5
$\begingroup$

I am curious if there are results available for the Kitaev model with a magnetic field -- in his 2006 paper, Kitaev obtains the form of the effective hamiltonian (Eq. 46 in https://arxiv.org/abs/cond-mat/0506438), however does not give the precise prefactors. Also wondering about the 2nd order contribution (which is proportional to the identity) -- I just would like to see what would be the correct counting.

Thanks!

$\endgroup$

1 Answer 1

6
+100
$\begingroup$

Yes, the precise result can be found in Eq. (30) of our paper arXiv:1109.4155:

$$\kappa = \frac{h_xh_yh_z}{8 u_0^2 J^2},$$

where $u_0$ is a numerical factor obtained by solving the mean field equation numerically. Kitaev mentioned in his paper that $\Delta E = |u_0J|\approx 0.27|J|$ (i.e. $u_0\approx -0.27$), and we provide a more accurate result $u_0\approx −0.262433$ in Eq. (27) of our paper. The explicit perturbation path is illustrated in Fig. 5(a) of our paper (which is a 3rd order perturbation, not a 2nd order one, as stressed in Kitaev's paper).


Let me briefly outline the derivation below. We start from an isotropic Kitaev honeycomb model with a perturbative Zeeman field ($|\boldsymbol{h}|\ll J$),

$$H=-J\sum_{\langle ij\rangle}S_i^{a}S_j^{a}-\sum_{i}\boldsymbol{h}\cdot\boldsymbol{S}_i,$$

where $a=1,2,3$ depends on the type ($x,y,z$) of the link $\langle ij\rangle$. Introduce four Majorana spinons $\chi_i^\alpha$ ($\alpha=0,1,2,3$) on each site $i$, defined by the anticommutation relation $\{\chi_i^\alpha, \chi_{j}^{\beta}\}=\delta_{ij}\delta_{\alpha\beta}$ (note the unusual normalization of the Majorana operator here). Under the gauge constraint (on-site constraint) $\chi_{i}^0\chi_{i}^1\chi_{i}^2\chi_{i}^3=1/4$, the spin operator $\boldsymbol{S}_i$ can be written in terms of the spinon bilinear form as

$$\boldsymbol{S}_i=\frac{i}{2}\Big(\chi_i^0\boldsymbol{\chi}_i-\frac{1}{2}\boldsymbol{\chi}_i\times\boldsymbol{\chi}_i\Big),$$

where the vector $\boldsymbol{\chi}_i=(\chi_i^1,\chi_i^2,\chi_i^3)$ is made of the last three components of the Majorana fermion. We can see that the $\chi^0$ ($c$-fermion) differs from $\chi^{1,2,3}$ ($b$-fermion) in this fractionalization scheme. This difference is also reflected in the mean-field Hamiltonian $H_\text{MF}$. In the unperturbed limit $\boldsymbol{h}=0$, $H_\text{MF}$ can be obtained by substitute the expression for $\boldsymbol{S}_i$ to the spin Hamiltonian and take the mean-field decomposition described by Kitaev:

$$H_\text{MF}=J\sum_{\langle ij\rangle}\big(\text{i}u_a \chi_i^0\chi_j^0+\text{i}u_0\chi_i^a\chi_j^a\big),$$

where the bond parameter $u_\alpha=\langle\text{i}\chi_{i}^\alpha\chi_{j}^\alpha\rangle$ (for $\alpha=0,1,2,3$) is determined self-consistently from the Majorana fermion correlation on the mean-field ground state (note $a=1,2,3$ is fixed by the link type, not a dummy index to be summed over). It is found that the mean-field solution reads $u_a=1/2$ and $$u_0=-\frac{1}{3N}\sum_{\boldsymbol{k}\in\text{BZ}}\big|e^{\text{i}k_y}+2e^{-\text{i}k_y/2}\cos(\sqrt{3}k_x/2)\big|\approx -0.262433,$$ where $N$ is the number of sites (we can evaluate the summation numerically on a finite lattice and then take the thermodynamic limit $N\to\infty$). The band structure of the spinon can be obtained by diagonalizing the mean-field Hamiltonian, as shown below:

enter image description here

One can see that the $\chi^0$ fermion is itinerant and has a gapless spectrum. But the $\boldsymbol{\chi}=(\chi^1,\chi^2,\chi^3)$ fermions are dimerized on the corresponding type of links and are therefore gapped (as the flat band). The energy gap for $\boldsymbol{\chi}$ fermions is $\Delta E=|u_0J|$.

If we are only interested in the low energy physics, we can neglect the high-energy $\boldsymbol{\chi}$ fermions. However, once the Zeeman field is introduced to the system, the mixing is turned on between the low-energy $\chi^0$ and high-energy $\boldsymbol{\chi}$ fermions (and also mixing among the components of $\boldsymbol{\chi}$). Thus a perturbation pathway illustrated below becomes possible:

enter image description here

which results in a 2nd nearest neighbor coupling between the low-energy $\chi^0$ fermion,

$$H_{\text{MF},0}=\text{i}u_a J\sum_{\langle ij\rangle} \chi_i^0\chi_j^0+\text{i}\kappa\sum_{\langle\!\langle ij\rangle\!\rangle}\chi_i^0\chi_j^0,$$

with the coefficient $\kappa$ given by the 3rd order perturbation (see this Wikipedia page for the 3rd order perturbation formula)

$$\kappa=\Big(\frac{h_x}{2}\Big)\frac{1}{u_0J}\Big(-\frac{h_z}{2}\Big)\frac{1}{u_0J}\Big(-\frac{h_y}{2}\Big)=\frac{h_xh_yh_z}{8u_0^2J^2}.$$

The 2nd neighbor coupling term $\kappa$ breaks the time-reversal symmetry and gaps out the low-energy fermion $\chi^0$. The gapless Kitaev spin liquid is then driven into the non-Abelian phase with the Ising topological order.

$\endgroup$
4
  • $\begingroup$ I just wonder in your expression for $\kappa$, for the parameter $J$, is it the one the same as in your starting Hamiltonian or is it the one used by Kitaev? because in Kitaev's paper, he wrote everything in terms of Pauli operators instead of spin operators, i.e., there can be a factor of 4 difference. $\endgroup$
    – Ogawa Chen
    Commented Sep 13, 2023 at 11:18
  • $\begingroup$ @OgawaChen I was using the spin operator convention. $\endgroup$ Commented Sep 14, 2023 at 13:09
  • $\begingroup$ I am a bit confused because you said "Kitaev mentioned in his paper that $\Delta E = u_0 J \approx 0.27 |J|$ (i.e.$ u_0 \approx 0.27$)", I think here the $J$ should be the one used in Kitaev's convention? In other words, the energy difference should be $0.27 |J|/4$ in terms of your convention. $\endgroup$
    – Ogawa Chen
    Commented Sep 14, 2023 at 13:22
  • $\begingroup$ just one more thing, should there be a combinational factor $3! = 6$ before? $\endgroup$
    – Ogawa Chen
    Commented Sep 16, 2023 at 7:48

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.