Literature DB >> 33319789

Effective field theories for interacting boundaries of 3D topological crystalline insulators through bosonisation.

Patricio Salgado-Rebolledo1, Giandomenico Palumbo2, Jiannis K Pachos3.   

Abstract

Here, we analyse two Dirac fermion species in two spatial dimensions in the presence of general quartic contact interactions. By employing functional bosonisation techniques, we demonstrate that depending on the couplings of the fermion interactions the system can be effectively described by a rich variety of topologically massive gauge theories. Among these effective theories, we obtain an extended Chern-Simons theory with higher order derivatives as well as two coupled Chern-Simons theories. Our formalism allows for a general description of interacting fermions emerging, for example, at the gapped boundary of three-dimensional topological crystalline insulators.

Entities:  

Year:  2020        PMID: 33319789      PMCID: PMC7738686          DOI: 10.1038/s41598-020-77966-3

Source DB:  PubMed          Journal:  Sci Rep        ISSN: 2045-2322            Impact factor:   4.379


Introduction

Time-reversal-invariant topological insulators are among the most well studied topological phases of matter. In three dimensions, they are characterised by suitable topological numbers in the bulk that guarantee the existence of topologically protected massless Dirac fermions on the boundary[1,2]. Although the topological invariant is a number, on the slab geometry, it has been shown that robust surface states are given by an odd number of Dirac fermions per boundary[3]. The situation changes in the case of three-dimensional topological crystalline insulators (TCIs), namely topological insulators characterised by further crystalline symmetries, such as mirror and rotation symmetries[4-11]. In particular, for three-dimensional TCIs protected by a single mirror symmetry, one can define the so called mirror Chern number on a given two-dimensional plane, which is invariant under the mirror symmetry. These phases host Dirac cones on each boundary[5]. Recently, mirror-invariant boundary interactions in these systems have been intensively studied by employing several approaches, such as non-linear sigma models[12,13], the coupled-wire method for [14], Higgs phases for [15] and symmetry arguments for [16]. Bosonisation represents another important quantum-field-theory approach to study interacting Dirac fermions. It was originally formulated in 1+1 dimensions to map the massive Thirring model to the Sine-Gordon theory[17,18] and then extended in higher-dimensional relativistic systems under the name of functional bosonisation[19]. Although this method has numerous implications that are relevant to condensed matter physics, it has been mainly employed in interacting systems involving a single emergent gauge field. The goal of this work is to analyse gapped and mirror-broken boundary states in presence of quartic contact interactions between several pieces of fermions. We assume that the interactions are exclusively acting on the boundary, while the bulk of the system is descried by free topological insulating phase. This is similar to the case of 2D time-reversal-invariant topological insulators, where the helical Luttinger liquids appear on the interacting boundary of the system while the 2D bulk states are still related to the free-fermion models[20]. We introduce then an external magnetic field orthogonal to the surface to induce a Dirac mass that breaks both time-reversal and mirror symmetries and consider generic intra- and inter-species interactions. For simplicity, we fix , as in[14], and employ functional bosonisation. This approach will allow us to map the self-interacting fermion model to free bosonic models. We are interested in obtaining the low energy topological properties of these effective bosonic models for various configuration of inter and intra-species interactions of the original fermionic model. Our analysis shows that all the resulting effective models contain topological Chern–Simons terms that usually emerge in a variety of T-broken systems such as the quantum Hall states[21,22], surface states of three-dimensional topological insulators[2] and graphene coupled to external magnetic fields[23]. However, differently from these previous works, we show the existence of a new exotic phase, characterised by a higher-derivative Chern–Simons term[24], when one of the intra-species interaction is switched off. This phase supports a massive U(1) boson and a ghost mode, which is completed decoupled from the bosonic mode, and thus it is a “good ghost”[25,26]. Moreover, we show the existence of another phase in which the bosonic theory comprises two massive bosons, with a mutual Chern–Simons term, which generalises the well-know Chern–Simons–Maxwell theory to multi-field gauge fields. The topological sector of this phase resembles the effective action studied in Ref.[27] in the context of thin-film topological insulators. Importantly, our approach is quite general and can be directly extended to . This will allow to identify novel topological crystalline phases in presence of very general contact interactions.

Two-fermion interacting system

The starting point of our construction is a -dimensional system of two interacting fermion species and living on the boundary of 3D topological crystalline insulator with bulk mirror Chern number . The corresponding effective action is given bywhere is the time-reversal broken mass induced by an external magnetic field orthogonal to the surface of the 3D TCI defined on the xy-plane. Here, we use the convention for the Minkowski metric . The gamma matrices are defined in terms of the Pauli matrices as , , and , the Dirac conjugate is , and the Clifford algebra has the form . For convenience in the presentation we choose the intra-species coupling constants to be given by and , where , and are real constants, and , while is the inter-species coupling constants. To analytically determine the behaviour of this interaction system we employ functional bosonisation. This is a powerful approach that will allows us to identify the equivalent bosonic theory describing our model in the low-energy regime. By defining , , the corresponding generating functional has the formIn order to integrate out the fermion field , we follow[19,28] (see also[29-31]) and express the third term in the action aswhere is an Hubbard–Stratonovich vector field. By replacing this back into the generating functional Z, we obtainWe now integrate out to obtain an effective bosonic action . In the large mass limit, it can be approximated as[32-34]where and is the (2  1)-dimensional Levi-Civita symbol with . Therefore we can writewhereThe action (7) holds for general values of the parameters , , and . The first two terms are purely given in terms of the vector field . They correspond to the self dual action resulting from a single fermionic species, , introduced in[35,36]. The total action is an extension to that self-dual model having the field coupled to a self-interacting fermionic field . In the following we consider specific configurations of these couplings and extract the behaviour of the model in each case.

Pauli term and higher-derivative Chern–Simons action

In this section we show that for a particular value of the couplings the fermionic system (1) can be described in the low energy limit by a single fermion field non-minimally coupled to an effective U(1) gauge field. In particular, this coupling configuration gives rise to a higher-derivative Chern–Simons theory[24]. We start our analysis of (7) by considering the interpolating actionwhich is given in terms of the Dirac fermion field , the vector field and a new gauge field . The path integral of is equivalent to the functional integral associated to the effective action (7). By integrating out the field in (8), we obtainwhere is given by (7). On the other hand, by integrating out the vector field in the interpolating action (8) we findwhere the action is given byIn this dual effective action the field is the field strength associated to the gauge potential and we have also used the standard definition . The action is dual to , so it faithfully describes the original system (1). The advantage of is that it is given in terms of the gauge field rather than the vector field and thus it is easier to identify its topological character. From (11) we observe that the effective action (7) can be dualised to a Chern–Simons–Maxwell model coupled to the fermion field by means of the Pauli term. Note that, after using the interpolating action, the coupling of the self interaction for has been shifted back to the original value it had in (1). By directly comparing (1) and (11), we see that we can interpret the Pauli coupling as the low energy description of the mixed interaction term . The duality between (11) and (7) has been previously established on-shell by eliminating the field or from the interpolating action (8) by means of their corresponding field equations[37,38].

Higher derivative Chern–Simons theory: the case

We now consider the action (7) for the case where and , which corresponds to in (1). In that case, the action (7) does not describe free fermions so that they cannot be integrated out. Moreover for we cannot employ a similar relation to (3) in order to linearise the interactions. In this case the self interaction in the dual action (11) vanishes and the effective theory takes the form of a fermion non-minimally coupled to the field strength by means of the Pauli term. Defining the Hodge dual of the curvature the action takes the formIn other words, the Pauli term couples the magnetic moment of the fermions with the magnetic field[39-41]. Note that in dimensions the magnetic moment is a scalar leading to the coupling term seen in (12). To analyse the properties of (12) we integrate out the fermions . As shown in[42-44] the Pauli term can be obtained starting from the standard minimal coupling in the Dirac action and shifting the gauge field into the generalised connection . We can then integrate out in (12) by using the result (5) for the generalised connection and then set , i.eBy using this result, which is also compatible with Ref.[45], the corresponding effective action takes the formIt is important to remark that, even though the higher-derivative term in the above action looks like a Chern–Simons form, it is not topological as it depends on the space-time metric. Indeed, as shown in[24], up to boundary terms one can writeThus, this term leads to a non-vanishing contribution to the energy-momentum tensor, which is a signature of its non-topological nature. As it has been shown in[24], the action (14) includes a ghost mode. Now we will show that this model admits a description in which the ghost is decoupled from the physical degree of freedom. In order to do so, we follow[46] and decompose the vector potential in terms of new variables X and Y as followsThe effective action (14) then becomesWe can now integrate out the field X in the corresponding partition function , which yields an effective action for Y given byThis higher derivative scalar field action can be expressed in terms of two Klein–Gordon fields defined byThe action then takes the form[25,26]Hence, the field redefinition (19) allows us to express (14) as the action for two decoupled massive Klein–Gordon fields, and . The field is a physical Klein–Gordon field, while is a ghost. Since the ghost fields is totally decoupled from the physical degree of freedom, the physical spectrum is not affected by it. In this sense we have a “good” ghost[25] emerging in our theory. From (19) we see that for any values of and . On the other hand, can be positive or negative depending on the values of the couplings and , implying that the ghost can be also a tachyon. Thus, this theory shares similar features with the Chern–Simons–Maxwell theory[47] that describes a single propagating massive bosonic mode. In our case, the effect of the higher-derivative term is to renormalise the topological mass of the boson.

Single and mutual Chern–Simons theories

In this section we show that, besides the Chern–Simons and Maxwell terms, suitable choices of the parameters in the starting action (1) lead to an effective description of the system that includes a mutual Chern–Simons term[48].

Single Chern–Simons theory: the case

The case corresponds to interaction couplings in (1) that satisfy . In this case we can define the four-spinor and the corresponding current , where . The generating functional (2) then boils down towhere . Since this is a standard Thirring model for we can linearise the interactions by introducing a vector field [19], so that by means of Gaussian integration we implement the replacement in (21). Using (5) to integrate out , the low energy behaviour of this system is captured by the following effective actionThis result can be also obtained from (7) by setting in and subsequently integrating out . Following[35,36], we can dualise this action to a Chern–Simons–Maxwell theoryHence, for the specific case where the system becomes formally equivalent to a single species self-interacting fermion that gives rise to a Chern–Simons theory with coupling . This theory describes massive bosons that only mediate short-range interactions[47].

Mutual Chern–Simons theories: the , case

The choice of parameters and corresponds to the action (1) with . In this case the effective action (7) becomesso one can integrate out directly. Following similar steps as above, we use (3) to linearise the self interaction in the path integral associated to (24) by introducing a new vector field . Subsequently, using (5) with the replacement leads toIn this action both and are vector fields. In order to turn them into gauge fields we employ the interpolating action procedure. Consider the interpolating path integral (see appendix A in supplementary information)where the masses , and are to be fixed in term of the couplings constants in (25). Integrating out the fields and leads exactly to the functional integral of the action (25), i.e.provided the masses , and are given byThe dual theory is obtained by integrating out the vector fields and in (26), which yieldswhere we have introduced a second field strength, . Interestingly, for , there appears an emergent symmetry that exchanges the gauge fields, i.e.This theory describes two massive bosons and generalises the Chern–Simons–Maxwell theory[47], which is defined for a single U(1) gauge field and the double-Maxwell-BF theory[49-51]. The latter, defined for , has been employed to study the Meissner effect in two-dimensional superconductors/superfluids that preserve time-reversal symmetry. In this context, the two massive bosons can be interpreted as massive modes related to an effective London penetration length[50]. Here, we give a physical interpretation of our model by neglecting the Maxwell terms and focusing on the topological sectorwhich is dominant at large distances. By a suitable rescaling of the gauge fields, this topological action formally coincides with that one derived in Ref.[27] in thin-film topological insulators. In this context, our T-broken action would describe an emergent quantum anomalous Hall state induced by interactions. In fact, the presence of in all the three coefficients , and is the signature of the presence of a common Chern number encoded in those terms that changes sign when the external Zeeman field is flipped. There are however important physical differences with respect to Ref.[27]. In that work, the field is identified with an external electromagnetic field and the two fermion species live on different boundaries, such that only in the thin-film limit the effective 2D model for the boundary contain both species. Finally, note that the effective action in (31) can be further reduced by integrating out the gauge field , which yields the Chern–Simons action we met in (23) with the same mass . Therefore, at the level of the topological affective action, integrating out is equivalent to set in the original action (1). On the other hand, if we choose to integrate the gauge field , we obtain the Chern–Simons term of (23) for the field with . This result corresponds to setting in (1), which eliminates the interaction between and and keeping only the Thirring self-interaction for with coupling .

Response to external electromagnetic field

In this subsection we probe the system in Eq. (24) by introducing an external electromagnetic potential . This requires to modify action (1) by minimally coupling the fermions and to as followsRepeating the steps outlined in the previous section leads to the generalization of the effective action (25) to the following oneThe interpolating path integral (26) is then modified asand the dual action (31) is generalized toIntegrating out the fields and in the corresponding path integral reduces (35) to the following actionwhich is given only in terms of . Varying this action with respect to we obtain the currentThis clearly shows a topological Hall response of our system in presence of an external electromagnetic field. This current is sensitive to the sign of the generated mass m, but it is insensitive to the particular values of the interaction couplings, , and .

Domain walls and chiral bosons

We now show that our effective topological field theory in Eq. (31) allows us to describe the 1D gapless modes trapped along defect lines (namely, 1D domain walls) that we can add on the 2D gapped boundary. In fact, defect lines behave as an effective spacial boundary for the 2  1-D bosonic model in Eq. (31) and the CS/CFT correspondence[52] allows us to derive the chiral boson action associated to the 1D modes[53]. For this action, we can define the following new fieldsIn this way, the effective action takes the form of two decoupled Chern–Simons termswhere we have definedFollowing[52], we adopt coordinates (t, x, y) and consider the generalized axial gaugeThe Gauss law leads to locally pure gauge configurationswhich can be implemented directly in the action (39) and leads toTherefore, the 1D dynamics is described by two chiral bosons, which are determined by the parameters and velocity v[53]. Importantly, these chiral modes trapped along the line defects can be eventually measured in experiments.

Conclusions

In this article we have studied the effect interactions have on two Dirac fermions in dimensions. As we are interested in the topological properties of this system we employed the bosonisation method in order to obtain the corresponding effective gauge theories. As we vary the fermion couplings with intra-species interactions, and , and inter-species interactions we obtain a variety of topological theories that correspond to different phases of the model. When one of the fermionic species does not self-interact, , then the system is described by a Chern–Simons theory with a higher-derivative term. With the appropriate field reparametrisation this theory can be written in terms of a physical scalar field and a “good ghost” that completely decouples from the physical spectrum. Hence, it gives a well behaved topological theory, which shares similar features with the Chern–Simons–Maxwell theory and the topological mass is renormalised by the higher-derivative term. Beyond this particular regime, when we take the action is given in terms of two coupled Chern–Simons theories that describes two propagating massive bosons. In this case the system is described by an emergent quantum anomalous Hall state induced by interactions and the two interacting massive Dirac fermions can be mapped to the two massive bosons. Moreover, for a particular choice of the coupling constants, there appears an emergent symmetry. In terms of physical observables, we have shown that by coupling the interacting model to an external electromagnetic field, the semiclassical currents are related to a topological Hall response. Moreover, by adding suitable domain walls on the gapped boundary, there appear propagating 1D modes trapped along the domain walls (i.e. defect lines). This is due to the well-known CS/CFT correspondence, where the CFT describes the 1D chiral modes, which can be in principle measured in experiments. Our method does not have a simple interpretation in the case where so an alternative approach needs to be taken. We leave this case for a future investigation. Finally, note that our approach can be naturally generalised in various ways. One can consider multi-species interactions described by multi-U(1) gauge fields. This paves the way to study the interacting boundaries of 3D topological crystalline insulators for through functional bosonisation. Moreover, one can consider multi-SU(N) non-Abelian generalisation of the gauge fields along the lines of Ref.[54]. Supplementary Informations.
  8 in total

1.  Composite-fermion approach for the fractional quantum Hall effect.

Authors: 
Journal:  Phys Rev Lett       Date:  1989-07-10       Impact factor: 9.161

2.  Order parameter and Ginzburg-Landau theory for the fractional quantum Hall effect.

Authors: 
Journal:  Phys Rev Lett       Date:  1989-01-02       Impact factor: 9.161

3.  Anomalous Crystal Symmetry Fractionalization on the Surface of Topological Crystalline Insulators.

Authors:  Yang Qi; Liang Fu
Journal:  Phys Rev Lett       Date:  2015-12-02       Impact factor: 9.161

4.  Helical liquid and the edge of quantum spin Hall systems.

Authors:  Congjun Wu; B Andrei Bernevig; Shou-Cheng Zhang
Journal:  Phys Rev Lett       Date:  2006-03-14       Impact factor: 9.161

5.  Topological insulators in three dimensions.

Authors:  Liang Fu; C L Kane; E J Mele
Journal:  Phys Rev Lett       Date:  2007-03-07       Impact factor: 9.161

6.  The birth of topological insulators.

Authors:  Joel E Moore
Journal:  Nature       Date:  2010-03-11       Impact factor: 49.962

7.  Topological crystalline insulators.

Authors:  Liang Fu
Journal:  Phys Rev Lett       Date:  2011-03-08       Impact factor: 9.161

8.  Skyrmion Superfluidity in Two-Dimensional Interacting Fermionic Systems.

Authors:  Giandomenico Palumbo; Mauro Cirio
Journal:  Sci Rep       Date:  2015-06-17       Impact factor: 4.379

  8 in total

北京卡尤迪生物科技股份有限公司 © 2022-2023.