betaFeedback

WorksPhysics and Astronomy

Average density of Bloch electrons in a homogeneous magnetic field: A second-order response

Abstract

We compute the average density of a three-dimensional multiband crystal of arbitrary symmetry, metal or insulator, to first and second order in a weak homogeneous magnetic field. To linear order and for insulators, the density follows the well-known Streda formula, but for metals there is an extra contribution from the orbital magnetic moments at the Fermi surface. To second order the average density depends on several microscopic processes. Among these, the quantum metric tensor plays an important role by generating a pseudo-magnetic moment resulting from the rotation of the Bloch wave functions in the complex projective plane. We also discuss the implications of our results for the volume and pressure. The method we develop is explicitly gauge invariant, considers intraband and interband processes on equal footing, accommodates relaxation processes, and can be readily extended to other observables.

Introduction and main results

Electronic systems in magnetic fields give rise to many novel phenomena [1, 2, 3, 4, 5, 6, 7] [8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22]. Strong-field responses have long inspired new ideas, particularly in connection with quantum Hall effects, but the weak-field regime has received comparatively less attention. In this work, we study how the average electronic density is modified by the presence of a weak static homogeneous magnetic field in a clean three-dimensional (3D) crystal of arbitrary symmetry, whether metal or insulator. We focus on the regime in which the field is strong enough to trigger interesting nonlinear responses [23–31] but weak enough that Bloch wave functions remain a valid description of the electronic states.

Generally, the average particle density can be expanded in powers of the magnetic field as

n=n(0)+ςa(1)Ba+ςab(2)BaBb+⋯ .(1)n = n^{(0)} + \varsigma^{(1)}_{a}B^{a} + \varsigma^{(2)}_{ab}B^{a}B^{b} + \cdots. \tag*{(1)}

The first term on the right-hand side represents the density in the absence of magnetic field, the second term is the linear response, and the third is the nonlinear response. Since the density is even under time reversal while the magnetic field is odd, the linear response vanishes if time reversal symmetry (TRS) is present.

If TRS is broken, the linear response is dominant, and we expect the (pseudo)tensor ς(1)\varsigma^{(1)} to follow the well-known Streda formula [4, 5, 6]

(∂n∂B)μ=−1eσH,(2)\left(\frac{\partial n}{\partial B}\right)_{\mu} = -\frac{1}{e}\sigma_H, \tag*{(2)}

which relates the average density to the anomalous Hall conductivity σH\sigma_H when the chemical potential lies within an energy gap. This relation was originally derived in the context of strong magnetic fields but later found to apply across a wide range of field strengths, even in the presence of electron interactions. It predicts that the average density changes in discrete steps when the anomalous Hall conductivity is quantized with nonzero Chern number. Indeed, it has been used as a diagnostic tool to measure the Chern numbers [7] [8] and has applications within the semiclassical theory of electron dynamics [9], orbital magnetization [10], Fermi liquid theory [11], Chern insulators [12], and out of equilibrium [14] and localization physics [15, 16], among others.

First, we compute the response tensor ς(1)\varsigma^{(1)} for a system of Bloch electrons in Sec. IV. We find that, indeed, the Streda formula holds for Bloch electrons when the chemical potential lies within an energy gap, but there is a correction when the chemical potential crosses an energy band [13],

∂n∂B=−1eσH+1V∑nkfn′mn.(3)\frac{\partial n}{\partial\mathbf{B}} = -\frac{1}{e}\sigma_H + \frac{1}{V}\sum_{n\mathbf{k}} f'_n\mathbf{m}_n. \tag*{(3)}

The notation used can be found in Appendix H. The second term arises from the interaction of the magnetic field with the orbital magnetic moments mn\mathbf{m}_n (H38) at the Fermi surface (FS). fn′f'_n is the first derivative of the Fermi function (H17), which is nonzero only near the Fermi level. Equation (3) implies that particles can be transferred to (and from) the FS and the edge states (assuming the expression can be applied in the Hall regime).

When TRS is present, the second-order response is the leading contribution. We derive a general expression for ς(2)\varsigma^{(2)} in terms of Bloch wave functions in § 1.4. It is not as compact as (3) since it includes FS, non-FS, one-band, two-band, and three-band terms and must generally be analyzed on a case-by-case basis. There is a contribution that deserves special attention (Sec. VI). We find that the Fubini-Study quantum metric tensor gnabg_n^{ab} which characterizes the distance between Bloch states [32–36] induces quadratic magnetic moments such as

mn(2)zz∼e2gnxxvnyvny,(4)m_n^{(2)zz} \sim e^2 g_n^{xx}v_n^y v_n^y, \tag*{(4)}

where vnav_n^a is the band velocity in direction aa. The origin of this magnetic moment is purely geometric, meaning it does not arise from spin or orbital angular momentum but, rather, from angular momentum due to rotations of Bloch wave functions in the complex projective plane (Figure 1). The quantum geometric magnetic moment appears only in the nonlinear response and is distinct from known mechanisms that generate magnetic field-like effects without an actual magnetic field, such as those found in strained graphene[18–20] or in the flat-band fractional quantum Hall effect [21].

Figure 1. Schematic representation of the mapping from parameter space k\mathbf{k} to the complex projective plane of Bloch wave functions. A small shift by g\mathbf{g} rotates the vector ∣unk⟩|u_{n\mathbf{k}}\rangle by an angle dθnkd\theta_{n\mathbf{k}} [see Eq. (29)].

In Sec. VII, we argue that a change in density implies a change in volume or pressure (however small) since these thermodynamic qualities can also be expanded in powers of the magnetic field

V=V(0)+v(1)aBa+v(2)abBaBb+⋯ ,(5)V = V^{(0)} + v^{(1)a}B^{a} + v^{(2)ab}B^{a}B^{b} + \cdots, \tag*{(5)}
P=P(0)+p(1)aBa+p(2)abBaBb+⋯ ,(6)P = P^{(0)} + p^{(1)a}B^{a} + p^{(2)ab}B^{a}B^{b} + \cdots, \tag*{(6)}

and, when the chemical potential lies within an energy gap, v(1)av^{(1)a} and p(1)ap^{(1)a} are also proportional to the anomalous Hall conductivity.

In Sec. VIII we show that in a simple two-band model the intraband contribution is largely due to geometric magnetic moments

ζintra(2)∼1V∑nkfn′′mn(2)zz,(7)\zeta_{\mathrm{intra}}^{(2)} \sim\frac{1}{V}\sum_{n\mathbf{k}} f^{\prime\prime}_{n}m_{n}^{(2)zz}, \tag*{(7)}

and results in a fractional density change in the order, n(2)/n(0)∼10−4%n^{(2)}/n^{(0)} \sim10^{-4}\%. More accurate values require large-scale numerical simulations that incorporate detailed crystal structure information. The analytical expressions obtained in this work are well suited for that purpose.

Particle density

We consider non-interacting electrons on a 3D lattice whose single-particle Hamiltonian is

h=12m(p−eA)2+U,(8)h = \frac{1}{2m}(\mathbf{p}-e\mathbf{A})^{2}+U, \tag*{(8)}

where e<0e<0 is the electron charge and UU is the ionic periodic potential. Without loss of generality, the magnetic field is taken along the zz axis and represented by the vector potential A=y^(Bz/q)sin⁡(g⋅r)\mathbf{A}=\hat{\mathbf{y}}(B^{z}/q)\sin(\mathbf{g}\cdot\mathbf{r}). The magnetic field is spatially modulated by a wave vector g=qx^\mathbf{g}=q\hat{\mathbf{x}}. We introduce the parameter qq to regularize the singular nature of the position operator in the Bloch basis [22]. Since we are interested in the response to a homogenous magnetic field, we take the q→0q\to0 limit at the end of the calculation. In this limit, the magnetic field is homogeneous:

B=z^Bz.(9)\mathbf{B}=\hat{\mathbf{z}}B^{z}. \tag*{(9)}

In addition, at every step of our calculation our expressions are invariant under the phase transformation of Bloch states, ∣nk⟩→eiα(k)∣nk⟩|n\mathbf{k}\rangle\to e^{i\alpha(\mathbf{k})}|n\mathbf{k}\rangle (we also call it gauge invariance).

The particle density operator

n^=ψ†ψ(10)\hat{n}=\psi^{\dagger}\psi \tag*{(10)}

can be expanded in Bloch operators anka_{n\mathbf{k}} as ψ=∑nkϕnkank\psi=\sum_{n\mathbf{k}}\phi_{n\mathbf{k}}a_{n\mathbf{k}}, where ϕnk\phi_{n\mathbf{k}} is a Bloch wave function with band nn and crystal momentum k\mathbf{k}. Its expectation value is then

n=∑nkmk′ϕnk∗ϕmk′ρmk′nk,(11)n=\sum_{n\mathbf{k}m\mathbf{k}^{\prime}}\phi_{n\mathbf{k}}^{*}\phi_{m\mathbf{k}^{\prime}}\rho_{m\mathbf{k}^{\prime}n\mathbf{k}}, \tag*{(11)}

where ρmk′nk≡⟨ank†amk′⟩\rho_{m\mathbf{k}^{\prime}n\mathbf{k}}\equiv\langle a_{n\mathbf{k}}^{\dagger}a_{m\mathbf{k}^{\prime}}\rangle is the density matrix in the Bloch basis (see Appendix § 2). Because the magnetic field varies spatially, it is convenient to expand the density in its spatial Fourier modes as

n=n0+ngcos⁡(g⋅r)+n2gcos⁡(2g⋅r)+⋯ ,(12)n=n_{0}+n_{\mathbf{g}}\cos(\mathbf{g}\cdot\mathbf{r})+n_{2\mathbf{g}}\cos(2\mathbf{g}\cdot\mathbf{r})+\cdots, \tag*{(12)}

where

nG=2V∫dr ncos⁡(G⋅r)(G≠0).(13)n_{\mathbf{G}}=\frac{2}{V}\int d\mathbf{r}\,n\cos(\mathbf{G}\cdot\mathbf{r})\qquad(\mathbf{G}\ne0). \tag*{(13)}

The ⋯\cdots in (12) represents higher-order Fourier modes. The expansion also contains sine modes, but as shown in Appendix C, they vanish. Here, n0n_{0} is the density in the absence of a magnetic field, ngn_{\mathbf{g}} corresponds to the linear correction, and n2gn_{2\mathbf{g}} corresponds to the quadratic correction. Substituting (11) into (13), we obtain

nG=2VRe⁡∑nmk⟨unk∣um,k−G⟩ρm,k−G,nk,(14)n_{\mathbf{G}}=\frac{2}{V}\operatorname{Re}\sum_{nm\mathbf{k}}\langle u_{n\mathbf{k}}|u_{m,\mathbf{k}-\mathbf{G}}\rangle\rho_{m,\mathbf{k}-\mathbf{G},n\mathbf{k}}, \tag*{(14)}

which is valid to all orders in BzB^{z} and for arbitrary G\mathbf{G}. The average density is then

n−n0=lim⁡q→0∑GnG,(15)n-n_{0}=\lim_{q\to0}\sum_{\mathbf{G}}n_{\mathbf{G}}, \tag*{(15)}

where the sum runs over G=g,2g\mathbf{G}=\mathbf{g},2\mathbf{g}.

Zero order

To zero order, only n0n_{0} contributes:

n(0)≡n0=1V∑nkfnk,(16)n^{(0)}\equiv n_{0}=\frac{1}{V}\sum_{n\mathbf{k}}f_{n\mathbf{k}}, \tag*{(16)}

where we used ρnknk(0)=fnk\rho^{(0)}_{n\mathbf{k}n\mathbf{k}}=f_{n\mathbf{k}}.

First order

To first order, only ngn_{\mathbf{g}} contributes:

n(1)=lim⁡q→0ng(1).(17)n^{(1)}=\lim_{q\to0}n_{\mathbf{g}}^{(1)}. \tag*{(17)}

In Appendix § 2, we calculate ρ(1)\rho^{(1)} in the static and clean limits (τ→∞\tau\to\infty). Substituting this into (14) gives

ng(1)=ζg(1)aBa.(18)n_{\mathbf{g}}^{(1)}=\zeta_{\mathbf{g}}^{(1)a}B^{a}. \tag*{(18)}

where the zz component of the response tensor is

ςg(1)z=2V(e2q)Im⁡∑nlk⟨unk∣Plk−gv^ky∣unk⟩flk−g−fnkεlk−g−εnk.(19)\varsigma_{\mathbf{g}}^{(1)z}=\frac{2}{V}\left(\frac{e}{2q}\right)\operatorname{Im}\sum_{n l k}\langle u_{n k}|P_{l k-\mathbf{g}}\hat{v}_{k}^{y}|u_{n k}\rangle\frac{f_{l k-\mathbf{g}}-f_{n k}}{\varepsilon_{l k-\mathbf{g}}-\varepsilon_{n k}}. \tag*{(19)}

All quantities are defined in Appendix H: fnkf_{n k} is the Fermi function (at zero temperature) of electrons at energy εnk\varepsilon_{n k}, and Pnk=∣unk⟩⟨unk∣P_{n k}=|u_{n k}\rangle\langle u_{n k}| is the projector onto band nn at point k\mathbf{k}. In deriving (19), we used v^k−gy=v^ky\hat{v}_{k-\mathbf{g}}^{y}=\hat{v}_{k}^{y} because g=x^q\mathbf{g}=\hat{x}q is perpendicular to v^ky\hat{v}_{k}^{y}; this also follows from (H24). Note that (19) is explicitly gauge invariant- the sum is the trace over a gauge-invariant product of operators.

The operator Plk−gv^kyP_{l k-\mathbf{g}}\hat{v}_{k}^{y} is interesting because its q→0q\to0 limit, ∼(∂xPl)v^y\sim(\partial_xP_l)\hat{v}^{y} (k omitted), could be interpreted as a quantum momentum tensor (QMT) operator. To see this, we compute its average over a Bloch state and note that it yields all possible momentum components,

References

  1. [1]N. W. Ashcroft and N. D. Mermin, Solid State Physics (Thomson Learning, Toronto, 1976).
  2. [2]J. Ziman, Electrons and Phonons: The Theory of Transport Phenomena in Solids, Oxford Classics Series in the Physical Sciences (Oxford University Press, 2001).
  3. [3]N. Nagaosa, J. Sinova, S. Onoda, A. H. MacDonald, and N. P. Ong, Anomalous hall effect, Rev. Mod. Phys. 82, 1539 (2010).DOI
  4. [4]A. Widom, Thermodynamic derivation of the hall effect current, Physics Letters A 90, 474 (1982).DOI
  5. [5]P. Streda, Theory of quantised hall conductivity in two dimensions, Journal of Physics C: Solid State Physics 15, L717 (1982).
  6. [6]A. H. MacDonald and P. Streda, Quantized hall effect and edge currents, Phys. Rev. B 29, 1616 (1984).DOI
  7. [7]F. Xu, Z. Sun, J. Li, C. Zheng, C. Xu, J. Gao, T. Jia, K. Watanabe, T. Taniguchi, B. Tong, L. Lu, J. Jia,

Paper details

vlnc:W_au7rx6qj
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 6

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