Вы находитесь на странице: 1из 17

Spin origami with Weyl and Dirac line nodes in

distorted kagome antiferromagnets


Krishanu Roychowdhury1 , D. Zeb Rocklin1 , and Michael J.
Lawler1,2,3
arXiv:1705.00015v1 [cond-mat.str-el] 28 Apr 2017

1
Laboratory of Atomic And Solid State Physics, Cornell
University, Ithaca, NY 14853, USA.
2
Department of Physics, Binghamton University, Binghamton, NY,
13902, USA.
3
Kavli Institute for Theoretical Physics, University of California
Santa Barbara, CA 93106-4030

1 Introduction
Recent work has identified distinct topological states of systems at
the edge of mechanical stability, among them origami and kirigami
(cut origami) sheets [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12]. The zero-
energy modes of such a sheet are those which reorient the edges
of the sheet without stretching them or tearing any of the faces,
reminiscent of the zero modes that result from constrained spin dy-
namics in frustrated magnetic systems [13, 14, 15]. We present an
exact mapping from frustrated antiferromagnetic systems onto gen-
eral origami sheets and a criterion for when this mapping holds. This
allows a direct application of topological phonon band and rigidity
theories to the spin systems. Most immediately, flat origami sheets
have many zero modes, while crumpled origami acquires mechani-
cal rigidity but typically retains a reduced residual entropy. We argue
this entropy follows from a topological number that enables both me-
chanical and spin origami to have Weyl lines. We then identify two
experimental candidates for spin origami materials: Cs2 ZrCu3 F12 and
Cs2 CeCu3 F12 . The former has a flattenable origami structure while
the latter has a crumpled origami structure with a point group sym-
metry. Surprisingly, the additional symmetry promotes the Weyl line
nodes in Cs2 CeCu3 F12 to doubly degenerate Dirac line nodes.
Frustrated spin systems are endowed with a vast quantity of zero modes har-
boring various exotic states of matter with unusual magnetic responses, some
of which having intriguing topological imprints [16, 17]. Prominent examples
include classical spin liquid and spin ice systems which are characterized by a
Corresponding author.

1
large manifold of ground states an accidental degeneracy purely as a result of
frustration (and not protected by any symmetry) that precludes conventional
ordering and makes the zero modes distinct from the Goldstone modes associ-
ated with spontaneous symmetry breaking. We will add to these examples spin
origami systems whose ground state spin order resembles a folded sheet of paper.
Because of their unique low-energy physics, all of these systems have encour-
aged research focused in part on finding appropriate materials where such novel
phenomena can be detected in real experiments. Here we claim spin origami
will arise in Cs2 ZrCu3 F12 and Cs2 CeCu3 F12 .
Separately from frustrated magnets, mechanical lattice systems are now fu-
eling research into exotic properties associated with their zero modes. Remark-
ably, they possess an integer known as the topological polarization related to
the locations of these modes [5]. They are also Maxwell in that their number
of degrees of freedom, Nd.o.f. , and constraints, Nc , are equal and fall under a
special case of the Maxwell-Calladine topological index[5] that demands

Nd.o.f. Nc = Nz Nss = 0, (1)

where Nz is the number of zero modes and Nss is the number of self stresses
states consisting of redundant combinations of constraints. These systems
possess a rich range of behaviors owing to this topological polarization and
index [18, 19] when a change in these numbers requires the energy gap to close.
Many of them bear connections to phenomena in electronic band insulators and
other condensed matter systems, yet their precise analogs in solid state context
are hardly accessible.
Recently, the topological numbers associated with mechanical zero modes
were shown to also exist in spin systems [20]. Many kagome Heisenberg anti-
ferromagnets, for example, obey Eq. 1. This suggests solid state realizations
of topological mechanics are possible. Here we will focus primarily on realiza-
tions related to spin origami, known for decades to relate the low-temperature
spin dynamics of the ideal kagome Heisenberg antiferromagnet (KHAF) to the
folding of a triangulated origami sheet[13, 14]. But by virtue of the topologi-
cal physics we seek, these realizations need not rely on the ideal KHAF model.
Here we expand the notion of spin origami so that it applies to distorted KHAFs
with potential for a rich variety of origami ground states and identify topological
invariants and striking topological line nodes in spin wave dispersions.
For each vertex of a two-dimensional triangulated sheet embedded in a three-
dimensional space, there are three bond constraints and three spatial degrees
of freedom. So the sheet is a Maxwell lattice and its zero modes are associated
with either missing bonds or self stresses. The constraint associated with a
given triangles edge of length li and a unit orientation vector ei between the
triangle vertices located at ri1 , ri2 is ri1 ri2 = li ei . If one interprets the ei
as unit spinPvectors Si , it immediately implies a condition that for each trian-
gular face j4 li Si S4 = 0. These forms of constraints exist in geometri-
cally frustrated spin models in which the underlying lattice has corner-sharing
simplices [21]. In particular, on a kagome lattice the quadratic form in S4 ,
{x, y, z}
1 X 0 X
H= S4 J 4,4 S40 = Jij Si Sj + const. (2)
2
4,40 hi,ji

2
Figure 1: (a) A kagome Star of David with nonuniform interactions which
on the exterior bonds satisfy the condition: J1 J3 J5 J7 J9 J11 = J2 J4 J6 J8 J10 J12 ,
necessary for a generic spin system to possess an origami analog. The parameters
1 , , 6 are the scale factors explained in the methods section. (b) Mapping
from one constrained spin configuration on this star to an origami where the
spins in the former represent the edge vectors in the latter drawn in dotted lines.

0
is a Heisenberg model with ground states satisfying S = 0 so long as J ,
is a positive definite matrix. For models with only nearest neighbor (nn) in-
0 4,40
teractions, J 4,4 = J, yielding Jij = Jli lj (note for the ideal KHAF
all li = 1). For models with further neighbor interactions, the relationship of
Eq. 2 is harder to establish and so we focus on materials dominated by nearest
neighbor exchanges. Each ground state of such a system then corresponds to a
zero-energy configuration of an origami sheet (in general non-flattenable) whose
triangular faces have shapes determined by the coupling constants Jij .
Given a generic spin model defined by Eq. 2, i.e., involving an arbitrary
set of anisotropic nn interactions (Jij ) on the kagome lattice [see Fig. 1 (a)],
an origami analog does not exist unless an important condition is met by the
interactions. We call it the star condition since it applies for each of the kagome
Star of David structures, as shown in Fig. 1 (b). As described in the methods
section, the requirement that the six origami faces depicted possess well-defined
sizes as well as shapes yields the following requirement on the antiferromagnetic
coupling constants:

J1 J3 J5 J7 J9 J11 = J2 J4 J6 J8 J10 J12 . (3)

This condition can be achieved, for example, when the Hamiltonian possesses a
point group symmetry that relates the even Ji s to the odd Ji s such as one of
the mirror planes of the kagome lattice.
The ideal KHAF model trivially satisfies Eq. 3. It corresponds to a flatten-
able origami made of equilateral triangles. However, other kagome spin models
that include anisotropic interactions and satisfy the star condition also possess
origami analogs. Depending on the systems symmetries, these can be crum-
pled (i.e., non-flattenable), with distinct features in their zero-mode physics.
We illustrate this point by identifying two such materials that have distorted
kagome geometry, a Cs2 ZrCu3 F12 compound [22, 23] and a Cs2 CeCu3 F12 com-
pound [24], shown in the left panels of Fig. 2 (a) and (b) respectively. Both the
compounds satisfy the star condition of Eq. 3 because of the specific patterns

3
of the exchange couplings shown in Fig. 2. The q = 0 spin configuration in
the former corresponds to a flattenable origami consisting of isosceles triangles
(Fig. 2, (a) right panel) and features a flat band of spin wave zero modes like
the ideal KHAF. In contrast, in the latter compound (Fig. 2(b), right panel),
the origami surface develops vertices with finite curvatures and becomes non-
flattenable, lifting the flat band degeneracy. The crumpling corresponding to
the non-coplanar spin configurations in Cs2 CeCu3 F12 reduces the number of
zero modes from O(L2 ) to O(L) (for an L L system), thus having a crucial
impact on relieving frustration.
We can understand this effect on frustration by borrowing the concept of self
stresses from topological mechanics. In the mechanical analog of the flat spin
origami sheet, we can add tensions to the twelve edges of the six triangular faces
adjoining a given vertex while preserving mechanical equilibrium regardless of
the shapes of the coplanar faces. According to the index theorem given in
Eq. 1, these self stresses necessarily generate zero modes, which in this case
are displacements of individual vertices in the direction perpendicular to the
faces. For crumpled origami with non-coplanar edges, as in in the Cs2 CeCu3 F12
compound, many of these self stresses are no longer possible the rigidity of the
sheet has become fundamentally enhanced via its geometry in a process akin to
corrugation. This then has the effect of lifting the zero-energy band of phonons
(lattice vibrations) from the origami system and magnons from the analogous
spin system.
Because the Cs2 CeCu3 F12 origami has mirror and inversion symmetries,
pairs of adjacent triangular faces become coplanar, creating an origami whose
ground state consists effectively of diamond-shaped (rhombus) faces, as shown
in Fig. 2 and calculated in the methods section. Flat-foldable quadrilateral
mesh origami often possesses global folding mechanisms [25] enabled by sym-
metry. Similarly, our current non-flat-foldable origami sheet has, rather than a
unique ground state, a one-dimensional continuum connected by a folding an-
gle of its mirror-symmetric vertices, resembling the well-known Miura origami
fold [26]. This diamond pattern and folding mechanism, reminiscent of the
Guest modes observed in other Maxwell lattices [27, 28, 1, 29], reveal physics
of the Cs2 CeCu3 F12 distinct from systems with generic interactions.
The linear spin-wave theory of spin origami, which describes the magnon
band structures of small spin rotations about the ground state parameterized by
canonical variables xi = (q i , pi ), is encoded in the so-called rigidity matrix[5]

S
R,i = . (4)
xi
In terms of it, the Hamiltonian matrix governing the spin waves is HSW = RT R
and so if Det[R] = 0, a zero mode exists. Fourier transforming, this fact also
holds at each wave vector. But since R(k) is complex, so is its determinant in
general. So the winding number around any closed loop C
I
1
w(C) = d(arg Det[R(k)]), (5)
2 C

is protected under distortions of R(k). It either measures the circulation of


isolated Weyl point nodes C encloses [30, 10] or characterizes the topological
polarization if C is a non-contractible loop across the torus [5]. But remarkably

4
Figure 2: (a) Left: Distorted kagome lattice structure of Cs2 ZrCu3 F12 with
interactions that satisfy the star condition mentioned in the text. Right: The
origami corresponding to a q = 0 state is a flat surface consisting of isosceles
triangles. The dark blue and the purple faces correspond respectively to the
blue-black and red-blue triangles of the kagome lattice in the left panel. (b)
Left: Another compound, Cs2 CeCu3 F12 , with a distorted kagome structure and
interactions that satisfy the star condition. Right: The spin-origami for a q = 0
state in this material is a crumpled surface (i.e. non-coplanar spin configura-
tions) belonging to a one-dimensional family of zero-energy states. Different
members in the family can be identified by a continuous parameter b0 (see the
methods section), and the configuration for b0 = 0.06 is shown as a reference.
The blue and dark red origami faces correspond respectively to the blue-red and
red-black spin triangles in the left panel. The ground states of the system consist
of coplanar pairs of triangles outlined in green, functioning as diamond-shaped
origami faces.

5
Figure 3: (a) Weyl line nodes (thick red lines) appearing in the plot of the
topological number (Eq. 6) over the BZ of a generic KHAF corresponding to a
crumpled origami surface. The lines separate the two regions of positive (yellow)
and negative (blue) values of denoted by + and - respectively. The quantity
t0 produces a combined effect of stretching and shearing the origami sheet,
thus, tuning the coupling constants of the spin model (see the methods section
for details). A continuous deformation of the sheet by tuning t0 sequentially
leads to (b)(c)(d) in which first a pair of line nodes vanishes in (b) (at
the BZ boundary the vanishing takes place at the M points), then the - region
gradually shrinks down as in (c) till in (d), the nodes disappear altogether except
the point which remains always gapless owing to the global spin rotation
symmetry.

6
for spin origami we find Det[R(k)] is a real number up to an over all constant
phase in the Brillouin zone. It obeys the mysterious realness condition previ-
ously observed for the rigidity matrices of triangulated mechanical origami [12].
The winding number w(C) therefore vanishes for all C. After eliminating this
constant phase by choosing a gauge, however, this realness condition defines
another topological number:

(k) = sign Det[R(k)] (6)

that demands two regions in the BZ with different (k) are separated by a line of
zero modes the Weyl line nodes (see Fig. 3). Under appropriate tuning of the
coupling constants of the KHAF model in Eq. 2, the line nodes vanish pairwise
[Fig. 3 (a)-(b)] and in certain situations, can be decimated entirely [Fig. 3 (d)].
These considerations therefore group KHAFs by their topological numbers.
KHAFs with spin origami ground states are characterized by and typically
have Weyl line nodes. Other materials that have a square rigidity matrix
but with complex determinant are characterized by w and typically have Weyl
points. Those with further neighbor exchange interactions have negative Maxwell-
Calladine indices (Eq. 1) and non-square rigidity matrices and are hence char-
acterized by neither of these numbers. These topological numbers therefore
govern the spin wave zero modes of these systems and so also the degree of their
frustration.
The situation is even more intriguing for Cs2 CeCu3 F12 . As shown in Fig.
4, is found to have the same sign throughout the Brillouin zone (BZ) but
line nodes still appear. Remarkably they are doubly degenerate. Such Dirac
line nodes separate regions of the same . Existence of these type of line nodes
have been previously reported in certain topological insulators (see Ref.[31] and
references therein), however, not in magnetic systems. Note: the use of the
phrase Dirac line node here is not related to any linearity features of the
dispersion across the node, just its doubly degenerate character and topological
protection. In fact we find the magnon dispersions observed in the material
Cs2 CeCu3 F12 to be quadratic ( k 2 ) unlike the line nodes in Ref.[31] that
feature linear dispersions ( k).
It turns out, the Dirac line nodes follow directly from Cs2 CeCu3 F12 s non-
trivial point group symmetry. It is this symmetry that forms pairs of the tri-
angular origami faces into diamond-shaped faces as shown in Fig. 2 (b). As
explained in the methods section, it is Z2 Z2 and include anti-unitary ele-
ments due to the complex conjugation K symmetry of the parameters of the
spin model. Writing the rigidity matrix in the basis that diagonalizes both a
unitary and antiunitary symmetry of this point group, we find it is real and takes
the block-diagonal form with two blocks R+ (k) and R (k). Since the matrices
are real, we then have two topological invariants + = sign Det[R+ (k)] and
= sign Det[R (k)] with = + . The symmetry also demands does not
change so long as the symmetry is preserved and that both + and change
sign across the line node. Hence, the line node is doubly degenerate with one
zero mode associated with the changing of each topological number.
The Dirac lines in Cs2 CeCu3 F12 survive over a broad range of distortion
made to the interactions J1 , J2 , J3 and J4 in Cs2 CeCu3 F12 , although their
locations differ. This depends also on the parameter b0 that characterizes dif-
ferent members of the family as seen in Fig. 4 (a) and (b) (and discussed in the

7
Figure 4: (a) and (b) are the two members of the one dimensional fam-
ily of origami configurations pertinent to the KHAF model in Cs2 CeCu3 F12
parametrized by the folding parameter b0 (see the methods section), and both
featuring Dirac line nodes separating zones of different values of the topological
invariant + (yellow and blue correspond to + and - respectively). The insets
of (a) and (b) are the plots of Det[R+ ] over a circle in the BZ shown in dotted
line. The locations of the lines are decided by the condition Det[R(k)] = 0 and
depend on b0 . Like Fig. 3 here also the parameter t0 , when set away from 1,
provides a combined effect of stretching and shearing the origami sheet mim-
icking distortions to the compound by means of external strain or pressure (see
the methods section for details). Under such deformations, each Dirac line
splits into two Weyl lines, as shown in (c), in presence of which, the topological
invariant in Eq. 6 changes sign. The inset of (c) is the plot of Det[R] over a
circle in the BZ. Some of the lowest magnon frequencies (k) in Cs2 CeCu3 F12
are shown in (d) as plotted along a high-symmetry path in the BZ shown in the
inset and corresponding to b0 = 0.06.

8
methods section). However, allowing for perturbations that break the point
group symmetry would split each of the Dirac lines into two Weyl lines as shown
in Fig. 4 (c), and generically lead to the situations portrayed in Fig. 3. Fig. 4
(d) displays some of the lowest magnon frequencies plotted over high symmetry
points in the BZ for the KHAF in this material. It shows doubly degenerate
bands like those appearing in inversion and time reversal symmetric electronic
band structures.
The above discoveries are ultimately a manifestation of geometric frustra-
tion physics that naturally arises in materials. Like other forms of geometric
frustration, it (essentially) depends on short distance (i.e. nearest neighbor)
spin exchanges dominating over longer distance (second or third neighbor) spin
exchanges. The flat band or lines of zero modes are likely gapped by these small
perturbations. Reducing this gap is then an engineering problem and the gap
will characterize the quality of a given material. Probing it on energy scales
above this gap, however, should reveal all the predicted phenomena discussed
above.
Thus we argue spin origami can exist in solid state systems and that by
identifying real materials associated with it, surprises lurk. Specifically, the
star condition efficiently enables this search and leads us to two materials:
Cs2 ZrCu3 F12 and Cs2 CeCu3 F12 . They are each associated with qualitatively
different origami analogs: the former flattenable origami and the latter crumpled
origami. Both share with the origami system the condition of realness, and the
topological number that typically leads to curved lines of zero modes. But for
the compound Cs2 CeCu3 F12 with a remarkable point group symmetry, the line
nodes are found to be of Dirac type in that they are doubly degenerate. These
a novel phenomena should be readily detected in the magnon spectrum of these
materials provided it is dominated by their nearest neighbor spin exchanges.
Acknowledgements We thank Tom C. Lubensky, James P. Sethna, and
Itai Cohen for useful discussion. MJL acknowledges supported in part by the
National Science Foundation under Grant No. NSF PHY-1125915. DZR grate-
fully acknowledges support from the the ICAM postdoctoral fellowship, the
Bethe/KIC Fellowship, and the National Science Foundation Grant No. NSF
DMR-1308089.

2 Methods
Derivation of the star condition: We begin our derivation of the star con-
dition with the following observation. Generic nearest neighbor KHAF models
can be described by the Hamiltonian

J X 2
H= S4 + const. ,
2
4

and
P characterized by the winding number w given in Eq. 5. Here S4 =
i a4,i Si with a4,i depending on the triangle 4. This observations follows by
writing H back in the traditional form of the spin model with spin exchanges:
(
4,i a4,j , if i, j 4
Ja
Jij =
0, otherwise

9
and by observing that the map between Jij and the set of parameters J, a,i is
one-to-one. Now consider two adjacent triangles of the kagome Star of David
shown in Fig. 1 (b). These triangles are associated with the constraints:
S4 = a4,2 S2 + a4,3 S3 + a4,11 S11 = 0,
S40 = a40 ,3 S3 + a40 ,4 S4 + a40 ,10 S10 = 0.
Following the notion of Jij mentioned above for spin S3 , whichp is common
to both pthe constraints, the weights are respectively a4,3 = J16 J3 /JJ4 and
a40 ,3 = J17 J2 /JJ1 .
Construction of an origami immediately follows when a4,3 = a40 ,3 , for which
S3 would represent a unit edge vector with the edge length l3 as it does in Eq. 2.
If not, one can assign scale factors 1 and 2 to the adjacent triangles 4 and
40 respectively [see Fig. 1 (a)] such that the weights a4,3 and a40 ,3 , when
multiplied by these scale factors appropriately, become equal:
1 a4,3 = 2 a40 ,3 l3 . (7)
These factors merely rescale the triangles and do not alter the constraints. We
refer to Eq. 7 as a length-consistency condition which can be continued for
all five sites that constitute the hexagon associated with the kagome Star of
David shape, assigning different scale factors for different triangles as shown
in Fig. 1 (a). For the remaining site, however, we run out of freedom to rescale
and must impose a condition on the Jij s for Eq. 7 to hold for all sites. We call
this condition the star condition presented in Eq. 3. It ensures closure of the
origami hexagon comprising six adjacent triangles joined at a common vertex.
As such, when this condition is satisfied for each Star of David of the spin
system, its ground state configurations may be mapped onto the ground state
configurations of a particular crumpled triangulated origami sheet.

The rigidity matrix and magnon dispersions: The crumpled origami sheet
to which the ground state of KHAF in Cs2 CeCu3 F12 corresponds has three dif-
ferent edge lengths la , lb and lc [see Fig. 5 (a)] that are related to the interactions
J1 , J2 , J3 and J4 as la /lb = J3 /J2 and lc /lb = J4 /J1 . The origami unit cell
contains four distinct vertices characterized by four basis vectors b1 , b2 , b3 and
b4 and two lattice vectors L1 and L2 [see Fig. 5 (a)]. The unit vectors along the
edges are the spins given in Fig. 5 (b) whose orientations are obtained through
the following procedure.
One can readily show that, for a given basis, the requirements that edge
from b1 to b4 is equal
The vertices at b1 , b2 , and b4 (and spins S9 , S11 and S7 ) form an isosceles
triangle [triangle 1 in Fig. 5 (b)] which can bep defined entirely on a plane with
b1 = (0, 0, 0), b2 = (la , 0, 0) and b4 = (la /2, lb2 la2 /4, 0). Now the origami
configuration demands the following set of equations
||b3 b4 || = lc
||b3 b2 || = lb
||b1 + L1 b2 || = la
||b1 + L1 b3 || = lb
||b4 + L1 b3 || = lc

10
Figure 5: (a) The origami unit cell with edge lengths and vertices specified.
(b) The spin assignment with arrows denoting the orientation associated with
the definition of the spin vector positive along the direction of the arrow (e.g.
S11 = (b2 b1 )/||b2 b1 || with ||b2 b1 || = la ).

to solve for b3 and L1 , which together have six variables (x, y and z component
of each). This leads us to define a tuning parameter bz3 = b0 (in Fig. 4) in a
range [0.12, 0.12] to allow for real solutions, thus, defining a one-dimensional
family of configurations for the crumpled sheet as mentioned in the text. We
further notice that setting L2 = b3 + b4 b1 b2 satisfies all other constraints
on the sheet and lead to the symmetries associated with the spin configuration
resulting in diamond-shaped faces of the origami noted in the main text. For a
given b0 with lb chosen to be 1, spin configuration of the ground state is then
uniquely specified through the vectors b1 , b2 , b3 , b4 , L1 , and L2 . Distortions
can be made to the sheet by introducing a parameter t0 to modify the definition
of L2 as L2 = b3 + b4 b1 t0 b2 . When t0 is set away from 1, it produces a
combined effects of shearing and stretching the sheet as noted in Fig. 3.
Since we are dealing with spins as unit vectors, each spin can be parametrized
by two canonical variables, xj and pj as
q q
Sj = cos(q j ) 1 p2j , sin(q j ) 1 p2j , pj ,


with the Poisson bracket {q i , pj } = ji . The rigidity matrix R is defined by


linearizing the constraints about the ground state as S4 = R4,j xj , where
j 4 and xj (q j , pj ).
An explicit construction of R follows by considering the basis 1 = [q0 , q1 , ,
q11 , p0 , p1 , , p11 ]T corresponding to the twelve spins S0 , S1 , , S11 in the
unit cell and 2 = [4x0 , , 4x7 , 4y0 , , 4y7 , 4z0 , , 4z7 ]T corresponding to
the eight faces in the unit cell shown in Fig. 5 (b) such that 2 = R 1 . We
present this explicit construction in the form of a python script as a supplemen-
tal material.
To translate to the momentum space, one needs to consider a Fourier trans-
formation of R. We chose the convention
1 X 0
R4,j = RD,d (k) ek(Ri Ri ) , (8)
N k

11
Figure 6: Plotting spin ground state pattern to reveal symmetries. Left: the
x-component of the spin vectors plotted on the kagome lattice. This reveals the
C2site symmetry about sites 0, 3, 8 and 11. Right: a sphere plot of the spins
within the unit cell consisting of placing the tail of each spin at the origin of a 3
dimensional Euclidean space. This reveals a mirror symmetry M in spin space.

where N is the number of sites in a unit cell; Ri and R0i denote the position
vectors of the unit cells which contain the triangle 4 at Ri + D and the site
j at R0i + d respectively. Our particular choice of unit cells is shown in Fig. 5
(b). The sites associated with phases z1 = ek1 a , z2 = ek2 a , z11 , z21 lying in
neighboring unit cells. Here k1 a = k T1 , k2 a = k T2 , T1 and T2 are Bravais
lattice vectors for the ordering pattern. With this choice of Fourier transform
convention and unit cells, we find that Det[R(k)] is real everywhere in the BZ.
Dispersions of the spin waves (the magnon frequencies) are associated with
0 0
the equation of motion of xi . Considering J 4,4 = 4,4 for models with
only nn interactions these equations read
0
xi = i,j RTj,4 4,4 R40 ,k xk x = RT Rx, (9)

where i,j = {xi , xj } = ij  is the Poisson bracket tensor with  the


two-dimensional Levi-Civita tensor. Using the momentum space representation
following Eq. 8, the magnon frequencies (k) are calculated [and plotted in
Fig. 4 (d)] by diagonalizing the matrix RT (k)R(k), cf. Eq. 9.

Symmetry analysis of Cs2 CeCu3 F12 ground state: The unexpected Dirac
line nodes (doubly degenerate line nodes) appear together with a point group
symmetry in our analysis of Cs2 CeCu3 F12 . The generic crumpled origami pre-
sented in Fig. 3 of the main text has singly degenerate line nodes and no point
group symmetry. So it is natural to expect the double degeneracy follows from
the added symmetry. Here we prove that this is the case.
To reveal the symmetry of Cs2 CeCu3 F12 s periodic crumpled origami sheet,
we have plotted in Fig. 6, one component of its spin structure in real space
as well constructed a sphere plot of the 12 spins in its unit cell. The sphere
plot places the tail of each spin at the origin of three dimensional Euclidean
space and illuminates the possibility of a symmetry in spin space that might be
combined with a space group transformation.
The plots show the following symmetries exist:

12
A 180o C2site spatial rotation symmetry about four of the sites within the
unit cell that sends r R+d r. These are all the same transformation
up to translations (i.e. the same transformation within the point group).

A mirror symmetry M in spin space followed by either a second C2hex


1/2
rotation about the center of a hexagon or a translation T1 by half the
unit cell of the ordering pattern.
1/2
In total, the three transformations C2site , M C2hex and M T1 together with the
identity make the group Z2 Z2 . Because all variables are real in real space,
there is also a complex-conjugation symmetry K as pointed out in Ref. [5] of
the main text. This additional symmetry plays a similar role as time reversal
symmetry in magnetic space groups. It can be combined with any of the above
four unitary symmetries. Hence there are eight elements to the point group.
In k-space, the symmetry becomes more interesting. Remarkably, four of
the eight elements of the preceding symmetry group transform a given point
in k space into itself. This is because C2 sends k k and so does K. The
symmetry group at a k point is then
1/2
G(k) = {e, KC2site (k), KM C2hex , M T1 },

which is again Z2 Z2 .
The consequences of this symmetry are the following:

1. The physical frequencies have doubly degenerate bands and a line node
must have double degeneracy.
2. The rigidity matrix R(k) can be made real and placed in a block diagonal
1/2
form with two blocks R+ (k) and R (k), one for M T1 = +1 and one
1/2
for M T1 = 1.
3. Either + sign Det[R+ (k)] or sign Det[R (k)] serves as a topo-
logical invariant that demands the gap closes if it changes. They must
change sign at the Dirac line nodes.

4. The previous topology characterized by the topological invariant is ren-


dered trivial for = + never changes sign in the BZ and so never
changes so long as this symmetry is preserved.
The remaining parts of this section prove these consequences.
To prove that the physical frequencies are doubly degenerate we start with
the following observation. If the mirror transformation M in the spin space is
chosen to be the xy-plane, then when acting on the spin deviations x(k) it takes
the form of a z Pauli matrix. Since the Poisson bracket tensor in this space is
= y we discover
1/2
{KM C2hex , RT (k)R(k)} = 0 ; {M T1 , RT (k)R(k)} = 0.

Namely, transformations involving the mirror symmetry M in the spin space


anti-commute with the matrix entering the eigenvalue problem for the physical
frequencies. This in turn implies the eigenvalues of this matrix come in pairs
with opposite signs since if v is an eigenvector of RT (k)R(k) with eigenvalue

13
1/2
then M T1 v is an eigenvector with eigenvalue . But the physical
frequency this eigenmode corresponds to is still the same positive number .
Hence the bands are doubly degenerate.
To prove that the rigidity matrix takes a block diagonal form, we need to
transform to the basis where the representation of the group that acts on R(k)
in k-space is irreducible. We carry this out in two steps. First we diagonalize
1/2
matrices corresponding to the M T1 transformation. Since the square of this
symmetry is the identity, it has eigenvalues 1 as noted above. Thus in the
basis that diagonalizes this matrix, R(k) forms two blocks. Next we transform
from this basis to a new basis which diagonalizes the antiunitary transforma-
tion KC2site (k). Since the square of this symmetry is also the identity, we can
transform to a basis where R(k) is real. Because the symmetry is antiunitary,
however, we can find a complete basis where all eigenvectors have eigenvalue
KC2site (k) = +1. We do so as follows. Starting with one eigenvector v1 of
1/2
M T1 we create w1 = v1 + KC2site (k) v1 if this is non-zero otherwise we set
w1 = v1 . We then take another eigenvector v2 and create w2 in a similar way.
But we further apply Gram-Schmidt to w2 following Wigner [32] to ensure it is
1/2
orthogonal to w1 . We then continue with each eigenvector v in M T1 . In the
1/2
end, the procedure produces a basis {wi } which has eigenvalues M T1 = 1
site
and KC2 (k) = 1. Since there are no further symmetries to use because
KM C2hex is just the product of these two symmetries, this is the eigenbasis of
the symmetry group at a k point.
To define a continuous basis throughout the BZ, one needs to adopt a smooth
gauge of parallel transport by comparing this basis to the one similarly produced
at a nearby k-point k0 . The procedure involves the non-abelian form of the Berry
connection, specifically the overlap matrix between the two bases at adjacent
points k and k0 as Bij = vi (k) vj (k0 ) [33]. One then computes the singular
value decomposition B = U V and carries out a unitary transformation of
the basis at k0 as X
vi (k0 ) = (V U )ji vj (k0 )
j

1/2 1/2
This is done separately for the eigenvalue M T1 = 1 and M T1 = 1 bases.
The resulting matrix B 0 = U U is Hermitian and thus defines a parallel
transport of the vectors and a continuous basis in the BZ as required.
Finally, the antisymmetry and the block-diagonal form of R additionally
explains the double degeneracy of the Dirac line nodes, the behavior of the new
topological invariants + , and the behavior of the old topological invariant .
Consider an eigenvector v of RT (k)R(k) with eigenvalue . The eigenvector
1/2
M T1 v has eigenvalue by the antisymmetry. This further implies in this
1/2
subspace of two vectors that the transformation M T1 = x , where x is the
usual Pauli matrix. But this matrix has two eigenvalues 1 and so in the
basis that places R(k) in block diagonal form and makes it real, one zero mode
1/2 1/2
belongs to the M T1 = 1 sector and one belongs to the M T1 = 1 sector.
Hence, as passes through zero at a line node, both + and change sign
and their product always remains the same throughout the BZ.

14
References
[1] Kai Sun, Anton Souslov, Xiaoming Mao, and TC Lubensky. Surface
phonons, elastic response, and conformal invariance in twisted kagome lat-
tices. Proceedings of the National Academy of Sciences, 109(31):12369
12374, 2012.
[2] Sahab Babaee, Jongmin Shim, James C Weaver, Elizabeth R Chen, Nikita
Patel, and Katia Bertoldi. 3d soft metamaterials with negative poissons
ratio. Advanced Materials, 25(36):50445049, 2013.
[3] Mark Schenk and Simon D Guest. Geometry of miura-folded metamate-
rials. Proceedings of the National Academy of Sciences, 110(9):32763281,
2013.
[4] Zhiyan Y Wei, Zengcai V Guo, Levi Dudte, Haiyi Y Liang, and L Mahade-
van. Geometric mechanics of periodic pleated origami. Physical review
letters, 110(21):215501, 2013.

[5] CL Kane and TC Lubensky. Topological boundary modes in isostatic lat-


tices. Nature Physics, 10(1):3945, 2014.
[6] Sicong Shan, Sung H Kang, Pai Wang, Cangyu Qu, Samuel Shian, Eliza-
beth R Chen, and Katia Bertoldi. Harnessing multiple folding mechanisms
in soft periodic structures for tunable control of elastic waves. Advanced
Functional Materials, 24(31):49354942, 2014.
[7] Jayson Paulose, Anne S Meeussen, and Vincenzo Vitelli. Selective buckling
via states of self-stress in topological metamaterials. Proceedings of the
National Academy of Sciences, 112(25):76397644, 2015.

[8] D Rocklin, Shangnan Zhou, Kai Sun, and Xiaoming Mao. Transformable
topological mechanical metamaterials. arXiv preprint arXiv:1510.06389,
2015.
[9] TC Lubensky, CL Kane, Xiaoming Mao, A Souslov, and Kai Sun. Phonons
and elasticity in critically coordinated lattices. Reports on Progress in
Physics, 78(7):073901, 2015.
[10] D Zeb Rocklin, Bryan Gin-ge Chen, Martin Falk, Vincenzo Vitelli, and
TC Lubensky. Mechanical weyl modes in topological maxwell lattices.
Physical review letters, 116(13):135503, 2016.
[11] Hamed Abbaszadeh, Anton Souslov, Jayson Paulose, Henning Schomerus,
and Vincenzo Vitelli. Sonic landau-level lasing and synthetic gauge fields
in mechanical metamaterials. arXiv preprint arXiv:1610.06406, 2016.
[12] Bryan Gin-ge Chen, Bin Liu, Arthur A Evans, Jayson Paulose, Itai Cohen,
Vincenzo Vitelli, and CD Santangelo. Topological mechanics of origami
and kirigami. Physical review letters, 116(13):135501, 2016.

[13] EF Shender, VB Cherepanov, PCW Holdsworth, and AJ Berlinsky.


Kagome antiferromagnet with defects: Satisfaction, frustration, and spin
folding in a random spin system. Physical review letters, 70(24):3812, 1993.

15
[14] P Chandra, P Coleman, and I Ritchey. The anisotropic kagome antifer-
romagnet: a topological spin glass? Journal de Physique I, 3(2):591610,
1993.

[15] Michael J Lawler. Emergent gauge dynamics of highly frustrated magnets.


New Journal of Physics, 15(4):043043, 2013.
[16] Claudine Lacroix, Philippe Mendels, and Frederic Mila. Introduction
to Frustrated Magnetism: Materials, Experiments, Theory, volume 164.
Springer Science & Business Media, 2011.
[17] HT Diep. Frustrated spin systems. World Scientific, 2013.
[18] Emil Prodan and Camelia Prodan. Topological phonon modes and
their role in dynamic instability of microtubules. Physical review letters,
103(24):248101, 2009.

[19] Vincenzo Vitelli. Topological soft matter: Kagome lattices with a twist.
Proceedings of the National Academy of Sciences, 109(31):1226612267,
2012.
[20] Michael J Lawler. Supersymmetry protected topological phases of isostatic
lattices and kagome antiferromagnets. Physical Review B, 94(16):165101,
2016.
[21] R Moessner and JT Chalker. Low-temperature properties of classical ge-
ometrically frustrated antiferromagnets. Physical Review B, 58(18):12049,
1998.

[22] T Ono, K Morita, M Yano, H Tanaka, K Fujii, H Uekusa, Y Narumi,


and Koichi Kindo. Magnetic susceptibilities in a family of s= 1 2 kagome
antiferromagnets. Physical Review B, 79(17):174407, 2009.
[23] Lewis James Downie, Cameron Black, EI Ardashnikova, CC Tang,
AN Vasiliev, AN Golovanov, PS Berdonosov, VA Dolgikh, and Philip Light-
foot. Structural phase transitions in the kagome lattice based materials cs 2-
x rb x sncu 3 f 12 (x= 0, 0.5, 1.0, 1.5). CrystEngComm, 16(32):74197425,
2014.
[24] T Amemiya, M Yano, K Morita, I Umegaki, T Ono, H Tanaka, K Fujii,
and H Uekusa. Partial ferromagnetic ordering and indirect exchange inter-
action in the spatially anisotropic kagome antiferromagnet cs 2 cu 3 cef 12.
Physical Review B, 80(10):100406, 2009.
[25] Thomas A Evans, Robert J Lang, Spencer P Magleby, and Larry L How-
ell. Rigidly foldable origami gadgets and tessellations. Royal Society open
science, 2(9):150067, 2015.

[26] Koryo Miura. Method of packaging and deployment of large membranes in


space. title The Institute of Space and Astronautical Science report, 618:1,
1985.
[27] SD Guest and JW Hutchinson. On the determinacy of repetitive structures.
Journal of the Mechanics and Physics of Solids, 51(3):383391, 2003.

16
[28] RG Hutchinson and NA Fleck. The structural performance of the periodic
truss. Journal of the Mechanics and Physics of Solids, 54(4):756782, 2006.

[29] D. Zeb Rocklin, Shangnan Zhou, Kai Sun, and Xiaoming Mao. Trans-
formable topological mechanical metamaterials. Nature Communications,
8(14201):19, 2017.
[30] Hoi Chun Po, Yasaman Bahri, and Ashvin Vishwanath. Phonon analog of
topological nodal semimetals. Phys. Rev. B, 93:205158, May 2016.

[31] Youngkuk Kim, Benjamin J Wieder, CL Kane, and Andrew M Rappe.


Dirac line nodes in inversion-symmetric crystals. Physical review letters,
115(3):036806, 2015.
[32] Eugene P Wigner. Normal form of antiunitary operators. Journal of Math-
ematical Physics, 1(5):409413, 1960.

[33] Alexey A Soluyanov and David Vanderbilt. Smooth gauge for topological
insulators. Physical Review B, 85(11):115415, 2012.

17

Вам также может понравиться