of the
65: original bosons. It shows that Mott insulator, proximate to the phase
66: fluctuating boson condensate, can be viewed as the
67: Abrikosov vortex lattice of the dual superconductor. This deep connection
68: has been exploited in modeling systems ranging from quantum spins to
69: fractional quantum Hall effect, and most recently cuprate superconductors.
70:
71: In cuprates such considerations are motivated by the experimental
72: findings of static checkerboard patterns in the charge density of very
73: underdoped samples \cite{vershinin1,hanaguri1,hashimoto1} which have
74: been interpreted as evidence for a Cooper pair Wigner crystal (PWC)
75: \cite{chen1}. The latter can be most naturally understood by appealing
76: to the Lee-Fisher duality \cite{tesanovic1,anderson1,balents1}.
77: However, recent analysis of the vibrational modes
78: of such a PWC \cite{tami1,taillefer1} indicates that it is 3-dimensional
79: (in the sense that vibrations propagate in all 3
80: space dimensions) and it is thus unclear how the inherently two-dimensional
81: Lee-Fisher duality applies to this situation.
82: The problem can be stated as follows. A key role in the formulation
83: of the Lee-Fisher duality is
84: played by vortices which appear (in pairs of opposite vorticity)
85: near the transition to the Mott insulating
86: phase as quantum fluctuations of the
87: system. The dual relationship between bosons and vortices however
88: exists only in two spatial dimensions where the latter can be regarded
89: as {\em point particles}.
90: In three space dimensions vortices form oriented loops and can no longer
91: be thought of as particles.
92: %For this reason Lee-Fisher
93: %duality does not admit a straightforward generalization to higher dimensions.
94: The question thus arises how to understand the formation of a PWC in the
95: three dimensional phase fluctuating superconductor indicated by experiments
96: \cite{vershinin1,hanaguri1,hashimoto1}.
97:
98: In this Letter we point a way out of this conundrum by constructing a
99: (3+1) dimensional implementation of vortex-boson duality using a
100: representation of vortex loops as relativistic bosonic strings. We then
101: show that string condensation indeed produces a ground state that can be
102: characterized as an insulating crystal of Cooper pairs and discuss some of
103: its unique properties.
104:
105: We remark that the lattice formulation of such (3+1)D duality has been given
106: long time ago \cite{peskin1,savit1} and was used recently to study
107: exotic fractionalized phases \cite{wen1,senthil1} in (3+1)D. Here, by contrast,
108: we formulate a continuum version which shows that a boson (Cooper pair)
109: crystal can emerge from a phase fluctuating superfluid (superconductor) even
110: in the absence of any underlying lattice structure. This is exactly the limit
111: apparently relevant to cuprates \cite{tami1}. On the formal side this
112: continuum approach also reveals an intimate connection to the theory of
113: bosonic strings and enables us to employ in our calculations some of the
114: string theory technology.
115: %In addition the physics of the (3+1)D duality, which on the lattice tends to
116: %get obscured by the unwieldy notation, becomes transparent in the continuum
117:
118: We begin by considering a continuum theory of a superconductor in (3+1)
119: space-time dimensions defined by the Euclidean
120: partition function $Z=\int{\cal D}[\Psi,\Psi^*]\exp(
121: -\int_0^{\beta} d\tau\int d^3 x {\cal L})$ with $\Psi=|\Psi|e^{i\theta}$ a
122: scalar order parameter and the Lagrangian density
123: %
124: \begin{eqnarray}\label{L0}
125: {\cal L} = {1\over 2}\tilde K\left|\left(\partial_\mu -
126: {2ie}A_\mu\right)\Psi\right|^2 + U(|\Psi|^2).
127: \end{eqnarray}
128: %
129: The Greek index $\mu=0,1,2,3$
130: labels the temporal and spatial components of (3+1) dimensional vectors,
131: and we use natural units with $\hbar=c=1$. $U$ is a potential
132: function that sets the value of the order parameter $\Psi$ in the
133: superconducting state in the absence of fluctuations.
134: The electromagnetic vector potential $A$ is explicitly displayed in order
135: to track the charge content of various fields. If we allowed $A$ to fluctuate
136: then Eq.\ (\ref{L0}) would coincide with the well known Abelian Higgs model.
137:
138: We now focus on the fluctuations in the phase $\theta$ by fixing the amplitude
139: $|\Psi|=\Psi_0$ at the minimum of $U$,
140: %
141: \begin{eqnarray}
142: {\cal L} = {1 \over 2} K
143: \left(\partial_\mu \theta - {2e}A_\mu\right)^2,
144: \end{eqnarray}
145: %
146: where $K=\tilde{K}\Psi_0^2$ represents the phase stiffness. The first few
147: steps of the duality mapping proceed just as in (2+1)D. We first
148: decouple the quadratic term with a real auxiliary field, $W_\mu$, using the
149: familiar Hubbard-Stratonovich transformation, obtaining
150: %
151: \begin{equation}\label{eq:W}
152: {\cal L} = {1 \over 2K}W^2_\mu +iW_\mu(\partial_\mu\Theta - {2e}A_\mu)+
153: iW_\mu(\partial_\mu\theta_s).
154: \end{equation}
155: %
156: We have also decomposed the phase into a
157: smooth part $\theta_s$ and singular part $\Theta$ containing vortex
158: lines.
159:
160: Gaussian integration over $\theta_s$ leads to a constraint
161: %
162: \begin{equation}\label{con}
163: \partial_\mu W_\mu=0,
164: \end{equation}
165: %
166: which reflects conservation of electric charge. In (2+1)D one enforces this
167: constraint by expressing $W_\mu$ as a curl of an auxiliary gauge field.
168: The curl operation, however, is meaningful only in 3 dimensions and herein
169: lies the difficulty with higher dimensional duality. In (3+1)D we may
170: enforce the constraint (\ref{con}) by writing
171: %
172: \begin{equation}\label{res}
173: W_\mu=\epsilon_{\mu\nu\alpha\beta}\partial_\nu B_{\alpha\beta},
174: \end{equation}
175: %
176: where $\epsilon_{\mu\nu\alpha\beta}$ is the totally antisymmetric tensor and
177: $B_{\alpha\beta}$ is antisymmetric rank-2 tensor gauge field.
178: Substituting Eq.\ (\ref{res}) back into the Lagrangian and performing
179: integration by parts in the term containing $\Theta$ we obtain
180: %
181: \begin{equation}\label{L2}
182: {\cal L} = {H^2_{\alpha\beta\gamma} \over 3K}
183: -iB_{\alpha\beta}(\epsilon_{\alpha\beta\mu\nu}\partial_\mu\partial_\nu\Theta)
184: -i2e(\epsilon_{\mu\nu\alpha\beta}\partial_\nu B_{\alpha\beta})A_\mu,
185: \end{equation}
186: %
187: where $H_{\alpha\beta\gamma}=\partial_\alpha B_{\beta\gamma}
188: +\partial_\beta B_{\gamma\alpha}+\partial_\gamma B_{\alpha\beta}$ is the
189: tensorial field strength which should be thought of as a generalization
190: of the Maxwell field strength $F_{\mu\nu}$.
191:
192: The above Lagrangian exhibits several notable features. First, it
193: possesses invariance under the gauge transformation
194: %
195: \begin{equation}\label{gauge}
196: B_{\alpha\beta} \to B_{\alpha\beta}+\partial_{[\alpha}\Lambda_{\beta]}
197: \end{equation}
198: %
199: for an arbitrary smooth vector function $\Lambda_\mu$. The square brackets
200: represent antisymmetrization, e.g.\ $\partial_{[\alpha}\Lambda_{\beta]}=
201: \partial_{\alpha}\Lambda_{\beta}-\partial_{\beta}\Lambda_{\alpha}$. This
202: gauge invariance reflects conservation of vorticity
203: in the original model. Second, from the last term in $\cal L$ we may
204: immediately deduce that the electric four-current is related to $B$ by
205: $j_\mu=2e(\epsilon_{\mu\nu\alpha\beta}\partial_\nu B_{\alpha\beta})$.
206: The charge density, in particular, can be written as
207: %
208: \begin{equation}\label{e}
209: \rho=j_0=2e(\epsilon_{ijk}\partial_iB_{jk}),
210: \end{equation}
211: %
212: where Roman indices run over spatial components only.
213:
214: The second term in $\cal L$ informs us that field $B$ is minimally
215: coupled to the ``vortex loop current''
216: %
217: \begin{equation}\label{vor}
218: \sigma_{\alpha\beta}(x)=
219: \epsilon_{\alpha\beta\mu\nu}\partial_\mu\partial_\nu\Theta(x),
220: \end{equation}
221: %
222: which is an antisymmetric rank-2 tensor quantity. For a smooth function the
223: right hand side of Eq.\ (\ref{vor}) would vanish since the derivatives would
224: commute. In 3 spatial dimensions, single valuedness of $e^{i\Theta(\bx)}$
225: permits {\em line singularities} in $\Theta(\bx)$ such that it varies
226: by an integer multiple of $2\pi$ along any line that encircles the singularity.
227: These are the vortex loops.
228:
229: To implement the duality transformation we now shift our point
230: of view from the phase field $\Theta(x)$ to the vortex loop {\em worldsheets}.
231: These describe the evolution of vortex loops in imaginary time and should be
232: thought of in analogy with worldlines of point particles. The worldsheets
233: are specified by a set of 2-parameter vector functions
234: $X_{\mu}^{(n)}(\sigma_1,\sigma_2)$ where $n$ labels the individual loops.
235: %For fixed $\sigma_1$ and $\sigma_2$ $X_{\mu}^{(n)}$ simply gives a point on the
236: %worldsheet of the $n$-th string.
237: We take $\sigma_1$ to be time-like, and correspondingly vary between
238: 0 and the inverse temperature $\beta$, and $\sigma_2$ space-like, which by
239: convention varies from 0 to $2\pi$ for closed loops. (Since vortices can only
240: terminate on magnetic monopoles we consider only closed vortex loops here.)
241: Clearly, given a set of worldsheets $X_{\mu}^{(n)}$ one can
242: reconstruct the phase field $\Theta(x)$ up to any smooth contribution.
243:
244: A surface element of a worldsheet is characterized by a rank-2 antisymmetric
245: tensor
246: %
247: \begin{equation}\label{sig}
248: \Sigma_{\mu\nu}^{(n)}={\partial X_{[\mu}^{(n)}\over \partial\sigma_1}
249: {\partial X_{\nu]}^{(n)}\over \partial\sigma_2}.
250: \end{equation}
251: %
252: It is straightforward to show that the loop current (\ref{vor}) is related
253: to the worldsheet by
254: %
255: \begin{equation}\label{sig2}
256: \sigma_{\mu\nu}(x)=2\pi\sum_n\int d^2\sigma \Sigma_{\mu\nu}^{(n)}
257: \delta\left( X^{(n)}-x\right).
258: \end{equation}
259: %
260: This relation allows us to rewrite the partition function as a functional
261: integral over the vortex loop worldsheets $X_{\mu}^{(n)}$.
262: We thus have $Z=\int{\cal D}[X] \exp(-{\cal S})$ with
263: %
264: \begin{eqnarray}\label{action}
265: {\cal S}&=&\sum_n \int d^2\sigma \left[{\cal T}\sqrt{
266: \Sigma_{\mu\nu}^{(n)}\Sigma_{\mu\nu}^{(n)}}
267: -2\pi i\Sigma_{\mu\nu}^{(n)}B_{\mu\nu}(X^{(n)})
268: \right]\nonumber \\
269: &+&{1\over 3K}\int d^4x H_{\alpha\beta\gamma}^2 +{\cal S}_{\rm int}+
270: {\cal S}_{\rm Jac}.
271: \end{eqnarray}
272: %
273: We recognize the first line as the celebrated Nambu-Goto action
274: \cite{nambu-goto} for
275: bosonic strings propagating in the presence of a background Kalb-Ramond
276: gauge field $B_{\mu\nu}$ \cite{kalb1}. The first term can be interpreted as the
277: reparametrization invariant surface area of the string worldsheet with
278: the string tension ${\cal T}$. Although such term does not appear
279: explicitly in Eq.\ (\ref{L2}) it would arise in a more careful treatment of
280: our starting theory (\ref{L0}) had we retained the cost of the suppression
281: of the order parameter amplitude $|\Psi|$ near the vortex core.
282: The second and the third terms follow directly from Eq.\ (\ref{L2}) and
283: describe long range interactions between strings mediated by the superflow,
284: now represented by the Kalb-Ramond gauge field.
285:
286: ${\cal S}_{\rm int}$ contains short range interactions between strings that
287: would also arise from a more careful treatment of the core physics. Finally,
288: ${\cal S}_{\rm Jac}$ represents the Jacobian of the transformation from
289: phase variable $\Theta$ to string worldsheets $X_{\mu}^{(n)}$. This last term
290: plays an important role in the quantization of our string theory. It is well known
291: that a fundamental string can be consistently quantized only in the critical
292: dimension, which for bosonic string is $D=26$ \cite{crit}. A question then
293: arises as to how we quantize our vortex strings in (3+1)D; after all we
294: started
295: from a well defined field theory (\ref{L0}) and we expect the string theory
296: (\ref{action}) derived from it to also be well behaved. The answer lies in
297: the fact that our strings are {\em not} fundamental; rather they are
298: Nielsen-Olesen type strings \cite{nielsen1} with intrinsic
299: thickness defined by the core size. It was shown by Polchinski and
300: Strominger \cite{pol1} that terms in ${\cal S}_{\rm Jac}$, which would be
301: absent in the case of a fundamental string, precisely cancel the conformal
302: anomaly responsible for the high critical dimension. Vortex strings are indeed
303: well behaved in the physical dimension.
304:
305: We are now ready to complete the duality mapping. Our main goal will be
306: to understand the string-condensed phase, analogous to the vortex-condensed
307: phase in the (2+1)D Lee-Fisher duality.
308: To this end we must pass to second quantized string theory, a ``string-field
309: theory''. This can be done rigorously in the so called light-cone gauge
310: \cite{kaku1} or using the Becchi-Rouet-Stora-Tyutin (BRST) procedure
311: \cite{brst}. Here we opt for a less rigorous but physically much more
312: transparent procedure devised in Ref.\ \cite{rey1} which provides a
313: straightforward route towards the description of the string-condensed phase.
314:
315: The central concept in the string-field description is the wave {\em functional}
316: $\Phi[X]$, a complex-valued functional defined on the space of one-parameter
317: string trajectories $\{X_\mu(\sigma_2),\sigma_2=(0,2\pi)\}$, which we regard
318: as cross sections of the string worldsheet $X_\mu(\sigma_1,\sigma_2)$
319: at fixed value of $\sigma_1$. The physical significance of $\Phi[X]$ is
320: most readily visualized in the so called static parametrization: in a
321: chosen Lorentz
322: frame of reference take $X_0=\sigma_1=\tau$ and $\bX=\bX(\tau,\sigma)$,
323: with $\tau$ the imaginary time. $\Phi[\tau,\bX]$ then represents the
324: quantum mechanical amplitude for finding the string in configuration
325: $\{\bX(\sigma),\sigma=(0,2\pi)\}$ at time $\tau$.
326:
327: The second
328: quantized action for the string functional takes the form \cite{rey1}
329: %
330: \begin{eqnarray}\label{action2}
331: {\cal S}&=& \int{\cal D}[X]\int d\sigma \sqrt{h}
332: \bigl[\big|(\delta/\delta\Sigma_{\mu\nu}-2\pi iB_{\mu\nu})\Phi[X]\big|^2
333: \nonumber \\
334: &+&{\cal M}_{\rm eff}^2\big|\Phi[X]\big|^2\bigr]
335: +{1\over 3K}\int d^4x H_{\alpha\beta\gamma}^2 +{\cal S}'_{\rm int}.
336: \end{eqnarray}
337: %
338: The plaquette derivative $\delta/\delta\Sigma_{\mu\nu}$ quantifies the
339: variation of the functional $\Phi[X]$ upon modifying the path $X$ by an
340: infinitesimal loop $\Delta X$ which sweeps the surface element
341: $\delta\Sigma_{\mu\nu}$. The loop space metric
342: $h=(\partial X_\mu(\sigma)/\partial\sigma)^2$ is needed to preserve the
343: reparametrization invariance of the action.
344: ${\cal S}'_{\rm int}$ represents short range string interactions and
345: contains terms cubic and higher order in $|\Phi|$. All contributions
346: quadratic in $|\Phi|$ have been folded into the effective string mass
347: ${\cal M}_{\rm eff}$. The action (\ref{action2}) remains invariant
348: under the gauge transformation (\ref{gauge}) if we require $\Phi$ to transform
349: as
350: %
351: \begin{equation}\label{gauge2}
352: \Phi[X]\to \Phi[X]e^{-4\pi i\int dX_\mu\Lambda_\mu}.
353: \end{equation}
354: %
355:
356: String condensation occurs when ${\cal M}_{\rm eff}^2$ becomes negative.
357: The string functional then develops nonzero vacuum expectation value,
358: $\langle 0| \Phi[X] |0\rangle \neq 0$. The simplest case is that of a
359: uniform string condensate,
360: %
361: \begin{equation}\label{unif}
362: \langle 0| \Phi[X] |0\rangle=\Phi_0={\rm const}.
363: \end{equation}
364: %
365: Physically, this simply means that {\em any} string configuration is
366: equally probable. This ansatz, however, cannot describe a phase
367: disordered superconductor. To see this note that substituting Eq.\ (\ref{unif})
368: into action (\ref{action2}) produces a mass term for the Kalb-Ramond gauge
369: field. Such a mass term then leads to the Meissner effect: the gauge
370: field is expelled from the interior of the sample, $B_{\mu\nu}=0$.
371: In view of Eq.\ (\ref{e}), this corresponds to complete expulsion of charge
372: from the system, which is not the situation we are interested in.
373:
374: What we seek is the analog of the Abrikosov vortex state in which
375: the field can penetrate in quantized increments.
376: We thus consider a more general ansatz, which allows both the amplitude and
377: the phase of $\Phi[X]$ to vary:
378: %
379: \begin{equation}\label{abr}
380: \langle 0| \Phi[X] |0\rangle=\Phi_0 e^{\int d\sigma
381: [\zeta\sqrt{X'^2}\ln f(X)+2\pi iX'_\mu\cdot\Omega_\mu(X)]
382: }.
383: \end{equation}
384: %
385: Here $X=X(\sigma)$, $X'=\partial_\sigma X(\sigma)$, $f(x)$ and
386: $\Omega_\mu(x)$ are real scalar and vector functions
387: parametrizing the functional, and $\zeta$ is a parameter with the dimension of
388: inverse length. $f(x)$ is nonnegative and should be thought of as the
389: space-time-dependent amplitude of the string condensate. Specifically,
390: $f=1$ corresponds to uniform condensate amplitude while $f>1$ ($f<1$)
391: describes its local enhancement (depletion).
392: $\Omega_\mu(x)$ determines the phase of the string condensate. Substituting
393: Eq.\ (\ref{abr}) to (\ref{action2}) we obtain
394: ${\cal S}=\int d^4x{\cal L}$ with
395: %
396: \begin{eqnarray}\label{action3}
397: {\cal L}&=& {\Phi_0^2\over 2}
398: \left[\pi^2f^2(\partial_{[\mu}\Omega_{\nu]}-2B_{\mu\nu})^2
399: +\zeta^2(\partial_\mu f)^2 +{\cal V}(f^2)\right] \nonumber \\
400: &+&
401: {1\over 3K} H_{\alpha\beta\gamma}^2.
402: \end{eqnarray}
403: %
404: We observe that any smooth part of $\Omega$ can be eliminated by
405: the gauge transformation (\ref{gauge2}). Thus, only the singular part of
406: $\Omega$ has physical significance. Indeed we note that it is
407: permissible for $\Omega$ to be {\em multiply valued} as long as the wave
408: functional (\ref{abr}) remains single valued.
409: A configuration of specific interest to us contains {\em monopoles}
410: in the spatial part of $\Omega=(\Omega_0,\Bomega)$,
411: %
412: \begin{equation}\label{mon}
413: \nabla\cdot(\nabla\times\Bomega)=\sum_a Q_a\delta^{(3)}(\bx-\bx_a),
414: \end{equation}
415: %
416: where $\bx_a$ and $Q_a$ label the position and the charge of the
417: $a$-th monopole.
418: Single valuedness of (\ref{abr}) demands that $Q_a$ be
419: {\em integer}. We shall see that such
420: singularities represent sources for $B_{\mu\nu}$, just as
421: vortices in a superconductor act as sources for the magnetic field.
422:
423: We now analyze the action (\ref{action3}) in the presence of static monopole
424: configurations in $\Bomega$. To this end we adopt a dual mean-field
425: approximation (DMFA) which neglects quantum fluctuations of all the fields. We
426: emphasize that in terms of the original phase degrees of freedom, DMFA
427: describes a highly nontrivial quantum fluctuating state. In addition, we
428: perform a dual ``London'' approximation, $f(\bx)=1$, which should be
429: adequate as long as the monopoles are relatively dilute. (This approximation
430: fails in the small region near the monopole center where $f\to 0$.)
431: The ground state energy of the system can then
432: be written as
433: %
434: \begin{equation}\label{ge}
435: {\cal E}= {1\over 2}\int d^3x
436: \left[\pi^2\Phi_0^2(\partial_{[i}\Omega_{j]}-2B_{ij})^2
437: +{1\over K}(\epsilon_{ijk}\partial_i B_{jk})^2
438: \right].
439: \end{equation}
440: %
441: Minimizing with respect to $B_{ij}$ leads to the Euler-Lagrange equation
442: %
443: \begin{equation}\label{el}
444: \pi^2\Phi_0^2(2B_{ij}-\partial_{[i}\Omega_{j]})
445: -{1\over 2eK}\epsilon_{ijk}\partial_k\rho=0,
446: \end{equation}
447: %
448: where we used Eq.\ (\ref{e}). Next, acting on all terms by
449: $\epsilon_{ijl}\partial_l$ and defining a dual ``penetration depth''
450: $\lambda_d^{-2}=2\pi^2\Phi_0^2K$,
451: we obtain an equation for charge density $\rho(\bx)$
452: %
453: \begin{equation}\label{el1}
454: \rho-\lambda_d^2\nabla^2\rho=2e\nabla\cdot(\nabla\times\Bomega).
455: \end{equation}
456: %
457: This equation resembles the London
458: equation for the $z$-component of magnetic field in the presence of an
459: Abrikosov lattice of vortices and can be analyzed by similar methods.
460: The key difference is that, in light of
461: Eq.\ (\ref{mon}), the right hand side describes a collection of {\em
462: point sources} in three space dimensions whereas Abrikosov vortices
463: are line singularities described by $\delta^{(2)}$. Below we briefly
464: summarize some main results of this analysis and the relevant
465: details will be given elsewhere \cite{franz00}.
466:
467:
468: Eq.\ (\ref{el1}) can be solved for an arbitrary arrangement
469: of monopole positions and charges to obtain
470: %
471: \begin{equation}\label{rho1}
472: \rho(\bx)=2e\sum_a Q_a {e^{-|\bx-\bx_a|/\lambda_d}\over 4\pi\lambda_d^2
473: |\bx-\bx_a|}.
474: \end{equation}
475: %
476: It is easy to show that the total electric charge associated with a monopole
477: is $2eQ_a$; the charge is quantized in the units of $2e$, as expected.
478: At finite charge density, monopoles with like charges repel by Yukawa
479: potential $\sim e^{-r/\lambda_d}/r$ and
480: the ground state is a Bravais lattice of elementary ($Q_a=1$) monopoles.
481: This leads to periodic modulation in $\rho(\bx)$ with charge $2e$ per unit
482: cell: a pair Wigner crystal in three space dimensions.
483:
484: An appealing overall picture thus emerges. Vortex loops in a
485: (3+1)-dimensional superconductor (or superfluid) can be efficiently described
486: as bosonic strings interacting through a rank-2 tensorial Kalb-Ramond gauge
487: field $B_{\mu\nu}$. In the non-superconducting phase, strings proliferate and
488: condense, producing Higgs mass for the gauge
489: field. In the Higgs phase the only way for $B_{\mu\nu}$ to penetrate into the
490: bulk of the system is to set up quantized monopole-like singularities in the
491: phase
492: $\Omega_\mu$ of the string condensate wave functional. These singularities
493: then act as point sources for $B_{\mu\nu}$. The associated electric charge
494: density $\rho$, which is closely related to the Kalb-Ramond field strength
495: $H_{\mu\nu\lambda}$, is then governed by a London-like equation (\ref{el1}).
496: %For a single point-like source the charge density forms a cloud with a
497: %distribution characteristic of a test charge placed in a metal with screening
498: %length given by the dual penetration depth $\lambda_d$ and the total charge
499: %quantized in units of $2e$.
500: For a periodic array
501: of point sources, such as will form at finite charge density, a 3-dimensional
502: pair Wigner crystal emerges with charge distribution given by Eq.\
503: (\ref{rho1}).
504:
505: The duality discussed above establishes vortex-loop condensation as a
506: concrete mechanism for the formation of a pair Wigner crystal in a
507: 3-dimensional quantum phase fluctuating superconductor. It explains how a
508: 3d PWC can form in underdoped cuprates and allows for detailed computations
509: of its structure and vibrational modes \cite{franz00} which are of direct
510: experimental interest.
511:
512:
513: The author is indebted to T. Davis, M.P.A. Fisher, S.-S. Lee, T. Pereg-Barnea,
514: C. Weeks
515: and Z. Te\v{s}anovi\'c for stimulating discussions and correspondence. This
516: work was supported by NSERC, CIAR and the A.P. Sloan Foundation.
517:
518:
519:
520: \begin{thebibliography}{0}
521: \bibitem{kramers1} \Name{H.A. Kramers and G.H. Wanier}
522: \REVIEW{ Phys.\ Rev.}{60}{1941}{252}.
523: \bibitem{dasgupta} \Name{C. Dasgupta and B.I. Halperin}
524: \REVIEW{Phys.\ Rev.\ Lett.}{47}{1981}{1556}
525: \bibitem{nelson} \Name{D.R. Nelson}
526: \REVIEW{Phys.\ Rev.\ Lett.}{60}{1988}{1973}
527: \bibitem{lee_fisher} \Name{M.P.A. Fisher and D.H. Lee}
528: \REVIEW{ Phys.\ Rev. B}{39}{1989}{2756}
529: \bibitem{vershinin1} \Name{M.\ Vershinin {\em et al.}}
530: % S.~Misra, S.~Ono, Y.~Abe, Y.~Ando, A.~Yazdani,
531: \REVIEW{Science}{303}{2004}{1995}
532: \bibitem{hanaguri1} \Name{T.\ Hanaguri {\em et al.}}
533: %C.~Lupien, Y.~Kohsaka, D.~-H. Lee, M.~Azuma, M.~Takano, H.~Takagi, J.~C.~Davis,
534: \REVIEW{Nature}{430}{2004}{1001}
535: \bibitem{hashimoto1} \Name{A.~Hashimoto {\em et al.}}
536: %N.~Momono, M.~Oda and M.~Ido,
537: \REVIEW{cond-mat/0512496}{}{unpublished}{}
538: \bibitem{chen1} \Name{H.-D. Chen {\em et al.}}
539: \REVIEW {Phys.\ Rev.\ Lett.}{93}{2004}{187002}
540: \bibitem{tesanovic1} \Name{Z.~Te\v{s}anovi\'c}
541: \REVIEW{Phys.\ Rev.\ Lett.}{93}{2004}{217004};
542: \Name{A. Melikyan and Z.~Te\v{s}anovi\'c}
543: \REVIEW{Phys.\ Rev.\ B}{71}{2005}{214511}
544: \bibitem{anderson1} \Name{P.W.\ Anderson}
545: \REVIEW{ cond-mat/0406038}{}{unpublished}{}
546: \bibitem{balents1}\Name{L.~Balents, L.~Bartosch, A.~Burkov, S.~Sachdev,
547: K.~Sengupta}
548: \REVIEW{Phys.\ Rev.\ B}{71}{2005}{144508}
549: \bibitem{tami1} \Name{T. Pereg-Barnea and M. Franz}
550: \REVIEW{Phys.\ Rev.\ B}{74}{2006}{014518}
551: \bibitem{taillefer1} \Name{N. Doiron-Leyraud {\it et al.}}
552: \REVIEW{cond-mat/0606645}{}{unpublished}{}
553: \bibitem{peskin1} \Name{M.E. Peskin}
554: \REVIEW{Ann.\ Phys.}{113}{1978}{122}
555: \bibitem{savit1} \Name{R. Savit}
556: \REVIEW{Rev.\ Mod.\ Phys.}{ 52}{1980}{453}
557: \bibitem{senthil1} \Name{O.I.\ Motrunich and T.\ Senthil}
558: \REVIEW{Phys.\ Rev.\ B}{71}{2005}{125102}
559: \bibitem{wen1}\Name{M.\ Levin and X.-G.\ Wen}
560: \REVIEW {Phys.\ Rev.\ B}{73}{2006}{035122}
561: \bibitem{nambu-goto} For an excellent review see e.g.\ \Name{B. Zwiebach}
562: \Book{A First Course in String Theory} \Publ{Cambridge Univ.\ Press, 2004}
563: \bibitem{kalb1} \Name{M. Kalb and P. Ramond}
564: \REVIEW{Phys.\ Rev.\ D}{9}{1974}{2273}
565: \bibitem{crit} \Name{P. Goddard, J. Goldstone, C. Rebbi and C.B. Thorn}
566: \REVIEW{ Nucl.\ Phys.\ B}{56}{1973}{109}
567: \bibitem{nielsen1} \Name{H.B.\ Nielsen and P.\ Olesen}
568: \REVIEW{Nucl.\ Phys.\ B}{61}{1973}{45}
569: \bibitem{pol1} \Name{J. Polchinski and A. Strominger}
570: \REVIEW{Phys.\ Rev.\ Lett.}{67}{1991}{1681}
571: \bibitem{kaku1} \Name{M. Kaku and K. Kikkawa}
572: \REVIEW{Phys.\ Rev.\ D}{10}{1974}{1110}
573: \bibitem{brst} \Name{E. Witten}
574: \REVIEW{ Nucl. Phys. B}{268}{1986}{253}
575: \bibitem{rey1} \Name{S.-J. Rey}
576: \REVIEW{Phys.\ Rev.\ D}{40}{1989}{3396}
577: \bibitem{franz00} \Name{M. Franz} (unpublished).
578:
579: \end{thebibliography}
580:
581: \end{document}
582:
583:
584:
585:
586:
587:
588:
589:
590:
591:
592:
593: \bibitem{}
594: \Name{}
595: \REVIEW{}{}{}{}.
596: \bibitem{}
597: \Name{}
598: \REVIEW{}{}{}{}.
599: \bibitem{}
600: \Name{}
601: \REVIEW{}{}{}{}.
602: \bibitem{}
603: \Name{}
604: \REVIEW{}{}{}{}.
605:
606: \bibitem{b.b}
607: \Name{Author F. \and Author S.}
608: \Book{Some Book of Interest}
609: \Editor{A. Editor}
610: \Vol{9}
611: \Publ{Publishing house, City}
612: \Year{1939}
613: \Page{666}.
614:
615: \bibitem{b.c}
616: \Editor{Editor A.}
617: \Book{Some Book of Interest}
618: \Vol{9}
619: \Publ{Publishing house, City}
620: \Year{1939}
621: \Section{A}.
622:
623: