Arxiv:1807.01984V2 [Hep-Th] 7 Nov 2018 ﬁlms [14] and Have Become Ever Since a Subject of an Intense Study, See [15–17] and References Therein

Conﬁnement and asymptotic freedom with Cooper pairs

M. C. Diamantini,1 C. A. Trugenberger,2 and V. M. Vinokur3 1NiPS Laboratory, INFN and Dipartimento di Fisica e Geologia, University of Perugia, via A. Pascoli, I-06100 Perugia, Italy 2SwissScientific Technologies SA, rue du Rhone 59, CH-1204 Geneva, Switzerland 3Materials Science Division, Argonne National Laboratory, 9700 S. Cass Ave., Lemont, IL 60439, USA One of the most profound aspects of the standard model of particle physics, the mechanism of confinement binding quarks into hadrons, is not sufficiently understood. The only known semiclassical mechanism of confinement, mediated by chromo-electric strings in a condensate of magnetic monopoles still lacks experimental evidence. Here we show that the infinite resistance superinsulating state, which emerges on the insulating side of the superconductor-insulator transition in superconducting films offers a realization of confinement that allows for a direct experimental access. We find that superinsulators realize a single-color version of quantum chromodynamics and establish the mapping of quarks onto Cooper pairs. We reveal that the mechanism of superinsulation is the linear binding of Cooper pairs into neutral “mesons” by electric strings. Our findings offer a powerful laboratory for exploring and testing the fundamental implications of confinement, asymptotic freedom, and related quantum chromodynamics phenomena via the desktop experiments on superconductors.

INTRODUCTION a

The standard model of particle physics is extraordinarily successful at explaining many facets of the physical realm. Yet, one of its profound aspects, the mechanism of confinement binding quarks into hadrons, is not sufficiently under- b stood. The only known semiclassical mechanism of confinement is mediated by chromo-electric strings in a condensate of magnetic monopoles [1–3] but its relevance for quantum chromodynamics still lacks experimental evidence. This suggests a quest for systems that could allow for direct experi- c mental tests of the string confinement mechanism. To iden- tify such a system we follow a brilliant insight of ‘t Hooft [4], who appealed to a solid state physics analogy in a Gedanken- experiment to explain quark confinement. He demonstrated that it is realized in a phase which is a dual twin to superconductivity, in a sense that it has zero particle mobility, FIG. 1. Dual Mandelstam–‘t Hooft–Polyakov confinement. and called hence this phase a “superinsulator.” The infinite- a: quark confinement by chromo-electric strings. b: magnetic tube resistance superinsulating state was indeed first predicted to (Abrikosov vortex) that forms in a superconductor between two mag- emerge in Josephson junction arrays (JJA) [5] and then in netic monopoles. c: electric string that forms in a superinsulator be- disordered superconducting films [6,7] at the insulating side tween the Cooper pair and anti-Cooper pair. The lines are the force of the superconductor-insulator transition (SIT) [8–12]. Ex- lines for magnetic and electric fields respectively. In all cases the energy of the string (the binding energy) is proportional to the distance perimentally, superinsulators were observed in titanium ni- between either the monopoles or the charges. tride (TiN) film [7, 13] and, albeit under different name, InO arXiv:1807.01984v2 [hep-th] 7 Nov 2018 films [14] and have become ever since a subject of an intense study, see [15–17] and references therein. dynamic (QCD) vacuum, in which Cooper pairs play the role Originally, the idea of superinsulation [5,7] grew from of quarks. We thus find that the Cooper pair binding mech- the supposed 2D logarithmic Coulomb interactions between anism in a superinsulator, leading to the infinite resistance at Cooper pairs in the critical vicinity of the SIT realized in lat- finite temperatures, is the linear, rather than logarithmic, con- eral Josephson junction arrays [5, 12]. Here we show that, finement of charges into neutral “mesons” due to Polyakov’s starting with the uncertainty principle for Cooper pairs [7] and electric strings [3, 18], arising in the vortex condensate. The building solely on the most general locality and gauge invari- Abelian character of the compact QED, albeit a strong cou- ance principles, one constructs the effective action for superin- pling gauge theory, allows for an analytical derivation of the sulators, which is exactly Polyakov’s compact quantum elec- linear confinement by electric strings, at variance to the QCD trodynamic (QED) action [3, 18]. Accordingly, superinsula- whose complexity requires heavy numerical computations. tion emerges as an explicit realization of the Mandelstam – Since linear confinement by strings is not restricted to ‘t Hooft S-duality [1,2] in materials that harbor Cooper pairs 2D, we establish that superinsulation is a distinct genuine and constitutes a single-color version of the quantum chromo- state of matter that appears in both 2D or 3D realizations 2 and calculate the deconfinement temperature that marks the action assumes the form phase transition of superinsulators into conventional insu- Z n √ √ lators and which, in 2D, coincides with the Berezinskii- S CS = d3 x i a ∂ b + i na Q + i nb M , 2π µ µαν α ν µ µ µ µ Kosterlitz-Thouless (BKT) transition temperature. Finally we (1) also unearth a Cooper pair analogue of the asymptotic free- where is the completely antisymmetric tensor, and dom effect [19], which suggests that systems smaller than the µαν string scale appear in a quantum metallic state. Our findings Z (i) X dxqµ(τ) 3 (i) offer thus an easy access tool for testing fundamental implica- Qµ = dτ δ x − xq (τ) , (i) dτ tions of confinement, asymptotic freedom, and related QCD i xq Z (i) phenomena via desktop experiments on superconductors. X dxmµ(τ) 3 (i) Mµ = dτ δ x − xm (τ) , (2) (i) dτ i xm are the world-lines of elementary charges and vortices labeled (i) (i) by the index i, parametrized by the coordinates xq and xm , re- RESULTS spectively, n is the dimensionless charge, and Greek subscripts run over the Euclidean three dimensional space encompassing Action in two-dimensional systems the 2D space coordinates and the Wick rotated time coordi- nate. Equation (1) defines the mixed Chern-Simons (CS) action [21] and represents the local formulation of the topologi- We start by showing how dual superconducting and su- cal interactions between charges and vortices, where the ABC perinsulating states can be understood from the uncertainty (i) phases are encoded in the Gauss linking number of the {xq } principle, ∆N∆ϕ > 1 between the number of charges, N = and {x(i)} world-lines. The CS action is invariant under the 2|Ψ|2, and the phase ϕ of the Cooper pairs quantum field m gauge transformations aµ → aµ + ∂µλ and bµ → bµ + ∂µχ, re- Ψ = N exp(iϕ), bound by the commutation relation [N, ϕ] = flecting the conservation of the charge and vortex numbers and i [15, 20]. At zero temperature, superconductors correspond is the dominant contribution to the action at long distances, to fixed ϕ, hence indefinite N. Inversely, fixed N and indef- since it contains only one field derivative. In this represen- inite ϕ characterizes the superinsulating state. As a Cooper √ √ tation jµ = ( n/2π)µαν∂αbν and φµ = ( n/2π)µαν∂αaν are pair is a charge quantum, while a vortex carries the 2π phase the continuous charge and vortex number current fluctuations, quantum, the SIT is driven by the competition between charge while Qµ and Mµ stand for integer point charges and vortices. (Cooper pairs) and vortex degrees of freedom, in accordance We use natural units c = 1, ~ = 1 but restore physical units with early ideas [11]. when necessary. Also, from now on we set the charge unit We turn now to the construction of the action of the Cooper n = 2 for Cooper pairs. pair-vortex system near the SIT, where both degrees of free- The next-order terms in the effective action of the SIT con- dom are to be included on an equal footing. The key contri- tain two field derivatives. Gauge invariance requires that they bution is the infinite-range (i.e. non-decaying with distance) be constructed in terms of the “electric” and “magnetic” fields Aharonov-Bohm-Casher (ABC) Cooper pair-vortex topolog- corresponding to the two gauge fields. Introducing the dual ical interaction, embodying the quantum phase acquired ei- field strengths fµ = µαν∂αbν and gµ = µαν∂αaν one identifies ther by a charge encircling a vortex or by a vortex encircling the magnetic fields as f0 and g0 and the electric fields as fi and a charge. To ensure a local formulation of the action, we gi, where ”0” denotes the Wick rotated time and Latin indices must introduce two emergent gauge fields, aµ and bµ medi- denote purely spatial components. We thus arrive at the full ating these ABC interactions. Then the topological part of the action

Z 1 1 ε 1 ε √ √ S d3 x i a ∂ b f 2 P f 2 g2 P g2 i a Q i b M . 2D = µ µαν α ν + 2 0 + 2 i + 2 0 + 2 i + 2 µ µ + 2 µ µ (3) π 2evµP 2ev 2eqµP 2eq

Here µ is the magnetic permeability and ε is the electric per- der of the expansion with respect to derivatives is perfectly P P √ mittivity [20], which define the speed of light vc = 1/ µPεP dual under the mutual exchange of charge and vortex degrees 2 2 in the material. The two coupling constants, eq = e /d and of freedom and the corresponding coupling constants. The 2 2 2 ev = π /(e λ⊥) are the characteristic energies of a charge and charge-vortex duality is expressed by the action symmetry a vortex in the film, respectively [20]. Here d is the thick- with respect to the transformation g ≡ ev/eq ↔ 1/g. Thus 2 g is the tuning parameter driving the system across the SIT, ness of the film, λ⊥ = λL /d is the Pearl length, and λL is the London length of the bulk. The effective action in this or- and the SIT itself corresponds to g = gc = 1. The possi- 3 ble duality breaking is a higher order effect. In field theory, JJA. This provides a crosscheck for our general result. this duality goes under the name of S-duality (strong-weak coupling duality). Note that the addition of kinetic terms gen- erates the topological Chern-Simons mass mT for both gauge Superinsulator fields. In the relativistic case, µP = εP = 1, and the CS mass becomes mT = eqev/π [21]. In the non-relativistic case the CS mass is modified to m = µ e e /π and the dispersion relation We are now equipped to discuss the nature of the superin- p T P q v 2 4 2 2 sulating state. To that end, we couple the charge current jµ becomes E = mT vc + vc p , see Methods, Lattice Chern- Simons operator. We stress here that we derived the action to the physical electromagnetic gauge field Aµ by adding to (3) describing the system of interacting Cooper pairs and vor- the action the minimal coupling term 2eAµ jµ. Setting Qµ = 0, tices using solely symmetry and gauge invariance considera- since charges are dilute, integrating out the gauge fields aµ and tions. Importantly, the action describing Josephson junction bµ, and summing over the condensed vortices Mµ, we arrive at the effective action S (A ) describing the electromagnetic arrays [5, 12] is a special case of the same action with εP = 1, eff µ 2 response of an ensemble of charges in a superinsulator. On a µP → ∞, eq → 4EC, ev → 2π EJ, where EC and EJ are the charging energy and the Josephson energy of a single junc- discretized lattice with spacing `, see Methods, Lattice Chern- tion, respectively, see Supplementary note 1, Gauge theory of Simons action, the effective action takes a form, See Sup- plementary note 2, Effective action for the superinsulator, in which one immediately recognizes a non-relativistic version of the Polyakov action for the compact QED model [3, 18]:

  γ2 X h i X 1 h i S (A ) = S 2D =  v 1 − cos 2e`2F + 1 − cos 2e`2F  . (4) eﬀ µ compact 2π2  c 0 v i   x x,i c 

Here the summation runs over the lattice grid {x}, Fµ = kµνAν around the string is the inverse of the photon mass mγ [22], is the dual electromagnetic field strength, kµν is the lattice wstring = 1/(vcmγ). The typical “meson” size instead, is given Chern-Simons operator µαν∂α, see Methods, Lattice Chern- by the string tension σ. In the 2D relativistic (vc = 1) model 2 Simons operator, and γ = Cηg/vc with C being a numerical these are given by [23] κ, vc) characterizes the 2)ל(constant. The quantity η = (1/α γ −γ2/2π strength of quantum fluctuations, see Supplementary note 3, mγ = √ e , πvc` Quantum phase structure. Here κ = λ⊥/ξ is the Ginzburg- 2 2 3/2 Landau parameter of the film, ξ is the superconducting coher- π mγvc π vc 2 σ = = e−γ /2π . (5) ence length, taking on the role of the ultraviolet cutoff `, and, 2D 4`γ2 4`2 2 finally, α = e /(~c) ≈ 1/137 is the fine structure constant. Unlike vortices, however, long strings are unstable: it is en- The physics of a superinsulator is governed by the spon- ergetically favorable to break a string into a sequence of seg- taneous proliferation of instantons [18] M = ∂µ Mµ, corre- ments via the creation of charge-anticharge pairs, see Fig. (2). sponding to magnetic monopoles, so that the vortex number This process corresponds to the√ creation of neutral “mesons” is not conserved in the vortex condensate. Then, in a mirror with the typical size dstring = vc/σ. From the dependence of 2 analogue to the formation of Abrikosov vortices in supercon- mγ and σ on γ , one finds for the nonrelativistic case ! ductors due to the Meissner effect mediated by the Cooper gη d ' ` K , pair condensate, the magnetic monopole condensate constricts string exp 2 (6) electric field lines connecting the charge-anticharge pair into vc electric strings [3, 18] confining Cooper pairs in superinsula- where K is a numerical constant. Near the SIT, where g ≈ 1/η √ tors into “mesons”, see Fig.1. Indeed, as seen from the ac- and vc = 1/ µPεP c due to the divergence of the elec- tion (4), at large γ, the dynamical fields get squeezed into the tric permittivity εP [7, 15], dstring `, and the electric string vicinity of the paths minimizing the action, to form quantized is a well-defined object. This establishes superinsulators as 2 fluxes ` Fµ. The quantized electric flux tubes are the ana- a single-color realization of QCD. Cooper pairs assume the logues of the strings mediating linear confinement of quarks role of quarks that are bound by electric strings into neutral into hadrons. Like Abrikosov vortices, for which the London mesons and this linear confinement is the origin of the infinite penetration depth, the inverse of the Anderson-Higgs photon resistance of superinsulators. As quarks cannot be observed mass, sets the spatial scale of the decay of encircling supercur- outside hadrons, Cooper pairs do not exists outside neutral rents and magnetic field associated with the vortex, the char- bound states, and the absence of free charge carriers causes acteristic lateral scale wstring for the decay of electric fields the infinite resistance. 4

Action and superinsulator in three-dimensional systems

2e –2e 2e –2e … 2e –2e

dstring The string confinement mechanism of superinsulation allows to generalize the concept of a superinsulator to higher dimen- FIG. 2. Splitting electric strings into neutral mesons. The for- sions, since linear confinement by electric strings is not spe- mation of a long string is energetically unfavorable, and small size cific to the 2D realm. Hence superinsulators can exist in 3D charge-anticharge pairs emerge, splitting the string into a sequence exactly as QCD exists in 3D. The 3D analogue of the topolog- of segments, each constituting a neutral meson. ical action (3) involves the so called BF term [24], combining the standard gauge field aµ with the Kalb-Ramond antisymmetric gauge field of the second kind [25] bµν,

√ Z 1 1 ε 1 ε √ 2 S d4 x i a ∂ b f 2 P f 2 b2 P e2 i a Q i b M . 3D = µ µναβ ν αβ + 2 0 + 2 i + 2 i + 2 i + 2 µ µ + µν µν (7) π 2evµP 2ev 2eqµP 2eq 2

√ Here ei = ∂0ai − ∂ia0 and bi = i jk∂ jak are the usual elec- fluctuation number current jµ = ( 2/2π) fµ retain their point tric and magnetic fields associated with the gauge field aµ, charge character. In 3D, eq is a dimensionless parameter, while fµ = (1/2)µναβ∂νbαβ is the dual field strength associ- eq = O(e), while ev has the dimension of mass, ev = O(1/λ), ated wiht the antisymmetric gauge field bµν. In addition to the with λ being the bulk London length of the material. The topo- gauge symmetry under transformations aµ → aµ + λ, this ac- logical mass arising from the BF coupling [26] maintains the tion is invariant under gauge symmetries of the second rank, same form as in 2D, mT = µPeqev/π. bµν → bµν + ∂µχν − ∂νχµ, in which the gauge function itself is a vector. In 3D, vortices are one-dimensional extended ob- The derivation of the effective action for a superinsulator in jects and their world-surfaces are described by the two-index 3D follows exactly the same steps as in 2D, Supplementary antisymmetric tensor Mµν. Cooper pairs, Qµ, and the related note 2, Effective action for the superinsulator, with the result

  γ2 X h i X 1 h i S SI (A ) = S 3D =  2v 1 − cos 2e`2F˜ + 1 − cos 2e`2F˜  . (8) eﬀ µ compact 2π2  c 0i v i j   x,i x,i, j c 

where F˜µν = kµναAα is the 3D dual field strength (kµνα be- exp (−S axion) is invariant under shifts θ → θ + 2π. Time re- ing the 3D lattice BF term, see Methods, Lattice BF term). versal, T , maps θ → −θ. So the only values of θ compatible This is again a relativistic version of Polyakov’s compact QED with T -invariance are θ = 0 and θ = π, modulo 2π. For model, this time in 3D [3, 18], with the relativistic (vc = 1) θ = π the string becomes fermionic [18], acquiring a topo- string tension given by [27] logical contribution (−1)ν in the partition function, where ν is √ ! the signed self-intersection number of the world-sheet in four- v z σ c K γ , dimensional Euclidean space-time. The (relativistic) string 3D = 2 0 (9) 64π` 4π tension changes to [27] where K0 is the McDonald function and z is the monopole fu- gacity. Equations (4) and (8) are our key results, establishing an exact mapping between QCD and the physics of superin- √ ! sulators, both in 2D and 3D. v z σ = c K (10) Finally, let us mention that, unlike in 2D, in 3D, the min- 3D 64π`2 0 16γ imal coupling of charges to electromagnetism can be com- R √ plemented by a topological coupling d4 x i(θ/8π 2) φ F √ µν µν of the vortex current φµν = ( 2/2π)µναβ∂αaβ to the electromagentic field strength Fµν. This leads to an axion term [28] Because the factor γ is now in the denominator, the fermionic R 4 2 S axion = d x i(θ/16π )FµνF˜µν in the electromagnetic effec- Cooper pair mesons are large also in the deep superinsulating tive action. This is a surface term, since the partition function region, where ηg 1 and v = O(1). 5 a coherence length ξ & d [17] and that of the InO film, where ξ d [16], one sees that the NbTiN film shows the BKT- while the InO film exhibits the VFT divergence, in compli- ance with our predictions about 3D superinsulation.

The deconfinement transition can be realized as a quantum dynamical phase transition driven by an applied electric field b E that would tear the electric strings. The threshold voltage, Vt ∝ σL, corresponding to the pair-breaking critical current in superconductors, breaks down the neutral meson chains, and a strip of ‘normal’ insulator forms along the former string path, carrying the current. This pretty much resembles the conventional dielectric breakdown where the electric field burns a conducting channel in otherwise insulating environment and triggers avalanche-like current jumps. The dielectric breakdown is usually accompanied by current noise. Such a noise FIG. 3. Deconfinement transition. a: Finite temperature deconfinement transition from a superinsulator (magnetic numbers M = has indeed been recently observed in InO films [31]. Exper- ±1 fall into the interior of the ellipse, while electric numbers Q = ±1 iments demonstrating the linear dependence of the threshold remain outside) to an insulator (no non-trivial quantum numbers fall voltage on the sample size in films are still to come. Yet the within the ellipse). b: The finite-temperature scaling factor that de- evidence for linear confinement was provided by the analysis termines the critical temperature for the superinsulator deconfine- of the superinsulating behavior in the ultrathin TiN films [32], ment transition, v is the light velocity in the material. c which revealed that the magnetic field dependence of Vt is exactly that of the 1D Josephson ladder.

Finite temperatures In QCD, the flip side of the string confinement mecha- Now we turn to the finite temperature behavior and the de- nism is asymptotic freedom, i.e. the unconstrained dynamics confinement transition at which string confinement of Cooper of quarks at spatial scales smaller than the string size [19]. pairs ceases to exist and the superinsulator transforms to a While, strictly speaking, asymptotic freedom refers to the ‘conventional’ insulator. This happens at the critical temper- running of the dimensionless gauge coupling to zero in the ature Tdc where the linear tension of the string turns to zero. ultraviolet limit, it can be viewed, from the string point of While it is known that, in 2D, Tdc ≡ TBKT [29], we can cal- view, as the “slackening” of the string so that quarks feel culate Tdc straightforwardly as the temperature of disappear- only weak short-range potentials at small scales. One would ance of the vortex condensate. This is done in Methods, Finite thus expect that, in superinsulators, asymptotic freedom, in temperature deconfinement transition, with the result that the this string sense, should map onto the unconstrained motion superinsulator experiences a direct deconfinement transition of the Cooper pairs at scales smaller than dstring. The ra- to an insulating state at the critical deconfinement tempera- tio of the string width to the string length is wstring/dstring ∝ ture determined by the equation 1/(gη) = S (T ) where the 2 2 dc (vc/γ )exp(Kγ /vc) with K being a numerical constant. For function S (T) is derived by a geometric condition for the two systems with small K and large γ2 this ratio is small. At competing condensations (see Supplementary note 3, Quan- scales wstring < r < dstring, Cooper pairs do not feel the string tum phase transitions) and is shown in Fig.3. This equation tension anymore but neither do they feel Coulomb interac- uniquely determines the deconfinement temperature as a functions screened by the photon mass. Hence, one can expect a tion of material parameters. metallic-like low-temperature behavior of small samples that should have turned superinsulating had their size exceeded the typical dimension of the confining string, estimated as Experimental implications dstring . ~vc/kBTBKT. Using the TiN films parameters[7, 15] one obtains dstring . 60 µm. Remarkably, the study of the size To explore the far reaching experimental implications of the dependence of superinsulating properties in TiN films [33] confining string theory of superinsulation note first that the de- revealed that in films with lateral sizes, of 20 µm and less, confinement criticality depends on the space dimension [30]. the insulating, thermally activated behavior saturates to the In 2D it coincides with that√ of the BKT transition [29], and the metallic one upon cooling to ‘superinsulating temperatures.’ resistance R2D ∝ exp(b/ |T/TBKT − 1|). In 3D, instead, the re- This complies with the expected asymptotic freedom behav- sistance exhibits the so-called Vogel-Fulcher-Tamman (VFT) ior. However, it would be premature to take it as a conclusive 0 criticality, R3D ∝ exp[b /|T/Tdc −1|][30]. Juxtaposing the crit- evidence for the asymptotic freedom in superinsulators, and ical behaviors of the NbTiN film, having a superconducting further experimental research is needed. 6

DISCUSSION APPENDIX

We conclude by pointing out a close connection of the Lattice Chern-Simons operator string confinement mechanism to concepts of many-body- localization (MBL) [34]. It was recently shown that MBL- The formulation of a gauge-invariant lattice Chern-Simons like behaviour may arise without exogenous disorder, due to term requires particular care. Following [5] we introduce first strong interactions alone [35], and that, in gauge theories, this the forward and backward derivatives and shift operators on a is due to the endogenous disorder embodied by the mixing three-dimensional Euclidean lattice with sites denoted by {x}, of superselection sectors [36], this process being identified as directions indicated by Greek letters and lattice spacing `, a transport-inhibiting mechanism due to confinement in the f (x + `µˆ) − f (x) Schwinger model in 1D. In our setting, it is the Polyakov d f (x) = , S f (x) = f (x + `µˆ) , µ ` µ monopole instantons that play the role of endogenous spon- f (x) − f (x + `µˆ) taneous disorder. Accordingly, our summation over the in- dˆ f (x) = , Sˆ f (x) = f (x − `µ(11)ˆ) . µ ` µ stanton gas configurations acts as averaging over endogenous disorder [3, 18]. Importantly, the instanton formulation de- Summation by parts on the lattice interchanges both the two scribes not only 1D, but the 2D and 3D physical dimensions as derivatives (with a minus sign) and the two shift operators. well. This spontaneous disordering mechanism has the same Gauge transformations are defined by using the forward lattice effect, that of mixing, in this case, the flux superselection sec- derivative. In terms of these operators one can then define two tors, leading to the survival of only the neutral charge sector as lattice Chern-Simons terms the physical state, while all other, charged states are localized ˆ ˆ ˆ on the string scale. Hence inhibition of the charge transport kµν = S µµανdα , kµν = µανdαS ν , (12) and the infinite resistance. The same confinement mechanism where no summation is implied over equal indices. Summa- that prevents the observation of quarks is thus responsible for tion by parts on the lattice interchanges also these two op- the absence of charged states and the infinite resistance in su- erators (without any minus sign). Gauge invariance is then perinsulators. guaranteed by the relations

kµαdν = dˆµkαν = 0 , kˆµνdν = dˆµkˆµν = 0 . (13) ACKNOWLEDGEMENTS Note that the product of the two Chern-Simons terms gives the lattice Maxwell operator We are delighted to thank N. Nekrasov, M. Vasin, and 2 Ya. Kopelevich for illuminating discussions. M.C.D. thanks kµαkˆαν = kˆµαkαν = −δµν∇ + dµdˆν , (14) CERN, where she completed this work, for kind hospitality. 2 The work at Argonne (V.M.V.) was supported by the U.S. De- where ∇ = dˆµdµ is the 3D Laplace operator. The discrete partment of Energy, Oﬃce of Science, Basic Energy Sciences, version of the mixed Chern-Simons gauge theory can thus be Materials Sciences and Engineering Division. formulated as

X `3 `3 `3ε `3 `3ε √ √ S = i a k b + f 2 + P f 2 + g2 + P g2 + i` 2a Q + i` 2b M , π µ µν ν 2 0 2 i 2 0 2 i µ µ µ µ x 2evµP 2ev 2eqµP 2eq

where the discrete dual ﬁeld strengths are given by Lattice BF operator

fµ = kµνbν , gµ = kµνaν . (15) As we show below, this action describes two massive modes with dispersion relation and mass given The formulation of a discrete 3D lattice BF model [24] can q be achieved along the same lines as in 2D. Following [5] we µPeqev E = m2v4 + v2k2 , m = . (16) T c c T π introduce the lattice BF operators √ where vc = 1/ µPεP is the light velocity in the medium. This is the non-relativistic version of the celebrated Chern-Simons mass [21]. kµνρ ≡ S µµανρdα kˆµνρ ≡ µναρdˆαSˆ ρ , (17) 7 where two shift operators; gauge transformations are deﬁned using the forward lattice derivative. Also the two lattice BF opera- f (x + `µˆ) − f (x) tors are interchanged (no minus sign) upon summation. More- dµ f (x)≡ , S µ f (x) ≡ f (x + `µˆ) , ` over they are gauge invariant, in the sense that they obey the f (x) − f (x − `µˆ) dˆ f (x)≡ , Sˆ f (x) ≡ f (x − `µˆ) ,(18) following equations: µ ` µ kµνρdν = kµνρdρ = dˆµkµνρ = 0 , are the forward and backward lattice derivative and shift op- kˆµνρdρ = dˆµkˆµνρ = dˆνkˆµνρ = 0 . (19) erators, respectively. Summation by parts on the lattice interchanges both the two derivatives (with a minus sign) and the Finally, they satisfy also the equations

2 kˆµνρkρλω = − δµλδνω − δµωδνλ ∇ + δµλdνdˆω − δνλdµdˆω + δνωdµdˆλ − δµωdνdˆλ , 2 kˆµνρkρνω = kµνρkˆρνω = 2 δµω∇ − dµdˆω , (20)

2 where ∇ = dˆµdµ is the lattice Laplacian. The Euclidean lattice BF model in 3D is then given by the action √ X `4 `4 `4ε `4 `4 √ 2 S = i a k b + b2 + P e2 + µ f 2 + P f 2 + i` 2a Q + i`2 b M , π µ µαβ αβ 2 i 2 i 2 P 0 2 i µ µ µν µν x 2eqµP 2eq 2ev 2ev 2

where the dual field strengths are now defined by must be also quantized in the integer multiples of 2π/β. This gives 1 ˜ ˆ fµ = kµνρbνρ , fµν = kµνρaρ , (21) Z 2πvc n=b ! 2 ` X 2π 2πv n dk0 f k0 → f c , (22) 0 β b` and ei = d0ai − dia0 and bi = f˜0i are the usual electric and n=0 magnetic fields associated with the gauge field a . The dis- µ where β = b`/v and the factor within the sum represents persion relation and mass remain identical to the 2D formulas. c the density of states. The integers n in the summation are In this case they are the non-relativistic generalizations of the known as Matsubara frequencies. Typically, however mo- BF mass [26]. menta integral are defined over the fundamental Brillouin zone [−πvc/`, πvc/`], rather then [0, 2πvc/`]. The corresponding finite temperature expression can be readily obtained from 0 0 (22) by the shift k → k − πvc/`, Finite Temperature Deconfinement Transition Z πvc k=b ! ` X π πv k dk0 f k0 → f c , (23) −πvc β b` ` k=−b In the field theory, the finite temperature T is introduced by formulating the action on a Euclidean time of finite length where k = 2n − b and thus correspondingly, the density of β = 1/T, with periodic boundary conditions (we have reab- states must be divided by a factor 2. sorbed the Boltzmann constant into the temperature). If the The finite temperature T > 0 affects primarily the parame- original field theory model is defined on a Euclidean lattice of ter η (see Supplementary note 3, Quantum phase structure) via spacing `, then β is quantized in integer multiples of `/vc. This the coefficient G(m`vc). At the zero temperature this is given representation of the finite-temperature field theory holds as by long as vcβ `, or, equivalently, if the temperature is much Z π 1 4 1 lower than the UV cutoff, T vc/`, as expected. Because G(m`vc) = d k . (24) 4 i 2 of the lattice structure, energies are defined only within a Bril- (2π) −π 2 P3 k (m`vc) + i=0 4 sin 2 louin zone of length 2vcπ/`, due to the periodic boundary condition in the Euclidean time direction, however the energy k0 At finite temperatures it has to be modified according to (23),

1 kX=+b π Z π dk1dk2dk3 G(m`vc, T) = , (25) π 4 b πk 2 P3 ki 2 (2 ) − −π 2 k= b (m`vc) + 4 sin 2b + i=1 4 sin 2 8

where T = vc/b`. As we have veriﬁed over 3 orders of with the relevant coupling constant taking the role of tem- magnitude (m`vc = 0.001 to m`vc = 1) the ratio S (T) = perature. The ﬁrst two terms in the action represent the two

G(m`vc, T)/G(m`vc) does not depend on the parameter m`vc Coulomb gases for charges with density ρQ and vortices with but is rather a function of the temperature alone. As a con- density ρV, the third is a kinetic term for the charges and the sequence, η and the semiaxes of the ellipse determining the final term represents the Aharonov-Bohm topological inter- phase structure, see Supplementary note 3, Supplementary action between charges and vortices. The only term which Equations (33), scale with the inverse of the function S (T). breaks perfect duality between charges and vortices in this ex- This means that with the increasing temperature the whole pression is the kinetic term for charges, which encodes the ellipse shrinks by the scale factor S (T). Magnetic quantum Josephson currents. The self-dual approximation, originally numbers M = ±1 that are within the ellipse at T = 0, will exit introduced in [5], consists in adding a corresponding kinetic its interior at some critical temperature defined by the condi- term for vortices and modifying thus the action as follows tion Z 1 1 S = d3 x 4E ρ ρ + 2π2E ρ ρ 1 C Q −∇2 Q J V −∇2 = S (Tc) , (26) gη Z 1 1 π2 1 d3 x ρ ρ ρ ρ + ˙Q 2 ˙Q + ˙V 2 ˙V assuming that the quantity on the left-hand side is larger than 2EJ −∇ 4EC −∇ Z Z Z one (i.e. there is a superinsulator at T = 0). Since the mag- +i dt d2x d2y ρ (t, x)Θ(x − y)˙ρ (t, y) , (31) netic semiaxis is always longer and thus no electric quantum Q V numbers may appear within the ellipse interior when the magnetic ones have fallen outside, the superinsulator experiences Note that this is a harmless modification, since such a kinetic a direct deconfinement transition into a topological insulator term for vortices is anyhow radiatively induced by integration over the charge dynamics, as is derived, e.g. in Eq. (34) at T = Tc. Correspondingly, superconductors undergo a phase of [12]. transition to topological insulators at T˜c defined by In order to proceed with the gauge theory derivation we g consider the action formulated in the Minkowski space, with = S (T˜c) . (27) η the only change of missing “i” in the interaction term between the charges and vortices, SUPPLEMENTARY NOTES Z 1 1 S = d3 x 4E ρ ρ + 2π2E ρ ρ M C Q −∇2 Q J V −∇2 V Gauge theory of Josephson junction arrays (JJA) Z 1 1 π2 1 d3 x ρ ρ ρ ρ + ˙Q 2 ˙Q + ˙V 2 ˙V 2EJ −∇ 4EC −∇ In this note we demonstrate that the general topological Z Z Z 2 2 action of the Cooper pairs-vortex system in two dimensions, + dt d x d y ρQ(t, x)Θ(x − y)˙ρV(t, y) , (32) given by Eq. (3) of the main text, naturally arises for the lateral Josephson junction array (JJA). Our starting point is the Now we combine the two independent variables encoded in coupled Coulomb gas description of [12] (Eq. (31) there). We the charge density and its time derivative in a single dual will consider first the continuum formulation of the coupled gauge field strength by introducing for charges and vortices Coulomb gas, which is given by the Euclidean action two fictitious gauge fields aµ and bµ, Z " # 1 1 1 1 i i j i j ˙ S d3 x E C ρ ρ π2E ρ ρ ρ ρ f = ∂ jb0 − b j , = 4 C Q 2 Q + 2 J V 2 V + ˙Q 2 ˙Q −∇ −∇ 2EJ −∇ i i j i j Z Z Z g = ∂ ja0 − a˙ j . (33) 2 2 + i dt d x d y ρQ(t, x)Θ(x − y)˙ρV(t, y) , (28) In this representation we take the spatial gauge fields ai and i j i j 2 b j as transverse, ai = ∂ jη and bi = ∂ jχ, since a longi- where EJ is the Josephson coupling, EC = 2e /C (with C the tudinal part can be reabsorbed by a redefinition of a0 and b0. junction capacitance) is the charging energy of the array and Rewriting the action as we have used Z    1 1 √ √  2π S = d3 x  f i f i + gigi + 2 a ρ + 2 b ρ  − ln|x| = δ(x) , (29) M  2 2 0 Q 0 V −∇2 2ev 2eq Z Z Z 2 2 and + dt d x d y ρQ(t, x)Θ(x − y)˙ρV(t, y) , (34) ! x2 Θ(x) = arctan . (30) with x1 2 2 As usual in statistical field theory, this action plays the same ev = 2π EJ , 2 role as the Hamiltonian of a 3D statistical mechanics model, eq = 4EC , (35) 9 and realizing that two Lagrange multipliers into the Chern-Simons term [21] µνα Z Z " # (1/π)aµ ∂νbα. This gives the total action 3 1 i i 3 1 2 1 2 d x f f = d x b0(−∇ )b0 + χ˙(−∇ )χ ˙ , Z   2e2 2e2 2e2  1 1 1  v v v S = d3 x  f i f i + gigi + i a µνα∂ b  , (44) Z Z    2 2 µ ν α 1  1 1  2ev 2eq π d3 x gigi = d3 x  a (−∇2)a + η˙(−∇2)˙η(36), 2  2 0 0 2  2eq 2eq 2eq where we have rotated back to the original Euclidean space. one obtains the two Coulomb interactions for charges and vortices by eliminating the non-dynamical Lagrange multipliers Effective action for the superinsulator a0 and b0 after having solved the Gauss law constraints √ 2 2 To explore the implications of vortex condensation we shall ∇ b0 = 2ev ρV , √ focus on the 3D case and consider the SIT effective action ∇2 2 a0 = 2eq ρQ . (37) in its lattice formulation. The corresponding superinsulator We can now pack also the charges into a complete gauge partition fuction is given by Z √ theory formulation by rewriting X −S a ,b i`2 2 b M D D ( µ µν) 2 µν µν √ ZSI = aµ bµ e e , (45) 0 aµ,bµν ρ = ( 2/2π) f , {Mµν} Q √ ρ = ( 2/2π)g0 , (38) V where S aµ, bµν represents the pure gauge part of the action where f 0 and g0 are the time components of three-dimensional and we have set Qµ = 0 since charges are dilute in this phase. dual field strengths The integers Mµν are the lattice representation of the world- surfaces spanned by the one-dimensional vortices. For closed µ 1 µνα µνα vortices, we can express these integers as in the form Mµν = f = fνα = ∂νbα , 2 `kˆµναnα with nα ∈ Z,

µ 1 µνα µνα Z √ g = gνα = ∂νaα . (39) X 3 2 −S (aµ,bµν) i` 2 nµkµαβbαβ 2 ZSI = DaµDbµ e e . (46) aµ,bµ In the gauge theory formulation these represent the conserved {nµ} charge and vortex currents Finally, we can turn the sum over nµ into an integral by the √ usual Poisson formula µ 2 µ j = f , X X Z 2π i2πnµkµ √ f nµ = dnµ f nµ e , (47) 2 nµ kµ φµ= gµ . (40) 2π where the new integer link variables {kµ} must satisfy dˆµkµ = 0 µ µ The Bianchi identities for ∂µ f and ∂µg then yield in order to guarantee gauge invariance under transformations √ nµ → nµ + `dµt. In principle, this gauge invariance should be 2 2 ρ˙Q = (−∇ )χ ˙ , gauge fixed in (46) by introducing, e.g. a quadratic term in nµ. √2π Removing the gauge fixing in the final result gives back the 2 same result obtained by naive integration, ρ˙ = (−∇2)˙η . (41) V 2π Z X √ D D −S (aµ,bµν) 3 − Substituting these expressions in (34) via (36) one obtains also ZSI = aµ bµ e δ ` kµαβbαβ 22πkµ . aµ,bµ the kinetic terms in (32). {kµ} (48) Finally, using The important consequence of this result√ is that gauge fields √ √ 2 2 2 ` bµν become quantized in units of (2π 2) so that the orig- ρ = f 0 = i j∂ b , Q 2π 2π i j inal gauge symmetry with gauge group R is broken√ down Z ∂ η = −i ja , to , with allowed gauge functions λµ = (2π 2)iµ, with i j ∈ ˆ i j iµ Z. This means that magnetic monopoles dµ Mµν are al- ∂iχ = − b j , (42) lowed at the boundaries of magnetic surfaces, since exp(−S ) and the identity is invariant under the restricted gauge transformations. The magnetic world-surfaces are actually open, reflecting the fact 2 ! ∂ j 1 x that the vortex number is not conserved in a condensate. In i j δ2(x) = − ∂ arctan , (43) ∇2 2π i x1 3D, the magnetic monopoles are particles, in 2D, they are tunneling events at the end of vortex world-lines, i.e. in- one can transform the last topological term in (32) into stantons [18]. We can now unravel the nature of the superin- R 3 i j 4 d x (−1/π)bi a˙ j which can be recombined with the sulator by adding to (45) the minimal coupling i` 2eAµ jµ of 10

Cooper pairs to the external electromagnetic field A , where where ∇2 denotes the 2D Laplacian and Q and M are the √ √ µ 2 µ µ 2 2 ˆ integer link variables describing the charge and vortex world- jµ = 2π hµ = 2π kµνρbνρ and by computing the effective action SI lines, respectively. S eff(Aµ). Integrating out the gauge fields and retaining only the dominant self-interactions we obtain In order to proceed we follow the standard arguments of [37] to retain only the self-interaction terms in (52). Near SI exp −S eff Aµ = the transition where large loops and long strings condense,   the2 typical configurations of the fields Q and M are very X  X 2 ˜ !2 X e`2F˜ !  µ µ  2  2e` F0i 1 2 i j  = exp −γ  2vc − M0i + − Mi jrare.(49) Therefore,, one may expect that the forces on every bond   2π vc 2π  Mµν  x,i x,i, j due to its neighbours (both on the same loop and on other ones) cancel out. Consider a closed string made of N bonds, where with integer quantum numbers Q = Q and M = M on all gηµ µ µ γ2 = A , (50) the lattice bonds forming the string and zero elsewhere. This vc corresponds to a fluctuation in which a charge-anticharge or vortex-antivortex pair is created from the vacuum, lives for a and F˜µν = kˆµναAα is the dual electromagentic field strength, g and η are the two parameters governing the phase struc- “time” proportional to its length in the 0 direction and is then annihilated in the vacuum again. We are interested in long- ture and µA = O(1) is a numerical parameter related to the entropy of surfaces. Since, as shown above, the condens- living fluctuations, in which the dominant contribution to the action comes from the “time” terms, first and third terms in ing magnetic excitations Mµν over which we have to sum in the partition function are unconstrained integers, allowing for (52). These fluctuations can be assigned an energy (equiva- magnetic monopoles, this is nothing else than a Villain for- lent to Euclidean action in statistical field theory) " # mulation of the non-relativistic version of the famed Polyakov 2 eq 2 ev 2 S top = πmT`vcG(mT`vc) Q + M N , (53) compact QED model [18]: ev eq γ2 X h i where G(m `v ) is the diagonal element of the lattice ker- S SI A v − e`2F˜ T c eff µ = 2 2 c 1 cos 2 0i 2π nel G(mT`v , x − y) representing the inverse of the operator x,i c (`2/v2)(m2v4 − d dˆ − v2∇2). The kernel G(m `v , x) is defined γ2 X 1 h i c T c 0 0 c 2 T c + 1 − cos 2e`2F˜ . (51) by the equation 2π2 v i j x,i, j c 2 2 2 4 ˆ 2 2 (` /vc)(mT vc − d0d0 − vc∇2) G(mT`vc, x) = δx,0 . (54) In 2D, calculations follow exactly the same lines and give the Defining the Fourier transform G(mT`vc, x) = analogous result with the entropy of surfaces µA substituted by R +πvc/` R +π/` 2 dk0 d k G(mT`vc, k) exp(ik · x) we obtain the entropy of lines µ ≈ ln 5. −πvc/` −π/`

Z +πvc/` Z +π/` 2 2 2 2 4 ˆ 2 2 ik·x dk0 d k G(mT `vc, k)(` /vc)(mT vc − d0d0 − vc∇2)e −πvc/` −π/` Z π Quantum phase structure 1 d3k ik·x . = 3 e (55) (2π) π In this Supplementary Note we derive the structure of the 2 2 2 4 Applying finally the finite difference operator (` /vc)(mT vc − phases in the vicinity of the SIT so that one could posit the su- ˆ 2 2 d0d0 − vc∇2) to the exponential in the Fourier transform and perinsulating state with respect to superconducting and Bose rescaling momenta gives the final result metal phases in the phase diagram and introduce the graphic Z π ellipsoid technique for determining the phase that realizes at 1 3 1 G(mT`vc) = d k . (56) 3 i 2 the given values of the parameters. This technique is used to (2π) −π 2 P2 k (mT`vc) + i=0 4 sin 2 calculate the temperature of the deconfinement transition in the main text. We start with integrating out the gauge fields in The string entropy, however is also proportional to their Eq. (12) of Appendix of the main text in order to obtain an ef- length, being given by µN with µ ≈ ln 5 since at each step fective Euclidean action for point charges and vortices alone. the non-backtracking strings can choose among 5 possible di- The real part of this action, the one that enters the determina- rections on how to continue. One can thus assign the free tion of the phase structure, is given by energy √ " # eq ev 1 µP 2 2 2 real 2 F = πmT`vcG(mT`vc) Q + M − N , (57) S top = vc √ × ev eq η εP`   X e2v (e2/v ) δ to a string of length L = `N carrying electric and magnetic  q c q c i j  Q0 Q0 + Qi Q j + quantum numbers Q and M, respectively. Here we have intro-  4 2 − ˆ − 2∇2 4 2 − ˆ − 2∇2  x vcmT d0d0 vc 2 vcmT d0d0 vc 2 duced the dimensionless parameter  2 2  X  evvc (ev/vc) δi j  2   πmT`v G(mT`vc) M0 M0 + Mi M(52)j , c  4 2 − ˆ − 2∇2 4 2 − ˆ − 2∇2  η = , (58) x vcmT d0d0 vc 2 vcmT d0d0 vc 2 µ 11 which, together with the ratio g = ev/eq fully determines the quantum phase structure, as we now show. The ground state of the quantum model is found by min- [1] Mandelstam, S. Vortices and quark confinement in non-Abelian imizing its free energy as a function of N. When the en- gauge theories. Phys. Rep. 23, 245-249 (1976). ergy term in (57) dominates, the free energy is positive and [2] ‘t Hooft, G. In High Energy Physics. Zichichi, A. Ed., Editrice consequently minimized by short closed loop configurations. Compositori, Bologna (1976). When, instead, the entropy dominates, the free energy is neg- [3] Polyakov, A. M. Compact gauge fields and the infrared catas- ative and minimized by large strings and long closed loops. trophe. Phys. Lett. 59, 82-84 (1975). The condition for condensation of long strings with integer [4] ‘t Hooft, G. On the phase transition towards permanent quark confinement. Nucl. Phys. B 138, 1-25 (1978). . quantum numbers Q and M is thus given by [5] Diamantini, M. C., Sodano, P. & Trugenberger, C. A. Gauge theories of Josephson junction arrays. Nuclear Physics B474, eq e η Q2 + η v M2 < 1 . (59) 641 – 677 (1996). ev eq [6] Kramer,¨ A. & Doniach, S. Superinsulator phase of two- dimensional superconductors. Physical Review Letters 81, 3523 If two or more condensations are allowed, one has to choose – 3527 (1998). the one with the lowest free energy. This condition describes [7] Vinokur, V.M. et al. Superinsulator and quantum synchroniza- the interior of an ellipse with the semiaxes tion. Nature 452, 613 – 615 (2008). [8] Efetov, K. B. Phase transition in granulated superconductors. s Sov. Phys. JETP 51, 1015 – 1022 (1980). ev 1 [9] Haviland, D., Liu, Y., & Goldman, A. Onset of superconductiv- rQ = , eq η ity in the two-dimensional limit. Phys. Rev. Lett. 62, 2180 – s 2183 (1989). eq 1 [10] Hebard. A. & Paalanen, M. A. Magnetic-field-tuned rM = , (60) superconductor-insulator transition in two-dimensional ev η films. Phys. Rev. Lett. 65, 927 – 930 (1990). [11] Fisher, M. P.A., Grinstein, G. & Girvin, S. M. Presence of quan- on a square lattice of integer electric and magnetic charges. tum diffusion in two dimensions: Universal resistance at the The phase diagram is consequently found by simply recording superconductor-insulator transition. Phys. Rev. Lett. 64, 587 – which integer charges lie within the ellipse when the semi- 590 (1990). axes are varied, [12] Fazio, R. & Schon,¨ G. Charge and Vortex Dynamics in Arrays  of Tunnel Junctions. Phys. Rev.B 43, 5307 – 5320 (1991). g > 1 , electric condensation = superconductor , [13] Baturina, T. I., Mironov, A. Yu., Vinokur, V.M., Bak- η< 1 →   lanov, M. R., Strunk, C. Localized superconductivity in the g < 1 , magnetic condensation = superinsulator , quantum-critical region of the disorder-driven superconductor-  insulator transition in TiN thin films. Phys. Rev. Lett. 99, g > η , electric condensation = superconductor ,  257003 (2007). η> 1 → η > g > 1 , no condensation = quantum metal ,  η [14] Sambandamurthy, G., Engel, L. M., Johansson A., Peled, E. &  1 Shahar, D. Experimental evidence for a collective insulating g < η , magnetic condensation = superinsulator , state in two-dimensional superconductors. Phys. Rev. Lett. 94, This approach immediately shows that the possible interme- 017003 (2005). [15] Baturina T. I. & Vinokur, V.M. Superinsulator–superconductor diate phase opening up for η > 1, typically called a Bose (or duality in two dimensions. Ann. Phys. 331, 236 – 257 (2013). quantum) metal [38], is nothing else than a bosonic topolog- [16] Ovadia, M., et al. Evidence for a finite-temperature insulator. ical insulator [39], with a topological long-distance effective Scientific Reports 5, 13503 (2015). action [40] (mixed Chern-Simons in 2D, BF in 3D). Note that, [17] Mironov, A. Yu. et al. Charge Berezinskii-Kosterlitz-Thouless for η < 1, there is a coexistence region for electric and mag- transition in superconducting NbTiN films. Scientific Reports, netic charges when g lies between η and 1/η. This indicates 8, 4082 (2018). that the direct transition from a superinsulator to a supercon- [18] Polyakov, A. M. Gauge Fields and Strings, Harwood Academic Publisher, Chur (Switzerland) (1987). ductor is actually a first-order transition with metastable co- [19] Gross, D. Twenty Five Years of Asymptotic Freedom, Nucl. existence of phases in a region around it. The parameter η, Phys. B: Proceedings Supplements 74 426 – 446 (1998). determining the possible appearance of an intermediate topo- [20] Diamantini, M. C., Trugenberger, C. A., Lukyanchuk, I., & Vi- logical insulator phase, acquires a particularly telling form in nokur, V.M. Gauge Topological Nature of the Superconductor- the vicinity of the SIT where eq ≈ ev. In this case Insulator Transition. arXiv:1710.10575 (2017). [21] Deser, S., Jackiw, R. & Templeton, S. Three-Dimensional Mas- ! 1 (v /c)2 π2 (v /c) sive Gauge Theories. Phys. Rev. Lett. 48, 975 (1982). η = c G c , (61) [22] Caselle, M., Panero, M. & Vadacchino, D. Width of the flux α k µ αk tube in compact U(1) gauge theory in three dimensions. JHEP 02, 180 (2016). where we have reinstated ~ and c, α = e2/(~c) ≈ 1/137 is the [23] Kogan, I. I. & Kovner, A. Compact QED3 -‘ A Simple Example fine structure constant and k = λ⊥/ξ is the Landau parameter of a Variational Calculation in a Gauge Theory. Phys. Rev.D 51, of the superconducting film. 1948 – 1955 (1995). 12

[24] Birmingham, D., Blau, M., Rakowski, M., & Thompson, G. Junction Arrays. Phys. Rev. Lett. 100, 086805 (2008). Topological Field Theory. Phys. Rep. 209, 129 – 340 (1991). [33] Kalok, D. et al. Intrinsic non-linear conduction in the super- [25] Kalb, M. & Ramond, P. Classical Direct Interstring Action. insulating state of thin TiN films. arXiv:1004.5153 (2010). Phys. Rev. D9, 2273 – 2284 (1974). [34] Basko, D. M., Aleiner, I. L. & Altshuler, B. L. Metal-insulator [26] Allen, T., Bowick, M. & Lahiri, A. Topological Mass Genera- transition in a weakly interacting many-electron system wihth tion in 3+1 Dimensions. Mod. Phys. Lett. A6, 559-571 (1991). localized single-particle states. Ann. Phys. 321, 1126 (2006). [27] Diamantini, M. C., Quevedo, F., & Trugenberger, C. A. Con- [35] Kormos, M., Collura, M., Takacs, G. & Calabrese, P. Real-time fining Strings with Topological Term. Phys. Lett. B 396, 115 – confinement following a quantum quench to a non-integrable 121 (1997). model. Nat. Phys. 13 246-249 (2017). [28] Wilczek, F. Two Applications of Axion Electrodynamics. Phys. [36] Brenes, M., Dalmonte, M., Heyl, M., & Sardicchio, A. Many- Rev. Lett. 58, 1799-1802 (1987). body localization dynamics from gauge invariance. Phys. Rev. [29] Svetitsky, B. & Yaffe, Critical behavior at finite temperature Lett. 120, 030601 (2018). confinement transitions. Nucl. Phys. B 210 [FS6], 423 – 447 [37] Banks, T., Myerson, R. & Kogut, J. Phase Transitions in (1982). Abelian Lattice Gauge Theories. Nuclear Physics B 129 493- [30] Diamantini, C. M., Trugenberger, C. A., Vinokur, V.M. 510 (1977) Vogel-Fulcher-Tamman criticality of 3D superinsulators. [38] Das, D., and Doniach, S. Phys. Rev. B 64 134511 (2001). arXiv:1710.10575. [39] For a review see: Hasan, M. Z. & Kane, C. L. Topological In- [31] Tamir, I., Levinson, T., Gorniaczyk, F., Doron, A., Lieb, J. & sulators. Rev. Mod. Phys. 82 3045 (2010). Shahar, D. Excessive Noise as Test fo Many-Body Localization. [40] Cho, G. Y. & Moore, J. E. Topological BF Field Theory De- arXiv:1806.09492. scription of Topological Insulators. Ann. Phys. 326 1515-1535 [32] Fistul, M. V., Vinokur, V.M. & Baturina, T. I. Collective (2011). Cooper-Pair Transport in the Insulating State of Josephson-