0% found this document useful (0 votes)
73 views10 pages

Lattice Mechanics of Origami Tessellations

This document presents a lattice theory for examining the mechanics of origami tessellations. The theory treats origami structures as crystalline materials by analyzing the kinematics and mechanics of periodic crease patterns. It introduces a framework for modeling the behavior of a single origami vertex and how to formulate the kinematics of deformation and energetics for a periodic lattice of vertices. As a case study, the theory is applied to analyze properties of the well-known Miura-ori origami pattern and recovers previously known aspects as well as identifies new features.

Uploaded by

大哥法庭
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
73 views10 pages

Lattice Mechanics of Origami Tessellations

This document presents a lattice theory for examining the mechanics of origami tessellations. The theory treats origami structures as crystalline materials by analyzing the kinematics and mechanics of periodic crease patterns. It introduces a framework for modeling the behavior of a single origami vertex and how to formulate the kinematics of deformation and energetics for a periodic lattice of vertices. As a case study, the theory is applied to analyze properties of the well-known Miura-ori origami pattern and recovers previously known aspects as well as identifies new features.

Uploaded by

大哥法庭
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 10

PHYSICAL REVIEW E 92, 013205 (2015)

Lattice mechanics of origami tessellations

Arthur A. Evans,1,* Jesse L. Silverberg,2 and Christian D. Santangelo1


1
Department of Physics, UMass Amherst, Amherst, Massachusetts 01003, USA
2
Department of Physics, Cornell University, Ithaca, New York 14853, USA
(Received 18 March 2015; published 27 July 2015)
Origami-based design holds promise for developing materials whose mechanical properties are tuned by crease
patterns introduced to thin sheets. Although there have been heuristic developments in constructing patterns with
desirable qualities, the bridge between origami and physics has yet to be fully developed. To truly consider
origami structures as a class of materials, methods akin to solid mechanics need to be developed to understand
their long-wavelength behavior. We introduce here a lattice theory for examining the mechanics of origami
tessellations in terms of the topology of their crease pattern and the relationship between the folds at each vertex.
This formulation provides a general method for associating mechanical properties with periodic folded structures
and allows for a concrete connection between more conventional materials and the mechanical metamaterials
constructed using origami-based design.

DOI: 10.1103/PhysRevE.92.013205 PACS number(s): 46.25.Cc, 05.50.+q, 62.20.−x, 81.05.Zx

I. INTRODUCTION bulk material is imbued with an effective mechanical response.


In contrast to conventional composites engineering, wherein
While for hundreds of years origami has existed as an
methods generally rely on designing response based on the
artistic endeavor, recent decades have seen the application of
interaction between the constituent parts that compose the
folding thin materials to the fields of architecture, engineering,
material, origami-based design injects novelty at the atomic
and material science [1–7]. Controlled actuation of thin mate-
level; even single vertices of origami behave as engineering
rials via patterned folds has led to a variety of self-assembly
mechanisms [14], providing novel functionality such as com-
strategies in polymer gels [8] and shape-memory materials [4],
plicated bistability [15–17] and auxetic behavior [6,11–13].
as well elastocapillary self-assembly [9], leading to the design
This generic property inspires the identification of origami
of a new category of shape-transformable materials inspired
tessellations with mechanical metamaterials or a composite
by origami design. The origami repertoire itself, buoyed by ad-
whose effective properties arise from the structure of the unit
vances in the mathematics of folding and the burgeoning field
cell. Although originally introduced to guide electromagnetic
of computational geometry [10], is no longer limited to designs
waves [18], rationally designed mechanical metamaterials
of animals and children’s toys that dominate the art in popular
have since been developed that control wave propagation
consciousness, but now includes tessellations, corrugations,
in acoustic media [19,20], thin elastic sheets, and curved
and other nonrepresentational structures whose mechanical
shells [21–24] and harness elastic instabilities to generate
properties are of interest from a scientific perspective. These
auxetic behavior [25–29].
properties originate from the confluence of geometry and
Traditional metamaterials invoke the theory of linear
mechanical constraints that are an intrinsic part of origami and
response in wave systems, but currently there is no general
ultimately allow for the construction of mechanical metamate-
theory for predicting the properties of origami-inspired designs
rials using origami-based design [1–4,6,11–13]. In this paper
on the basis of symmetry and structure. In the following we
we formulate a general theory for periodic lattices of folds in
propose a general framework for analyzing the kinematics and
thin materials and combine the language of traditional lattice
mechanics of an origami tessellation as a crystalline material.
solid mechanics with the geometric theory underlying origami.
By treating a periodic crease pattern, we naturally connect the
A distinct characteristic of all thin materials is that geo-
geometric mathematics of origami to the more conventional
metric constraints dominate the mechanical response of the
analysis of elasticity in solid state lattice structures. In Sec. II
structure. Because of this strong coupling between shape and
we outline the general formalism required to find the kinematic
mechanics, it is far more likely for a thin sheet to deform by
solutions for a single origami vertex. In Sec. III we discuss the
bending without stretching. Strategically weakening a material
general formulation for a periodic lattice, including both the
with a crease or fold, and thus lowering the energetic cost of
kinematics of deformation modes and energetics for a periodic
stretching, allows complex deformations and reordering of
crease pattern. In Sec. IV we examine the well-known case
the material for negligible elastic energy cost. This vanishing
study of the Miura-ori pattern. Our analysis here recovers
energy cost, especially combined with increased control over
known aspects of the Miura-ori pattern and identifies key
microscopic and nanoscopic material systems, indicates great
features that have not been quantitatively discussed previously.
promise for structures whose characteristics depend primarily
on geometry, rather than material composition.
By patterning creases, hinges, or folds into an otherwise II. SINGLE ORIGAMI VERTEX
flat sheet (be it composed of paper, metal, or polymer gel), the Many of the design strategies for self-folding materials
involves a single fold, an array of nonintersecting folds, or
an array of folds that intersect only at the boundary of the
*
[email protected] material [9,30–33]. From a formal standpoint, we define a

1539-3755/2015/92(1)/013205(10) 013205-1 ©2015 American Physical Society


EVANS, SILVERBERG, AND SANTANGELO PHYSICAL REVIEW E 92, 013205 (2015)

FIG. 1. (Color online) (a) Graph for a single vertex. This degree-6 vertex has its graph determined by the six sector angles αi . Each crease
has a dihedral angle fi associated with it. In the flat case every fi = π or, equivalently, every fold angle is identically zero, since the fold angle
is defined as the supplement of the dihedral angle. (b) By assigning fold angles to each crease, a three-dimensional embedding of the vertex (i.e.,
the folded form of the origami) is fully determined. Every face must rotate rigidly about the defined creases and the sector angles must remain
constant. There is a limited set of fold angles that will solve these conditions. (c) Schematic projection of the curve of intersection between
the unit sphere and the folded form origami. For an N -degree vertex this projection generates a spherical N -gon. To proceed, the N -gon is
divided into N − 2 spherical triangles and the interior angles (i.e., the fi ) follow as a result of applying the rules of spherical trigonometry. All
three-dimensional origami structures are visualized using Tessellatica, a freely available online package for Mathematica [34].

fold as a straight line demarcating the boundary between two partition the angle into sectors by subdividing the spherical
flat sheets of unbendable, unstretchable material. These sheets, N -gon into N − 2 triangles [Fig. 1(c)]. We label the angles
in isolation, are allowed to rotate around the fold so that the that lead from f1 to fi as i , where 1 = α1 and N−1 = αN are
structure behaves mechanically like a simple hinge. If the fold sector angles. All the angles αi are spherical polygon edges and
is produced by plastically deforming a piece of material, rather since origami structures allow only isometric deformations,
than functioning as a hinge the fold has a preferred angle and these angles are constant. The i are the angles subtended
is more precisely called a crease. Herein we shall use the terms by drawing a geodesic on the encapsulating sphere from f1
interchangeably, since the kinematic motions of a fold and the to fi+1 ; expressions for relating the i to the fold angles fi
energetics involved for a crease can be described separately. An are found by using the spherical law of cosines around the
important, and arguably defining, characteristic of an origami vertex [35]:
structure is that it requires that more than one fold meet at
a vertex. While each fold individually allows for unrestricted 
N−2  
cos αi+1 − cos i+1 cos i
rigid body rotation of a sheet, geometrical constraints arise f1 = cos−1 , (1)
sin i+1 sin i
when several folds coincide at a vertex. These constraints are i=1
 
what provide origami structures with their mechanical novelty cos 2 − cos α1 cos α2
f2 = cos−1 , (2)
and ultimately are why deployable structures and mechanical sin α1 sin α2
metamaterials display exotic and tunable properties.  
cos N−2 − cos αN−1 cos αN
A vertex of degree N is defined as a point where N fN = cos−1 , (3)
sin αN−1 sin αN
straight creases meet. Figure 1(a) shows the crease pattern for  
cos i−2 − cos αi−1 cos i−1
a schematic six-degree vertex, with sectors defined by planar fi = cos−1
angles αi . The three-dimensional folded form of this vertex is sin i−1 sin αi−1
 
found by supplying fold angles to each of the creases, subject to cos i − cos αi cos i−1
the constraints mentioned previously [35,36]. This procedure + cos−1 . (4)
sin i−1 sin αi
is an exercise in spherical trigonometry.
One way to visualize the constraints is to surround each These expressions are essentially all that is required to deter-
vertex with a sphere and consider the intersection between it mine the folding of a single vertex, although the associated
and the surface [Fig. 1(b)]. In this construction, the side lengths solutions are generically multivalued. These results imply that
of the spherical polygon are the angles between adjacent folds, there are multiple branches of configuration space for any
which must remain fixed, and the dihedral fold angles are the given spherical polygon.
internal angles of the polygon on the sphere. Since an N -sided To specify the internal state of each vertex we define an
polygon has N − 3 continuous degrees of freedom, each vertex N − 3 component vector s. Given the internal state of a vertex,
does as well. These N − 3 degrees of freedom can be thought all N of the dihedral fold angles are determined, which we
of, for example, as the angles between a fixed fold and the collect in the vector f(s). In practice, computations are vastly
remaining nonadjacent folds. simplified by choosing the appropriate degrees of freedom;
Starting with a general vertex containing dihedral angles for example, for a degree-6 vertex of the type displayed in
fi , we use spherical trigonometry to calculate these angles in Fig. 1, we choose s = {3 ,f2 ,f6 } and the fold vector is given
terms of the N − 3 degrees of freedom. To calculate f1 we by f = {f1 ,f2 ,f3 ,f4 ,f5 ,f6 }.

013205-2
LATTICE MECHANICS OF ORIGAMI TESSELLATIONS PHYSICAL REVIEW E 92, 013205 (2015)

III. GENERAL LATTICE THEORY row of D corresponds to a fold connecting a pair of vertices
in the origami tessellation while each column corresponds
To determine the mechanical properties of an origami
to a component of F . Analysis of this construction is the
tessellation we begin by examining how many vertices are
essence of origami mechanics and lies at the heart of the
connected together in a crease pattern. When constructing a
difficulty in determining general properties of tessellations and
real piece of origami, artists and designers specify “mountain”
corrugations. Finding the null vectors of D amounts to finding
and “valley” creases in the pattern to encode instructions for
all of the possible solutions for the fold angles and thus all of
how the structure will fold. In our formulation we will treat the
the kinematically allowed motions of the rigid origami. While
crease pattern as a simple connected graph, where each unique
computational methods have been developed for simulating
crease is an edge that connects two vertices to one another.
the kinematics of origami and linkage structures [2,6,11,12],
there has been no general analytical study that seeks to identify
A. Kinematically allowed deformations mechanical properties based solely on the crease pattern.
In addition to the origami constraints discussed above for The functions F (s) are, in general, nonlinear. To proceed
a single vertex, joining multiple vertices together generates analytically, we expand s about a state s0 that solves the
further constraints on the folds. Consider a crease pattern that constraint equations. That is, if F (s0 ) = F 0 then DF 0 ≡ 0.
consists of P vertices. Each vertex vp , with p ∈ {1, . . . ,P }, A trivial choice for s0 has every entry identically equal to
p p p
has Np folds, collected in the vector f p = (f1 f2 · · · fNp )T . If π , indicating that the piece of origami is unfolded. The more
we collect all the folds into the vector F , given by common, and more interesting, scenario involves a folded state
⎛ 1⎞ where the values of the internal vector s0 are known. Assuming
f1 that such a state exists, we write s = s0 + δs, with δs a small
⎜ f1 ⎟
⎜ 2 ⎟ perturbation, and then have
⎜ .. ⎟
⎜ . ⎟
⎜ ⎟ DJδs ≡ Rδs = 0, (7)
⎜f 1 ⎟
⎜ N1 ⎟
⎜ 2⎟ where the Jacobian of the fold angles for each vertex J ≡
⎜ f1 ⎟
⎜ ⎟ ∂F /∂s|s0 is a block diagonal matrix defining the small
⎜ . ⎟
F = ⎜ .. ⎟, (5) deviations from the ground state s0 and R is a rigidity matrix
⎜ 2 ⎟
⎜ fN2 ⎟ that informs on the infinitesimal isometric deformations of
⎜ ⎟
⎜ .. ⎟ the origami structure [37,38]. This formulation is convenient
⎜ . ⎟
⎜ P⎟ since it separates the effects of the crease pattern topology
⎜ f1 ⎟
⎜ ⎟ (contained entirely in D) from the constrained motion of a
⎜ .. ⎟
⎝ . ⎠ single vertex (contained entirely in J). We can thus solve for
fNPP each of these matrices individually.
To find D, we first exploit the periodicity of the lattice to
then we have the following constraint equation for the folds: decompose the vector F and matrix D in a Fourier basis such
DF = 0, (6) that F = n,m eiq·x F q + c.c. Here q is a two-dimensional
wave vector and x = na1 + ma2 is the two-dimensional
where D is a sparse rectangular matrix that enforces the position vector of the fundamental unit cell on the crease
condition that if two vertices vq ,vp are adjacent and two pattern lattice, where (n,m) indexes this position in terms of
folds Fi ,Fj connect vq ,vp , then Fi = Fj (see Fig. 2 for the lattice vectors a1,2 . Since F ≈ Jδs and J is independent
an example). This constraint enforces the connectivity of the of the lattice position, we also have δs = n,m eiq·x δsq + c.c.,
graph, since each unique crease clearly must have a compatible where F q = Jδsq . In this representation the constraints given
fold angle associated with the vertices that connect it. Each in Eq. (6) are

D(q)F q = D(q)Jδsq = 0. (8)


(a) (b)
Now, instead of a matrix operation over all the vertices,
f21
the size of D(q) is vastly simplified. For a pattern with p
f31 f11 f21 f22 distinct vertices per unit cell, each of degree Np , D(q) is a
p p
i=1 (Ni /2) × i=1 Ni matrix. In Fourier space, D(q) is the
f41 f31 f11 f32 f12
complex-valued constraint matrix for the graph of the unit
f22 cell vertices and folds. Specifically, each fold of the unit cell
f41 f42
is represented by a row in D(q) having only two nonzero
f32 f12
entries. Those entries all have the form ±eiq·a1 , ± eiq·a2 , ±1,
f42 depending on whether the fold connects to an adjacent unit
cell along a1,2 or is internal to the unit cell.
FIG. 2. (Color online) (a) Two degree-4 vertices with labeled The formulation in terms of the matrix R(q) is completely
folds. (b) Graph for the crease pattern consisting of these two vertices general for any origami tessellation. The rectangular matrix
contains a single crease that is shared by both vertices. In this case the D(q) carries all of the topological information regarding the
constraint equation DF = 0 simply becomes the scalar relationship fold network, while the Jacobian J carries the information
f11 = f32 . about the type of vertex that has been specified; J will be

013205-3
EVANS, SILVERBERG, AND SANTANGELO PHYSICAL REVIEW E 92, 013205 (2015)

block diagonal with one block for each vertex of a unit cell, (a) (b)
but does not depend on q for a regular tessellation.

B. Origami energetics β+
φ+
While the R matrix determines the kinematically isometric
deformation to leading order, these constraints are generally θ− θ+
not the end of the story for real materials. Creases in folded β+ φ+ β− φ−
paper, thermoresponsive gels with programmed folding angles,
and elastocapillary hinges all balance energetic considerations
θ− θ+
with geometric constraints. In many cases these creases and α
hinges act as torsional springs, while the bending of faces have
additional elastic energy content [7,39,40]. β− φ−
The energy associated with the entire structure may be
written, to quadratic order in the dihedral vectors, as FIG. 3. (Color online) (a) While the crease pattern of a Miura-ori
generally introduces only four folds per vertex, the bending of faces
E = 12 (F − F 0 )T A(F − F 0 ), (9) acts to allow two extra folds per vertex, so the crease pattern we
consider is a triangulated lattice. At each vertex the dihedral angles
where A is a general stiffness matrix and F 0 is a reference
contained in f are determined by specifying the state vector s and
fold angle. For linear response this is the most generic form
satisfying the geometric constraints. (b) Single-vertex origami with
for the energy. In the simplest of cases A is constant over the enclosing sphere to visualize the constraints between f and s.
lattice and diagonal with respect to F ; this models each crease
as a torsional spring with uniform spring constant [7,13,40]. A
small amplitude response is found by examining the origami fact be done with little effort [43]. To incorporate these extra
structure near the ground state, that is, when F = F 0 . When degrees of freedom into Miura-ori, we assume that there are
the energy is expanded about the ground state E0 we find two extra folds per vertex to account for face bending. While in
E = E0 + 12 δsT JT AJδs, (10) the extreme case of the creases being perfectly rigid these extra
folds would actually take the form of stretching ridges [44],
or in the Fourier decomposition many real applications involve fabrication processes that will
LW  † allow the face to be well approximated as perfect bending.
E= δsq Mδsq , (11) Each unit cell in the tessellation has four six-valent vertices
2 q (Fig. 3) so there are 12 degrees of freedom per unit cell. In this
where L is the length of the tessellation in the a1 direction, example the fold vector for the ith vertex is given by f i =
W is the width in the a2 direction, and M = JT AJ is a (θ+i ,φ+i ,β+i ,θ−i ,β−i ,φ−i )T and the vector F = (f 1 f 2 f 3 f 4 )T .
matrix operator that is independent of wave number. Since the There are three degrees of freedom per vertex that define the
null space of R(q) will determine the modes of deformation, internal state s, which we parametrize using three angles: ,
the solution to this problem lies in finding the kinematically the angle between folds labeled θ± in Fig. 3, and the angles
allowed deformations and then any energetic description will φ± representing the bending of the faces. Using the geometric
simply involve a change of basis to a system of deformations relationships between the angles [35], we find the general
that diagonalize the operator M. nonlinear relationship for a single vertex and then expand about
the ground state s0 = { + δ ,π + δφ+ ,π + δφ− } to find the
matrix J; here ∈ [π − 2α,π + 2α]. This expansion naturally
IV. MIURA-ORI follows from assuming that the faces are nearly flat and that the
As an example of this formulation, we consider inhomoge- Miura-ori has been folded into the standard configuration. The
neous deformations of a particular origami metamaterial, the Jacobian J = diag(J0 −J0 J0 −J0 ) is a 24 × 12 diagonal
Miura-ori. First introduced as a framework for a deployable block matrix formed from four identical blocks
surface, the design appears often in nature, from plant ⎛ ⎞
A C C
leaves [41] to animal viscera [42]. Additionally, theoretical
⎜ 0 1 0⎟
calculations and experiments have suggested the Miura-ori as ⎜ ⎟
⎜B C 0⎟
a canonical, origami-based, auxetic metamaterial [6,7,11–13]. J0 = ⎜ ⎟, (12)
⎜−A 0 0 ⎟
Its ubiquity may be related to its simplicity: The Miura-ori is ⎝B 0 C ⎠
determined from a single crease angle α and the mountain and
0 0 1
valley assignments of the pattern shown in Fig. 3. Conventional
origami mathematics considers that each Miura-ori vertex where
is degree 4 and thus there is only one degree of freedom.
However, casual experimentation with a real Miura-ori quickly
demonstrates that it has far more than one degree of freedom, A = cos α csc( /2)/ sin2 ( /2) − cos2 α, (13)
indicating an array of soft modes enabled by the bending of
the individual faces. This breakdown of the assumptions of B = sin( /2)/ sin2 ( /2) − cos2 α, (14)
mathematical origami is well known and there are many crease
patterns that are mathematically impossible to fold that can in C = csc(α/2)/2. (15)

013205-4
LATTICE MECHANICS OF ORIGAMI TESSELLATIONS PHYSICAL REVIEW E 92, 013205 (2015)

(a) (b)
3 2 9 8
1 7
4 10
b 5
6
11
12

18 17 24 23
16 22
13 19
y 14
15
20 21
x
a

(c) ei(qx a/2) ei(qy b/2) ei(qx a/2+qy b/2)


3 8

22
13 15
14

FIG. 4. (Color online) (a) Miura-ori, without the assignment of mountain and valley folds, has a simple directed graph structure with a
unit cell composed of four vertices. By tessellating these four vertices, the entire pattern emerges. Note that the tessellation is rectangular,
with lattice vectors a1 = a x̂ and a2 = bŷ. (b) Each vertex has six folds, labeled in the fashion shown here. (c) In Fourier space, translations
associated with connecting these folds together throughout the tessellation merely amounts to a phase factor associated with the appropriate
wave number and lattice vector. Shown on the left is translating in the x direction. The middle shows translating in the y direction. The right
shows that connecting the extra folds involves a diagonal translation across the unit cell. Note that the five internal folds have a phase factor
identically equal to one.

To calculate the constraint matrix, we note that there are 12 are interconnected. For internal folds the constraint matrix has
unique folds per unit cell, so D(q) is a 12 × 24 rectangular a value of ±1, while folds that leave the unit cell have a phase
matrix. It has a row for each bond in Fig. 4 with two factor associated with it. For convenience of computation we
nonzero columns indicating which folds of each vertex have symmetrized these phase factors, and the full matrix is
given by

⎛ ⎞
1 0 0 0 0 0 0 0 0 0 0 0
⎜0 e iqy /2
0 0 0 0 0 0 0 0 0 0⎟
⎜ ⎟
⎜0 0 eiqy /2 0 0 0 0 0 0 0 0 0⎟
⎜ ⎟
⎜0 0 0 −e−iqx /2 0 0 0 0 0 0 0 0⎟
⎜0 −1 0⎟
⎜ 0 0 0 0 0 0 0 0 0 ⎟
⎜0 −1 0⎟
⎜ 0 0 0 0 0 0 0 0 0 ⎟
⎜0 eiqx /2 0⎟
⎜ 0 0 0 0 0 0 0 0 0 ⎟
⎜0 eiqx /2+iqy /2 0⎟
⎜ 0 0 0 0 0 0 0 0 0 ⎟
⎜0 eiqy /2 0⎟
⎜ 0 0 0 0 0 0 0 0 0 ⎟
⎜−1 0 0 0 0 0 0 0 0 0 0 0⎟
⎜ ⎟
⎜0 0 0 0 0 0 0 0 −1 0 0 0⎟
⎜ ⎟
⎜0 0 0 0 0 0 0 0 0 eiq/2 0 0⎟
DT (q) = ⎜
⎜0
⎟.
⎜ 0 0 0 0 0 0 0 0 0 −e−iq/2 0⎟⎟
⎜0 0 0 0 0 0 −e−iqx /2−iqy /2 0 0 0 0 0⎟
⎜ ⎟
⎜0 0 −e−iqy /2 0 0 0 0 0 0 0 0 0⎟
⎜ ⎟
⎜0 0 0 0 0 0 0 0 0 0 0 1⎟
⎜ ⎟
⎜0 0 0 0 1 0 0 0 0 0 0 0⎟
⎜ ⎟
⎜0 0 0 0 0 0 0 0 0 −e−iqx /2 0 0⎟
⎜ ⎟
⎜0 0 0 0 0 0 0 0 0 0 0 −1⎟
⎜ ⎟
⎜0 −e−iqy /2 0 0 0 0 0 0 0 0 0 0⎟
⎜ ⎟
⎜0 0 0 0 0 0 0 −e−iqy /2 0 0 0 0⎟
⎜ ⎟
⎜0 0 0 0 0 0 0 0 0 0 eiqx /2 0⎟
⎝0 0 0 0 0 0 0 0 1 0 0 0⎠
0 0 0 0 0 1 0 0 0 0 0 0
(16)

013205-5
EVANS, SILVERBERG, AND SANTANGELO PHYSICAL REVIEW E 92, 013205 (2015)

A. Bulk deformation creases, but the energy scale for bending of the faces will
The combination D(q)J is square such that Eq. (7) has a depend on the material properties of the structure [7]. The
nontrivial solution whenever det[D(q)J] = 0. We nondimen- energy for bending can be treated as an effective torsional
sionalize the wave number by the physical lengths of the lattice spring constant kb and thus the energy can be written in terms
vectors such that qx a → qx and qy b → qy and the resulting of the ratio kb /k ≡ . Nondimensionalizing the energy by
dispersion relation is kLx Ly , we find the energy eigenvalues λ in terms of the
null vectors. Decomposing the internal variable deformation
cos2 α δs = i ai ψi , where ψi = i /| i | is the normalized null
sin2 (qx /2) + sin2 (qy /2) = 0. (17)
sin4 ( 0 /2) vector with i ∈ {I,II,III}, we write Eq. (11) as
The only real solution to this equation is q = 0, indicating Lx Ly T
that an infinite origami tessellation does not admit spatially E= a Ma, (19)
2
inhomogeneous solutions; only uniform deformations are ⎛ ⎞
allowed. The null space of R is three dimensional here, aI
corresponding to three uniform deformation modes of the a = ⎝ aII ⎠, (20)
Miura-ori. These zero modes are given by the vectors i : aIII
⎛ ⎞ ⎛ ⎞ ⎛ C⎞ ⎛ ⎞
1 0 −2 A ψIT MψI ψIT MψII ψIT MψIII
⎜0⎟ ⎜−1⎟ ⎜ 1 ⎟ ⎜ ⎟
⎜ ⎟ ⎜ ⎟ ⎜ ⎟ M = ⎝ ψIIT MψI ψIIT MψII ψIIT MψIII ⎠. (21)
⎜0⎟ ⎜1⎟ ⎜ 1 ⎟
⎜ ⎟ ⎜ ⎟ ⎜ ⎟ T
ψIII MψI T
ψIII MψII T
ψIII MψIII
⎜1⎟ ⎜0⎟ ⎜ 0 ⎟
⎜ ⎟ ⎜ ⎟ ⎜ ⎟
⎜0⎟ ⎜−1⎟ ⎜ 1 ⎟ Each matrix element of M represents overlaps between the
⎜ ⎟ ⎜ ⎟ ⎜ ⎟ null vectors ψi and the energy matrix M; only in exceptional
⎜0⎟ ⎜1⎟ ⎜ 1 ⎟
I = ⎜ ⎟ , II = ⎜ ⎟ , III = ⎜ ⎟. (18)
⎜1⎟ ⎜0⎟ ⎜−2 CA ⎟ circumstances will M be diagonal in the null basis. In general
⎜0⎟ ⎜−1⎟ ⎜ ⎟
⎜ ⎟ ⎜ ⎟ ⎜ 1 ⎟ it is given by
⎜0⎟ ⎜1⎟ ⎜ 1 ⎟ ⎛ √ ⎞
⎜ ⎟ ⎜ ⎟ ⎜ ⎟
⎜1⎟ ⎜0⎟ ⎜ 0 ⎟ 2(A2 + B 2 ) 0 2(A−B)BC

⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ C +A
2 2

⎝0⎠ ⎝−1⎠ ⎝ 1 ⎠ ⎜ 0 C2 + 0 ⎟
M = ⎜ √2(A−B)BC 2 ⎟.
⎝ √ 2 2 +(3A −2BA+2B ⎠
2 2 2
0 1 1 0 A )C
C +A A2 +C 2
These infinitesimal deformations of the unit cell correspond
to a uniform contraction, a twisting mode, and a saddlelike (22)
deformation, respectively (see Fig. 5). To describe the kine- An example for when M is diagonal is given by α = π/3, =
matics of deformation, all we require are the null vectors of π/2 (see Fig. 5), for which M becomes
the constraint equations, but for examining energy associated ⎛ ⎞
with the creases we need to calculate the eigenvalues of 8 0 0
the matrix M = JT AJ. In general, it is not unreasonable M = ⎝0 1 + 0 ⎠. (23)
to assume that a creased and folded Miura-ori will have a 0 0 2
3
(3 + )
crease stiffness k that is approximately equal for all patterned
Note that for this particular combination of parameters the
uniform expansion mode has a flat stiffness over all ranges of
(a) (b)
λ since there is no face bending for this deformation. In the
regime where face bending is relatively inexpensive (  1),
I 1000
the out-of-plane deformation modes are correspondingly softer
100 than the uniform deformation. These results are in agreement
10 with previous numerical research done on the structural
III I
II 1
mechanics of Miura-ori [6,11,12]. Should other values of (α, )
II be chosen, the energy matrix is not necessarily diagonal and
0.1 thus eigensolutions mix the null vectors.
0.01
III
0.001 B. Inhomogeneous deformation
0.01 0.1 1 10 100
Γ For a finite tessellation, the deformation is fundamen-
tally different, since some folds reach the boundary and,
FIG. 5. (Color online) Shapes and energy eigenvalues for the consequently, do not yield constraints. Since the tessellation
three uniform modes for = π/2 and α = π/3. (a) The three uniform mechanics are determined by the allowable deformations,
null vectors correspond to a uniform mode (I), a twisting mode (II), which are determined by the constraint equations, the presence
and a saddle mode (III). These are identical to the modes determined of free boundaries allows much more flexibility and the
numerically in previous studies [6,12]. (b) Eigenvalues associated Miura-ori develops additional degrees of freedom. These
with each of the three bulk modes as a function of face stiffness . localized edge states are reminiscent of evanescent waves
Note that over a wide range the softest mode is the twisting mode in electromagnetism, boundary layers in elastic lattices [45],
(II), since it involves purely face bending. and Rayleigh surface waves [46]. Letting qx ≡ q (where q is

013205-6
LATTICE MECHANICS OF ORIGAMI TESSELLATIONS PHYSICAL REVIEW E 92, 013205 (2015)

FIG. 6. (Color online) Experimental observations of deformation localization in an 8 × 8 Miura-ori tessellation. (a) An undeformed
Miura-ori shows a regular periodic pattern. Under (b) small deformations, (c) large deformations, and (d) in the presence of a pop-through
defect (PTD) [7], the lattice distorts to accommodate the induced strain. (e) Qualitatively, the amount of deformation localization can be easily
seen by a simple image subtraction between the deformed and undeformed state. (f) Measuring strain along the horizontal axis as a function of
unit cell position n relative to the location of the disturbance shows a rapid decay for all three scenarios (points). For small and large amplitudes,
the decays can be readily fit to an exponential function with decay length  [upper (red) and lower (black) lines], whereas for a PTD, the decay
length can be estimated to within 100%. Because the PTD induces an extensional distortion rather than a compression, the strain is oppositely
signed. The inset is a plot of the decay length against an approximate measure of the distortion wave vector q showing that the larger wave
vector decays much more rapidly than the shorter wave vectors. Within error bars, this measurement is consistent with an inverse relationship
between decay length and wave vector. The solid line is the theoretical prediction from Eq. (24) for = π/2 and α = π/3.

real), Eq. (17) yields qy = ±iκ(q), where deformations decay δs(x) into a sum of upward (in y) decaying and downward
away from the boundaries of constant y with a length scale decaying modes
 ≡ 1/κ(q), with
δs = eiqx [(u1 χ 1 e−k(q)y + u2 χ 2 e−k(q)y )
1
(q) = −1
. (24) + (d1 η1 ek(q)y + d2 η2 ek(q)y )] + c.c. (25)
2|sinh [cos α sin(q/2)/(sin2 /2)]|
This localization length is readily observed in deformation The vectors χ 1,2 correspond to the upward decaying modes
experiments on Miura-ori sheets (see Fig. 6). Using laser-cut and η1,2 the downward decaying modes. Note that, since the
sheets of paper, an 8 × 8 Miura-ori is constructed by folding values of the angles must be real, η1,2 (q) = χ̄ 1,2 (−q). In the
the whole sheet using a planar angle of α = π/3 into the long-wavelength limit, i.e., q  1, we have
ground state given by = π/2 [Fig. 6(a)]. Inhomogeneous ⎛ CAq+CB(q−2i) ⎞ ⎛ 2C ⎞
− AB −A
deformations are created using both an external indenter to ⎜ ⎟ ⎜ 0 ⎟
apply a displacement [Figs. 6(b) and 6(c)] and by placing ⎜ 0 ⎟ ⎜ ⎟
⎜ ⎟ ⎜ ⎟
reversible pop-through defects [Fig. 6(d)] [7]. The strain γn ⎜ 2q ⎟ ⎜ 0 ⎟
⎜ −CAq+CB(q+2i) ⎟ ⎜ 2C ⎟
at each unit cell n is measured such that γn = wn /w̄, where ⎜ ⎟ ⎜− ⎟
⎜ AB ⎟ ⎜ A ⎟
ww is the change in width of the nth cell and w̄ is the average ⎜ 0 ⎟ ⎜−2iq ⎟
⎜ ⎟ ⎜ ⎟
width for an undisturbed cell. As shown in Fig. 6, the strain ⎜ ⎟ ⎜ 0 ⎟
⎜ 0 ⎟ ⎜ ⎟
decays exponentially away from the indenter with a decay χ 1 = ⎜ CAq−CB(q−2i) ⎟, χ 2 = ⎜ 2C ⎟. (26)
⎜ ⎟ ⎜ − ⎟
length that is consistent (within error) with our theoretical ⎜ AB ⎟ ⎜ A ⎟
⎜ 0 ⎟ ⎜−2iq ⎟
predictions. ⎜ ⎟ ⎜ ⎟
⎜ ⎟ ⎜ 0 ⎟
To examine these deformation modes more quantitatively, ⎜ 0 ⎟ ⎜ ⎟
we return to the dispersion relation given by Eq. (17). There are ⎜ CAq+CB(q+2i) ⎟ ⎜ 2C ⎟
⎜ ⎟ ⎜ − ⎟
two possible solutions to Eq. (17), corresponding to different ⎜ AB ⎟ ⎜ A ⎟
⎝ 0 ⎠ ⎝ 0 ⎠
decay directions, and thus the null space of R corresponding
to each of these branches is two dimensional. We decompose 2q 0

013205-7
EVANS, SILVERBERG, AND SANTANGELO PHYSICAL REVIEW E 92, 013205 (2015)

(a) (b)

15
4
I
10
Γ = 0.1 3 Γ = 0.1
λ 2
5
1

0.5 1.0 1.5 2.0 2.5 3.0 0.5 1.0 1.5 2.0 2.5 3.0
15

Γ=1
4 I
Γ=1 II
10 3
λ 2
II
5
1 III,IV
0.5 1.0 1.5 2.0 2.5 3.0 0.5 1.0 1.5 2.0 2.5 3.0 III, IV
15 5
4
10
λ 3
2
5
Γ = 10 1 Γ = 10
0.5 1.0 1.5 2.0 2.5 3.0 0.5 1.0 1.5 2.0 2.5 3.0

q q
FIG. 7. (Color online) Eigenvalues and mode shapes as a function of wave number for a given . (a) Shown on the left is the mode structure
for = 0.1, 1, and 10, with = π/2 and α = π/3. At long wavelengths the saddle mode I is the stiffest for a wide range of q, since it involves
both bending of the faces and deformation of the angles away from the reference state. Shown on the right is the mode structure for = 0.1, 1,
and 10, with = π/2 and α = 9π/20. The labeled modes are in ascending order from largest eigenvalue (mode I) to lowest eigenvalue (mode
IV). (b) Visualization of the basic modes for q = π/6.

The null space, and thus the number of elementary excitations, The two independent blocks of H are given by
for a finite-size Miura-ori is actually different than for the † †

limit q → 0. While this may seem counterintuitive, the nature χ 1 Mχ 1 χ 1 Mχ 2
H0 = † †
(30)
of the null vectors is inherently chiral, as indicated by the χ 2 Mχ 1 χ 2 Mχ 2
decomposition into upward and downward decaying solutions.
At q = 0, the dimensionality of the null space is smaller and
because there is no distinction between handedness for uniform † †

χ 1 Mη1 χ 1 Mη2
deformation. H1 = † †
. (31)
χ 2 Mη1 χ 2 Mη2
C. Miura-ori’s soft modes For finite wave number there are four modes of deformation.
The vectors χ 1,2 govern the kinematic deformations of Typical eigenvalues of H(q) are shown in Fig. 7. The largest
Miura-ori, giving the possible solutions to the constraint two eigenvalues are typically associated with changing ,
equations. For a tessellation with an associated torsional spring since there is an energetic cost even for very small . The
energy at each crease, the energy density per mode may be typically smallest two eigenvalues correspond to twisting
written in Fourier space as mode and a fourth mode that has no analog in the zero
wave number case. This mode has a qualitative shape that
Lx L y †
E= c (q)H(q)c(q), (27) is similar to the twisting mode and an energy that vanishes
2 as q → 0, much like an acoustic mode in a crystal. Previous
where analyses of inhomogeneous deformations have not found this
⎛ ⎞ mode, which we identify here as arising from the breaking
u1 (q)
of continuous symmetry when a boundary is added to one
⎜u (q)⎟
⎜ 2 ⎟ side of the tessellation. The acoustic mode corresponds to
c(q) = ⎜ ⎟ (28)
⎝d1 (q)⎠ an antisymmetric combination of upward and downward
d2 (q) decaying modes; consequently, as q becomes smaller, the
change in fold angles associated with the combination cancels
and H is the 2 × 2 Hermitian block matrix and only three modes appear at q = 0.
 The modes that are softest depend not only on the
H0 H1
H= (29) stiffness of face bending, but on the ground state defined
† † .
H1 H0 by 0 (see Fig. 7). This stiffness dependence is in accord

013205-8
LATTICE MECHANICS OF ORIGAMI TESSELLATIONS PHYSICAL REVIEW E 92, 013205 (2015)

with the previously predicted anisotropic in-plane stiffness crease pattern symmetry may lead to preferentially directed
response [6,13]. Additionally, since our analysis allows for chiral states.
arbitrary size and wave number, we are able to capture the A great deal of this analysis can be carried through to
response of the previously unidentified acoustic mode. other origami fold patterns. What is less clear, however, is
how the number of degrees of freedom, the null space of
R(q), changes for different fold patterns. At the outset it may
V. DISCUSSION seem coincidental that the matrix R(q) is square. In fact, this
behavior is likely more generic. In particular, the Miura-ori,
While there has been numerical analysis of tessellations
with additional folds across the faces, is composed of triangular
in the past, our theoretical formulation provides several
subunits. In any triangulated origami fold pattern, vertices will
key insights into the design and understanding of origami
tend to have, on average, six folds. Hence, for V vertices (with
mechanics. We not only analytically calculated expressions
V very large), we have 3V unique folds and 3V degrees of
for first-order inhomogeneous deformations, but we found
freedom per vertex. Consequently, R(q) will be a 3V × 3V
an additional acoustic mode of deformation that has not
square matrix for sufficiently large V .
been identified using numerics. Moreover, we have found an
Finally, a great advantage to this approach is the ability
analytical expression for a decay length that arises in Miura-ori
to separate the topological nature of the crease pattern from
and identified that these soft modes are edge states that cannot
the geometry of the vertex. The ability to isolate mechanical
occur in an infinite tessellation. Indeed, the appearance of
deformations or elementary excitations in exotic materials is
a single decay length and the ability to fully quantify the
of great interest in quantum condensed matter [38], amorphous
deformation modes using a single wave number indicates that
solids [53–55], and complex fluids [56]. Our theoretical
the boundaries of Miura-ori fully define the deformation state.
framework for origami tessellations bridges the gap between
We can directly conclude from this that, unlike normal solids,
the origami mechanics literature and a theory of origami
the number of degrees of freedom scales with the perimeter of
metamaterials by identifying the constraint-based nature of
a finite tessellation, rather than the area. This result suggests
the folding mechanisms and applying well-known methods of
that there are surface boundary states that can be used to
analysis from solid state physics and lattice mechanics.
probe the full deformation of the material and hints at the
connection between our work and recent studies on topological ACKNOWLEDGMENTS
mechanics [38]. In fact, our mathematical formalism shares
many parallels with the topological mechanics of linkages The authors acknowledge interesting and helpful discus-
[47–49], as well as the more conventional literature concerning sions with Tom Hull, Robert Lang, Tomohiro Tachi, Scott
topological insulators and semimetals [50–52]. It remains to Waitukaitis, Martin van Hecke, and Michael Assis. We also
be seen exactly how the symmetry and topology of the crease thank F. Parish for help with the laser cutter. This work was
pattern affect the nature of chiral modes in origami, but there funded by the National Science Foundation through Grant No.
is evidence to suggest that even slight modifications of the EFRI ODISSEI-1240441.

[1] T. Tachi, Symposium of the International Association for Shell [8] J.-H. Na, A. A. Evans, J. Bae, M. C. Chiappelli, C. D. Santangelo,
and Spatial Structures (50th. 2009. Valencia). Evolution and R. J. Lang, T. C. Hull, and R. C. Hayward, Adv. Mater. 27, 79
Trends in Design, Analysis and Construction of Shell and Spatial (2015).
Structures: Proceedings (Editorial de la Universitat Politécnica [9] C. Py, P. Reverdy, L. Doppler, J. Bico, B. Roman, and C. N.
de Valencia, Valencia, 2009). Baroud, Phys. Rev. Lett. 98, 156103 (2007).
[2] T. Tachi, in Proceedings of the International Association for [10] J. Solomon, E. Vouga, M. Wardetzky, and E. Grinspun,
Shell and Spatial Structures (IASS) Symposium, edited by A. Computer Graphics Forum (Wiley, New York, 2012), Vol. 31,
Domingo and C. Lazaro (Editorial de la Universitat Politécnica pp. 1567–1576.
de Valencia, Valencia, 2010), Vol. 12, pp. 458–460. [11] M. Schenk and S. Guest, Folded shell structures, Ph.D. thesis,
[3] T. Tachi, in Symposium of the International Association for Shell University of Cambridge, 2011.
and Spatial Structures (50th. 2009. Valencia). Evolution and [12] M. Schenk and S. D. Guest, Origami 5, 291 (2011).
Trends in Design, Analysis and Construction of Shell and Spatial [13] Z. Y. Wei, Z. V. Guo, L. Dudte, H. Y. Liang, and L. Mahadevan,
Structures: Proceedings (Editorial de la Universitat Politécnica Phys. Rev. Lett. 110, 215501 (2013).
de Valencia, Valencia, 2010). [14] K. Abdul-Sater, F. Irlinger, and T. C. Lueth, J. Mech. Robot. 5,
[4] E. Hawkes, B. An, N. Benbernou, H. Tanaka, S. Kim, E. 031005 (2013).
Demaine, D. Rus, and R. Wood, Proc. Natl. Acad. Sci. USA [15] S. Waitukaitis, R. Menaut, B. G.-g. Chen, and M. van Hecke,
107, 12441 (2010). Phys. Rev. Lett. 114, 055503 (2015).
[5] M. A. Dias, L. H. Dudte, L. Mahadevan, and C. D. Santangelo, [16] B. H. Hanna, J. M. Lund, R. J. Lang, S. P. Magleby,
Phys. Rev. Lett. 109, 114301 (2012). and L. L. Howell, Smart Mater. Struct. 23, 094009 (2014).
[6] M. Schenk and S. D. Guest, Proc. Natl. Acad. Sci. USA 110, [17] N. P. Bende, A. A. Evans, S. Innes-Gold, L. A. Marin, I. Cohen,
3276 (2013). R. C. Hayward, and C. D. Santangelo, arXiv:1410.7038.
[7] J. L. Silverberg, A. A. Evans, L. McLeod, R. C. Hayward, T. [18] J. B. Pendry, D. Schurig, and D. R. Smith, Science 312, 1780
Hull, C. D. Santangelo, and I. Cohen, Science 345, 647 (2014). (2006).

013205-9
EVANS, SILVERBERG, AND SANTANGELO PHYSICAL REVIEW E 92, 013205 (2015)

[19] M. Kadic, T. Bückmann, N. Stenger, M. Thiel, and M. Wegener, [37] R. Hutchinson and N. Fleck, J. Mech. Phys. Solids 54, 756
Appl. Phys. Lett. 100, 191901 (2012). (2006).
[20] S. Brûlé, E. H. Javelaud, S. Enoch, and S. Guenneau, Phys. Rev. [38] C. Kane and T. Lubensky, Nat. Phys. 10, 39 (2014).
Lett. 112, 133901 (2014). [39] F. Lechenault, B. Thiria, and M. Adda-Bedia, Phys. Rev. Lett.
[21] M. Farhat, S. Guenneau, and S. Enoch, Phys. Rev. Lett. 103, 112, 244301 (2014).
024301 (2009). [40] J. L. Silverberg, J.-H. Na, A. A. Evans, B. Liu, T. C. Hull, C.
[22] N. Stenger, M. Wilhelm, and M. Wegener, Phys. Rev. Lett. 108, D. Santangelo, R. J. Lang, R. C. Hayward, and I. Cohen, Nat.
014301 (2012). Mater. 14, 389 (2015).
[23] J. Shim, S. Shan, A. Košmrlj, S. H. Kang, E. R. Chen, J. C. [41] L. Mahadevan and S. Rica, Science 307, 1740 (2005).
Weaver, and K. Bertoldi, Soft Matter 9, 8198 (2013). [42] A. E. Shyer, T. Tallinen, N. L. Nerurkar, Z. Wei, E. S. Gil, D.
[24] A. A. Evans and A. J. Levine, Phys. Rev. Lett. 111, 038101 L. Kaplan, C. J. Tabin, and L. Mahadevan, Science 342, 212
(2013). (2013).
[25] Y. Zhang, E. A. Matsumoto, A. Peter, P.-C. Lin, R. D. Kamien, [43] E. D. Demaine, M. L. Demaine, V. Hart, G. N. Price, and T.
and S. Yang, Nano Lett. 8, 1192 (2008). Tachi, Graphs Combinator. 27, 377 (2011).
[26] E. A. Matsumoto and R. D. Kamien, Phys. Rev. E 80, 021604 [44] T. Witten, Rev. Mod. Phys. 79, 643 (2007).
(2009). [45] A. S. Phani and N. A. Fleck, J. Appl. Mech. 75, 021020 (2008).
[27] K. Bertoldi, P. M. Reis, S. Willshaw, and T. Mullin, Adv. Mater. [46] J. W. Strutt and L. Rayleigh, Proc. London Math. Soc. 17, 4
22, 361 (2010). (1885).
[28] E. A. Matsumoto and R. D. Kamien, Soft Matter 8, 11038 [47] B. G.-g. Chen, N. Upadhyaya, and V. Vitelli, Proc. Natl. Acad.
(2012). Sci. USA 111, 13004 (2014).
[29] J. T. B. Overvelde, S. Shan, and K. Bertoldi, Adv. Mater. 24, [48] J. Paulose, B. G.-g. Chen, and V. Vitelli, Nat. Phys. 11, 153
2337 (2012). (2015).
[30] C. Yoon, R. Xiao, J. Park, J. Cha, T. D. Nguyen, and D. H. [49] J. Paulose, A. S. Meeussen, and V. Vitelli, arXiv:1502.03396.
Gracias, Smart Mater. Struct. 23, 094008 (2014). [50] M. Z. Hasan and C. L. Kane, Rev. Mod. Phys. 82, 3045 (2010).
[31] Y. Liu, J. K. Boyles, J. Genzer, and M. D. Dickey, Soft Matter [51] X.-L. Qi and S.-C. Zhang, Rev. Mod. Phys. 83, 1057 (2011).
8, 1764 (2012). [52] H. C. Po, Y. Bahri, and A. Vishwanath, arXiv:1410.1320.
[32] L. Ionov, Soft Matter 7, 6786 (2011). [53] K. Sun, A. Souslov, X. Mao, and T. Lubensky, Proc. Natl. Acad.
[33] G. Stoychev, N. Puretskiy, and L. Ionov, Soft Matter 7, 3277 Sci. USA 109, 12369 (2012).
(2011). [54] X. Mao, N. Xu, and T. C. Lubensky, Phys. Rev. Lett. 104, 085504
[34] Tessellatica, http://www.langorigami.com/science/ (2010).
computational/tessellatica/tessellatica.php [55] M. Wyart, S. Nagel, and T. Witten, Europhys. Lett. 72, 486
[35] D. A. Huffman, IEEE Trans. Comput. 25, 1010 (1976). (2005).
[36] s.-m. belcastro and T. C. Hull, Linear Algebra Appl. 348, 273 [56] E. Lerner, G. Düring, and M. Wyart, Proc. Natl. Acad. Sci. USA
(2002). 109, 4798 (2012).

013205-10

You might also like