betaFeedback

WorksPhysics and Astronomy

Wiedemann-Franz law violation domain for graphene and nonrelativistic systems

Abstract

A systematic non-fluid to fluid transition framework and comparative research on Lorenz ratios for graphene and nonrelativistic systems have been studied to identify their Wiedemann-Franz law violation domain. Here, Lorenz ratio is defined as thermal conductivity divided by electrical conductivity times temperature times Lorenz number. In non-fluid framework, Lorenz ratio become exactly one, which means that the Wiedemann-Franz is obeyed within a Fermi Liquid domain. When one enters from Fermi Liquid to Dirac Fluid domain, Lorenz ratio becomes less than one in non-fluid framework but in fluid framework, it always remain greater than one for both domain. By compiling our outcomes and connecting with experimental data, a non-fluid to fluid transition framework is expected during the transition from Fermi Liquid to Dirac Fluid domain.

Introduction

In 1853, Gustav Wiedemann and Rudolph Franz experimentally discovered a universal or constant value of the ratio between thermal (κ\kappa) and electrical (σ\sigma) conductivity, which are approximately followed by all metals. Later, in 1872, Ludvig Lorenz theoretically realized that the ratio in terms of the universal constants kBk_B (Boltzmann constant) and ee (electric charge) is:

κσT=π23(kBe)2=L0=2.445×10−8wattΩK2,(1)\begin{aligned} \frac{\kappa}{\sigma T} = \frac{\pi^2}{3}\left(\frac{k_B}{e}\right)^2 \\ &= L_0 = 2.445 \times10^{-8} watt\frac{\Omega}{K^2}, \tag*{(1)} \end{aligned}

where L0L_0 is known as the Lorenz number. The fact is very well known from our text book Refs. [1, 2, 3, 4, 5]. In natural unit, ℏ=c=kB=1\hbar= c = k_B = 1 and e2=4π137e^2 = \frac{4\pi}{137}, we can express temperature TT in eV, σ\sigma in eV and κ\kappa in eV2^2. So Lorenz number will be a dimensionless ratio L0=137π12≈35.84L_0 = \frac{137\pi}{12} \approx35.84. In the present paper, we will mostly use the natural unit methodology for convenience, but we may go to the actual unit in some cases (whenever necessary).

The Wiedemann-Franz (WF) law has proven remarkably robust for many metallic systems, where electrons are the main electric and thermal charge transportation carriers. Due to the similar mechanism of transportation in free electron theory [1, 2, 3, 4, 5], the dimensionless (in natural unit) ratio of two transport coefficients - thermal and electrical conductivity becomes constant. However, deviations have been observed in many systems like Li0.9Mo6O17Li_{0.9}Mo_6O_{17} [6], CeCoIn5CeCoIn_5 [7], ZrZn2ZrZn_2 [8], YbRh2Si2YbRh_2Si_2 [9], (Pr,Ce)2CuO4(Pr,Ce)_2CuO_4 [10] and VO2VO_2 [11]. For our convenience, if we define the ratio of thermal and electrical conductivity as L=κσTL = \frac{\kappa}{\sigma T}, and if we call L/L0L/L_0 as Lorenz ratio, then the validity of the WF law will be established from the relations L=L0L = L_0 or LL0=1\frac{L}{L_0} = 1. On the other hand, relations LL0>1\frac{L}{L_0} > 1 or LL0<1\frac{L}{L_0} < 1 reflect violation of the WF law. A large diverging outcome LL0>104\frac{L}{L_0} > 10^4 is expected in one-dimensional exotic Tomonaga-Luttinger liquids [6]. On the other hand, a strong downward violation of the WF Law L<L0L < L_0 was observed in WP2WP_2 semimetal [12, 13] and MoP binary compound [14]. This downward violation is also observed in heavy fermion metals [7], marginal ferromagnetic metal [8], anti-ferromagnetic metal [9] and copper-oxide superconducting material [10] at low-temperature regions as well as in metallic vanadium dioxide [11] at high-temperature range (240-340°K). Similar violation is also observed for the nuclear matter, produced in heavy ion collision experiments [15, 16, 17, 18]. Understanding the WF law violation mechanism of that wide range of systems is a non-trivial task and challenge for the theoretical community to explain.

Recently, Crossno et al. [19] have found a similar kind of the WF law violation in graphene systems by tuning the doping concentration and temperature. Unlike standard metals, the Fermi level of graphene can be moved upwards or downwards from the relative Dirac point depending upon the n- or p-type doping of graphene. Fermi energy (ϵF\epsilon_F) or a chemical potential (μ\mu) at Dirac point is considered zero for undoped graphene and, via n- or p-type doping, will shift it towards the positive or negative direction. The net charge carrier concentration will be tuned from zero to non-zero when one experimentally goes from undoped to doped graphene cases, and theoretically, one goes from ϵF=0\epsilon_F = 0 to ϵF≠0\epsilon_F \ne0 cases. This kind of tuning possibility of ϵF\epsilon_F can not be expected for metal systems as it almost remains constant. Their typical values remain within the range ϵF=2−10\epsilon_F = 2 - 10 eV, for which one can expect the limit ϵF/T≫1\epsilon_F/T \gg1 at room temperature T≈0.025T \approx0.025 eV. In this limit, one can consider electrons in metal as Fermi gas and use Sommerfeld expansion, which provides a linear TT dependent on electron-specific heat. Fermi gas is completely based on the crude non-interacting assumption, but there is a theory of interacting Fermions system, which is popularly known as the Fermi Liquid (FL) Theory[20]. Originally, Landau [21, 22, 23] proposed this phenomenological theory for studying3^3He. In FL prescription, interaction is taken care of via the effective mass of electrons by assuming a quasi-particle picture. So, if we define ϵF/T>1\epsilon_F/T > 1 as the FL domain and ϵF/T≫1\epsilon_F/T \gg1 as the Fermi gas (FG) domain, then electron-doped graphene systems almost follow WF law. When we go towards undoped or clean graphene systems with ϵF/T<1\epsilon_F/T < 1, Fermi liquid theory becomes invalid (as it is commonly accepted). It is concluded from the experimental observation of WF law violation in this ϵF/T<1\epsilon_F/T < 1 or ϵF/T≪1\epsilon_F/T \ll1 domains, popularly called Dirac fluid (DF) or Dirac Liquid (DL) domain. Theoretical works [24, 25, 26, 27, 28, 29, 30, 31] are attempted to explain this WF law violation in the DF domain. In this regards, electron hydrodynamics (eHD) or fluid dynamics in graphene system, recently observed by Refs [32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45], is thought to be linked with this WF law violation. In the direction of material perspective [46], one can link eHD and WF law violation as follows. The electron-electron scattering length becomes larger than the electron-impurity or electron-phonon scatterings in DF domain but reverse is occured in FL or Fermi gas domain. We know that the relaxation time is inverse to scattering strength. So, the ee scattering dominating domain i.e. DF domain shows eHD behaviour, where electron relaxation time becomes much much smaller than the system size. In this context, the present work is planned to explore a transition from non-fluid to fluid type expressions of Lorenz ratio for a 2-dimensional (2D) graphene (G) system with linear energy-momentum relation (E∝pE \propto p) as well as a 3-dimensional (3D) metal system with nonrelativistic (NR) energy-momentum relation (E∝p2E \propto p^2). For mathematical completeness, we have demonstrated all possible cases like 2D-NR, 3D-NR, 2D-G, and 3D-G but our final destination is kept to show how fluid expressions of 2D-G case can be associated with WF law violation and how non-fluid expressions of 3D-NR case, followed by metal systems, obey WF law. In the direction of eHD, our earlier comparative research on ideal [47] and viscous dissipating [48] parts of energy-momentum tensor calculation clearly demonstrate that the graphene hydrodynamics is neither relativistic hydrodynamics nor nonrelativistic hydrodynamics. The present work has gone through similar comparative research on electrical conductivity, thermal conductivity, and Lorenz ratio for a systematic search of the WF law violation domain.

The article is organized as follows, in § 1, the formalism part of WF law calculations for different cases. The results and brief interpretation are in Sec. III and Sec. IV respectively. At the end, our study is concluded in § 1.4.

Formalism

Graphene is a 2-dimensional single atomic layer of carbon atoms tightly bound into a honeycomb lattice [49]. Near the Dirac point, electrons in graphene follow the linear dispersion relation

ϵ(k)=pvF,(2)\epsilon(k) = p v_F, \tag*{(2)}

where vFv_F is the Fermi velocity of electrons in graphene, whose values remain within the range 0.003c-0.01c [50]. In one direction, electrons in graphene do not follow a quadratic relation between energy and momentum (ϵ=p2/(2m)\epsilon= p^2/(2m)) like in the nonrelativistic case. On the other hand, its velocity (vFv_F) is not close to the speed of light (cc), so no relativistic corrections are expected. In that sense, electrons in graphene follow neither relativistic nor nonrelativistic. So, its different properties may not be the same as expected for traditional nonrelativistic or relativistic matters as shown in Refs. [47, 48], for thermodynamics [47] and dissipative viscous [48] properties. In this regard, thermal and electrical conductivity for the graphene (G) case may also be interesting to compare with nonrelativistic (NR) and ultrarelativistic (UR) cases.

Non-fluid description

For systematic study and understanding, we will calculate the Lorenz ratios for the 3D system and the 2D system. The expressions for 2D graphene are derived and analyzed in this section, highlighting its distinct transport characteristics. Meanwhile, the corresponding calculations for other systems, including 3D non-relativistic (NR), 3D graphene (G), and ultra-relativistic (UR) cases in both 3D and 2D, are provided in Appendix D for completeness. Then, we will also examine the limiting conditions using Sommerfeld expansion.

2D-G: For actual 2D graphene case, the electrical and thermal transport properties are:

σG2D=ne2τcvg2μ,(3)\sigma_G^{2D} = \frac{n e^2 \tau_c v_g^2}{\mu}, \tag*{(3)}
σQ3D=ne2τcc2μq,(4)\sigma_Q^{3D} = \frac{n e^2 \tau_c c^2}{\mu_q}, \tag*{(4)}
κG2D=12nvF2τc 2DG[CV]e,(5)\kappa_G^{2D} = \frac{1}{2} n v_F^2 \tau_c\,{}^{2D}G[C_V]_e, \tag*{(5)}

the ratio of κG2D\kappa_G^{2D} and σG2D\sigma_G^{2D} is given by

κG2DσG2D=12ϵFe2 2DG[CV]e.(6)\frac{\kappa_G^{2D}}{\sigma_G^{2D}} = \frac{1}{2}\frac{\epsilon_F}{e^2}\,{}^{2D}G[C_V]_e. \tag*{(6)}
κG2DσG2D=12ϵFe2 2DG[CV].(7)\frac{\kappa_G^{2D}}{\sigma_G^{2D}} = \frac{1}{2}\frac{\epsilon_F}{e^2}\,{}^{2D}G[C_V]. \tag*{(7)}

the ratio of κQ3D\kappa_Q^{3D} and σQ3D\sigma_Q^{3D} is given by

κQ3DσQ3D=13μqe2 3DQ[CV]e.(8)\frac{\kappa_Q^{3D}}{\sigma_Q^{3D}} = \frac{1}{3}\frac{\mu_q}{e^2}\,{}^{3D}Q[C_V]_e. \tag*{(8)}
κQ3DσQ3D=13μqe2 3DQ[CV].(9)\frac{\kappa_Q^{3D}}{\sigma_Q^{3D}} = \frac{1}{3}\frac{\mu_q}{e^2}\,{}^{3D}Q[C_V]. \tag*{(9)}

Using two possible expressions (see Appendix § 1.8) of specific heat for 2D-G system,

G12D[CV]e=2kB[3f3(A)f2(A)−2f2(A)f1(A)],(10){}^{2D}_{G1}[C_V]_e = 2k_B\left[3\frac{f_3(A)}{f_2(A)}-2\frac{f_2(A)}{f_1(A)}\right], \tag*{(10)}

and

G22D[CV]e=2kB[3f3(A)f2(A)−ϵFkBT],(11){}^{2D}_{G2}[C_V]_e = 2k_B\left[3\frac{f_3(A)}{f_2(A)}-\frac{\epsilon_F}{k_BT}\right], \tag*{(11)}

in Eq. (9), we get the Lorenz ratios

LG12DL0=κG12DσG12DTL0=3π2ϵFkBT[3f3(A)f2(A)−2f2(A)f1(A)](12)\frac{L^{2D}_{G1}}{L_0}=\frac{\kappa^{2D}_{G1}}{\sigma^{2D}_{G1}TL_0}=\frac{3}{\pi^2}\frac{\epsilon_F}{k_BT}\left[3\frac{f_3(A)}{f_2(A)}-2\frac{f_2(A)}{f_1(A)}\right] \tag*{(12)}

and

LG22DL0=κG22DσG22DTL0=3π2ϵFkBT[3f3(A)f2(A)−ϵFkBT](13)\frac{L^{2D}_{G2}}{L_0}=\frac{\kappa^{2D}_{G2}}{\sigma^{2D}_{G2}TL_0}=\frac{3}{\pi^2}\frac{\epsilon_F}{k_BT}\left[3\frac{f_3(A)}{f_2(A)}-\frac{\epsilon_F}{k_BT}\right] \tag*{(13)}

respectively. The SL of Eqs. (12) and (13) will be

LG12D=π23(kBe)2=L0,(14)L^{2D}_{G1}=\frac{\pi^2}{3}\left(\frac{k_B}{e}\right)^2=L_0, \tag*{(14)}

and

LG22D=2π23(kBe)2=2L0(15)L^{2D}_{G2}=\frac{2\pi^2}{3}\left(\frac{k_B}{e}\right)^2=2L_0 \tag*{(15)}

respectively.

An approach towards the fluid description

The previous subsection provides the expression of κ\kappa and σ\sigma using a standard solid state physics framework, where no fluid concept has been entered. So, we can call that the non-fluid description. In the present section, we will try to build an approach towards a fluid description by using relaxation time approximation-based Boltzmann’s transport equations. Here, we are using the terminology ”towards the fluid description” because the actual phenomenological fluid description [26] for explaining the Wiedemann-Franz law violation in the experimental 2D graphene system is quite rigorous. Aim of the present work is to zoom in the intermediate journey from non-fluid to fluid description. Interestingly, we will get a Wiedemann-Franz law violating domain in T−ϵFT-\epsilon_F plane for all cases. Let us first address the brief calculations of the Lorenz ratio for 2D-G systems; then, we will write down its final expressions for other possible systems like 2D-NR, 3D-NR, and 3D-G instead of repeating similar calculations (see Appendix § 1.10).

2D-G: Let us assume a local thermalization picture of electron fluid in a 2D-G system, where the equilibrium distribution function (see Appendix § 1.1)

f0(T(xμ))=1e(ϵ−ϵF)/kBT(xμ)+1,(16)f_0\left(T(x^\mu)\right)=\frac{1}{e^{(\epsilon-\epsilon_F)/k_BT(x^\mu)}+1}, \tag*{(16)}

will be xμx^\mu-dependent due to temperature profile T(xμ)T(x^\mu). Here, xμ=(ct,xi)x^\mu=(ct,x^i), a four-dimensional coordinate is considered for general notation, but we have to take care of i=1,2i=1,2 for the 2D system and i=1,2,3i=1,2,3 for the 3D system.

Now, let us first write down the macroscopic definitions for electrical and thermal conductivity

J⃗=σE⃗,\vec{J}=\sigma\vec{E},
Q⃗=κ∇⃗T,(17)\vec{Q}=\kappa\vec{\nabla}T, \tag*{(17)}

where electrical current vector J⃗\vec{J} and heat flow vector Q⃗\vec{Q} can have microscopic expressions:

Ji=ge∫d2ph2viδfσ,J_i=ge\int\frac{d^2p}{h^2}v_i\delta f_\sigma,
Qi=g∫d2ph2ϵviδfκ.(18)Q_i=g\int\frac{d^2p}{h^2}\epsilon v_i\delta f_\kappa. \tag*{(18)}

Here, we are assuming that the external electric field EiE_i and the temperature gradient ∇⃗T\vec{\nabla}T will create deviations δfσ\delta f_\sigma and δfκ\delta f_\kappa respectively from the equilibrium distribution f0f_0.

It is relaxation time approximation (RTA) based on Boltzmann’s transport equation (BTE) [51]

∂f∂t+∂xi∂t∂f∂xi+∂pi∂t∂f∂pi=−δfτc,(19)\frac{\partial f}{\partial t}+\frac{\partial x^i}{\partial t}\frac{\partial f}{\partial x^i}+\frac{\partial p^i}{\partial t}\frac{\partial f}{\partial p^i}=-\frac{\delta f}{\tau_c}, \tag*{(19)}

which will guide us to guess appropriate form of δfσ\delta f_\sigma and δfκ\delta f_\kappa. Considering f=f0+δf≈f0f=f_0+\delta f\approx f_0 in left hand side of Eq. (19) and local thermalization assumption of f0f_0, given in Eq. (16), we can simplify Eq. (19) as

vi∂f0∂xi+eEi∂f0∂pi=−δfστc−δfκτc,vi∂T∂xi∂ϵ∂T∂f0∂ϵ+eEi∂ϵ∂pi∂f0∂ϵ=−δfστc−δfκτc,vi∂T∂xi[CV]e∂f0∂ϵ+eEivi∂f0∂ϵ=−δfστc−δfκτc.(20)\begin{aligned} v^i\frac{\partial f_0}{\partial x^i}+eE^i\frac{\partial f_0}{\partial p^i}&=-\frac{\delta f_\sigma}{\tau_c}-\frac{\delta f_\kappa}{\tau_c},\\ v^i\frac{\partial T}{\partial x^i}\frac{\partial\epsilon}{\partial T}\frac{\partial f_0}{\partial\epsilon}+eE^i\frac{\partial\epsilon}{\partial p^i}\frac{\partial f_0}{\partial\epsilon}&=-\frac{\delta f_\sigma}{\tau_c}-\frac{\delta f_\kappa}{\tau_c},\\ v^i\frac{\partial T}{\partial x^i}[C_V]_e\frac{\partial f_0}{\partial\epsilon}+eE^iv_i\frac{\partial f_0}{\partial\epsilon}&=-\frac{\delta f_\sigma}{\tau_c}-\frac{\delta f_\kappa}{\tau_c}. \tag*{(20)} \end{aligned}

Here, we consider the approximation [CV]e≈∂ϵ∂T[C_V]_e\approx\frac{\partial\epsilon}{\partial T}, and one can expect again two possible definitions of specific heat as discussed in the earlier section. From Eq. (20), we can get the form of δfσ\delta f_\sigma and δfκ\delta f_\kappa as

δfσ=eEivi(−∂f0∂ϵ)τc,\delta f_\sigma=eE^iv_i\left(-\frac{\partial f_0}{\partial\epsilon}\right)\tau_c,
δfκ=vi(−∂f0∂ϵ)[CV]e(∂T∂xi)τc.(21)\delta f_\kappa=v^i\left(-\frac{\partial f_0}{\partial\epsilon}\right)[C_V]_e\left(\frac{\partial T}{\partial x^i}\right)\tau_c. \tag*{(21)}
δfκ=vi(−∂f0∂ϵ)[CV](∂T∂xi)τc(22)\delta f_{\kappa}=v^{i}\left(-\frac{\partial f_{0}}{\partial\epsilon}\right)[C_{V}]\left(\frac{\partial T}{\partial x^{i}}\right)\tau_{c} \tag*{(22)}

with

(−∂f0∂ϵ)=βf0(1−f0).(23)\left(-\frac{\partial f_{0}}{\partial\epsilon}\right)=\beta f_{0}(1-f_{0}). \tag*{(23)}

Using (21) in (18) and then comparing with (17), we will get the final expressions of electrical and thermal conductivity:

σ=ge2vF22τc∫d2ph2βf0(1−f0)⟹σG2D=2πkBTτc(ϵh)2f1(A)T,(24)\sigma=ge^{2}\frac{v_{F}^{2}}{2}\tau_{c}\int\frac{d^{2}p}{h^{2}}\beta f_{0}(1-f_{0}) \qquad\Longrightarrow\qquad \sigma_{G}^{2D}=2\pi k_{B}T\tau_{c}\left(\frac{\epsilon}{h}\right)^{2}f_{1}(A)T, \tag*{(24)}

and

κ=gevF22τc[CV]e∫d2ph2βf0(1−f0)⟹κG2D=4πkB2Tτch2[CV]ef2(A)T2.(25)\kappa=ge\frac{v_{F}^{2}}{2}\tau_{c}[C_{V}]_{e}\int\frac{d^{2}p}{h^{2}}\beta f_{0}(1-f_{0}) \qquad\Longrightarrow\qquad \kappa_{G}^{2D}=\frac{4\pi k_{B}^{2}T\tau_{c}}{h^{2}}[C_{V}]_{e}f_{2}(A)T^{2}. \tag*{(25)}
κF/NFG=τcπ[CV]F/NFf2(A)T2.(26)\kappa_{F/NF}^{G}=\frac{\tau_{c}}{\pi}[C_{V}]_{F/NF}f_{2}(A)T^{2}. \tag*{(26)}
κF/NFQ=24π2τc[CV]F/NFf3(A)T3.(27)\kappa_{F/NF}^{Q}=\frac{24}{\pi^{2}}\tau_{c}[C_{V}]_{F/NF}f_{3}(A)T^{3}. \tag*{(27)}

Now, using two different forms of specific heat, given in Eqs. (10) and (11) in Eq. (25), we get

κG12D=8πkB3Tτch2[3f3(A)f2(A)−2f2(A)f1(A)]f2(A)T2,(28)\kappa_{G1}^{2D}=\frac{8\pi k_{B}^{3}T\tau_{c}}{h^{2}}\left[3\frac{f_{3}(A)}{f_{2}(A)}-2\frac{f_{2}(A)}{f_{1}(A)}\right]f_{2}(A)T^{2}, \tag*{(28)}

and

κG22D=8πkB3Tτch2[3f3(A)f2(A)−ϵFkBT]f2(A)T2,(29)\kappa_{G2}^{2D}=\frac{8\pi k_{B}^{3}T\tau_{c}}{h^{2}}\left[3\frac{f_{3}(A)}{f_{2}(A)}-\frac{\epsilon_{F}}{k_{B}T}\right]f_{2}(A)T^{2}, \tag*{(29)}

and hence, corresponding Lorenz ratios will be

LG12DL0=12π2[3f3(A)f2(A)−2f2(A)f1(A)]f2(A)f1(A),(30)\frac{L_{G1}^{2D}}{L_{0}}=\frac{12}{\pi^{2}}\left[3\frac{f_{3}(A)}{f_{2}(A)}-2\frac{f_{2}(A)}{f_{1}(A)}\right]\frac{f_{2}(A)}{f_{1}(A)}, \tag*{(30)}

and

LG22DL0=12π2[3f3(A)f2(A)−ϵFkBT]f2(A)f1(A)(31)\frac{L_{G2}^{2D}}{L_{0}}=\frac{12}{\pi^{2}}\left[3\frac{f_{3}(A)}{f_{2}(A)}-\frac{\epsilon_{F}}{k_{B}T}\right]\frac{f_{2}(A)}{f_{1}(A)} \tag*{(31)}

respectively.

Results

In the above section, we have calculated all the general expressions of electrical and thermal conductivity and their ratios to check the validity of the Wiedemann-Franz law fore 2D graphene (G) system. Using those final expressions, the present section is intended to go with their numerical estimations. During the result generation, we used the natural unit kB=ℏ=c=1k_{B}=\hbar=c=1 for our convenience. Dimensionless quantities Lorenz ratio (L/L0)(L/L_{0}) and ϵF/T\epsilon_{F}/T will be taken as the y and x axes of all graphs.

In Figure 1 and Figure 2, we explored the Lorenz ratio (L/L0)(L/L_{0}) by using non-fluid (NF) type expressions. Figure 1(a) and (b) have shown the Lorenz ratio versus ϵF/T\epsilon_{F}/T for 3D-G and 2D-G systems respectively. By using the non-fluid (NF) type of (94) and its Sommerfeld limit (98), we plotted the black solid and dotted lines in Figure 1(a). We used the specific heat [CV]e1=∂∂T(UN)[C_{V}]_{e1}=\frac{\partial}{\partial T}\left(\frac{U}{N}\right), which is traditionally used in solid state or non-fluid systems. We are getting L/L0=1L/L_{0}=1 in the limit of ϵF/T≫1\epsilon_{F}/T\gg1 or Fermi Liquid (FL) regime of graphene case, which is as expected for most of the metal cases. The deviation of the Lorenz ratio from 1 has been started below ϵF/T≈10\epsilon_{F}/T\approx10 as we notice from the black solid line. We have also plotted the red solid line by using (95), where another definition of specific heat [CV]e2[C_{V}]_{e2} is used in the thermal conductivity expression. In the case of SL of 3D-G (the red dotted line), the Lorenz ratio is three instead of one. For the general case (the red solid line), the Lorenz ratio deviates from this constant line 3 below ϵF/T≈10\epsilon_{F}/T\approx10. For the 2D-G system, similar kinds of black and red solid lines are plotted in Figure 1(b) by using (12) and (13). Their respective SLs by using (14) and (15), are plotted as black and red horizontal dotted lines. We considered the surface area (SS) in place of volume (VV) in the 2D system. The Lorenz ratio when we take [CV]e2[C_{V}]_{e2}, the ratio becomes 2 (the red-dotted line).

Lorenz ratio regarding the chemical potential for non-fluid descriptions of graphene in 3D and 2D

Figure 1. The Lorenz ratio regarding the chemical potential for non-fluid descriptions of graphene case (a) for 3D and (b) for 2D expressions

Lorenz ratio regarding the chemical potential for non-fluid descriptions of nonrelativistic systems in 3D and 2D

Figure 2. The Lorenz ratio regarding the chemical potential for non-fluid descriptions of nonrelativistic case (a) for 3D and (b) for 2D expressions

The Fig. 2(a) and (b), similar to Fig. 1, display the Lorenz ratio in terms of ϵF/T\epsilon_{F}/T for non-fluid expressions of NR case. Though the real metals have a constant chemical potential within the range, ϵF=2\epsilon_{F}=2–1010 eV, we imagine that it can be tunable from 0 to ∞\infty. In Fig. 2(a), we notice that the WF law is valid in the SL of the 3D metal case, expressed by Eq. (D8) (green dot-dashed line). We have taken T=60∘T=60^\circ K (≈0.005\approx0.005 eV), so metal range ϵF=2\epsilon_{F}=2–1010 eV will give the ratio ϵF/T=400\epsilon_{F}/T=400–20002000 (≫1\gg1), which is quite good domain for SL. Therefore, being situated in the SL domain, metal always follows the WF law. A slight larger value L=1.5L0L=1.5L_{0} comes from Eq. (D12) (blue dotted line), when specific heat [CV]e2[C_{V}]_{e2} is taken. A similar kind of behavior can be seen in Fig. 2(b) for 2D-NR systems, but interestingly, SLs in both cases, given by Eqs. (D29) and (D30) are coincided at L=L0L=L_{0} (blue-dotted and green dot-dashed lines). It probably indicates that if one builds 2D metal systems, then it will also follow the WF law as their Fermi energy range is again expected to be situated in the SL domain.

Before going to the next graphical results, let us briefly analyze what we have learned from Figure 1(a), (b) and Figure 2(a), (b). Grossly, the non-fluid type expressions of Lorenz ratios for all cases - 3D-NR, 2D-NR, 3D-G, and 2D-G are saturated near 1 in the SL range, which means that the WF law is obeyed in the SL range. Among all four cases, 3D-NR and 2D-G are realistic cases. The former is applicable to metal systems, whose Fermi energies remain constant and within the SL range, so they always follow the WF law. Latter case - the non-fluid expressions of Lorenz ratio for 2D-G can be applicable for realistic graphene system with high doping concentration (ϵF\epsilon_F is quite large or ϵF/T≫1\epsilon_F/T \gg1), where the WF law is obeyed well again [19]. However, by decreasing the doping concentration, when the graphene system approaches the charge-neutral points (ϵF\epsilon_F is quite small or ϵF/T≪1\epsilon_F/T \ll1), the WF law violation domain is observed [19]. According to our non-fluid expressions, the WF law violation is also expected in the ϵF/T≪1\epsilon_F/T \ll1 domain but the outcome L/L0<1L/L_0 < 1 is not matching with experimental outcome L/L0>1L/L_0 > 1 by Crossno et al. [19]. A possible explanation of L/L0>1L/L_0 > 1 in the appearance of electron hydrodynamics [26] property in the domain ϵF/T<1\epsilon_F/T < 1, which is popularly called the Dirac fluid domain. Present work is firstly intended to highlight the overlooked possibility of Wiedemann-Franz law violation of any systems like NR, G or UR even we consider the non-fluid expression of κ\kappa and σ\sigma. Then, its next attempt is to show the transition from non-fluid (NF) to fluid (F) pictures in a systematic way. An approach towards the fluid description is addressed in formalism part Sec: (II B). Let us come to the numerical estimation for fluid-type formalism.

For numerical curves of fluid-type expressions, let us focus only on the 2D-G case. Although formalism of other cases with an Appendix is provided in the present article, one may go for result generation for graphene case. Using Eqs. (30) and (31), black and red solid lines are plotted in the left panel of Figure 3. They are saturated at the values of 2 and 4, respectively, in the high doping or ϵF/T≫1\epsilon_F/T \gg1 domain. It means that if the fluid type expression is valid in this high doping domain 2D-G system, then the WF law may not be obeyed. However, the experimental data [19] claims that the WF law is well obeyed in the high doping or ϵF/T≫1\epsilon_F/T \gg1 domain. This means that the non-fluid expression of the Lorenz ratio should be applicable to this domain instead of fluid-type expression. The SL of Eqs. (30) and (31) will also be

Lorenz ratio and comparison of specific heat and fluid aspect in 2D-G system

Figure 3. Lorenz ratio of 2D-G system by using fluid-type expression versus ϵF/T\epsilon_F/T (LEFT) and comparison of specific heat and fluid aspect in 2D-G system versus ϵF/T\epsilon_F/T (RIGHT)

LG12DL0=2+2π2kB2T23ϵF2,(32)\frac{L_{G1}^{2D}}{L_0}=2+\frac{2\pi^2 k_{\mathrm{B}}^2T^2}{3\epsilon_F^2}, \tag*{(32)}
LG22DL0=4+4π2kB2T23ϵF2,(33)\frac{L_{G2}^{2D}}{L_0}=4+\frac{4\pi^2 k_{\mathrm{B}}^2T^2}{3\epsilon_F^2}, \tag*{(33)}

which are plotted by black and red dotted lines, respectively, in the left panel of Figure 3. They blow up from their saturated values (2 and 4) when ϵF/T\epsilon_F/T or doping concentration decreases.

The experimental data of Crossno et al. [19] is pasted, where S1 and S2 represent the samples of the 2D-G system in terms of lower to higher values of minimum charge carrier density. Theoretically, zero charge carrier density can be reached at ϵF/T→0\epsilon_F/T \to0 but experimentally, a minimum charge carrier density [19] will be achieved. So, from Figure 3(a), we notice that our fluid-based expression of LG22DL0\frac{L_{G2}^{2D}}{L_0}, given in Eq.(30), (31), can’t be able to explain the experimental data points. It indicates that our formalism needs next level modification. Due to this step by step development of our formalism from (standard) non-fluid to (new directional) fluid description, we put the name of Sec: (II B) as ”towards the fluid description”. Now if we revisit our RTA based kinetic theory formalism, then we can find that we have systematically considered the local thermalization concept of fluid. In this concept, we consider fluid as a large number fluid elements, having velocity uμu^\mu, temperature TT and chemical potential ϵF\epsilon_F. Now, fluid elements will be at different position and time (xμx^\mu), so, considering uμ(xμ)u^\mu(x^\mu), T(xμ)T(x^\mu), ϵF(xμ)\epsilon_F(x^\mu) as function of xμx^\mu can map mathematically the local thermalization concept. Since present work is only focus on thermal and electrical conductivity, so only T(xμ)T(x^\mu) is considered and others uμ(xμ)u^\mu(x^\mu) and ϵF(xμ)\epsilon_F(x^\mu) are ignored. Reader can see this consideration in Eq. (20), where the information of temperature gradient ∂T∂xμ\frac{\partial T}{\partial x^\mu} is entered. Now the another term specific heat [CV]e≈∂u∂T[C_V]_e \approx\frac{\partial u}{\partial T}, which is basically considered as temperature derivative of average energy of electron. This average energy of electron information can be modified into enthalpy per electron h=u+PnTh = \frac{u+P}{nT} in fluid picture. Another point is that electron-hole plasma picture will be important when we go ϵFT<1\frac{\epsilon_F}{T}<1 domain (towards charge neutrality point). So, we should take enthalpy per net electron h=u+Pnh = \frac{u+P}{n}, where u=ue+uhu=u_e+u_h, P=Pe+PhP=P_e+P_h and n=ne−nhn=n_e-n_h will be respectively total internal energy density, pressure and net charge density. We have adopted the replacement

[CV]e≈∂u∂T→hT=u+PnT(34)[C_V]_e \approx\frac{\partial u}{\partial T} \to\frac{h}{T}=\frac{u+P}{nT} \tag*{(34)}

and using it in Eq. (25), we will get a better fluid-type expression of thermal conductivity:

κG2D=4πkB2Tch2(u+PnT)f2(A)T2(35)\kappa_G^{2D}=\frac{4\pi k_B^2 T_c}{h^2}\left(\frac{u+P}{nT}\right)f_2(A)T^2 \tag*{(35)}

and Lorenz ratio:

(LG2DL0)F/NF=6π2[CV]F/NFf2(A)f1(A).(36)\left(\frac{L_G^{2D}}{L_0}\right)_{F/NF}=\frac{6}{\pi^2}[C_V]_{F/NF}\frac{f_2(A)}{f_1(A)}. \tag*{(36)}

For comparison between two quantities CVeC_V^e and hT=u+PnT\frac{h}{T}=\frac{u+P}{nT}, we have plotted them in the right panel of Figure 3(b). Next, in Figure 4(a), we have drawn dashed-line by using Eq. (36). Here, one can notice that the Lorenz ratio shows a divergence trend in ϵFT<1\frac{\epsilon_F}{T}<1 domain, which in favor of experimental data S1S_1 and S2S_2. We have also plotted non-fluid(NF) expression of Lorenz ratio, given in Eq. (12), which is also in favor of experimental data in ϵFT>1\frac{\epsilon_F}{T}>1 domain. Based on this data favoring theoretical curves, we have classified three basic regions: (1) Fluid (F) domain, (2) Non-Fluid (NF) domain and (3) Mixed or (F+NF) domain, which are discussed one by one (below).

Lorenz ratio of fluid and non-fluid and their mixture for 2D-G system

Figure 4. Lorenz ratio of fluid and non-fluid for 2D-G system versus ϵF/T\epsilon_F/T (LEFT) and Lorenz ratio of the mixture of fluid and non-fluid with fitting parameters for 2D-G system versus ϵF/T\epsilon_F/T (RIGHT)

  • Fluid (F) domain: Fluid based Lorenz ratio, given in Eq. (36) valid well with experimental data in the range 0.2<ϵFT<0.70.2<\frac{\epsilon_F}{T}<0.7.

  • Non-Fluid (NF) domain: Non-Fluid based Lorenz ratio, given in Eq. (12), valid well with experimental data in the ranges (a) 2.5<ϵFT<102.5<\frac{\epsilon_F}{T}<10 and (b) 10<ϵFT<∞10<\frac{\epsilon_F}{T}<\infty. We can called subrange (b) as ohmic non-fluid or NF(Ohmic) domain, since we get WF law or LL0=1\frac{L}{L_0}=1 for this ϵFT>10\frac{\epsilon_F}{T}>10 domain, where Fermi-Dirac distribution function can be assumed as step function. However, for subrange (a), we get LL0<1\frac{L}{L_0}<1 because assumption of step function is not further valid here. Corresponding domains are also marked in Figure 3(b), where we can notice that CVe∝TϵFC_V^e\propto\frac{T}{\epsilon_F} follows in NF(Ohmic) domain (due to validation of electron step function like distribution) but a transition from CVe∝TϵFC_V^e\propto\frac{T}{\epsilon_F} to CVe=2kB[f3(A)f2(A)−2f2(A)f1(A)]C_V^e=2k_B\left[\frac{f_3(A)}{f_2(A)}-\frac{2f_2(A)}{f_1(A)}\right] is observed in NF domain (2.5<ϵFT<10)(2.5<\frac{\epsilon_F}{T}<10). Reader can understand this transition from Eq. (10).

  • Mixed or (F+NF) domain: The domain 0.7<ϵFT<2.50.7<\frac{\epsilon_F}{T}<2.5 is marked as mixed or (F+NF) domain as we have to go with replacement CVe→u+PnTC_V^e\to\frac{u+P}{nT} or transit from NF based quantity CVeC_V^e to fluid-based quantity u+PnT\frac{u+P}{nT}. This domain may be considered as a grey zone and we may put an open question for further research- How this transition happens? Interestingly, reader can notice a ϵFT\frac{\epsilon_F}{T} independent trend approximately of CVeC_V^e and u+PnT\frac{u+P}{nT} within this (F+NF) domain.

If we analyze the F domain of Figure 3(b) and Figure 4(a), then we can identify that the divergent trend of u+PnT\frac{u+P}{nT} and LL0\frac{L}{L_0} are inter-related. Net density nn trends to zero when ϵFT\frac{\epsilon_F}{T} trends to zero and that is why LL0∝1n\frac{L}{L_0}\propto\frac{1}{n} trends to infinite. So, we may connect the divergent trend of Lorenz ratio in Dirac fluid domain (ϵFT<1)(\frac{\epsilon_F}{T}<1) with the fluid properties of electron-hole plasma in graphene, where fluid quantity enthalpy per particle will be the main resource for that divergence. To make a bridge between the F and NF domain, we have used two switching functions.

aF(x)=1ebF(x−xF)+1,(37)a_F(x)=\frac{1}{e^{b_F(x-x_F)}+1}, \tag*{(37)}
aNF(x)=1ebNF(x−xNF)+1.(38)a_{NF}(x)=\frac{1}{e^{b_{NF}(x-x_{NF})}+1}. \tag*{(38)}

where x=ϵFTx=\frac{\epsilon_F}{T} and bFb_F, bNFb_{NF}, xFx_F, xNFx_{NF} are tuning parameters. Using these switching functions, we have prescribed a sandwich expression of Lorenz ratio as

LL0=(LL0)F×aF+(LL0)NF×aNF,(39)\frac{L}{L_0}=\left(\frac{L}{L_0}\right)_F\times a_F+\left(\frac{L}{L_0}\right)_{NF}\times a_{NF}, \tag*{(39)}

where NF and F Lorenz ratio can be written from Eq. (30) and (36) in a compact form.

(LL0)F/NFG=6π2[CV]F/NFf2(A)f1(A)(A=ex),(40)\left(\frac{L}{L_0}\right)^G_{F/NF}=\frac{6}{\pi^2}[C_V]_{F/NF}\frac{f_2(A)}{f_1(A)}\left(A=e^x\right), \tag*{(40)}
(LL0)F/NFQ=9π2[CV]F/NFf3(A)f2(A).(41)\left(\frac{L}{L_0}\right)^Q_{F/NF}=\frac{9}{\pi^2}[C_V]_{F/NF}\frac{f_3(A)}{f_2(A)}. \tag*{(41)}

with

[CVe]F=u+PnT,(42)\left[C_{V}^{e}\right]_{F}=\frac{u+P}{nT}, \tag*{(42)}
[CVe]NF=2kB[3f3(A)f2(A)−2f2(A)f1(A)].(43)\left[C_{V}^{e}\right]_{NF}=2k_{B}\left[3\frac{f_{3}(A)}{f_{2}(A)}-2\frac{f_{2}(A)}{f_{1}(A)}\right]. \tag*{(43)}

Using Eq. (41), we have plotted blue and red solid curves to fit experimental data point S1S_{1} (blue circles) and S2S_{2} (red circles). For blue line, tuning parameters are taken as xF=0.2x_{F}=0.2, xNF=0.8x_{NF}=0.8, bF=2b_{F}=2, bNF=0.7b_{NF}=0.7 and for red line, tuning parameters are taken as xF=0.9x_{F}=0.9, xNF=2x_{NF}=2, bF=2b_{F}=2, bNF=0.7b_{NF}=0.7. By using a better version of xx-dependent switching functions, one may go with better fitting of experimental data. However, instead of quantitative exact fitting, the core aim of the present work may be considered as its qualitative message or understanding on the F and NF domain and their transition.

This qualitative message of transition from NF to F is in agreement with earlier theoretical works by Lucas et al. [26]. Though final expressions of Lorenz ratio of Ref. [26] and the present work are little different but both works have similar qualitative message- NF to F transition is associated with the WF law violation and enhancement of Lorenz ratio more than one. There are other works also which provide little different reasons for getting WF law violations. Ref. [25] has linked this fact with presence of band gap and bipolar diffusion, while Ref. [24] has built thermodynamical expression of electron-hole plasma in terms of polylogarithm function, which can provide a greater than one values of Lorenz ratio. All of these theoretical results for LL0\frac{L}{L_{0}} along with present work are compiled in Figure 5 for a comparative visibility.

Lorenz ratio of 2D-G system versus number density for different theoretical works

Figure 5. Lorenz ratio of 2D-G system versus number density for different theoretical works by Rycerz [24], Tu et al. [25], Lucas et al. [26], including present work

We have done more comparison with experimental data and other earlier theory for better visibility of present work. In Figure 6 (a) and (Figure 6b), we have plotted experimental data points (circular) of electrical conductivity σ\sigma (in kΩ−1\mathrm{k}\Omega^{-1}) and thermal conductivity κ\kappa (in nW/K) from Ref. [19]. Here, we have added our results (solid line) by using 1 ps relaxation time [52], which show a non-fluid to fluid transition as we decrease the net electron density. We have also added two other theoretical results by Lucas et al. [26] (dash line) and Hartnoll et al. [53] (dotted line) which are quite close to experimental data. Here, σ\sigma of Hartnoll et al. [53] is built via eHD framework by adding charge neutrality conductivity aspect near the charge neutrality point. Therefore, their results are quite well matched at low nn. By extending the formalism of Ref. [53], Lucas et al. [26] additionally consider a parametrization due to local disorder and charge density and their theoretical curve of σ\sigma, κ\kappa are quite close to experimental data. In this context, present work is focused on non-fluid to fluid transition framework, so a qualitative matching can be found. However, our future plan is to extend this NF to F transition framework into a realistic quasi-particle modeling for better fitting of experimental data of Crossno et al. [19] as well as recent data by Majumdar et al. [54]. In future, our framework may be explored in other graphene like materials - germanene and silicene[55, 56] as well as in doped graphene at finite magnetic field [57].

Electrical and thermal conductivity versus number density

Figure 6. Our present theoretical results of electrical conductivity (LEFT) and thermal conductivity (RIGHT) with different theoretical predictions [26, 53] accompanied by experimental data [19] versus number density

Brief Interpretation of Results

In this section, we will quickly revisit our results and try to interpret them briefly in terms of the transformation from non-fluid to fluid description. Here, we have gone through comparative research on Lorenz ratios for 3D, 2D NR and G systems for a systematic search of the WF law violation domain. During our analysis, we found the WF law violation domain of Fermi energy (normalized by a fixed temperature), where the Lorenz ratio deviates from one. First, we have used standard solid state book-based non-fluid expressions of Lorenz ratio for the 3D-NR case, which are also applied for 2D-NR, 3D-G, and 2D-G cases. For all cases, the Lorenz ratio remains constant in the higher values of normalized Fermi energy. This fact is well agreement with the universal validity of the WF law in metal systems, which can be described by the 3D-NR case of the electron gas having larger Fermi energy (2–10 eV) or normalized Fermi energy 400–2000 at 60° K. Depending on the definition of specific heat at constant volume and Fermi energy, we will get two possible saturated values of Lorenz ratios - 1 and 2, where former is experimentally observed (average) value but one may mark the saturating domain as the WF law obeying domain. If one identifies two domains - less and greater than 10 value of normalized Fermi energy (i.e. ϵFT<10\frac{\epsilon_{F}}{T}<10 and ϵFT>10\frac{\epsilon_{F}}{T}>10), then the former domain of all cases show the deviation of WF law. This tuning of normalized Fermi energy is conveniently possible for 2D-G system only via tuning the electron doping. Recently measured Lorenz ratio by Crossno et al. [19] indicates the violation of the WF law in the lower normalized Fermi energy domain, which is popularly called as Dirac fluid (DF) domain. Using standard solid state or non-fluid theory for 2D-G system, we also find WF law violation in the lower normalized Fermi energy domain. However, non-fluid theory gets less than one values of Lorenz ratio, while experiment gets its greater than one values. It demands that we may have to transist towards a fluid-type from the standard non-fluid expressions of the Lorenz ratio. So far, to the best of our knowledge, this straightforward thermodynamical phase space analysis of Lorenz ratio, based on non-fluid theory is missing in the literature. Present work is fulfilling this gap, which may be considered one of the important contents of present article. A next level content of present article is its fluid based calculation,whose outcomes are summarized in the next paragraph.

Towards the fluid description, we have approached in two steps. In the first step, we have used Boltzmann transport equation for 2D-G system, where local thermalization concept is mainly implimented for fluid description. If we compare the NF and F expressions of Lorenz ratios, then we can find that only via specific heat, Fermi integral functions are entered into the NF expression of Lorenz ratio. On the other hand, F expressions of electric and thermal conductivity and hence Lorenz ratio carry additional Fermi integral functions, which are completely appeared because of local thermalization aspect of fluid description. During this first step attempt of fluid description, we get a little deviated outcomes of Lorenz ratio with respect to NF results. Sommerfeld limit of F-based Lorenz ratios give a divergent outcomes instead of constant values (it was found in its NF-based expression). This divergent trend of Lorenz ratio in lower normalized Fermi energy is experimentally observed, so this first step fluid description results may be considered as a preliminary hints of NF to F transition in the Dirac fluid domain or lower Fermi energy domain.

Observing the quantitative mismatch between experimental data and first step fluid theory, we have proceeded to second step extension of fluid theory. Here, we have gone through the replacement of specific heat by enthalpy per (divided by temperature). This replacement is based on the intuition for transition from quasi-particle concept to fluid concept. In NF picture, electron average energy is the quasi-particle concept and its temperature derivative gives specific heat concept. Whereas, enthalpy per net electron is the fluid concept for electron-hole plasma in Dirac fluid domain. Interestingly, we notice that this conceptual transformation and replacement provide an enhancement of Lorenz ratio in Dirac fluid domain and their theoretical values are quite close to experimental data points.

Conclusion

Present work has explored the WF law violation of electrons in graphene in terms of a non-fluid to fluid transition framework. After developing a step by step NF to F frameworks, we have identified four regions along the axis of normalized Fermi energy ϵFT\frac{\epsilon_F}{T} at T=60∘T = 60^\circ K:

  • ϵFT>10\frac{\epsilon_F}{T} > 10: Lorenz ratio is one and it can be called hardcore NF Ohmic domain. Sommerfeld limit of our NF expression can explain it easily.

  • Within 2.5<ϵFT<102.5 < \frac{\epsilon_F}{T} < 10: Lorenz raio become less than one and it can be called NF but non-ohmic type domain. Our NF expression of Lorenz ratio and experimental data both hint this domain.

  • 0.7<ϵFT<2.50.7 < \frac{\epsilon_F}{T} < 2.5: It is the transition domain from NF to F by showing less than one to greater than one values of Lorenz ratio.

  • ϵFT<0.7\frac{\epsilon_F}{T}<0.7: Here, Lorenz ratio becomes greater than one and it can be called Fluid domain as enhancing trend in fluid theory and experimental data towards lower doping range is noticed.

At the end, for final quantitative data matching, we have demonstrated a sandwich estimation of F and NF results. However, present article intend to draw the attention of reader on the qualitative message of NF to F transition. This association of enhancement Lorenz ratio observation in the Dirac fluid domain with the NF to F transition is agreement with earlier theoretical works.

Acknowledgments

This work was partly (T.Z.W. and C.W.A.) supported by the Doctoral Fellowship in India (DIA) program of the Ministry of Education, Government of India. The authors thank the other members of eHD club Sesha P. Vempati, Ashutosh Dwibedi, Narayan Prasad, Bharat Kukkar, and Subhalaxmi Nayak.

FERMI-DIRAC DISTRIBUTION FUNCTION

The Fermi-Dirac distribution function is

f0(ϵ)=1eβ(ϵ−ϵF)+1=1A−1eβϵ+1,(44)f_0(\epsilon)=\frac{1}{e^{\beta(\epsilon-\epsilon_F)}+1}=\frac{1}{A^{-1}e^{\beta\epsilon}+1}, \tag*{(44)}

where ϵ\epsilon = energy of fermions and ϵF\epsilon_F = fermi energy, and AA is the fugacity of the system described by

A=exp⁡(ϵFkBT).A=\exp\left(\frac{\epsilon_F}{k_B T}\right).

And the derivative of the distribution function with respect to energy is

−∂f0∂ϵ=βeβ(ϵ−ϵF)(eβ(ϵ−ϵF)+1)2=∂∂ϵF(1eβ(ϵ−ϵF)+1).(45)-\frac{\partial f_0}{\partial\epsilon}=\frac{\beta e^{\beta(\epsilon-\epsilon_F)}}{\left(e^{\beta(\epsilon-\epsilon_F)}+1\right)^2}=\frac{\partial}{\partial\epsilon_F}\left(\frac{1}{e^{\beta(\epsilon-\epsilon_F)}+1}\right). \tag*{(45)}

FERMI-DIRAC FUNCTION

  • The Fermi-Dirac Function

fν(A)=1Γ(ν)∫0∞xν−1A−1ex+1 dx,(46)f_\nu(A)=\frac{1}{\Gamma(\nu)}\int_0^\infty\frac{x^{\nu-1}}{A^{-1}e^x+1}\,dx, \tag*{(46)}

where fν(A)f_\nu(A) is known as the Fermi-Dirac function.

Case.1. When AA is small, then the Fermi-Dirac function can be written in a series form which is

fν(A)=A−A22ν+A33ν−A44ν+⋯ ,(47)f_\nu(A)=A-\frac{A^2}{2^\nu}+\frac{A^3}{3^\nu}-\frac{A^4}{4^\nu}+\cdots, \tag*{(47)}
=∑n=1∞(−1)n−1Annν.(48)=\sum_{n=1}^{\infty}(-1)^{n-1}\frac{A^n}{n^\nu}. \tag*{(48)}

Case.2. When AA is too much small (ϵF≪kBT\epsilon_F\ll k_B T), then the function becomes simplified as AA

fν(A)=A,(49)f_\nu(A)=A, \tag*{(49)}

but if we take ϵF=0\epsilon_F=0, the function becomes unity.

Case.3. When the temperature is very small, and the Fermi energy has some finite value, then the Fermi-Dirac function can be written according to Sommerfeld lemma, which gives the expression of the function as

fν(A)=ανΓ(ν+1)[1+ν(ν−1)π261α2+ν(ν−1)(ν−2)(ν−3)7π43601α4+⋯ ],(50)f_\nu(A)=\frac{\alpha^\nu}{\Gamma(\nu+1)}\left[1+\nu(\nu-1)\frac{\pi^2}{6}\frac{1}{\alpha^2}+\nu(\nu-1)(\nu-2)(\nu-3)\frac{7\pi^4}{360}\frac{1}{\alpha^4}+\cdots\right], \tag*{(50)}

Case.4. When (ϵF≫kBT\epsilon_F\gg k_B T), then the function can be written only using the zeroth order term of Sommerfeld lemma expression.

The derivative of Fermi function with respect to Fermi energy

∂fν(A)∂ϵF=βfν−1(A).(51)\frac{\partial f_\nu(A)}{\partial\epsilon_F}=\beta f_{\nu-1}(A). \tag*{(51)}

The derivative of Fermi function with respect to temperature

∂fν(A)∂T=1Afν−1(A)∂A∂T.(52)\frac{\partial f_{\nu}(A)}{\partial T}=\frac{1}{A}f_{\nu-1}(A)\frac{\partial A}{\partial T}. \tag*{(52)}

The one identity for the function can be written as

1A∂A∂T=−ϵFβ2kB=−ϵFkBT2.(53)\frac{1}{A}\frac{\partial A}{\partial T}=-\epsilon_F\beta^2 k_B=-\frac{\epsilon_F}{k_B T^2}. \tag*{(53)}

ELECTRONIC SPECIFIC HEAT

For 2D system graphene, which follows the linear dispersion relation, with the help of the density of state method, we can calculate the number density, total internal energy, and the electronic specific heat.

So, the number of energy states in energy range ϵ\epsilon to ϵ+dϵ\epsilon+d\epsilon and the surface area SS is written as

D(ϵ)dϵ=gS(2πℏ)22πvF2ϵdϵ.(54)D(\epsilon)d\epsilon=g\frac{S}{(2\pi\hbar)^2}\frac{2\pi}{v_F^2}\epsilon d\epsilon. \tag*{(54)}

Now, the total number of particles at any value of temperature can be calculated as

N=∫0∞D(ϵ)dϵf0(ϵ).(55)N=\int_{0}^{\infty}D(\epsilon)d\epsilon f_0(\epsilon). \tag*{(55)}

After plugging the value of D(ϵ)dϵD(\epsilon)d\epsilon in the above equation, we get

n=NS=g2πℏ2vF2∫0∞ϵA−1eβϵ+1dϵ.(56)n=\frac{N}{S}=\frac{g}{2\pi\hbar^2v_F^2}\int_{0}^{\infty}\frac{\epsilon}{A^{-1}e^{\beta\epsilon}+1}d\epsilon. \tag*{(56)}

After solving this integration, we get the final, more general expression of number density, which is given by

N=g2πSh2vF2Γ(2)β2f2(A)(57)N=g\frac{2\pi S}{h^2v_F^2}\frac{\Gamma(2)}{\beta^2}f_2(A) \tag*{(57)}
⟹n=gkB2πℏ2vF2f2(A)T2.(58)\Longrightarrow n=g\frac{k_B^2}{\pi\hbar^2v_F^2}f_2(A)T^2. \tag*{(58)}

The total internal energy of a system can be calculated as

U=∫0∞D(ϵ)dϵf0(ϵ)ϵ.(59)U=\int_{0}^{\infty}D(\epsilon)d\epsilon f_0(\epsilon)\epsilon. \tag*{(59)}

By substituting the value of D(ϵ)dϵD(\epsilon)d\epsilon in the above equation and solving the integration, we get the final, more general expression of total internal energy, which is given by

U=g2πSh2vF2Γ(3)β3f3(A)(60)U=g\frac{2\pi S}{h^2v_F^2}\frac{\Gamma(3)}{\beta^3}f_3(A) \tag*{(60)}
⟹u=US=g2πh2vF2Γ(3)β3f3(A).(61)\Longrightarrow u=\frac{U}{S}=g\frac{2\pi}{h^2v_F^2}\frac{\Gamma(3)}{\beta^3}f_3(A). \tag*{(61)}

Now, from the above equations, we get

un=UN=2kBTf3(A)f2(A).(62)\frac{u}{n}=\frac{U}{N}=2k_BT\frac{f_3(A)}{f_2(A)}. \tag*{(62)}

Let us define the specific heat capacity of the electron by taking the temperature derivative of internal energy per electron (U/N=u/n)(U/N=u/n) while keeping surface area SS and ϵF\epsilon_F as constants:

G12D[CV]e1=[∂∂T(UN)]S,ϵF.(63){}_{G_1}^{2D}[C_V]_{e1}=\left[\frac{\partial}{\partial T}\left(\frac{U}{N}\right)\right]_{S,\epsilon_F}. \tag*{(63)}

Using the (62), the specific heat is

G12D[CV]e1=2kB[f3(A)f2(A)+T∂∂T(f3(A)f2(A))].(64){}_{G_1}^{2D}[C_V]_{e1}=2k_B\left[\frac{f_3(A)}{f_2(A)}+T\frac{\partial}{\partial T}\left(\frac{f_3(A)}{f_2(A)}\right)\right]. \tag*{(64)}

Now, the derivative part after using the identity (B7) of the Fermi-Dirac function can be written as

∂∂T(f3(A)f2(A))=1f2(A)∂∂T(f3(A))−f3(A)∂∂T(1f2(A))=1Af2(A)f2(A)∂A∂T−f3(A)[f2(A)]2∂∂Tf2(A)=1A∂A∂T−f3(A)f2(A)f1(A)f2(A)1A∂A∂T.\begin{aligned} \frac{\partial}{\partial T}\left(\frac{f_3(A)}{f_2(A)}\right)=\frac{1}{f_2(A)}\frac{\partial}{\partial T}\left(f_3(A)\right)-f_3(A)\frac{\partial}{\partial T}\left(\frac{1}{f_2(A)}\right) \\ &=\frac{1}{A}\frac{f_2(A)}{f_2(A)}\frac{\partial A}{\partial T}-\frac{f_3(A)}{\left[f_2(A)\right]^2}\frac{\partial}{\partial T}f_2(A) \\ &=\frac{1}{A}\frac{\partial A}{\partial T}-\frac{f_3(A)}{f_2(A)}\frac{f_1(A)}{f_2(A)}\frac{1}{A}\frac{\partial A}{\partial T}. \end{aligned}

The final form, we get

∂∂T(f3(A)f2(A))=2T[f3(A)f2(A)−f2(A)f1(A)].(65)\frac{\partial}{\partial T}\left(\frac{f_3(A)}{f_2(A)}\right)=\frac{2}{T}\left[\frac{f_3(A)}{f_2(A)}-\frac{f_2(A)}{f_1(A)}\right]. \tag*{(65)}

After substituting the value from equation (C12) into equation (C11), we get a more general form of electronic specific heat given by

G12D[CV]e1=2kB[f3(A)f2(A)−2f2(A)f1(A)].(66){}_{G_1}^{2D}[C_V]_{e1}=2k_B\left[\frac{f_3(A)}{f_2(A)}-2\frac{f_2(A)}{f_1(A)}\right]. \tag*{(66)}

Let us go for another definition of specific heat per electron

G22D[CV]e2=1N∂U∂T∣S,ϵF.(67){}_{G_2}^{2D}[C_V]_{e2}=\left.\frac{1}{N}\frac{\partial U}{\partial T}\right|_{S,\epsilon_F}. \tag*{(67)}

From (60), the total internal energy is

G2DU=γT3f3(A),{}_{G}^{2D}U=\gamma T^3f_3(A),

where

γ=g2πSh2vF2Γ(3)kB3.\gamma=g\frac{2\pi S}{h^2v_F^2}\Gamma(3)k_B^3.

Then, we will take the derivative of just the above equation to TT at constant Fermi energy ϵF\epsilon_F and surface area SS; we get

∂U∂T∣S,ϵF=γ[3T2f3(A)+T3f2(A)1A∂A∂T](68)\left.\frac{\partial U}{\partial T}\right|_{S,\epsilon_F}=\gamma\left[3T^2f_3(A)+T^3f_2(A)\frac{1}{A}\frac{\partial A}{\partial T}\right] \tag*{(68)}
=γT2f2(A)[3f3(A)f2(A)+T1A∂A∂T].(69)=\gamma T^2f_2(A)\left[3\frac{f_3(A)}{f_2(A)}+T\frac{1}{A}\frac{\partial A}{\partial T}\right]. \tag*{(69)}

So,

2DG2[CV]e2=2kB[3f3(A)f2(A)−ϵFkBT].(70){}^{2D}G_2[C_V]_{e2}=2k_B\left[3\frac{f_3(A)}{f_2(A)}-\frac{\epsilon_F}{k_BT}\right]. \tag*{(70)}

By doing a similar type of calculation for 2D-NR case, we will get two different specific heat expressions as

2DNR1[CV]e1=kB[2f2(A)f1(A)−f1(A)f0(A)],(71){}^{2D}NR_1[C_V]_{e1}=k_B\left[2\frac{f_2(A)}{f_1(A)}-\frac{f_1(A)}{f_0(A)}\right], \tag*{(71)}

and

2DNR2[CV]e2=kB[2f2(A)f1(A)−ϵFkBT].(72){}^{2D}NR_2[C_V]_{e2}=k_B\left[2\frac{f_2(A)}{f_1(A)}-\frac{\epsilon_F}{k_BT}\right]. \tag*{(72)}

Next, for 3D-NR and 3D-G cases also, one can repeat the calculations and find the expressions of specific heat:

3DNR1[CV]e1=32kB[5f52(A)2f32(A)−3f32(A)2f12(A)],(73){}^{3D}NR_1[C_V]_{e1}=\frac{3}{2}k_B\left[\frac{5f_{\frac{5}{2}}(A)}{2f_{\frac{3}{2}}(A)}-\frac{3f_{\frac{3}{2}}(A)}{2f_{\frac{1}{2}}(A)}\right], \tag*{(73)}
3DNR2[CV]e2=32kB[5f52(A)2f32(A)−ϵFkBT],(74){}^{3D}NR_2[C_V]_{e2}=\frac{3}{2}k_B\left[\frac{5f_{\frac{5}{2}}(A)}{2f_{\frac{3}{2}}(A)}-\frac{\epsilon_F}{k_BT}\right], \tag*{(74)}
3DG1[CV]e1=3kB[4f4(A)f3(A)−3f3(A)f2(A)],(75){}^{3D}G_1[C_V]_{e1}=3k_B\left[4\frac{f_4(A)}{f_3(A)}-3\frac{f_3(A)}{f_2(A)}\right], \tag*{(75)}
3DG2[CV]e2=3kB[4f4(A)f3(A)−ϵFkBT].(76){}^{3D}G_2[C_V]_{e2}=3k_B\left[4\frac{f_4(A)}{f_3(A)}-\frac{\epsilon_F}{k_BT}\right]. \tag*{(76)}

NON-FLUID DESCRIPTION of OTHER SYSTEMS

3D-NR: The well-known Drude’s formula for the electrical conductivity of nonrelativistic electrons in 3D solids (metals) is

σNR3D=ne2τcm=ne2λmvF,(77)\sigma^{3D}_{NR}=\frac{ne^2\tau_c}{m}=\frac{ne^2\lambda}{mv_F}, \tag*{(77)}

and the thermal conductivity is

κNR3D=13nvFλ NR3D[CV]e,(78)\kappa^{3D}_{NR}=\frac{1}{3}nv_F\lambda\,{}^{3D}_{NR}[C_V]_e, \tag*{(78)}

where, nn is the number (electron) density, τc\tau_c is the relaxation time, mm is electron mass, λ=vFτc\lambda=v_F\tau_c is the mean free path, and [CV]e[C_V]_e is the electronic specific heat per particle. After taking the ratio of thermal and electrical conductivity, we get

κNR3DσNR3D=23ϵFe2 NR3D[CV]e,(79)\frac{\kappa^{3D}_{NR}}{\sigma^{3D}_{NR}}=\frac{2}{3}\frac{\epsilon_F}{e^2}\,{}^{3D}_{NR}[C_V]_e, \tag*{(79)}

where ϵF=μ=12mvF2\epsilon_F=\mu=\frac{1}{2}mv_F^2 is the Fermi energy. Here, in Eq. (79), [CV]e[C_V]_e is the main important thing to calculate the conductivity ratio. We will use the two definitions of [CV]e[C_V]_e:

  1. [CV]e1=∂∂T(UN)∣V,ϵF[C_V]_{e1}=\left.\frac{\partial}{\partial T}\left(\frac{U}{N}\right)\right|_{V,\epsilon_F},

  1. [CV]=∂∂T(UN)∣V,ϵF[C_V]=\left.\frac{\partial}{\partial T}\left(\frac{U}{N}\right)\right|_{V,\epsilon_F},

  1. [CV]e2=1N∂U∂T∣V,ϵF[C_V]_{e2}=\left.\frac{1}{N}\frac{\partial U}{\partial T}\right|_{V,\epsilon_F}.

Here, for subsequent simplicity, we will use the notations like NR1 and G1 concerning [CV]e1[C_V]_{e1} and NR2 and G2 dealing with [CV]e2[C_V]_{e2}. The former definition prescribes taking TT derivative of internal energy per particle u=U/Nu=U/N, while the latter definition says to take TT derivative of internal energy UU and then normalized by NN. The detailed calculation can be seen in Appendix C. Let us first put the former specific heat,

3DNR1[CV]e1=32kB[5f52(A)2f32(A)−3f32(A)2f12(A)],(80){}^{3D}NR_1[C_V]_{e1}=\frac{3}{2}k_B\left[\frac{5f_{\frac{5}{2}}(A)}{2f_{\frac{3}{2}}(A)}-\frac{3f_{\frac{3}{2}}(A)}{2f_{\frac{1}{2}}(A)}\right], \tag*{(80)}

in Eq. (D3)}, the Lorenz ratio for 3D-NR will be

LNR13DL0=3π2ϵFkBT[5f52(A)2f32(A)−3f32(A)2f12(A)].(81)\frac{L^{3D}_{NR_1}}{L_0}=\frac{3}{\pi^2}\frac{\epsilon_F}{k_BT}\left[\frac{5f_{\frac{5}{2}}(A)}{2f_{\frac{3}{2}}(A)}-\frac{3f_{\frac{3}{2}}(A)}{2f_{\frac{1}{2}}(A)}\right]. \tag*{(81)}

Here,

fν(A)=1Γ(ν)∫0∞xν−1A−1ex+1 dx,(82)f_\nu(A)=\frac{1}{\Gamma(\nu)}\int_0^\infty\frac{x^{\nu-1}}{A^{-1}e^x+1}\,dx, \tag*{(82)}

is the standard Fermi integral function (see details in Appendix B) with A=eϵF/kBTA=e^{\epsilon_F/k_BT}. Using Sommerfeld lemma, in the limit of ϵF/kBT=ln⁡A≫1\epsilon_F/k_BT=\ln A\gg1, the electronic specific heat becomes [1]

3DNR1[CV]e1=π22kB2TϵF,(83){}^{3D}NR_1[C_V]_{e1}=\frac{\pi^2}{2}\frac{k_B^2T}{\epsilon_F}, \tag*{(83)}

and then Eq. (D5)} becomes

LNR13D=π23(kBe)2=L0.(84)L^{3D}_{NR_1}=\frac{\pi^2}{3}\left(\frac{k_B}{e}\right)^2=L_0. \tag*{(84)}

which is the so-called Wiedemann-Franz law for metals. Next, using another definition of [CV]e2[C_V]_{e2} for the 3D-NR case, whose general expression will be (see Appendix C)

NR23D[CV]e2=32kB[52f5/2(A)f3/2(A)−ϵFkBT](85){}^{3D}_{NR2}[C_V]_{e2}=\frac{3}{2}k_B\left[\frac{5}{2}\frac{f_{5/2}(A)}{f_{3/2}(A)}-\frac{\epsilon_F}{k_BT}\right] \tag*{(85)}

we will get the Lorenz ratio as

LNR23DL0=3π2ϵFkBT[52f5/2(A)f3/2(A)−ϵFkBT].(86)\frac{L^{3D}_{NR2}}{L_0}=\frac{3}{\pi^2}\frac{\epsilon_F}{k_BT}\left[\frac{5}{2}\frac{f_{5/2}(A)}{f_{3/2}(A)}-\frac{\epsilon_F}{k_BT}\right]. \tag*{(86)}

In the limit of ϵF/T=ln⁡A≫1\epsilon_F/T=\ln A\gg1, Sommerfeld expansion of (85) becomes

NR23D[CV]e2=3π24kB2TϵF,(87){}^{3D}_{NR2}[C_V]_{e2}=\frac{3\pi^2}{4}\frac{k_B^2T}{\epsilon_F}, \tag*{(87)}

and then the (86) becomes

LNR23D=π22(kBe)2=1.5L0.(88)L^{3D}_{NR2}=\frac{\pi^2}{2}\left(\frac{k_B}{e}\right)^2=1.5L_0. \tag*{(88)}

Reader can see that Eq. (D8) and (D12), both are independent of TT, ϵF\epsilon_F and are made by universal constants like kBk_B and ee. So, the domain of T−ϵFT-\epsilon_F plane, where L/L0L/L_0 or LL becomes independent of TT, ϵF\epsilon_F, can be considered as Wiedemann-Franz law obeying domain and remaining will be Wiedemann-Franz law violating domain. Although, in terms of L0L_0, LL is 1.5 factor times larger for Eq. (D12).

3D-G: Now, we consider a hypothetical 3D system, which follows graphene-type dispersion relation ϵ=vFp\epsilon=v_Fp. In this case, the electrical conductivity

σG3D=ne2τcϵFvF2=ne2λϵFvF,(89)\sigma^{3D}_G=\frac{ne^2\tau_c}{\epsilon_F}v_F^2=\frac{ne^2\lambda}{\epsilon_F}v_F, \tag*{(89)}

and the thermal conductivity,

κG3D=13nvFλ G3D[CV]e,(90)\kappa^{3D}_G=\frac{1}{3}nv_F\lambda\,{}^{3D}_G[C_V]_e, \tag*{(90)}

will form the ratio

κG3DσG3D=13ϵFe2 G3D[CV]e,(91)\frac{\kappa^{3D}_G}{\sigma^{3D}_G}=\frac{1}{3}\frac{\epsilon_F}{e^2}\,{}^{3D}_G[C_V]_e, \tag*{(91)}

where the relation between the Fermi energy and the Fermi momentum will be ϵF=pFvF\epsilon_F=p_Fv_F for graphene. Using two possible expressions (see Appendix C) of specific heat for 3D-G system,

G13D[CV]e1=3kB[4f4(A)f3(A)−3f3(A)f2(A)](92){}^{3D}_{G1}[C_V]_{e1}=3k_B\left[4\frac{f_4(A)}{f_3(A)}-3\frac{f_3(A)}{f_2(A)}\right] \tag*{(92)}

and

G23D[CV]e2=3kB[4f4(A)f3(A)−ϵFkBT],(93){}^{3D}_{G2}[C_V]_{e2}=3k_B\left[4\frac{f_4(A)}{f_3(A)}-\frac{\epsilon_F}{k_BT}\right], \tag*{(93)}

in (91), we get the Lorenz ratios

LG13DL0=κG13DσG13DTL0=3π2ϵFkBT[4f4(A)f3(A)−3f3(A)f2(A)](94)\frac{L^{3D}_{G1}}{L_0}=\frac{\kappa^{3D}_{G1}}{\sigma^{3D}_{G1}TL_0}=\frac{3}{\pi^2}\frac{\epsilon_F}{k_BT}\left[4\frac{f_4(A)}{f_3(A)}-3\frac{f_3(A)}{f_2(A)}\right] \tag*{(94)}

and

LG23DL0=κG23DσG23DTL0=3π2ϵFkBT[4f4(A)f3(A)−ϵFkBT](95)\frac{L^{3D}_{G2}}{L_0}=\frac{\kappa^{3D}_{G2}}{\sigma^{3D}_{G2}TL_0}=\frac{3}{\pi^2}\frac{\epsilon_F}{k_BT}\left[4\frac{f_4(A)}{f_3(A)}-\frac{\epsilon_F}{k_BT}\right] \tag*{(95)}

respectively. In the Sommerfeld limit, Eqs. (D16) and (D17) will be converted to

G13D[CV]e1=π2kB2TϵF,(96){}^{3D}_{G1}[C_V]_{e1}=\pi^2\frac{k_B^2T}{\epsilon_F}, \tag*{(96)}

and

G23D[CV]e2=3π2kB2TϵF,(97){}^{3D}_{G2}[C_V]_{e2}=3\pi^2\frac{k_B^2T}{\epsilon_F}, \tag*{(97)}

and so, Eqs. (D18) and (D19) become

LG13D=π23(kBe)2=L0,(98)L^{3D}_{G1}=\frac{\pi^2}{3}\left(\frac{k_B}{e}\right)^2=L_0, \tag*{(98)}

and

LG23D=π2(kBe)2=3L0,(99)L^{3D}_{G2}=\pi^2\left(\frac{k_B}{e}\right)^2=3L_0, \tag*{(99)}

respectively.

2D-NR: Now, let us go for 2D cases. In the case of 2D nonrelativistic system, the ratio of thermal and electrical conductivity can be written as

κNR2DσNR2D=ϵFe2 NR2D[CV]e.(100)\frac{\kappa^{2D}_{NR}}{\sigma^{2D}_{NR}}=\frac{\epsilon_F}{e^2}\,{}^{2D}_{NR}[C_V]_e. \tag*{(100)}

Using two possible expressions (see Appendix C) of specific heat for 2D-NR system,

NR12D[CV]e1=kB[2f2(A)f1(A)−f1(A)f0(A)],(101){}^{2D}_{NR1}[C_V]_{e1}=k_B\left[2\frac{f_2(A)}{f_1(A)}-\frac{f_1(A)}{f_0(A)}\right], \tag*{(101)}

and

NR22D[CV]e2=kB[2f2(A)f1(A)−ϵFkBT],(102){}^{2D}_{NR2}[C_V]_{e2}=k_B\left[2\frac{f_2(A)}{f_1(A)}-\frac{\epsilon_F}{k_BT}\right], \tag*{(102)}

in (100), we get the Lorenz ratios

LNR12DL0=κNR12DσNR12DTL0=3π2ϵFkBT[2f2(A)f1(A)−f1(A)f0(A)](103)\frac{L^{2D}_{NR1}}{L_0}=\frac{\kappa^{2D}_{NR1}}{\sigma^{2D}_{NR1}TL_0}=\frac{3}{\pi^2}\frac{\epsilon_F}{k_BT}\left[2\frac{f_2(A)}{f_1(A)}-\frac{f_1(A)}{f_0(A)}\right] \tag*{(103)}

and

LNR22DL0=κNR22DσNR22DTL0=3π2ϵFkBT[2f2(A)f1(A)−ϵFkBT](104)\frac{L^{2D}_{NR2}}{L_0}=\frac{\kappa^{2D}_{NR2}}{\sigma^{2D}_{NR2}TL_0}=\frac{3}{\pi^2}\frac{\epsilon_F}{k_BT}\left[2\frac{f_2(A)}{f_1(A)}-\frac{\epsilon_F}{k_BT}\right] \tag*{(104)}

respectively. In the Sommerfeld limit (SL), Eqs. (103) and (104) will be converted to

LNR12D=π23(kBe)2=L0,(105)L^{2D}_{NR1}=\frac{\pi^2}{3}\left(\frac{k_B}{e}\right)^2=L_0, \tag*{(105)}
LNR22D=π23(kBe)2=L0.(106)L^{2D}_{NR2}=\frac{\pi^2}{3}\left(\frac{k_B}{e}\right)^2=L_0. \tag*{(106)}

3D-UR and 2D-UR: The expressions of electrical and thermal conductivity for 3D-UR and 2D-UR will be the same as those expressions for 3D-G and 2D-G if vFv_F is replaced by the speed of light cc. Readers can notice that specific heat and Lorenz ratios for 3D-G and 2D-G are independent of vFv_F, so those expressions can also be used for 3D-UR and 2D-UR systems. Since our focal quantity is the Lorenz ratio in the results, we will not discuss it further. The expressions of 3D-UR and 2D-UR systems coincide with those of 3D-G and 2D-G systems.

APPROACH TOWARDS THE FLUID DESCRIPTION OF OTHER SYSTEMS

2D-NR After doing the same calculation as the 2D G of towards fluid system for a 2-dimensional nonrelativistic system, the final general expressions of the electrical and two possible thermal conductivities and their corresponding Lorenz ratios LNR2D/L0L^{2D}_{NR}/L_0 are given by as below:

σNR2D=4πkBτc(eh)2f1(A)T,(107)\sigma^{2D}_{NR}=4\pi k_B\tau_c\left(\frac{e}{h}\right)^2 f_1(A)T, \tag*{(107)}
κNR12D=8πkB3Tτch2[2f2(A)f1(A)−f1(A)f0(A)]f2(A)T2,(108)\kappa^{2D}_{NR1}=\frac{8\pi k_B^3T\tau_c}{h^2}\left[\frac{2f_2(A)}{f_1(A)}-\frac{f_1(A)}{f_0(A)}\right]f_2(A)T^2, \tag*{(108)}
κNR22D=8πkB3Tτch2[2f2(A)f1(A)−ϵFkBT]f2(A)T2,(109)\kappa^{2D}_{NR2}=\frac{8\pi k_B^3T\tau_c}{h^2}\left[\frac{2f_2(A)}{f_1(A)}-\frac{\epsilon_F}{k_BT}\right]f_2(A)T^2, \tag*{(109)}
LNR12DL0=6π2[2f2(A)f1(A)−f1(A)f0(A)]f2(A)f1(A),(110)\frac{L^{2D}_{NR1}}{L_0}=\frac{6}{\pi^2}\left[2\frac{f_2(A)}{f_1(A)}-\frac{f_1(A)}{f_0(A)}\right]\frac{f_2(A)}{f_1(A)}, \tag*{(110)}
LNR22DL0=6π2[2f2(A)f1(A)−ϵFkBT]f2(A)f1(A).(111)\frac{L^{2D}_{NR2}}{L_0}=\frac{6}{\pi^2}\left[2\frac{f_2(A)}{f_1(A)}-\frac{\epsilon_F}{k_BT}\right]\frac{f_2(A)}{f_1(A)}. \tag*{(111)}

3D-G: For the 3D-G case, the expression of the electrical conductivity and two possible thermal conductivities and their corresponding Lorenz ratios LG3D/L0L^{3D}_{G}/L_0 are given as below:

σG3D=16πkB2τc(e23h3vF)f2(A)T2,(112)\sigma^{3D}_{G}=16\pi k_B^2\tau_c\left(\frac{e^2}{3h^3v_F}\right)f_2(A)T^2, \tag*{(112)}
κG13D=48πkB4τch3vF[4f4(A)f3(A)−3f3(A)f2(A)]f3(A)T3,(113)\kappa^{3D}_{G1}=\frac{48\pi k_B^4\tau_c}{h^3v_F}\left[4\frac{f_4(A)}{f_3(A)}-3\frac{f_3(A)}{f_2(A)}\right]f_3(A)T^3, \tag*{(113)}
κG23D=48πkB4τch3vF[4f4(A)f3(A)−ϵFkBT]f3(A)T3,(114)\kappa^{3D}_{G2}=\frac{48\pi k_B^4\tau_c}{h^3v_F}\left[4\frac{f_4(A)}{f_3(A)}-\frac{\epsilon_F}{k_BT}\right]f_3(A)T^3, \tag*{(114)}
LG13DL0=27π2[4f4(A)f3(A)−3f3(A)f2(A)]f3(A)f2(A),(115)\frac{L^{3D}_{G1}}{L_0}=\frac{27}{\pi^2}\left[4\frac{f_4(A)}{f_3(A)}-3\frac{f_3(A)}{f_2(A)}\right]\frac{f_3(A)}{f_2(A)}, \tag*{(115)}
LG23DL0=27π2[4f4(A)f3(A)−ϵFkBT]f3(A)f2(A).(116)\frac{L^{3D}_{G2}}{L_0}=\frac{27}{\pi^2}\left[4\frac{f_4(A)}{f_3(A)}-\frac{\epsilon_F}{k_BT}\right]\frac{f_3(A)}{f_2(A)}. \tag*{(116)}

3D-NR: Now again, the same type of calculation is done for a 3-dimensional nonrelativistic system; we get all expressions of the electrical, thermal conductivity, and the ratio LNR3D/L0L^{3D}_{NR}/L_0 in a more general form, which can be expressed as

σNR3D=2e2kB32τcm(2πmh2)32f32(A)T32,(117)\sigma^{3D}_{NR}=\frac{2e^2k_B^{\frac{3}{2}}\tau_c}{m}\left(\frac{2\pi m}{h^2}\right)^{\frac{3}{2}}f_{\frac{3}{2}}(A)T^{\frac{3}{2}}, \tag*{(117)}
κNR13D=15kB72Tτc2m(2πmh2)32[52f52(A)f32(A)−32f32(A)f12(A)]f52(A)T52,(118)\kappa^{3D}_{NR1}=\frac{15k_B^{\frac{7}{2}}T\tau_c}{2m}\left(\frac{2\pi m}{h^2}\right)^{\frac{3}{2}}\left[\frac{5}{2}\frac{f_{\frac{5}{2}}(A)}{f_{\frac{3}{2}}(A)}-\frac{3}{2}\frac{f_{\frac{3}{2}}(A)}{f_{\frac{1}{2}}(A)}\right]f_{\frac{5}{2}}(A)T^{\frac{5}{2}}, \tag*{(118)}
κNR23D=15kB72Tτc2m(2πmh2)32[52f52(A)f32(A)−ϵFkBT]f52(A)T52,(119)\kappa^{3D}_{NR2}=\frac{15k_B^{\frac{7}{2}}T\tau_c}{2m}\left(\frac{2\pi m}{h^2}\right)^{\frac{3}{2}}\left[\frac{5}{2}\frac{f_{\frac{5}{2}}(A)}{f_{\frac{3}{2}}(A)}-\frac{\epsilon_F}{k_BT}\right]f_{\frac{5}{2}}(A)T^{\frac{5}{2}}, \tag*{(119)}
LNR13DL0=454π2[52f52(A)f32(A)−32f32(A)f12(A)]f52(A)f32(A),(120)\frac{L^{3D}_{NR1}}{L_0}=\frac{45}{4\pi^2}\left[\frac{5}{2}\frac{f_{\frac{5}{2}}(A)}{f_{\frac{3}{2}}(A)}-\frac{3}{2}\frac{f_{\frac{3}{2}}(A)}{f_{\frac{1}{2}}(A)}\right]\frac{f_{\frac{5}{2}}(A)}{f_{\frac{3}{2}}(A)}, \tag*{(120)}
LNR23DL0=454π2[52f52(A)f32(A)−ϵFkBT]f52(A)f32(A).(121)\frac{L^{3D}_{NR2}}{L_0}=\frac{45}{4\pi^2}\left[\frac{5}{2}\frac{f_{\frac{5}{2}}(A)}{f_{\frac{3}{2}}(A)}-\frac{\epsilon_F}{k_BT}\right]\frac{f_{\frac{5}{2}}(A)}{f_{\frac{3}{2}}(A)}. \tag*{(121)}

References

  1. [1]Neil W. Ashcroft and N David Mermin. Solid State Physics. Cengage Learning, 2022.
  2. [2]V Devanathan. The Wiedemann-Franz law for electrical and thermal conduction in metals. J. Chennai Academy of Sciences, 4:1–26, 2021.
  3. [3]R. K. Pathria. Statistical Mechanics. Butterworth-Heinemann, 2nd ed. edition, 1996.
  4. [4]SO Pillai. Solid State Physics. New Age International, 2006.
  5. [5]RK Puri and VK Babbar. Solid State Physics. S Chand and Co Ltd, 1997.
  6. [6]Nicholas Wakeham, Alimamy F Bangura, Xiaofeng Xu, Jean-Francois Mercure, Martha Greenblatt, and Nigel E Hussey. Gross violation of the Wiedemann–Franz law in a quasi-one-dimensional conductor. Nat. Commun., 2(1):396, 2011.DOI
  7. [7]Makary A Tanatar, Johnpierre Paglione, Cedomir Petrovic, and Louis Taillefer. Anisotropic violation of the Wiedemann-Franz law at a quantum critical point. Science, 316(5829):1320–1322, 2007.
  8. [8]RP Smith, M Sutherland, GG Lonzarich, SS Saxena, N Kimura, S Takashima, M Nohara, and H Takagi. Marginal breakdown of the Fermi-Liquid state on the border of metallic Ferromagnetism. Nature, 455(7217):1220–1223, 2008.DOI
  9. [9]Heike Pfau, Stefanie Hartmann, Ulrike Stockert, Peijie Sun, Stefan Lausberg, Manuel Brando, Sven Friedemann, Cornelius Krellner, Christoph Geibel, Steffen Wirth, et al. Thermal and electrical transport across a magnetic quantum critical point. Nature, 484(7395):493–497, 2012.DOI
  10. [10]RW Hill, Cyril Proust, Louis Taillefer, P Fournier, and RL Greene. Breakdown of Fermi-Liquid theory in a copper-oxide superconductor. Nature, 414(6865):711–715, 2001.pubmed.ncbi.nlm.nih.gov/11742390
  11. [11]Sangwook Lee, Kedar Hippalgaonkar, Fan Yang, Jiawang Hong, Changhyun Ko, Joonki Suh, Kai Liu, Kevin Wang, Jeffrey J Urban, Xiang Zhang, et al. Anomalously low electronic thermal conductivity in metallic vanadium dioxide. Science, 355(6323):371–374, 2017.DOI
  12. [12]A. Jaoui, B. Fauque, C. W. Rischau, et al. Departure from the Wiedemann-Franz law in WP₂ driven by mismatch in T-square resistivity prefactors. npj Quantum Mater. 3, 64, 2018.
  13. [13]Johannes Gooth, F Menges, N Kumar, V Süß, C Shekhar, Yan Sun, Ute Drechsler, R Zierold, Claudia Felser, and Bernd Gottsmann. Thermal and electrical signatures of a hydrodynamic electron fluid in tungsten diphosphide. Nat. Commun., 9(1):4093, 2018.DOI
  14. [14]Nitesh Kumar, Yan Sun, Michael Nicklas, Sarah J Watzman, Olga Young, Inge Leeramakers, Jacob Hornung, Johannes Klotz, Johannes Gooth, Kaustuv Manna, et al. Extremely high conductivity observed in the triple point topological metal MoP. Nat. Commun., 10(1):2475, 2019.DOI
  15. [15]Kshitish Kumar Pradhan, Dushmanta Sahu, Ronald Scaria, and Raghunath Sahoo. Conductivity, diffusivity, and violation of the Wiedemann-Franz law in a hadron resonance gas with van - der waals interactions. Physical Review C, 107(1):014910, 2023.DOI
  16. [16]Pragati Sahoo, Raghunath Sahoo, and Swatantra Kumar Tiwari. Wiedemann-Franz law for hot qcd matter in a color string percolation scenario. Physical Review D, 100(5):051503, 2019.DOI
  17. [17]Rutuparna Rath, Sushanta Tripathy, Bhaswar Chatterjee, Raghunath Sahoo, Swatantra Kumar Tiwari, and Abhishek Nath. Violation of Wiedemann-Franz law for hot hadronic matter created at NICA, FAIR and RHIC energies using non-extensive statistics. The European Physical Journal A, 55:1–10, 2019.DOI
  18. [18]Kamaljeet Singh, Jayanta Dey, Raghunath Sahoo, and Sabyasachi Ghosh. Effect of time-varying electromagnetic field on Wiedemann-Franz law in a hot hadronic matter. Physical Review D, 108(9):094007, 2023.DOI
  19. [19]Jesse Crossno, Jing K Shi, Ke Wang, Xiaomeng Liu, Achim Harzheim, Andrew Lucas, Subir Sachdev, Philip Kim, Takashi Taniguchi, Kenji Watanabe, et al. Observation of the Dirac fluid and the breakdown of the Wiedemann-Franz law in graphene. Science, 351(6277):1058–1061, 2016.DOI
  20. [20]Matthew P Gochan. Fermi Liquid Properties of Dirac Materials. PhD thesis, Boston College, 2020.
  21. [21]L D. Landau. The Theory of a Fermi Liquid. J. Exptl. Theoret. Phys., 3(6):920–925, 1957.
  22. [22]L D. Landau. Oscillations in a Fermi Liquid. J. Exptl. Theoret. Phys., 5(1):101–108, 1957.
  23. [23]L D. Landau. On the theory of Fermi Liquid. J. Exptl. Theoret. Phys., 35 (8)(6):70–74, 1959.
  24. [24]Adam Zareyan. Wiedemann–Franz law for massless Dirac fermions with implications for graphene. Materials, 14(11):2704, 2021.
  25. [25]Yi-Ting Tu and Sankar Das Sarma. Wiedemann-Franz law in graphene. Phys. Rev. B, 107(8):085401, 2023.
  26. [26]Andrew Lucas, Jesse Crossno, Kin Chung Fong, Philip Kim, and Subir Sachdev. Transport in inhomogeneous quantum critical fluids and in the Dirac fluid in graphene. Phys. Rev. B, 93(7):075426, 2016.DOI
  27. [27]Raghu Mahajan, Maissam Barkeshli, and Sean A. Hartnoll. Non-Fermi liquids and the Wiedemann-Franz law. Phys. Rev. B, 88:125107, 2013.DOI
  28. [28]Andrew Lucas and Sankar Das Sarma. Electronic hydrodynamics and the breakdown of the Wiedemann-Franz and Mott laws in interacting metals. Phys. Rev. B, 97(24):245128, 2018.DOI
  29. [29]Mohammad Zarenia, Thomas Benjamin Smith, Alessandro Principi, and Giovanni Vignale. Breakdown of the Wiedemann-Franz law in AB-stacked bilayer graphene. Physical Review B, 99(16):161407, 2019.DOI
  30. [30]Alessandro Principi and Giovanni Vignale. Violation of the Wiedemann-Franz law in hydrodynamic electron liquids. Phys. Rev. Lett., 115:056603, Jul 2015.pubmed.ncbi.nlm.nih.gov/26274433
  31. [31]Ali Lavasani, Daniel Bulmash, and Sankar Das Sarma. Wiedemann-Franz law and Fermi Liquids. Phys. Rev. B, 99:085104, Feb 2019.DOI
  32. [32]Mark J. H. Ku et al. Imaging viscous flow of the Dirac Fluid in graphene. Nature, 583(7817):537–541, 2020.
  33. [33]Georgios Varnavides, Adam S Jermyn, Polina Anikeeva, Claudia Felser, and Prineha Narang. Electron hydrodynamics in anisotropic materials. Nat. Commun., 11(1):4710, 2020.DOI
  34. [34]Eddwi H. Hasdeo, Johan Ekström, Edvin G. Idrisov, and Thomas L. Schmidt. Electron hydrodynamics of two-dimensional anomalous hall materials. Phys. Rev. B, 103:125106, Mar 2021.
  35. [35]Aaron Hui, Vadim Og anesyan, and Eun-Ah Kim. Beyond Ohm’s law: Bernoulli effect and streaming in electron hydrodynamics. Phys. Rev. B, 103:235152, Jun 2021.DOI
  36. [36]Xiaoyang Huang and Andrew Lucas. Electron-phonon hydrodynamics. Phys. Rev. B, 103:155128, Apr 2021.DOI
  37. [37]Omid Tavakol and Yong Baek Kim. Artificial electric field and electron hydrodynamics. Phys. Rev. Res., 3:013290, Mar 2021.DOI
  38. [38]Domenico Di Sante, Johanna Erdmerger, Martin Greiter, Ioannis Matthaiakis, René Meyer, David Rodríguez Fernández, Ronny Thomale, Erik van Loon, and Tim Wehling. Turbulent hydrodynamics in strongly correlated Kagome metals. Nat. Commun., 11(1):3997, 2020.DOI
  39. [39]B. N. Narozhny, I. V. Gornyi, and M. Titov. Hydrodynamic collective modes in graphene. Phys. Rev. B, 103:115402, Mar 2021.DOI
  40. [40]Joseph A Sulpizio, Lior Ella, Asaf Rozen, John Birkbeck, David J Perello, Debarghya Dutta, Moshe Ben-Shalom, Takashi Taniguchi, Kenji Watanabe, Tobias Holder, et al. Visualizing Poiseuille flow of hydrodynamic electrons. Nature, 576(7785):75–79, 2019.
  41. [41]Patrick Gallagher, Chan-Shan Yang, Tairu Lyu, Fanglin Tian, Rai Kou, Hai Zhang, Kenji Watanabe, Takashi Taniguchi, and Feng Wang. Quantum-critical conductivity of the Dirac Fluid in graphene. Science, 364(6436):158–162, 2019.
  42. [42]A. I. Berdyugin, S. G. Xu, F. M. D. Pellegrino, R. Krishna Kumar, A. Principi, I. Torre, M. Ben Shalom, T. Taniguchi, K. Watanabe, I. V. Grigorieva, M. Polini, A. K. Geim, and D. A. Bandurin. Measuring hall viscosity of graphene’s electron fluid. Science, 364(6436):162–165, 2019.DOI
  43. [43]Lior Ella, Asaf Rozen, John Birkbeck, Moshe Ben-Shalom, David Perello, Johanna Zultak, Takashi Taniguchi, Kenji Watanabe, Andre K Geim, Shahal Ilani, et al. Simultaneous voltage and current density imaging of flowing electrons in two dimensions. Nat. Nanotechnol., 14(5):480–487, 2019.DOI
  44. [44]Denis A Bandurin, Andrey V Shytov, Leonid S Levitov, Roshan Krishna Kumar, Alexey I Berdyugin, Moshe Ben Shalom, Irina V Grigorieva, Andre K Geim, and Gregory Falkovich. Fluidity onset in graphene. Nat. Commun., 9(1):4533, 2018.DOI
  45. [45]D. A. Bandurin, I. Torre, R. Krishna Kumar, M. Ben Shalom, A. Tomadin, A. Principi, G. H. Auton, E. Khestanova, K. S. Novoselov, I. V. Grigorieva, L. A. Ponomarenko, A. K. Geim, and M. Polini. Negative local resistance caused by viscous electron backflow in graphene. Science, 351(6277):1055–1058, 2016.DOI
  46. [46]Alessandro Principi and Giovanni Vignale. Violation of the Wiedemann-Franz law in hydrodynamic electron liquids. Phys. Rev. Lett., 115(5):056603, 2015.pubmed.ncbi.nlm.nih.gov/26274433
  47. [47]Thandar Zaw Win, Cho Win Aung, Gaurav Khandal, and Sabyasachi Ghosh. Graphene is neither relativistic nor non-relativistic: thermodynamics aspects. Pramana, 99(1), Jan 2025.DOI
  48. [48]Cho Win Aung, Thandar Zaw Win, Gaurav Khandal, and Sabyasachi Ghosh. Shear viscosity expression for a graphene system in relaxation time approximation. Phys.DOI
  49. [49]Andre K Geim and Konstantin S Novoselov. The rise of graphene. Nat. Mater., 6(3):183–191, 2007.
  50. [50]DC Elias, RV Gorbachev, AS Mayorov, SV Morozov, AA Zhukov, P Blake, LA Ponomarenko, IV Grigorieva, KS Novoselov, F Guinea, et al. Dirac cones reshaped by interaction effects in suspended graphene. Nat. Phys., 7(9):701–704, 2011.DOI
  51. [51]LP Pitaevskii, EM Lifshitz, and John Bradbury Sykes. Course of Theoretical Physics: Physical Kinetics, volume 10. Elsevier, 2017.
  52. [52]Yi Zhan. Two distinct methods to evaluate graphene relaxation time and mobility in Boltzmann diffusive transport, considering ionized impurity scattering and Thomas-Fermi screening. arXiv preprint arXiv:1712.08965, 2017.arxiv.org/abs/1712.08965
  53. [53]Sean A. Hartnoll, Pavel K. Kovtun, Markus Müller, and Subir Sachdev. Theory of the Nernst effect near quantum phase transitions in condensed matter and in dyonic black holes. Phys. Rev. B, 76:144502, Oct 2007.DOI
  54. [54]Aniket Majumdar, Nisarg Chadha, Pritam Pal, Akash Gugnani, Bhaskar Ghawri, Kenji Watanabe, Takashi Taniguchi, Subroto Mukerjee, and Arindam Ghosh. Universality in quantum critical flow of charge and heat in ultra-clean graphene. arXiv preprint arXiv:2501.03193, 2025.DOI
  55. [55]Raad Chegel and Somayeh Behzad. Tunable electronic, optical, and thermal properties of two-dimensional germanene via an external electric field. Scientific Reports, 10(1):704, 2020.DOI
  56. [56]Raad Chegel. Remarkable thermopower property enhancement in two-dimensional SiC via B and N doping and magnetic field. Journal of Alloys and Compounds, 967:171682, 2023.DOI
  57. [57]Raad Chegel. Magneto-electronic and thermopower properties of b, n and si doped monolayer graphene. Diamond and Related Materials, 137:110154, 2023.DOI

Paper details

vlnc:W_2thdamyz
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents