betaFeedback

WorksPhysics and Astronomy

Classical Heisenberg and XY models on zigzag ladder lattices with nearest-neighbor bilinear-biquadratic exchange: Exact solution for the ground-state problem

Abstract

An exact and complete solution of the ground-state problem for the classical Heisenberg and XY models with nearest-neighbor bilinear-biquadratic exchange on two- and three-dimensional lattices composed of isosceles triangles is determined with the use of a cluster method. It is shown how the geometric frustration due to the presence of triangles as structural units leads to the emergence of a rich phase diagram with incommensurate spiral orderings and their collinear limits, as well as canted and noncoplanar (conical) structures. Surprisingly, there are two different spiral phases with both continuous and discontinuous phase transitions between them. One of these phases is degenerate on two-dimensional partially anisotropic triangular lattice. This degeneracy is lifted on three-dimensional lattices. Canted phase is highly degenerate and this degeneracy persists on three-dimensional lattices.

Introduction

Frustrated magnetic systems on geometrically constrained lattices, such as the triangular and kagomé lattices, Shastry-Sutherland lattice, pyrochlore lattice [1, 2], 3D honeycomb zigzag ladder lattice [3] etc. exhibit rich phase diagrams with exotic phases. The emergence of unusual phases is also due to complex interactions between magnetic moments, in particular to biquadratic exchange coupling.

Classical Heisenberg models with bilinear-biquadratic exchange interaction have been intensively studied during the last decades [4, 5, 6, 7]. The reason is that there are many magnets, especially 2D ones, where biquadratic exchange interactions play a key role and cannot be neglected (see for instance [7, 8]).

Ground states of the classical bilinear-biquadratic Heisenberg model on the isotropic triangular lattice were studied in Refs. [5, 6, 9]. In Ref. [9] four phases were identified in this model: a ferromagnetic phase, a nematiclike phase, a 120° phase (triangular phase with 120 degree angle among the nearest moments), and a phase with noncoplanar orientations of spins (see also figure 1 in Ref. [5]).

We studied noncoplanar ground states of classical Heisenberg spins with bilinear-biquadratic interaction on a fully anisotropic triangular lattice [10]. However, the investigation of the complete phase diagram for such a model with six interaction parameters is a difficult task. Therefore, in this paper, we present a complete phase diagram for a simpler lattice, namely for a partially anisotropic triangular lattice composed of isosceles triangles (a model with four parameter instead of six). We have already investigated ground states for classical Heisenberg spins on lattices of this type but with bilinear interaction only [11]. Here, in addition to two bilinear interaction parameters, we consider two biquadratic ones. The problem is reduced to the investigation of the ground states for one triangular plaquette. The solution makes it possible to construct global ground states for many lattices composed of identical isosceles triangles including three-dimensional ones.

Ground-state structures and diagrams for one triangle

Let us consider one isosceles triangular plaquette with classical spins (unit 3-vectors) at its vertices [see Figure 1(a)]. The spins interact via bilinear (KK and LL) and biquadratic (AA and BB) pairwise exchange interactions. Interaction parameters are given per one plaquette. If a bond is shared by nn plaquettes, then the parameters should be multiplied by nn to obtain the interactions between corresponding spins. The Hamiltonian of a zigzag ladder lattice composed of such plaquettes can be presented as a sum over all the plaquettes,

Figure 1. (a) An elementary triangular plaquette for zigzag ladder lattices with spins S1S_1, S2S_2, and S3S_3 (unit 3-vectors) at its vertices. There are linear (KK and LL) and biquadratic (AA and BB) pairwise interactions (per one plaquette) between the spins. (b) Tetrahedron of values for angles α\alpha, β\beta, and γ\gamma between the spins. Vertices of the tetrahedron correspond to three collinear spins; edges and faces to three coplanar ones; for edges two of three spins are collinear. (c) Cross section of the tetrahedron at a fixed value of α\alpha.

H=∑Δi{KSi1⋅Si2+L(Si1⋅Si3+Si2⋅Si3)−A(Si1⋅Si2)2−B[(Si1⋅Si3)2+(Si2⋅Si3)2]},(1)H = \sum_{\Delta_i} \left\{K\mathbf{S}_{i1}\cdot\mathbf{S}_{i2}+L(\mathbf{S}_{i1}\cdot\mathbf{S}_{i3}+\mathbf{S}_{i2}\cdot\mathbf{S}_{i3})-A(\mathbf{S}_{i1}\cdot\mathbf{S}_{i2})^2-B\left[(\mathbf{S}_{i1}\cdot\mathbf{S}_{i3})^2+(\mathbf{S}_{i2}\cdot\mathbf{S}_{i3})^2\right]\right\}, \tag*{(1)}

where Si1\mathbf{S}_{i1}, Si2\mathbf{S}_{i2}, and Si3\mathbf{S}_{i3} are the spins at the three vertexes of ii-th triangular plaquette.

Let α\alpha, β\beta, and γ\gamma are angles between the spins (0≤α,β,γ≤π0 \le\alpha, \beta, \gamma\le\pi). These angles should satisfy the following inequalities:

α+β+γ≤2π,−α+β+γ≥0,α−β+γ≥0,α+β−γ≥0.(2)\begin{aligned} \alpha+\beta+\gamma&\le2\pi,\\ -\alpha+\beta+\gamma&\ge0,\\ \alpha-\beta+\gamma&\ge0,\\ \alpha+\beta-\gamma&\ge0. \tag*{(2)} \end{aligned}

The solution of this set of inequalities is the tetrahedron shown in Figure 1(b). If at least one of the inequalities becomes an equality then the vectors are coplanar. This corresponds to a point on the surface of the tetrahedron. The coupling energy between the spins of the plaquette shown in Figure 1(a) reads

E=Kcos⁡α+L(cos⁡β+cos⁡γ)−Acos⁡2α−B(cos⁡2β+cos⁡2γ).(3)E=K\cos\alpha+L(\cos\beta+\cos\gamma)-A\cos^2\alpha-B(\cos^2\beta+\cos^2\gamma). \tag*{(3)}

To obtain the ground-state phase diagram for the XY model, one should minimize the energy (3) on the tetrahedron surface, i.e. at all vertices, edges, and faces. There are

four vertices (α,β,γ)(\alpha,\beta,\gamma):

(0,0,0);(0,π,π);(π,0,π);(π,π,0);(4)(0,0,0);(0,\pi,\pi);(\pi,0,\pi);(\pi,\pi,0); \tag*{(4)}

six edges:

α=0, γ=β;β=0, γ=α;γ=0, β=α;α+β=π, γ=π;α+γ=π, β=π;β+γ=π, α=π;(5)\begin{aligned} \alpha&=0,\ \gamma=\beta;\quad\beta=0,\ \gamma=\alpha;\quad\gamma=0,\ \beta=\alpha;\\ \alpha+\beta&=\pi,\ \gamma=\pi;\\ \alpha+\gamma&=\pi,\ \beta=\pi;\\ \beta+\gamma&=\pi,\ \alpha=\pi; \tag*{(5)} \end{aligned}

and four faces:

−α+β+γ=0,α−β+γ=0,α+β−γ=0,α+β+γ=2π.(6)\begin{aligned} -\alpha+\beta+\gamma&=0,\\ \alpha-\beta+\gamma&=0,\\ \alpha+\beta-\gamma&=0,\\ \alpha+\beta+\gamma&=2\pi. \tag*{(6)} \end{aligned}

If the energy attains its minimum in an intrinsic point of the tetrahedron [Figure 1(b)], then this minimum is determined from the following conditions:

∂E∂α=(−K+2Acos⁡α)sin⁡α=0,∂E∂β=(−L+2Bcos⁡β)sin⁡β=0,∂E∂γ=(−L+2Bcos⁡γ)sin⁡γ=0.(7)\begin{aligned} \frac{\partial E}{\partial\alpha}&=(-K+2A\cos\alpha)\sin\alpha=0,\\ \frac{\partial E}{\partial\beta}&=(-L+2B\cos\beta)\sin\beta=0,\\ \frac{\partial E}{\partial\gamma}&=(-L+2B\cos\gamma)\sin\gamma=0. \tag*{(7)} \end{aligned}

It follows from these equations

cos⁡α=K2A,cos⁡β=cos⁡γ=L2B.(8)\cos\alpha=\frac{K}{2A},\qquad\cos\beta=\cos\gamma=\frac{L}{2B}. \tag*{(8)}

Let us find the minimum of the energy (3) on the face γ=α−β\gamma=\alpha-\beta of the tetrahedron [see Figure 1(b)].

∂E∂α=sin⁡α[2Acos⁡α−K]+sin⁡(α−β)[2Bcos⁡(α−β)−L]=0,∂E∂β=L[sin⁡(α−β)−sin⁡β]−B[sin⁡2(α−β)−sin⁡2β]=0.(9)\begin{aligned} \frac{\partial E}{\partial\alpha}&=\sin\alpha\left[2A\cos\alpha-K\right]+\sin(\alpha-\beta)\left[2B\cos(\alpha-\beta)-L\right]=0,\\ \frac{\partial E}{\partial\beta}&=L[\sin(\alpha-\beta)-\sin\beta]-B[\sin2(\alpha-\beta)-\sin2\beta]=0. \tag*{(9)} \end{aligned}

We have from the last equation

Lsin⁡(α2−β)cos⁡α2−Bsin⁡(α−2β)cos⁡α=0(10)L\sin\left(\frac{\alpha}{2}-\beta\right)\cos\frac{\alpha}{2}-B\sin(\alpha-2\beta)\cos\alpha=0 \tag*{(10)}

or

sin⁡(α2−β)[Lcos⁡α2−2Bcos⁡(α2−β)cos⁡α]=0.(11)\sin\left(\frac{\alpha}{2}-\beta\right)\left[L\cos\frac{\alpha}{2}-2B\cos\left(\frac{\alpha}{2}-\beta\right)\cos\alpha\right]=0. \tag*{(11)}

This equation yields

β=α2orLcos⁡α2−2Bcos⁡(α2−β)cos⁡α=0.(12)\beta=\frac{\alpha}{2}\quad\text{or}\quad L\cos\frac{\alpha}{2}-2B\cos\left(\frac{\alpha}{2}-\beta\right)\cos\alpha=0. \tag*{(12)}

In the same way, we have for the face α+β+γ=2π\alpha+\beta+\gamma=2\pi

β=π−α2orLcos⁡α2−2Bcos⁡(α2+β)cos⁡α=0.(13)\beta=\pi-\frac{\alpha}{2}\quad\text{or}\quad L\cos\frac{\alpha}{2}-2B\cos\left(\frac{\alpha}{2}+\beta\right)\cos\alpha=0. \tag*{(13)}

For faces α−β+γ=0\alpha-\beta+\gamma=0 and α+β−γ=0\alpha+\beta-\gamma=0, equations are the same as for faces −α+β+γ=0-\alpha+\beta+\gamma=0 [Eq. (12)] and α+β+γ=2π\alpha+\beta+\gamma=2\pi [Eq. (13)], respectively.

Let us solve the equation

Lcos⁡α2−2Bcos⁡(α2±β)cos⁡α=0.(14)L\cos\frac{\alpha}{2}-2B\cos\left(\frac{\alpha}{2}\pm\beta\right)\cos\alpha=0. \tag*{(14)}

After simple transformations we obtain a quadratic equation with respect to cos⁡β\cos\beta

4B2sin⁡2α2(1−cos⁡2β)cos⁡2α=(Lcos⁡α2−2Bcos⁡α2cos⁡αcos⁡β)2.(15)4B^2\sin^2\frac{\alpha}{2}(1-\cos^2\beta)\cos^2\alpha =\left(L\cos\frac{\alpha}{2}-2B\cos\frac{\alpha}{2}\cos\alpha\cos\beta\right)^2. \tag*{(15)}

Table I. Ground-state configurations for the classical bilinear-biquadratic Heisenberg model on one isosceles triangle and their energies. The first, second and third groups of rows correspond, respectively, to the vertices, edges, and faces of the tetrahedron shown in Figure 1 [see also Eqs. (4–6)]. The last row corresponds to the interior points of the tetrahedron.

Angles α\alpha, β\beta, γ\gammaEnergyConditionsStructure
α=β=γ=0\alpha=\beta=\gamma=0E=K+2L−A−2BE=K+2L-A-2BF
α=0\alpha=0, β=π\beta=\pi, γ=π\gamma=\piE=K−2L−A−2BE=K-2L-A-2BAF1_1
α=π\alpha=\pi, β=0\beta=0, γ=π\gamma=\piE=−K−A−2BE=-K-A-2BAF21_{21}
α=π\alpha=\pi, β=π\beta=\pi, γ=0\gamma=0E=−K−A−2BE=-K-A-2BAF22_{22}
α=0\alpha=0, cos⁡β=L2B\cos\beta=\frac{L}{2B}, γ=β\gamma=\betaE=K−A+L22BE=K-A+\frac{L^2}{2B}∣L∣<2∣B∣|L|<2|B|CF
cos⁡α=K+L2(A+B)\cos\alpha=\frac{K+L}{2(A+B)}, β=0\beta=0, γ=α\gamma=\alphaE=(K+L)24(A+B)+L−BE=\frac{(K+L)^2}{4(A+B)}+L-B∣K+L∣<2∣A+B∣|K+L|<2|A+B|CF11_{11}
or β=α\beta=\alpha, γ=0\gamma=0CF12_{12}
cos⁡α=K−L2(A+B)\cos\alpha=\frac{K-L}{2(A+B)}, β=π\beta=\pi, γ=π−α\gamma=\pi-\alphaE=(K−L)24(A+B)−L−BE=\frac{(K-L)^2}{4(A+B)}-L-B∣K−L∣<2∣A+B∣|K-L|<2|A+B|CF21_{21}
or β=π−α\beta=\pi-\alpha, γ=π\gamma=\piCF22_{22}
α=π\alpha=\pi, β=π2\beta=\frac{\pi}{2}, γ=π2\gamma=\frac{\pi}{2}E=−K−AE=-K-ACF3_3
8Acos⁡3α2+2(B−K−2A)cos⁡α2−L=08A\cos^3\frac{\alpha}{2}+2(B-K-2A)\cos\frac{\alpha}{2}-L=0, β=γ=α2\beta=\gamma=\frac{\alpha}{2}E=4Acos⁡4α2+Lcos⁡α2−K−AE=4A\cos^4\frac{\alpha}{2}+L\cos\frac{\alpha}{2}-K-A0<cos⁡α2<10<\cos\frac{\alpha}{2}<1Sp11_{11}
8Acos⁡3α2+2(B−K−2A)cos⁡α2+L=08A\cos^3\frac{\alpha}{2}+2(B-K-2A)\cos\frac{\alpha}{2}+L=0, β=γ=π−α2\beta=\gamma=\pi-\frac{\alpha}{2}E=4Acos⁡4α2−Lcos⁡α2−K−AE=4A\cos^4\frac{\alpha}{2}-L\cos\frac{\alpha}{2}-K-A0<cos⁡α2<10<\cos\frac{\alpha}{2}<1Sp12_{12}
8ABcos⁡3α−4B(K+B)cos⁡2α+L2=08AB\cos^3\alpha-4B(K+B)\cos^2\alpha+L^2=0E=Acos⁡2α+L22Bcos⁡α+L24B−BE=A\cos^2\alpha+\frac{L^2}{2B\cos\alpha}+\frac{L^2}{4B}-B0<α≤arccos⁡L2+L4+32B2L216B20<\alpha\le\arccos\frac{L^2+\sqrt{L^4+32B^2L^2}}{16B^2}Sp21_{21}
β\beta is determined from Eq. (16), γ\gamma is determined from Eqs. (6)or arccos⁡L2−L4+32B2L216B2≤α<π\arccos\frac{L^2-\sqrt{L^4+32B^2L^2}}{16B^2}\le\alpha<\piSp22_{22}
cos⁡α=K2A\cos\alpha=\frac{K}{2A}, cos⁡β=L2B\cos\beta=\frac{L}{2B}, γ=β\gamma=\betaE=K24A+L22BE=\frac{K^2}{4A}+\frac{L^2}{2B}∣K∣<2∣A∣|K|<2|A|, ∣L∣<2∣B∣|L|<2|B|Noncoplanar

Table 1. Ground-state configurations for the classical bilinear-biquadratic Heisenberg model on one isosceles triangle and their energies.

It has the solution                                                                                                                             at the condition

cos⁡β=L(1+cos⁡α)4Bcos⁡α±(cos⁡α−1)[L2(1+cos⁡α)−8B2cos⁡2α]4Bcos⁡α(16)\cos\beta=\frac{L(1+\cos\alpha)}{4B\cos\alpha} \pm\frac{\sqrt{(\cos\alpha-1)\left[L^2(1+\cos\alpha)-8B^2\cos^2\alpha\right]}}{4B\cos\alpha} \tag*{(16)}
8B2cos⁡2α−L2cos⁡α−L2≥0,(17)8B^2\cos^2\alpha-L^2\cos\alpha-L^2\ge0, \tag*{(17)}

that is equivalent to the conditions

0<α≤arccos⁡L2+L4+32B2L216B2,∣L∣≤2∣B∣;0 < \alpha\le\arccos\frac{L^2+\sqrt{L^4+32B^2L^2}}{16B^2}, \qquad|L| \le2|B|;

\arccos \frac{L^2-\sqrt{L^4+32B^2L^2}}{16B^2} \le \alpha < \pi.

{#eq:18}

For all the faces, the expression for the energy can be written in the following form:

E=Kcos⁡α−Acos⁡2α+L(cos⁡β+cos⁡(α±β))−B(cos⁡2β+cos⁡2(α±β))=Kcos⁡α−Acos⁡2α+2Lcos⁡α2cos⁡(α2±β)−B[1+cos⁡α(2cos⁡2(α2±β)−1)].(18)\begin{aligned} E &= K\cos\alpha-A\cos^2\alpha+L\left(\cos\beta+\cos(\alpha\pm\beta)\right) \\ &\quad-B\left(\cos^2\beta+\cos^2(\alpha\pm\beta)\right) \\ &= K\cos\alpha-A\cos^2\alpha+2L\cos\frac{\alpha}{2}\cos\left(\frac{\alpha}{2}\pm\beta\right) \\ &\quad-B\left[1+\cos\alpha\left(2\cos^2\left(\frac{\alpha}{2}\pm\beta\right)-1\right)\right]. \tag*{(18)} \end{aligned}

Equation (12) yields

cos⁡(α2±β)=Lcos⁡α22Bcos⁡α.(19)\cos\left(\frac{\alpha}{2}\pm\beta\right)=\frac{L\cos\frac{\alpha}{2}}{2B\cos\alpha}. \tag*{(19)}

Using this equation, the expression (18) can be written as

E=−Acos⁡2α+Kcos⁡α+B(cos⁡α−1)+L2(cos⁡α+1)4Bcos⁡α.(20)E=-A\cos^2\alpha+K\cos\alpha+B(\cos\alpha-1)+\frac{L^2(\cos\alpha+1)}{4B\cos\alpha}. \tag*{(20)}

Finding the derivative of this function with respect to α\alpha,

∂E∂α=sin⁡α(8ABcos⁡3α−4B(B+K)cos⁡2α+L2)4Bcos⁡2α,(21)\frac{\partial E}{\partial\alpha}=\frac{\sin\alpha\left(8AB\cos^3\alpha-4B(B+K)\cos^2\alpha+L^2\right)}{4B\cos^2\alpha}, \tag*{(21)}

we obtain the equation

8ABcos⁡3α−4B(B+K)cos⁡2α+L2=0.(22)8AB\cos^3\alpha-4B(B+K)\cos^2\alpha+L^2=0. \tag*{(22)}

Taking into account this equation, we can write the expression for the energy in the following form:

E=Acos⁡2α+L22Bcos⁡α+L24B−B.(23)E=A\cos^2\alpha+\frac{L^2}{2B\cos\alpha}+\frac{L^2}{4B}-B. \tag*{(23)}

Equations (22) and (23) are the same for all the faces.

In a similar way we obtain the equations for α\alpha and the expressions for the energy for β=π2\beta=\frac{\pi}{2} and β=π−π2\beta=\pi-\frac{\pi}{2} [see Eqs. (12) and (13)].

All these results are presented in Table I. The first, second and third groups of rows correspond, respectively, to the vertices, edges, and faces of the tetrahedron show in Figure 1 [see also Eqs. (4)–(6)]. The last row corresponds to the interior points of the tetrahedron and has to be considered only for the Heisenberg model. In the first column of the Table I, three angles, α\alpha, β\beta, and γ\gamma, between three vectors are given. In the third group, cos⁡α\cos\alpha (or cos⁡α2\cos\frac{\alpha}{2}) is determined as a solution of a cubic equation.

In the second column, the energy (per one plaquette) of the spin structure is given as a function of the interaction parameters and the angle α\alpha. In the third column, necessary conditions for existence of the ground-state structure are indicated and, in the last column, the triangular plaquette spin structures are shown. Denotations F, AF, CF, and Sp mean “Ferromagnetic,” “Antiferromagnetic,” “Canted Ferromagnetic,” and “Spiral,” respectively.

It should be noted that all three spins can be rotated by the same angle in the 3D-space, since only the angles between pairs of spin matter. One can also swap the spins at the base of the isosceles triangle. The angles also do not change if all three spins are replaced by opposite ones. In the structures with two collinear spins (CF structures), one can rotate the third spin by an arbitrary angle with respect to the axis of the collinear ones.

Figure 4. Ground-state phase diagram for the XY model on one isosceles triangle at A=−1.0∣L∣A=-1.0|L|. Diagrams of these types arise when −496∣L∣<A<−12∣L∣-\frac{4}{9}\sqrt{6}|L|<A<-\frac{1}{2}|L|.

Using Table I one can construct the ground-state phase diagrams for the bilinear-biquadratic Heisenberg and XY models. We have constructed these diagrams in (K,B)(K,B)-plane. They are shown in Figure 2–Figure 6. It is sufficient to investigate the ground-state phase diagrams for L=1L=1 and L=0L=0 only. For L=−1L=-1 the diagrams are similar to those for L=1L=1. One just need to replace structure AF1 with structure F, structures CF21 and CF22 with structures CF11 and CF12, and structures Sp12 with structures Sp11 [see Table I and Figure 2(a) and Figure 6].

Figure 2. Ground-state phase diagrams for the bilinear-biquadratic Heisenberg and XY models on one isosceles triangle. (a) A=−0.2∣L∣A=-0.2|L| (Heisenberg model), (b) A=−0.2∣L∣A=-0.2|L| (XY model), (c) A=0A=0 (both models), (d) A=0.2∣L∣A=0.2|L| (both models). Black and red lines correspond to continuous and discontinuous transitions, respectively. Dotted lines correspond to A→+∞A \to+\infty. The diagram of type (a) for the Heisenberg model occurs if −496L≈−1.08866L<A<0-\frac{4}{9}\sqrt{6}L \approx-1.08866L < A < 0 and the diagram of type (b) for XY model occurs if −12<A<0-\frac{1}{2} < A < 0.

Figure 6. Ground-state phase diagram for the bilinear-biquadratic Heisenberg model on one isosceles triangle at A=−0.2A = -0.2, L=−1.0L = -1.0

The ground-state phase diagrams in the (K,B)(K,B)-plane for A>−496L≈−1.08866LA>-\frac{4}{9}\sqrt{6}L\approx-1.08866L (L>0L>0) are shown in Figure 2. In this case, there are four ground-state structures: collinear, AF1, AF2 (AF21 and AF22), coplanar, CF, Sp12, and, for the Heisenberg model at A<0A<0, a noncollinear. At the boundary of two phases the angles α\alpha, β\beta, and γ\gamma can change continuously or discontinuously. In Figure 2–Figure 6, lines of discontinuous transitions are shown with red color. Transitions between some phases can be both continuous and discontinuous, depending on values of parameters. In this case, there are tricritical points, where continuous and discontinuous transitions meet.

At A<−496LA<-\frac{4}{9}\sqrt{6}L (L>0L>0), in addition to the phases listed above, there is the phase Sp2 (Sp21 and Sp22). Between phases Sp2 and AF2 and between phases Sp2 and CF (for the XY model) the transition is continuous.

In the Heisenberg model, transitions between phases Sp2 and Sp12 are both continuous and discontinuous. Continuous transitions exist if A<−251032L≈−2.4705LA<-\frac{25\sqrt{10}}{32}L\approx-2.4705L (L>0L>0) (see Appendix). In this case, at the boundary of the phases, there are two tricritical points and the point of minimal value of the angle α\alpha for the phase Sp2 at fixed AA and LL [Figure 3(a)].

Figure 3. Ground-state phase diagram for the bilinear-biquadratic (a), (c) Heisenberg and (b), (d) XY models on one isosceles triangle at A=−3.0A=-3.0 and A=−2.0A=-2.0 (L=1.0L=1.0). Diagrams of these types occur when A<−496L≈−1.08866LA<-\frac{4}{9}\sqrt{6}L\approx-1.08866L (L>0L>0). Black and red lines correspond to continuous and discontinuous transitions, respectively. The tricritical points and the points of minimum α\alpha value for Sp2Sp_{2} structure are indicated with magenta color. At A=−3.0A=-3.0 and L=1.0L=1.0, α≈0.68673486445π\alpha\approx0.68673486445\pi [see Figs. (a) and (b)].

Phases CF1 and CF2 cannot have minimum energy in any region of the model parameter space. The phase CF3 can have minimum energy at L=0L=0 only (Figure 5).

Figure 5. Ground-state phase diagram for the bilinear-biquadratic (a), (c) Heisenberg and (b), (c) XY models on one isosceles triangle at L=0L = 0.

Global ground-state structures on zigzag ladder latticies

Using the ground-state structures and diagrams for one triangle, we can construct global ground-state structures for partially anisotropic triangular lattice. They are shown in Figure 7. Whatever the directions of the spins, the angles between spins should be the same for all the triangular plaquettes. Since each pair of spins belongs to two neighboring plaquettes on the lattice, the interaction parameters are equal to 2K2K, 2L2L, 2A2A, and 2B2B. For a distorted kagome lattice (Figure 8) the interaction parameters remain the same as for one triangle. We can also construct ground-state structures for three-dimensional zigzag ladder lattices on the base of planar bipartite lattices (see Figure 9 and Figure 10 as examples).

Figure 7. Ground-state structures for a partially anisotropic triangular lattice with a nearest-neighbor bilinear-biquadratic exchange. The interaction parameters are equal to 2K2K, 2L2L, 2A2A, and 2B2B (see Figure 1). (a), (b) Antiferromagnetic AF1_1 and AF2_2 structures. (c) Canted CF structures (β=γ=70∘\beta= \gamma= 70^\circ). All the spins are coplanar, however for the Heisenberg model they can be noncoplanar in general case. (d) Periodic spiral Sp12_{12} structure with α=0.8π\alpha= 0.8\pi, β=γ=π−α2=0.6π\beta= \gamma= \pi- \frac{\alpha}{2} = 0.6\pi. The magnetization is equal to zero. All Sp12_{12} structures are periodic in the direction perpendicular to the ladder legs with period of two ladders, but not necessarily periodic in the direction of the legs. (e) Incommensurate spiral Sp2_2 structure for K=3.5K = 3.5, L=1.0L = 1.0, A=−3.0A = -3.0, B=1.5B = 1.5. α≈142.06∘\alpha\approx142.06^\circ, β≈168.93∘\beta\approx168.93^\circ, γ≈26.87∘\gamma\approx26.87^\circ.

Figure 8. Distorted kagome lattice is composed with corner sharing isosceles triangles. This is not a zigzag ladder lattice. A part of ground-state spin configurations for this lattice can be obtained from anisotropic triangular lattice configurations by removing some of the lattice sites. The interaction parameters are equal to KK, LL, AA, and BB.

Figure 9. 3D honeycomb zigzag ladder lattice. The interaction parameters are equal to 3K3K, 2L2L, 3A3A, and 2B2B.

Figure 10. A 3D zigzag ladder lattice can be also constructed on the base of 4.824.8^2 Euclidean tiling. The interaction parameters are equal to 3K3K, 2L2L, 3A3A, and 2B2B.

The CF spin structure on one triangle gives rise to a highly degenerate global phase on zigzag ladder lattice (Figure 7(c)). All the spins on the same leg should have the same direction but only the angle between the spins on adjacent legs matters. Thus, giving a spin configuration on one leg of a zigzag ladder, there are two ways to construct the spin configuration for the other leg in the XY model and infinitely number of ways in the Heisenberg model. Noncollinear global CF structures for the Heisenberg model are possible and this leads to a high degeneracy. This degeneracy is not lifted on 3D lattices.

Let us consider the coplanar Sp2 spin structure. Giving a spin configuration on one leg of a zigzag ladder, there are in general two ways to construct the spin configuration on the other leg. This is shown in Figure 11. Thus, the global Sp2 spin structure is nontrivially degenerate. This degeneracy is lifted for 3D zigzag ladder lattices, for instance, the honeycomb zigzag ladder lattice (Figure 9) or a zigzag ladder lattice based on the 4.824.8^{2} (Figure 10), 444^{4} (square lattice), and 4.6.124.6.12 Euclidean tiling.

Figure 11. Two ways to construct an adjacent spin chain on the zigzag ladder. The interaction parameters are equal to KK, 2L2L, AA, 2B2B.

A distorted kagome lattice (Figure 8) is composed with corner sharing isosceles triangles although this is not a zigzag ladder lattice. A part of ground-state spin configurations for this lattice can be obtained from anisotropic triangular lattice configurations by removing every forth site. But some ground states of the kagome lattice are more degenerate, for instance CF structure. The interaction parameters are equal to KK, LL, AA, and BB.

All Sp12 structures are periodic in the direction perpendicular to the ladder legs with a period of two ladders, but not necessarily periodic in the direction of the legs. Thus, these structures generate ground-state structures for 3D zigzag ladder lattices as well.

The Sp2 phase is degenerate because the sequence of β\beta and γ\gamma that determines a ground-state structure [see the lower part of Figure 7(e) can be arbitrary. And only one sequence, ...β\beta, β\beta, β\beta,... (or ...γ\gamma, γ\gamma, γ\gamma, ...), gives a periodic structure with a period of two ladders and therefore generates a ground-state structure for 3D zigzag ladder lattices.

We do not consider here the noncoplanar ground-state phase. This was done for the isotropic triangular lattice in [5] and for the fully anisotropic triangular lattice in [10].

Conclusions

We have investigated an exact and complete ground-state phase diagram for the classical Heisenberg and XY models on various two- and three-dimensional zigzag ladder lattices with nearest-neighbor bilinear-biquadratic exchange. The problem is reduced to constructing ground-state phase diagram for one triangular plaquette. Nevertheless, two different biquadratic-exchange parameters together with two bilinear-exchange ones give rise to a complex ground-state phase diagram with collinear, coplanar, and noncoplanar spin structures. There are continuous and discontinuous transitions between phases. The most interesting result is the existence of two different spiral phases (Sp1_{1} and Sp2_{2}) between which there are both continuous and discontinuous transitions, and therefore, tricritical points at the boundary. The spiral phase Sp2_{2} on the anisotropic triangular lattice is nontrivially degenerate; the degeneracy is lifted on 3D lattices. The canted phase CF is highly degenerate on the anisotropic triangular lattice in the Heisenberg model and this degeneracy is not lifted on 3D lattices.

Acknowledgment

The author is gratefull for the hospitality of the Max Planck Institute for the Physics of Complex Systems where the work has been completed.

Appendix

We show here how to construct the ground-state phase diagrams for the model considered, that is we give equations for boundaries between different phases and for specific points on these boundaries. We also show for some triangular plaquette spin configurations that there are no regions in the parameter space where they have a minimum energy.

A < 0, L > 0 case

The equation for the boundary between AF2_2 and Sp12_{12} phases reads

B=−2Acos⁡4α2+L2cos⁡α2,K=−2A(cos⁡2α2−1)2+L(cos⁡2α2+1)2cos⁡α2.(24)\begin{aligned} B &= -2A\cos^4\frac{\alpha}{2}+\frac{L}{2}\cos\frac{\alpha}{2},\\ K &= -2A\left(\cos^2\frac{\alpha}{2}-1\right)^2+\frac{L\left(\cos^2\frac{\alpha}{2}+1\right)}{2\cos\frac{\alpha}{2}}. \tag*{(24)} \end{aligned}

Set of equations for the point where Sp12_{12}, Sp2_2, and CF phases meet (XY model)

16A2x6−48A2x4−8ALx3−6ALx+L2=0,B=3AL2x232A2x4+4ALx3+4ALx−L2,K=L24B+2A−B;x=cos⁡α2.(25)\begin{aligned} 16A^2x^6-48A^2x^4-8ALx^3-6ALx+L^2&=0,\\ B&=\frac{3AL^2x^2}{32A^2x^4+4ALx^3+4ALx-L^2},\\ K&=\frac{L^2}{4B}+2A-B;\qquad x=\cos\frac{\alpha}{2}. \tag*{(25)} \end{aligned}

The boundary between Sp12_{12} and Sp2_2 phases is determined from the following set of equations

8ABcos⁡3α−4B(K+B)cos⁡2α+L2=0,8Ax3+2(B−K−2A)x+L=0,Acos⁡2α+L22Bcos⁡α+L24B−B=4Ax4−Lx−A−K.(26)\begin{aligned} 8AB\cos^{3}\alpha-4B(K+B)\cos^{2}\alpha+L^{2}&=0,\\ 8Ax^{3}+2(B-K-2A)x+L&=0,\\ A\cos^{2}\alpha+\frac{L^{2}}{2B\cos\alpha}+\frac{L^{2}}{4B}-B&=4Ax^{4}-Lx-A-K. \tag*{(26)} \end{aligned}

The equation for the line of second order phase transition between Sp12\mathrm{Sp}_{12} and Sp2\mathrm{Sp}_{2} phases (XY model)

B=−Lx2(2x2−1),x=cos⁡α2,B=-\frac{Lx}{2(2x^{2}-1)},\qquad x=\cos\frac{\alpha}{2},
K=16Ax5−16Ax3+Lx2+4Ax−L2x(2x2−1).(27)K=\frac{16Ax^{5}-16Ax^{3}+Lx^{2}+4Ax-L}{2x(2x^{2}-1)}. \tag*{(27)}

Equations for the tricritical points read

cos⁡α=L24AB3,\cos\alpha=\sqrt[3]{\frac{L^{2}}{4AB}},
B=L24Acos⁡3α,K=3Acos⁡α−L24Acos⁡3α,B=\frac{L^{2}}{4A\cos^{3}\alpha},\qquad K=3A\cos\alpha-\frac{L^{2}}{4A\cos^{3}\alpha},
K=3AL24AB3−B.(28)K=3A\sqrt[3]{\frac{L^{2}}{4AB}}-B. \tag*{(28)}

Values of α\alpha for the tricritical points are solutions of the following equation

2A(2cos⁡2α2−1)2cos⁡α2+L=0.(29)2A\left(2\cos^{2}\frac{\alpha}{2}-1\right)^{2}\cos\frac{\alpha}{2}+L=0. \tag*{(29)}

If AL<−251032≈−2.4705\frac{A}{L}<-\frac{25\sqrt{10}}{32}\approx-2.4705 then this equation has three real positive roots less that 1 and therefore there are three tricritical points in the (K,B)(K,B)-plain for the XY model and two for the Heisenberg model.

The equation for the boundary between CF and Sp12\mathrm{Sp}_{12} reads

B=4Ax(x2−1)2−L(x2+1)4x+(x2−1)(4Ax(x+1)2−L)(4Ax(x−1)2−L)4x,B=\frac{4Ax(x^{2}-1)^{2}-L(x^{2}+1)}{4x} +\frac{(x^{2}-1)\sqrt{(4Ax(x+1)^{2}-L)(4Ax(x-1)^{2}-L)}}{4x},
K=8ABx4−2BLx−L24B;x=cos⁡α2.(30)K=\frac{8ABx^{4}-2BLx-L^{2}}{4B};\qquad x=\cos\frac{\alpha}{2}. \tag*{(30)}

The line of first order phase transition between Sp12\mathrm{Sp}_{12} and Sp2\mathrm{Sp}_{2} phases (XY model) is determined by the equation

16A2x6−16A2(y2+y+1)x4−8ALx3+4A2(−2y3−y2+2y+1)x2+2AL(y2−2y−2)x+L2=0,\begin{aligned} 16A^{2}x^{6}-16A^{2}(y^{2}+y+1)x^{4} -8ALx^{3}+4A^{2}(-2y^{3}-y^{2}+2y+1)x^{2}\\ +2AL(y^{2}-2y-2)x+L^{2}&=0, \end{aligned}
B=L2(y+2)x2y[8Ax5−8Ax3−2Lx2+2Ax(1−y2)−L],B=\frac{L^{2}(y+2)x}{2y[8Ax^{5}-8Ax^{3}-2Lx^{2}+2Ax(1-y^{2})-L]},
K=8Ax3−2(2A−B)x+L2x;0<x=cos⁡α12≤1,K=\frac{8Ax^{3}-2(2A-B)x+L}{2x};\qquad0<x=\cos\frac{\alpha_{1}}{2}\leq1,
y=cos⁡α2,−1<y≤L2−L4+32B2L216B2,y=\cos\alpha_{2},\qquad-1<y\leq\frac{L^{2}-\sqrt{L^{4}+32B^{2}L^{2}}}{16B^{2}},

or

L2+L4−32B2L216B2≤y<1.(31)\frac{L^{2}+\sqrt{L^{4}-32B^{2}L^{2}}}{16B^{2}}\leq y<1. \tag*{(31)}

The equation for the points where the phases Sp12\mathrm{Sp}_{12}, Sp2\mathrm{Sp}_{2}, and AF2\mathrm{AF}_{2} meet

16A2x6−16A2x4−8ALx3+2ALx+L2=0,16A^{2}x^{6}-16A^{2}x^{4}-8ALx^{3}+2ALx+L^{2}=0,
B=−2Ax4+L2x,B=-2Ax^{4}+\frac{L}{2}x,
K=−4Ax5−8Ax3−Lx2+4Ax−L2x.(32)K=-\frac{4Ax^{5}-8Ax^{3}-Lx^{2}+4Ax-L}{2x}. \tag*{(32)}

If A→−∞A\to-\infty, then K→1K\to1 and B→L−2AB\to L-2A (L>0L>0). At A=−496LA=-\frac{4}{9}\sqrt{6}L the first equation has two equal real solutions x=64x=\frac{\sqrt{6}}{4} (B=64LB=\frac{\sqrt{6}}{4}L, K=29636K=\frac{29\sqrt{6}}{36}). At −496L<A<0-\frac{4}{9}\sqrt{6}L<A<0 (L>0L>0) the ground state phase diagram is of the type shown in Figure 2(a).

The points where two first order phase transition lines meet (points of minimal value of α\alpha for Sp2\mathrm{Sp}_{2} phase) are determined from the following set of equations

4Ax[12x4−8(y2+y+1)x2−2y3−y2+2y+1]−L(12x2−y2+2y+2)=0,4Ax\left[12x^{4}-8(y^{2}+y+1)x^{2}-2y^{3}-y^{2}+2y+1\right]-L\left(12x^{2}-y^{2}+2y+2\right)=0,
A2[80x6−48(y2+y+1)x4−(8y3+4y2−8y−4)x2]−16ALx3−L2=0,x=cos⁡α12,y=cos⁡α2,A^{2}\left[80x^{6}-48(y^{2}+y+1)x^{4}-\left(8y^{3}+4y^{2}-8y-4\right)x^{2}\right]-16ALx^{3}-L^{2}=0,\quad x=\cos\frac{\alpha_{1}}{2},\quad y=\cos\alpha_{2},
B=L2x(y+2)2y[8Ax5−8Ax3−2Lx2−2A(y2−1)x−L],B=\frac{L^{2}x(y+2)}{2y\left[8Ax^{5}-8Ax^{3}-2Lx^{2}-2A\left(y^{2}-1\right)x-L\right]},
K=8Ax3−4Ax+2Bx+L2x.(33)K=\frac{8Ax^{3}-4Ax+2Bx+L}{2x}. \tag*{(33)}

At A=−496LA=-\frac{4}{9}\sqrt{6}L (L>0L>0), x=64x=\frac{\sqrt{6}}{4}, y=−1y=-1 and the region for Sp2\mathrm{Sp}_{2} is reduced to a point.

The equation for lines of fixed α\alpha for the phases Sp2\mathrm{Sp}_{2} reads

K=8ABcos⁡3α2−4B2cos⁡2α2+L24Bcos⁡2α2.(34)K=\frac{8AB\cos^{3}\frac{\alpha}{2}-4B^{2}\cos^{2}\frac{\alpha}{2}+L^{2}}{4B\cos^{2}\frac{\alpha}{2}}. \tag*{(34)}

Equation

8Acos⁡3α2+2(B−K−2A)cos⁡α2+L=0(35)8A\cos^{3}\frac{\alpha}{2}+2(B-K-2A)\cos\frac{\alpha}{2}+L=0 \tag*{(35)}

(for the phase Sp12\mathrm{Sp}_{12}) has two real roots 0<α<π0<\alpha<\pi for L>0L>0 if

27A+4(B−K−2A)3<0,27A+4(B-K-2A)^{3}<0,
A>0,10A+B−K>0,A>0,\quad10A+B-K>0,
4A+2B−2K+L>0.(36)4A+2B-2K+L>0. \tag*{(36)}

and one real root if

4A+2B−2K+L<0.(37)4A+2B-2K+L<0. \tag*{(37)}

If the phase with the energy E=−K−AE=-K-A exists, then

B<0,K>L+B,K>−L24B>0.(38)B<0,\quad K>L+B,\quad K>-\frac{L^{2}}{4B}>0. \tag*{(38)}

If A<0A<0, L>0L>0 this phase cannot exist in the region where Sp12\mathrm{Sp}_{12} phase is defined (4A+2B−2K+L<04A+2B-2K+L<0).

Let us prove that there is no minimum energy region for the phase CF1_1 (CF2_2). The energy difference between phases CF1_1 (CF2_2) and AF2_2 is equal to

(K±L)24(A+B)±L−B−(−K−A−2B)=(2A+2B+K±L)24(A+B).(39)\frac{(K\pm L)^{2}}{4(A+B)}\pm L-B-\left(-K-A-2B\right)=\frac{(2A+2B+K\pm L)^{2}}{4(A+B)}. \tag*{(39)}

This value is negative only when B<−AB<-A. But this is beyond the upper region where the phase CF1_1 (CF2_2) is defined (see Figure 12). Thus, this phase cannot have the minimum energy at B>−AB>-A.

Definition regions for the CF$_1$ and CF$_2$ phases

Figure 12. To the proof that the phases CF1_1 and CF2_2 cannot exist at A<0A<0. The definition regions for these phases (see Table I) are shown by blue and violet lines, respectively. The shaded area is the upper part of the definition region for the CF2_2 structure. The curves where energies of phases CF2_2 and Sp12\mathrm{Sp}_{12} (CF2_2 and CF) are equal are outside the region Sp12\mathrm{Sp}_{12} (CF).

At L>0L>0 the energy of the CF2_2 phase is less than the energy of the phase CF1_1 (in the region where both phases are defined).

A > 0, L > 0 case

The boundary between spiral Sp12\mathrm{Sp}_{12} and antiferro AF1_1 phases (a straight line) is determined by the following equations

4Ax3+8Ax2+4Ax−L=0,4Ax^{3}+8Ax^{2}+4Ax-L=0,
B=K−6Ax2−4Ax,x=cos⁡α2.(40)B=K-6Ax^{2}-4Ax,\quad x=\cos\frac{\alpha}{2}. \tag*{(40)}

If A→∞A\to\infty then B→K−LB\to K-L, α→π\alpha\to\pi.

References

References

  1. [1]R. Moessner and A. P. Ramirez. Geometrical frustration, Physics Today 59, 24 (2006).DOI
  2. [2]H. T. Diep, ed., Frustrated Spin Systems (World Scientific, Singapore, 2020) p. 752.
  3. [3]Y. I. Dublenych and O. A. Petrenko, An Ising model on a 3D honeycomb zigzag-ladder lattice: a solution to the ground-state problem, SciPost Phys. Core 5, 047 (2022).DOI
  4. [4]L. X. Hayden, T. A. Kaplan, and S. D. Mahanti, Frustrated Classical Heisenberg and XY Models in Two Dimensions with Nearest-Neighbor Biquadratic Exchange: Exact Solution for the Ground-State Phase Diagram, Phys. Rev. Lett. 105, 047203 (2010).
  5. [5]S. E. Korshunov, F. Mila, and K. Penc, Degeneracy and ordering of the noncoplanar phase of the classical bilinear-biquadratic Heisenberg model on the triangular lattice, Phys. Rev. B 85, 174420 (2012).DOI
  6. [6]S. Wenzel, S. E. Korshunov, K. Penc, and F. Mila, Zero-temperature Monte Carlo study of the noncoplanar phase of the classical bilinear-biquadratic Heisenberg model on the triangular lattice, Phys. Rev. B 88, 094404 (2013).DOI
  7. [7]A. Kartsev, M. Augustin, R. F. L. Evans, K. S. Novoselov, and E. J. G. Santos, Biquadratic exchange interactions in two-dimensional magnets, npj Computational Materials 6, 150 (2020).DOI
  8. [8]J. Y. Ni, X. Y. Li, D. Amoroso, X. He, J. S. Feng, E. J. Kan, S. Picozzi, and H. J. Xiang, Giant Biquadratic Exchange change in 2D Magnets and Its Role in Stabilizing Ferromagnetism of NiCl2 Monolayers, Phys. Rev. Lett. 127, 247204 (2021).
  9. [9]H. Kawamura and A. Yamamoto, Vortex-Induced Topological Transition of the Bilinear-Biquadratic Heisenberg Antiferromagnet on the Triangular Lattice, Journal of the Physical Society of Japan 76, 073704 (2007).DOI
  10. [10]Y. I. Dublenych, Ground states of a system of classical spins on an anisotropic triangular lattice and the spin-liquid problem in NiGa2S4 and FeGa2S4 compounds, Phys. Rev. B 96, 140401 (2017).DOI
  11. [11]Y. I. Dublenych, Incommensurate single-angle spiral ordering of classical Heisenberg spins on zigzag ladder lattices, Phys. Rev. B 93, 054415 (2016).DOI

Paper details

vlnc:W_8kx4mpda
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