Additive Manufacturing 46 (2021) 102126

Contents lists available at ScienceDirect

Additive Manufacturing
journal homepage: [Link]/locate/addma

Research paper

Harnessing fractal cuts to design robust lattice metamaterials for

energy dissipation
Zhennan Zhang a, Fabrizio Scarpa b, Brett A. Bednarcyk c, Yanyu Chen a, *
a
Department of Mechanical Engineering, University of Louisville, Louisville, KY 40292, USA
b
Bristol Composites Institute and Dynamics and Control Research Group, University of Bristol, BS8 1TR Bristol, UK
c
NASA Glenn Research Center, Cleveland, OH 44135, USA

A R T I C L E I N F O A B S T R A C T

Keywords: Lattice metamaterials exhibit remarkable mechanical properties and novel functionalities, such as high specific
Metamaterials stiffness, fracture toughness, tunable vibroacoustic properties, and energy absorption. This study explores a
Fractal group of lattice metamaterials with fractal cuts that feature energy dissipation via structural sliding friction
Friction
mechanisms and intrinsic material damping. Lattice metamaterials with three types of fractal cuts patterns were
Energy dissipation
designed and fabricated using additive manufacturing. Three-point bending tests and numerical analysis have
3D printing
been combined to investigate their bending behavior. Experimental results show that the structural bending
compliance of the metamaterials can be sharply enhanced by increasing the fractal orders, while at the same time
keeping the shape recoverability during cyclic loading. Loss factors associated with the cyclic bending of these
metamaterials are almost independent of the fractal order used, which is attributed to the synergistic effect
between friction and viscoelasticity. It is further demonstrated via validated finite element models that the
sliding friction plays a critical role by investigating the effects of the bending displacement and sample thickness
on the flexural behavior. Results suggest that the magnitude of the maximum bending displacement has a
negligible effect on the loss factors, and a power scaling law exists between bending stiffness and sample
thickness. This study suggests that fractal lattice metamaterials have unique energy dissipation properties, with
potential applications in many industrial sectors such as defense, energy, and transportation.

1. Introduction provide toughness to the intrinsically brittle and low elastic limit base
material. These have been shown to undergo large deformations of up to
The integration of architectural paradigms into traditional material 80% compressive strain without catastrophic failure [18]. Hollow
science expands the material-property design space [1–3]. Architected metallic lattice topologies can fully recover from a compressive strain up
materials have demonstrated unique mechanical features and perfor­ to 50% while dissipating a considerable portion of the elastic strain
mance due to their rationally designed periodic architectures using both energy [19]. In these lattice structures, energy dissipation mechanisms
monolithic and multiphase constituents [4–7]. A well-known engi­ include local beam buckling, friction, plastic deformation, and intrinsic
neered architecture, the Eiffel Tower, demonstrates the possibility of material damping. Because of their low relative densities, these light­
architectural scaling-up with a relatively low density and a robust me­ weight lattice structures are desirable for protective applications where
chanical performance because of the hierarchical lattice design [8,9]. energy dissipation at controlled load levels is needed [20,21]. The
Lattice configurations are typical of metamaterials with novel properties development of lattice metamaterials remains, however, a challenging
and multifunctionalities, including high stiffness, tunable vibroacoustic task because of the limitations and expense of traditional manufacturing
properties, enhanced energy absorption, and shape recoverability processes. In this context, recent breakthroughs in additive
[10–13]. For example, architected materials with re-entrant structure, manufacturing [22–24] now offer unprecedented freedom to design and
chiral structure, or bistable structure have shown novel mechanical produce structures with complex geometries across multiple length
properties, such as negative Poisson’s ratio and mechanical program­ scales.
mability [14–17]. Incorporating lattice structures within ceramics can Fractal lattices can be described by self-similarity: subdivision into

* Corresponding author.
E-mail address: [Link]@[Link] (Y. Chen).

[Link]
Received 26 February 2021; Received in revised form 14 June 2021; Accepted 16 June 2021
Available online 24 June 2021
2214-8604/© 2021 Elsevier B.V. All rights reserved.
Z. Zhang et al. Additive Manufacturing 46 (2021) 102126

small sections yields pieces with geometries that resemble the whole ribs have been shown as the main deformation modes, because cuts
[25]. They broadly exist in natural materials, with examples like bone, divide the surface into a series of slender segments, which can easily be
wood, and glass sponge skeletons. Fractal configurations have attracted twisted and are interconnected at fixed points [47]. Although there is
considerable interest because of their mechanical robustness and dam­ some work in open literature related to the mechanical properties of
age tolerance [8,26–30]. Notably, stiffness becomes less sensitive to fractal lattice structures, little is known about their energy dissipation
errors in assembly with additional levels of hierarchy [31]. Fractal lat­ performance.
tices with self-similar structures across multiple length scales can be Here the energy dissipation behavior of a group of fractal chiral
exploited to maximize the surface area coverage and promote mechan­ lattice metamaterials is investigated using a combined experimental and
ical stability under large strain deformations [32,33]. When employed numerical approach. These lattices have been designed with three orders
as reinforcing elements, additively manufacturing fractal lattice struc­ (N1 , N2 , N3 ) of structural hierarchy and then produced by additive
tures can significantly enhance the first crack strength and the residual manufacturing. Three-point bending tests have been performed on these
loading capacity of cement composites [34]. A fractal design possesses samples to evaluate the effects of the fractal orders, the maximum
high structural efficiency and can be tailored to increased buckling deflection, and sample thickness. Through a combination of digital
stability and potentially develop ultralow density materials [35–37]. image correlation (DIC) monitoring and finite element simulations, the
Stretching-dominated fractal lattices show near-linear scaling of evolutions of strains and stresses in each sample have been captured.
strength and stiffness down to a density of 0.01%, while the corre­ The effective bending stiffness, energy absorption, energy dissipation,
sponding first-order lattice has scaling exponents as high as 2.7 below a and their corresponding loss factors of each of these lattices have been
density of 0.1% [18]. Furthermore, the scaling law of lattice structures determined and compared. Compared to findings from other studies, the
can be manipulated beneficially by varying the hierarchy order [38]. A energy dissipation of this structure occurs due to stick-slip friction and
type of hollow alumina patterned into self-similar unit cells in a fractal viscoelastic damping. The density of the lattice presented in this work
lattice can recover up to 98% of its original height after compression up decreases gradually with the increase of the fractal geometry orders. The
to 50% strain [8]. A recently developed chiral lattice is based on energy dissipation performance, however, improves significantly with
structural chirality, which introduces deformation features typical of the fractal order; this indicates the possibility to manufacture light­
rotating nodes and bent ligaments of chiral lattices into a fractal design weight and high-energy dissipation materials. By introducing topologies
[39]. Chiral topologies are commonly developed by connecting ribs with fractal cuts, brittle materials can be transformed into very ductile
tangentially to cylinders or quadratic units [40]. This microstructural structures while still maintaining satisfactory shape recovery properties.
design leads to multifunctional properties such as vibration attenuation,
impact energy absorption and negative coefficient of thermal expansion 2. Materials and methods
[41–46]. A representative fractal chiral hinge lattice generates internal
rotations by creating hinges and an overall chiral structural configura­ 2.1. Fractal lattice metamaterial design
tion. This structure shows a very high in-plane shear stiffness with a
quasi-isotropic behavior compared to other structural chiral configura­ A square flat sheet with three orders of fractal cut patterns has been
tions [39]. For this fractal chiral hinge lattice, torsion and bending of the considered (Fig. 1(a)-(c)). This cut topology produces symmetrical

Fig. 1. Design and manufacturing of the proposed fractal lattice metamaterials. (a)-(c) Fractal unit cell designs for N1 , N2 , and N3 , respectively. (d) 3D printed
samples for three-point bending testing.

2
Z. Zhang et al. Additive Manufacturing 46 (2021) 102126

motifs under all rotations about the center. The unit cell can be central deflections were applied to each specimen, from dmin = 10
described by the parameters ai , bi and ti , where ai represents the side mm to dmax = 32.5mm with an interval of 2.5 mm.
length of the unit cell, bi stands for the thickness of the ribs, ti is the (iii) Sample thickness: samples N3 with ten different thicknesses
width of the slit, and the subscript irepresents the fractal order level. The (between 10 and 30 mm, interval 2 mm) were used in these tests.
relation between these parameters is expressed as ai = mi × (bi + ti ), in Like test (i), all specimens were loaded to a maximum central
which mi is an even integer that determines the cut motifs. Compared deflection of 20 mm.
with bulk materials, the relative density ρi of N1 , N2 , N3 structures can be
calculated as: A digital image correlation (DIC) system was used to measure the
( ) full-field displacements to get the strain εx . An ARAMIS 4 M system
m2 b2 + bi ti m2i + 4 + 2t2i
ρi = i i (1) (GOM GmbH, Germany) was used to control the capture of the images,
2
[mi (bi + ti )] perform the image correlation, and process the displacements into
Based on the printing resolution, the side length of the unit cells ai strains. Each experiment was repeated three times.
was set to 25 mm. The gap width is fixed at ti = 0.625 mm, while the mi
values are 6, 10, and 14 for the three unit cells, respectively. Therefore, 2.4. Finite element analysis
for the parameters in this paper, the relative densities are ρ1 = 0.865,ρ2
= 0.759 and ρ3 = 0.656 for N1 , N2 , and N3 , respectively. Unless other­ To better understand the effect of the geometric cut motif, finite
wise specified, the out-of-plane thickness T is 20 mm. element simulations were performed using the commercial package
ABAQUS 6.14 (SIMULIA, Dassault Systems, USA). The dynamic implicit
solver with adaptive time-stepping was employed. Models from N1 toN3
2.2. 3D printing of fractal lattice metamaterials were discretized using C3D8R (eight-node linear bricks with reduced
integration and hourglass control) solid elements. The head and sup­
The lattice metamaterials have been fabricated via a multi-material ports of the three-point bending fixture model were defined as analytical
3D printer Objet260 Connex3 (Stratasys Ltd., USA). This machine is rigid with a contact surface radius of 5 mm (similarly to the experi­
based on the layer-wise PolyJet printing process in which the jetting mental setup). The dissipative losses caused by structural contact fric­
nozzles in the print head switch between multiple materials, including tion and material damping were considered by applying a friction
the support material for overhanging features. Material is then cured by coefficient between fixture and specimen of 0.2. A contact friction co­
ultraviolet light using a source situated on-board of the printing head, efficient of 0.35 was adopted to model the specimen self-contact [54,
which immediately causes the liquid resin to react and solidify, then 55]. VeroWhitePlus was modeled as isotropic with a Young’s modulus
roller smooths out the deposited material and prepares the surface for E0 = 2400 MPa, a Poisson’s ratio of v = 0.33 and density ρ = 1174
the next layer [48–50]. To ensure the structural strength of the fractal kg/m3 [56,57]. To consider the intrinsic material damping, a visco­
lattice configurations, a rigid polymer, VeroWhitePlus, has been used as elastic constitutive model was selected. Prony series coefficients were
the base constituent material. It is a thermoset acrylic resin that allows used to characterize the viscous parameters determined based on the
the creation of rigid models with a great degree of accuracy in either a Maxwell material model. The relaxation function E(t)can be expressed
matte or glossy finish. The Vero resin is cured with UV light and does not as [58,59]:
require a resin wash or post-curing. VeroWhitePlus provides samples ⎡ ⎛ ⎞⎤
with good overall properties when using the PolyJet process [51]. The ∑n

fabricated fractal specimens have an array of 2 × 10 unit cells. Following

⎣
E(t) = E0 1 − ⎝
gi 1 − e− τt ⎠⎦
i (2)
the dimensions prescribed by the ASTM D790-17 standard, the speci­
i=1

mens have an overall side length of L = 250 mm and side width of W =

where E(t)is a relaxation function, E0 is an instantaneous modulus of
50mm. Fig. 1(d) shows 3D printed specimens with an identical thick­
elasticity, t is time, gi and τi are Prony constants. For simplicity, only a
ness T = 20 mm from the first-order (top) to the third-order (bottom).
single term series of Prony coefficients was considered here. To char­
acterize the parameters, a series of loading-unloading three-point
2.3. Three-point bending tests bending tests have been performed on 3D printed bulk specimens. It was
found that a single term in the Prony (with g1 = 0.3, τ1 = 0.6, E0 =
The specifications of the ASTM standard D790–17 recommend the 2400MPa) accurately described the loading-unloading behavior of the
use of specimens of sufficient length with thickness larger than 1.6 mm, samples. The single coefficient Prony series was subsequently applied to
to allow an overhang on each end of at least 10% of the support span. simulate samples N1 and N2 . Although the resolution of the printer has a
The overhang should be in no case less than 6.4 mm [52,53]. Therefore, feature size of 16 µm [60], there are some defects when the sample
three-point bending experiments on a support span S = 150 mm have thickness is below 1 mm. These defects will significantly affect the
been carried out using a universal testing machine (Instron 5569A, USA) stiffness of the materials [61,62]. To consider this effect, we choose E0 =
with a maximum load capacity of 5 kN. The three-point bending 1700 MPa for N3 sample, as the minimum thickness b3 = 1.161 mm is
extension rate was controlled by Bluehill 2.0 software (Instron, USA) at close to 1 mm.
2.5 mm/min. The performance of the 3D printed specimens was inves­
tigated by considering the effects of the geometric cut motif, the 3. Results and discussion
amplitude of the central deflection, and the sample thickness. Towards
this end, three sets of three-point bending experiments were conducted 3.1. Role of fractal order on the bending response
under displacement control while varying the following:
The bending behavior of representative specimens with different
(i) Fractal order: specimens with three fractal orders (N1 , N2 , N3 , T = fractal orders is described by normalized load-deflection curves (Fig. 2
20 mm) were loaded to a maximum central deflection of d = (a)). For clarity, curves related to N2 and N3 have been plotted sepa­
20mm, then unloaded with the same loading rate. A loading- rately in Fig. 2(b) because the peak force in N1 is significantly larger
unloading force-displacement curve was recorded in every than the other cases. The force and displacements are respectively
experiment. normalized as:
(ii) Maximum central deflection: we selected sample N3 with T = 20 Pn = Pρi /T (3)
mm for these loading-unloading experiments. Ten maximum

3
Z. Zhang et al. Additive Manufacturing 46 (2021) 102126

Fig. 2. Bending behavior of the fractal lattice metamaterials. (a)-(b) Loading-unloading load-deflection relations for N1 , N2 , and N3 . Here the normalized load is
defined as the ratio of force to sample thickness and relative density. The relative density for N1 , N2 , and N3 are ρ1 = 0.865,ρ2 = 0.759 and ρ3 = 0.656. (c) Loading-
unloading deformation of N3 . (d) Strain evolutions of N3 . Here d1 = 0, d2 = 14, d3 = 20, d4 = 7, and d5 = 2.3 mm.

dn = d/S (4) shape recovery ratioRr can be calculated as follows [63]:

dmax − dres
where Pis the force, ρi is the relative density, Tis the sample thickness, d Rr = × 100% (5)
dmax
is the deflection, and S is the span. We normalize the force and
displacement with respect to relative density and sample thickness, To further understand the bending response of the fractal lattices, we
because these two geometric features have a significant impact on the illustrate the evolution of the deformation patterns for sample N2 . Fig. 2
mechanical properties. As such, each influencing factor can be explained (c) shows that the specimen maintained excellent structural integrity
in a comparable way. The normalized peak loads for N1 , N2 , and N3 are without apparent damage during the loading-unloading cycle. With the
41.59, 3.51, and 0.64 N/mm, respectively. This is because the rib successive increase of the maximum central deflection the adjacent ribs
thickness becomes smaller in high-order fractal lattices. These curves are are less and less parallel to each other and form angles that open grad­
almost linear within the initial loading and the later unloading stages, ually from the upper to the lower surface up to the final maximum
with almost the same slope. No obvious plastic deformation can be deflection of d3 = 20 mm (d/S = 0.133). For a specific central deflection,
observed, even within large deformations. This suggests that no local the largest opening angle appears in the middle of the sample and de­
damage/failure occurs during bending. Compared with samples N1 , a creases gradually to both sides. To better demonstrate the evaluation of
linear elastic behavior in N2 and N3 persists up to deflections of ~13 mm the opening angles, the DIC was used to detect the strain distribution in
(d/S = 0.187) and 11 mm (d/S = 0.073), after which the slopes of the sample N2 (Fig. 2(d)). As one would expect, the largest strain appears in
curves indicate a slight rise. The initially increased slope in the load- the middle and lower parts of the sample. Some small residual strain (~
deflection data arises from the initial contact between the upper parts 2%) is present at the bottom of the specimen.
of the ribs on the top compressive side of the specimen when the Fig. 3 (a), (c), and (e) shows the comparison between the numerical
deflection becomes large. After unloading, the residual deflections for simulations and bending experiments. The loading-unloading responses
N1 , N2 , and N3 are almost the same (~2.64, 2.47, and 2.56 mm, d/S = up to 20 mm (d/S = 0.133) for N1 , N2 , and N3 agree quite well with the
0.018, 0.016, 0.017). All specimens show an almost 88% shape recovery experimental observations, in which, E0 for N1 and N2 are predictions
from the initial undeformed deflection, and there is no obvious shape while E0 for N3 is adjusted. The von Mises stress distributions for sam­
recovery under the free stress condition after unloading. Based on the ples N1 , N2 , and N3 are displayed in Fig. 3(b), (d), and (f). One promi­
maximum loading deflectiondmax and the residual deflectiondres , the nent feature shown by the computations is that the local stress first
appears at the middle part of the sample and then gradually extends to

4
Z. Zhang et al. Additive Manufacturing 46 (2021) 102126

Fig. 3. Finite element simulations. (a), (c), and (e): Comparison of loading-unloading curves between simulations and experiments for N1 , N2 , and N3 . (b), (d), and
(f): Bending response for each sample.

both sides with the increase of the central deflection. For the maximum considering the hypothesis underpinning the models and experimental
stress concentrations, they are highly localized on the joint of the hor­ errors. The experimental results also confirm the marked decrease in
izontal and vertical ribs, which can be especially observed in Fig. 3(b) terms of bending stiffness of the samples with architectures passing from
while d/S = 0.067. Comparing the findings in N1 with those of orders N2 N1 to N3 .
and N3 , it was found that increased levels of fractal order led to a Another remarkable point here is the energy absorption performance
noticeable decrease of this von Mises stress and, therefore, to more of the three architectures. Comparisons between experimental and nu­
compliant structures in bending. This is mainly attributed to the merical energy absorption, energy dissipation and loss factors of the
different rib thicknesses. Meanwhile, other factors make some contri­ three types of fractal beams are summarized in Fig. 4(b)-(d). The input
butions, such as the upper rib contact, the bottom shear stress, as well as energy Uis here defined as the energy necessary to deform a given
the sliding friction between structural slits (the contribution of which is specimen up to the maximum central deflection and can be evaluated by
quantified below). It should be pointed out that the focus of this work is integrating the area under the loading force-displacement curve [18]. As
to experimentally understand the bending behavior of the proposed illustrated in Fig. 4(b), the structural energy absorption capacity
fractal lattice metamaterials, which has been confirmed by finite monotonically decreases with the increase of the fractal orders, and this
element simulations. Our ongoing work includes establishing a micro­ can be attributed to the higher stiffness of the lower-order fractal
polar theory to predict the bending response of the proposed lattice structures. During unloading, the absorbed energies cannot be entirely
configurations and compare with the results of experiments and finite transformed into strain energy and energy dissipation (ΔU) occurs.
element simulations. Dissipated energy per unit mass can be expressed as the area encom­
Fig. 4(a) compares the bending stiffness between the three-point passed by the loading and unloading load-deflection curves. Here, the
bending experiments and the finite element simulations. Measured area bounded by the loading and unloading curves is taken as an indi­
average bending stiffness values are 35.1, 2.3, and 0.32 N/mm for cator of the dissipated energy. Compared to the energy absorption, a
samples N1 , N2 , and N3 , respectively. The bending stiffness values of N1 similar trend is observed in Fig. 4(c). The loss factor of the structure η
and N2 obtained via FE are somewhat smaller than those from the cor­ can be determined from [64,65]:
responding experiments, while the numerical bending stiffness of sam­
ΔU
ple N3 is slightly larger. One can, however, conclude that the η= (6)
2π⋅U
experimental and numerical results are in quite good agreement when

5
Z. Zhang et al. Additive Manufacturing 46 (2021) 102126

Fig. 4. Comparison of mechanical properties. (a) Bending stiffness. (b) Energy absorption. (c) Energy dissipation. (d) Loss factor.

Fig. 4(d) shows a comparison of the loss factor from experiments and larger maximum deflection.
simulations. Interestingly, there is no strong correlation between loss All samples show a linear increase of the force at the beginning of the
factors and fractal cut orders, with an almost constant value of close to loading, however the slope increases after a critical central deflection
4%. These results are likely to be attributed to several physical mecha­ around 11 mm (d/S = 0.073). The increased slope at this loading can
nisms. In general, energy dissipation and structural damping can be again be attributed to the initial contact between the upper ribs of the
affected by phenomena like fracture energy and plastic work at the fractal architecture. The stiffness of the N3 structure at different
microscale level, elastic buckling, macroscale friction, microscale fric­ maximum deflections is illustrated in Fig. 5(b). As expected, the effec­
tion or intrinsic material damping [19]. Based on the known properties tive bending stiffness is independent of the maximum deflection. These
of VeroWhitePlus [66–68], it is likely that the intrinsic material damping stiffness points are fitted with a straight line kB = 0.3125N/mm, where
and macroscale friction are the primary damping mechanisms due to the kB is the effective bending stiffness.
viscoelastic nature of the base material along with the stick-slip behavior Fig. 5(c) illustrates the evolution of energy absorption and energy
by frictional interactions. When compressed to large strains, the lattice dissipation. One can observe a gradual rise of the energy growth rate
ribs contact each other and dissipate energy via stick-slip friction under maximum central deflection ranging from 10 mm to 32.5 mm
[69–71]. To further explore the contribution of these two factors, the (d/S = 0.067 ∼ 0.217). The corresponding loss factors are shown in
friction dissipated energy from the finite element simulation was Fig. 5(d), indicating that the peak value is associated with the sample
extracted. Those energy values are 0.11, 0.015, and 0.003 J for N1 , N2 , with the maximum deflection. The result could be partially caused by
and N3 , respectively. The corresponding ratio of friction dissipated en­ the increased contact area between ribs at a higher central deformation,
ergy to total dissipated energy are 9.9%, 19.1%, and 21.9% for N1 , N2 , resulting in more stick-slip friction. The evolution of the deformations
and N3 , respectively. Although the value of the loss factor between detected via DIC in specimen with two representative maximum de­
fractal orders is close, this result suggests that stick-slip friction plays, a flections (10 and 32.5 mm) are shown in Fig. 5(e) and (f). The maximum
more critical role in increasing fractal orders, which is caused by the εx strain appears at the bottom of the sample and is distributed from the
increased contact area between ribs. middle to both sides. As expected, the strain shows a marked increase
with larger central deflections, but the structure still maintains good
structural integrity with no indications of local damage/failure.
3.2. Effect of maximum deflection on bending response

The influence of the central maximum deflection has been assessed 3.3. Effect of the sample thickness on the bending response
by performing a series of additional three-point bending experiments on
the N3 lattice structures. As illustrated in Fig. 5(a), the maximum To explore how the bending response changes with the thickness of
normalized loads range from 0.257 to 1.381 N/mm. The residual de­ the specimen, additional three-point bending experiments were per­
flections are between 1.1 mm and 4.1 mm, leading to a recovery rate formed up to a maximum deflection of 20 mm for sample N3 with
ranging from 89.0% to 87.4%. The maximum central deflection reaches thickness varying from 10 mm to 30 mm. Several interesting features
32.5 mm (d/S = 0.217) without apparent failure and forms a complete are evident in Fig. 6(a). First, even for the sample with a thickness of
loop after unloading. As expected, the maximum von Mises stress in­ 30 mm, it can form a complete loading-unloading curve, revealing its
creases monotonically with the increase of central deflection; mean­ excellent recovery. No significant difference was present in terms of
while, the recovery rate slightly decreases, likely due to the hysteresis residual depth after unloading between samples having different
provided by the material viscoelasticity. Hysteresis and lower recover­ thicknesses, with their structural recovery rate being close to 88%.
ability after unloading are more significant in samples subjected to the Compared with the effect of different applied maximum central

6
Z. Zhang et al. Additive Manufacturing 46 (2021) 102126

Fig. 5. Effect of the central deflection of sample N3 on bending response. (a) Load-deflection relations for different depths. (b) Effective bending stiffness. (c) Energy
absorption and dissipation. (d) Loss factor. (e)-(f) Deformation evolutions for maximum depth from 10 mm to 32.5 mm. Here d1 = 3, d2 = 6, d3 = 10, d4 = 11, d5
= 22 and d6 = 32.5 mm, respectively.

deflections (Section 3.2), in which the recoverability is strongly related thicknesses, where a power scaling law of the type kB ∼ T 2.4 is observed.
to maximum central deflection, the finding in this section indicates that It is well known that the bending stiffness of a bulk plate scales by its
thickness will not significantly affect recoverability. As one would thickness cubed, kB ∼ T 3 [72]. Compared with the power scaling law of a
expect, the sample thickness also impacts significantly the maximum bulk plate, it can be shown that the proposed fractal lattice design has
normalized load, which increases from 0.2452 N/mm at T = 10 mm to improved compliance and hence flexibility.
2.084 N/mm at T = 30 mm. For all thicknesses, the initial part of the Next, the influence of the thickness of the architected fractal beams
loading curve is linear. The force-displacement slopes are, of course, over their energy-related properties is illustrated. Fig. 6(c) shows that
larger in thick samples. After the initial linear response, samples with increased sample thickness results in higher energy absorption and
T ≥ 18mm exhibit stiffening points in the force-displacement curves, dissipation, and this effect becomes more significant for higher thickness
while others (10 ≤ T ≤ 18mm) increase linearly until the maximum values. To better explore this effect, loss factors for beams as a function
central deflection of 20 mm. From the stiffening point onwards, the of thicknesses are plotted in Fig. 6(d). The minimum loss factor is found
slope of the force-displacement curves shows nonlinear growth in the for a thickness of 18 mm. This result is somewhat counterintuitive, and
later stage and finally increases steadily with a fixed slope. The main several factors could explain this phenomenon. The intrinsic material
reason for this nonlinear behavior is that the upper ribs of the sample are damping associated with the viscoelasticity of the material contributes
in gradual contact with each other; when the contact area between ribs to a part of the energy dissipation. With the increase of central deflec­
reaches a peak, the curve then returns linear. A strong link between tion, stick-slip friction energy loss increases. It is apparent in Fig. 6(d)
stiffening point positions and sample thickness is also evident. The that the contribution of these two dissipation mechanisms provides a
stiffening points of the curves appear earlier in the case of thick samples, minimum at the thickness of 18 mm (η = 0.035). Of the thicknesses
varying from 16 mm (d/S = 0.107) at T= 18–6 mm (d/S = 0.4) at considered, the maximum value of loss factor is η = 0.057 with a
T= 30 mm. This can be attributed to an early rib contact present in the thickness of 10 mm. This dimensionless damping index quantitatively
case of large thickness samples. shows the excellent energy dissipation performance for this structure
Fig. 6(b) displays the relation between bending stiffness and sample [73–75].

7
Z. Zhang et al. Additive Manufacturing 46 (2021) 102126

Fig. 6. Effect of thickness of sample N3 on bending response. (a) Load-deflection relations for samples with different thicknesses. (b) Effective bending stiffness. (c)
Energy absorption and dissipation. (d) Loss factor. (e)-(f) Deformation evolutions for sample thickness of 10 mm and 30 mm. Here d1 = d4 = 3 mm, d2 = d5 = 5mm,
d3 = d6 = 20 mm, respectively.

To help better illustrate the experimental observations, Fig. 6 (e) and contact friction. Based on the numerical simulations, friction has been
(f) provide a comparison of the evolution of the εx strains for repre­ shown to play a more critical role in architected fractal beams with
sentative samples with T = 10 and T = 30 mm, respectively. These DIC increased fractal order, which is attributed to the augmented contact
results show that there is a larger and wider εx strain distribution for the area. For different thickness specimens, it has been shown that the
thicker sample (T = 30mm). contribution of these two mechanisms reaches a maximum for an in­
termediate thickness value of 18 mm. In addition to the three orders of
4. Conclusions fractal level in this work, the proposed fractal cutting strategy can be
employed to create lattice metamaterials with higher orders. It is ex­
In this work, the bending behavior of fractal lattice metamaterials pected that with further increase of fractal orders, lattice metamaterials
with self-similar unit cells over three orders has been investigated. can be engineered with lightweight, improved structural compliance,
Through a combination of three-point bending experiments and and energy dissipation. This study suggests fractal lattice metamaterials
nonlinear finite element computations, it has been shown that fractal have great potential to provide unique, tailorable energy dissipation,
architectures impart a unique combination of properties into the pro­ and shape recoverability properties that could find applications in many
posed lattice metamaterials, including energy dissipation, shape recov­ industrial sectors, such as defense, energy, and transportation.
erability, and tailorable compliance. Experimental results show that the
increased fractal order leads to a noticeable decrease in bending stiffness CRediT authorship contribution statement
under the same deformation; meanwhile, the structural shape recover­
ability can be enhanced sharply with no impact on structural integrity. Zhennan Zhang: Methodology, Formal analysis, Investigation, Data
Specimens with different fractal orders show almost 88% recovery upon Curation; Fabrizio Scarpa and Brett A. Bednarcyk: Investigation,
unloading. Although the energy absorption and dissipation values ob­ Writing - original draft; Yanyu Chen: Conceptualization, Writing - re­
tained from the bending tests on these different order fractal lattices are view & editing, Supervision. All authors have approved the final version
significantly different, their loss factors are very close, which is attrib­ of the manuscript.
uted to the synergistic effect between the intrinsic material damping and

8
Z. Zhang et al. Additive Manufacturing 46 (2021) 102126

Declaration of Competing Interest [27] Y. Cho, J.H. Shin, A. Costa, T.A. Kim, V. Kunin, J. Li, S.Y. Lee, S. Yang, H.N. Han, I.
S. Choi, D.J. Srolovitz, Engineering the shape and structure of materials by fractal
cut, Proc. Natl. Acad. Sci. USA 111 (49) (2014) 17390–17395.
The authors declare that they have no known competing financial [28] E. Narducci, K.Y. Lee, S.T. Pinho, Realising damage-tolerant nacre-inspired CFRP,
interests or personal relationships that could have appeared to influence J. Mech. Phys. Solids 116 (2018) 391–402.
the work reported in this paper. [29] V. Kunin, S. Yang, Y. Cho, P. Deymier, D.J. Srolovitz, Static and dynamic elastic
properties of fractal-cut materials, Extrem. Mech. Lett. 6 (2016) 103–114.
[30] J.A. Fan, W.H. Yeo, Y.W. Su, Y. Hattori, W. Lee, S.Y. Jung, Y.H. Zhang, Z.J. Liu, H.
Acknowledgments Y. Cheng, L. Falgout, M. Bajema, T. Coleman, D. Gregoire, R.J. Larsen, Y.G. Huang,
J.A. Rogers, Fractal design concepts for stretchable electronics, Nat. Commun. 5
(2014) 1–8.
YC gratefully acknowledges the NASA KY Space Grant and the start- [31] J.A. Michel, P.J. Yunker, Structural hierarchy confers error tolerance in biological
up fund from the Department of Mechanical Engineering at the Uni­ materials, Proc. Natl. Acad. Sci. USA 116 (8) (2019) 2875–2880.
versity of Louisville. FS acknowledges the support of the Faculty of [32] R.S. Farr, Fractal design for an efficient shell strut under gentle compressive
loading, Phys. Rev. E Stat. Nonlin Soft Matter Phys. 76 (5) (2007), 056608.
Engineering of the University of Bristol. The National Aeronautics & [33] D. Rayneau-Kirkhope, R.S. Farr, Y. Mao, Fractal-like dependence in the designs of
Space Administration (NASA), (No: 3200003095-21-068). efficient pressure-bearing structures, EPL 93 (3) (2011) 34002.
[34] I. Farina, R. Goodall, E. Hernández-Nava, A. di Filippo, F. Colangelo, F. Fraternali,
Design, microstructure and mechanical characterization of Ti6Al4V reinforcing
References elements for cement composites with fractal architecture, Mater. Des. 172 (2019),
107758.
[1] X. Zhang, A. Vyatskikh, H.J. Gao, J.R. Greer, X.Y. Li, Lightweight, flaw-tolerant, [35] D. Rayneau-Kirkhope, Y. Mao, R. Farr, Ultra-light hierarchical meta-materials on a
and ultrastrong nanoarchitected carbon, Proc. Natl. Acad. Sci. USA 116 (14) (2019) body-centred cubic lattice, EPL 119 (1) (2017) 14001.
6665–6672. [36] X.F. Man, Z. Luo, J. Liu, B.Z. Xia, Hilbert fractal acoustic metamaterials with
[2] D.R. Reid, N. Pashine, J.M. Wozniak, H.M. Jaeger, A.J. Liu, S.R. Nagel, J.J. de negative mass density and bulk modulus on subwavelength scale, Mater. Des. 180
Pablo, Auxetic metamaterials from disordered networks, Proc. Natl. Acad. Sci. USA (2019), 107911.
115 (7) (2018) 1384–1390. [37] H.J. Gao, Application of fracture mechanics concepts to hierarchical biomechanics
[3] O.R. Bilal, A. Foehr, C. Daraio, Bistable metamaterial for switching and cascading of bone and bone-like materials, Int. J. Fract. 138 (1–4) (2006) 101–137.
elastic vibrations, Proc. Natl. Acad. Sci. USA 114 (18) (2017) 4603–4606. [38] D. Rayneau-Kirkhope, Y. Mao, R. Farr, Ultralight fractal structures from hollow
[4] S. Yin, J.N. Li, H.Y. Chen, R.O. Ritchie, J. Xu, Design and strengthening tubes, Phys. Rev. Lett. 109 (20) (2012), 204301.
mechanisms in hierarchical architected materials processed using additive [39] W.J. Zhang, R. Neville, D.Y. Zhang, F. Scarpa, L.F. Wang, R. Lakes, The two-
manufacturing, Int. J. Mech. Sci. 149 (2018) 150–163. dimensional elasticity of a chiral hinge lattice metamaterial, Int. J. Solids Struct.
[5] J.R. Greer, J. Park, Additive manufacturing of nano- and microarchitected 141 (2018) 254–263.
materials, Nano Lett. 18 (4) (2018) 2187–2188. [40] C.S. Ha, E. Hestekin, J.H. Li, M.E. Plesha, R.S. Lakes, Controllable thermal
[6] J.T.B. Overvelde, K. Bertoldi, Relating pore shape to the non-linear response of expansion of large magnitude in chiral negative Poisson’s ratio lattices, Phys.
periodic elastomeric structures, J. Mech. Phys. Solids 64 (2014) 351–366. Status Solidi B 252 (7) (2015) 1431–1434.
[7] D.M. Kochmann, J.B. Hopkins, L. Valdevit, Multiscale modeling and optimization [41] I. Fernandez-Corbaton, C. Rockstuhl, P. Ziemke, P. Gumbsch, A. Albiez,
of the mechanics of hierarchical metamaterials, MRS Bull. 44 (10) (2019) 773–781. R. Schwaiger, T. Frenzel, M. Kadic, M. Wegener, New twists of 3D chiral
[8] L.R. Meza, A.J. Zelhofer, N. Clarke, A.J. Mateos, D.M. Kochmann, J.R. Greer, metamaterials, Adv. Mater. 31 (26) (2019) 1807742.
Resilient 3D hierarchical architected metamaterials, Proc. Natl. Acad. Sci. USA 112 [42] M. Upmanyu, H.L. Wang, H.Y. Liang, R. Mahajan, Strain-dependent twist-stretch
(37) (2015) 11502–11507. elasticity in chiral filaments, J. R. Soc. Interface 5 (20) (2008) 303–310.
[9] J.R. Greer, V.S. Deshpande, Three-dimensional architected materials and [43] P. Celli, S. Gonella, V. Tajeddini, A. Muliana, S. Ahmed, Z. Ounaies, Wave control
structures: design, fabrication, and mechanical behavior, MRS Bull. 44 (10) (2019) through soft microstructural curling: bandgap shifting, reconfigurable anisotropy
750–757. and switchable chirality, Smart Mater. Struct. 26 (3) (2017), 035001.
[10] X. Zhang, Y.J. Wang, B. Ding, X.Y. Li, Design, fabrication, and mechanics of 3D [44] W.W. Wu, W.X. Hu, G.A. Qian, H.T. Liao, X.Y. Xu, F. Berto, Mechanical design and
micro-/nanolattices, Small 16 (15) (2020) 1902842. multifunctional applications of chiral mechanical metamaterials: a review, Mater.
[11] Y. Chen, T. Li, F. Scarpa, L. Wang, Harnessing out-of-plane deformation to design Des. 180 (2019), 107950.
3D architected lattice metamaterials with tunable Poisson’s ratio, Sci. Rep. 7 (2) [45] Z.J. Wang, L.Q. Jing, K. Yao, Y.H. Yang, B. Zheng, C.M. Soukoulis, H.S. Chen, Y.
(2017) 8949. M. Liu, Origami-based reconfigurable metamaterials for tunable chirality, Adv.
[12] Y. Chen, T. Li, Z. Jia, F. Scarpa, C.W. Yao, L. Wang, 3D printed hierarchical Mater. 29 (27) (2017) 1700412.
honeycombs with shape integrity under large compressive deformations, Mater. [46] H. Bahaloo, Y.N. Li, Micropolar modeling of auxetic chiral lattices with tunable
Des. 137 (2018) 226–234. internal rotation, in: J. Appl. Mech. T. ASME, 86, 2019, 041002.
[13] M.C. Fernandes, J. Aizenberg, J.C. Weaver, K. Bertoldi, Mechanically robust [47] A. Holterman, Pattern Kerfing for Responsive Wooden Surfaces: A formal approach
lattices inspired by deep-sea glass sponges, Nat. Mater. 20 (2) (2020) 237–241. to produce flexible panels with acoustic performance, TU Delft, 2018.
[14] N. Ghavidelnia, M. Bodaghi, R. Hedayati, Idealized 3D Auxetic Mechanical [48] J.W. Boley, W.M. van Rees, C. Lissandrello, M.N. Horenstein, R.L. Truby,
Metamaterial: An Analytical, Numerical, and Experimental Study, Materials (Basel) A. Kotikian, J.A. Lewis, L. Mahadevan, Shape-shifting structured lattices via
(2021) 993. multimaterial 4D printing, Proc. Natl. Acad. Sci. USA 116 (42) (2019)
[15] T. Chen, M. Pauly, P.M. Reis, A reprogrammable mechanical metamaterial with 20856–20862.
stable memory, Nature 589 (7842) (2021) 386–390. [49] T.S. Lumpe, J. Mueller, K. Shea, Tensile properties of multi-material interfaces in
[16] N. Ghavidelnia, M. Bodaghi, R. Hedayati, Femur auxetic meta-implants with tuned 3D printed parts, Mater. Des. 162 (2019) 1–9.
micromotion distribution, Mater. (Basel) 14 (1) (2020) 114. [50] P. Tran, T.D. Ngo, A. Ghazlan, D. Hui, Bimaterial 3D printing and numerical
[17] L.R. Akhmetshin, Numerical study of the influence of relative parameters of the analysis of bio-inspired composite structures under in-plane and transverse
metamaterial structure on its mechanical behavior, IOP Conf. Ser. Mater. Sci. Eng. loadings, Compos. B. Eng. 108 (2017) 210–223.
1093 (1) (2021), 012001. [51] A. Mitchell, U. Lafont, M. Holynska, C. Semprimoschnig, Additive manufacturing -
[18] J. Bauer, L.R. Meza, T.A. Schaedler, R. Schwaiger, X.Y. Zheng, L. Valdevit, a review of 4D printing and future applications, Addit. Manuf. 24 (2018) 606–626.
Nanolattices: an emerging class of mechanical metamaterials, Adv. Mater. 29 (40) [52] ASTM, D790 - Flexural Properties of Unreinforced and Reinforced Plastics and
(2017) 1701850. Electrical Insulating Materials, ASTM Standard, 2017, p. 12.
[19] L. Salari-Sharif, T.A. Schaedler, L. Valdevit, Energy dissipation mechanisms in [53] C.Y. Smith, K.C. White, S. Suzuki, M. Chain, C.F. Cox, Application of Astm D790-92
hollow metallic microlattices, J. Mater. Res. 29 (16) (2014) 1755–1770. for flexural testing of unreinforced dental polymers, J. Dent. Res. 73 (1994), 322-
[20] T. Siegmund, F. Barthelat, R. Cipra, E. Habtour, J. Riddick, Manufacture and 322.
mechanics of topologically interlocked material assemblies, Appl. Mech. Rev. 68 [54] C. Majidi, R.E. Groff, Y. Maeno, B. Schubert, S. Baek, B. Bush, R. Maboudian,
(4) (2016), 040803. N. Gravish, M. Wilkinson, K. Autumn, R.S. Fearing, High friction from a stiff
[21] A.L. Ruschel, F.W. Zok, A bi-material concept for periodic dissipative lattices, polymer using microfiber arrays, Phys. Rev. Lett. 97 (7) (2006), 076103.
J. Mech. Phys. Solids 145 (2020), 104144. [55] I. Vu, Fracture and Friction Characterization of Polymer Interfaces, Virginia Tech,
[22] Q. Ge, H.J. Qi, M.L. Dunn, Active materials by four-dimension printing, Appl. Phys. 2015.
Lett. 103 (13) (2013), 131901. [56] G.A. Khalid, H. Bakhtiarydavijani, W.R. Whittington, R. Prabhu, M.D. Jones,
[23] M. Mirkhalaf, F. Barthelat, Design, 3D printing and testing of architectured Material response characterization of three poly jet printed materials used in a high
materials with bistable interlocks, Extrem. Mech. Lett. 11 (2017) 1–7. fidelity human infant skull, Mater. Today Proc. 20 (2020) 408–413.
[24] B. Bhushan, M. Caspers, An overview of additive manufacturing (3D printing) for [57] T.T. Li, Y.Y. Chen, L.F. Wang, Enhanced fracture toughness-in architected-
microfabrication, Microsyst. Technol. 23 (4) (2017) 1117–1124. interpenetrating phase composites by 3D printing, Compos. Sci. Technol. 167
[25] J.A. Fan, W.H. Yeo, Y. Su, Y. Hattori, W. Lee, S.Y. Jung, Y. Zhang, Z. Liu, H. Cheng, (2018) 251–259.
L. Falgout, M. Bajema, T. Coleman, D. Gregoire, R.J. Larsen, Y. Huang, J.A. Rogers, [58] A. Bula, M. Kozlowski, J. Hulimka, B. Chmielnicki, Analysis of methyl methacrylate
Fractal design concepts for stretchable electronics, Nat. Commun. 5 (2014) 3266. adhesive (MMA) relaxation with non-linear stress strain dependence, Int. J. Adhes.
[26] R. Oftadeh, B. Haghpanah, D. Vella, A. Boudaoud, A. Vaziri, Optimal fractal-like Adhes. 94 (2019) 40–46.
hierarchical honeycombs, Phys. Rev. Lett. 113 (10) (2014), 104301. [59] E.R. Wallace, T. Chaise, D. Nelias, Three-dimensional rolling/sliding contact on a
viscoelastic layered half-space, J. Mech. Phys. Solids 143 (2020), 104067.

9
Z. Zhang et al. Additive Manufacturing 46 (2021) 102126

[60] Stratasys, Objet260 3D Printer System User Guide (2014). [69] N.S. Pandya, H. Shah, M. Molana, A.K. Tiwari, Heat transfer enhancement with
[61] S. Wickramasinghe, T. Do, P. Tran, FDM-based 3D printing of polymer and nanofluids in plate heat exchangers: a comprehensive review, Eur. J. Mech. B
associated composite: a review on mechanical properties, defects and treatments, Fluids 81 (2020) 173–190.
Polymers (Basel) (2020) 1529. [70] L. Han, J. Yin, L.F. Wang, K.K. Chia, R.E. Cohen, M.F. Rubner, C. Ortiz, M.C. Boyce,
[62] S.E. Zeltmann, N. Gupta, N.G. Tsoutsos, M. Maniatakos, J. Rajendran, R. Karri, Tunable stimulus-responsive friction mechanisms of polyelectrolyte films and tube
Manufacturing and security challenges in 3D printing, JOM 68 (7) (2016) forests, Soft Matter 8 (33) (2012) 8642–8650.
1872–1881. [71] A. Takaku, R. Arridge, The effect of interfacial radial and shear stress on fibre pull-
[63] Q. Zhao, H.J. Qi, T. Xie, Recent progress in shape memory polymer: New behavior, out in composite materials, J. Phys. D Appl. Phys. 6 (17) (1973) 2038–2047.
enabling materials, and mechanistic understanding, Prog. Polym. Sci. 49–50 [72] S. Poincloux, T. Chen, B. Audoly, P.M. Reis, Bending response of a book with
(2015) 79–120. internal friction, Phys. Rev. Lett. 126 (21) (2021), 218004.
[64] F. Orban, Damping of materials and members in structures, J. Phys. Conf. Ser. 268 [73] J. Zhang, R.J. Perez, E.J. Lavernia, Documentation of damping capacity of metallic,
(2011), 012022. ceramic and metal-matrix composite-materials, J. Mater. Sci. 28 (9) (1993)
[65] A. Treviso, B. Van Genechten, D. Mundo, M. Tournour, Damping in composite 2395–2404.
materials: properties and models, Compos. B Eng. 78 (2015) 144–152. [74] A.P. Garland, K.M. Adstedt, Z.J. Casias, B.C. White, W.M. Mook, B. Kaehr, B.
[66] H. Jiang, H. Ziegler, Z.N. Zhang, H. Meng, D. Chronopoulos, Y.Y. Chen, Fabrication H. Jared, B.T. Lester, N.S. Leathe, E. Schwaller, B.L. Boyce, Coulombic friction in
and characterization of high molecular keratin based nanofibrous membranes for metamaterials to dissipate mechanical energy, Extrem. Mech. Lett. 40 (2020),
wound healing, Colloids Surf. B Biointerfaces 194 (2020), 111158. 100847.
[67] L. Liu, A Triaxiality-Dependent Constitutive Model For Predicting Damage-Induced [75] D.D.L. Chung, Structural composite materials tailored for damping, J. Alloys
Softening in 3D-Printed Soft Suture Layer, University of New Hampshire, 2018. Compd. 355 (1–2) (2003) 216–223.
[68] M. Hossain, R. Navaratne, D. Peric, 3D printed elastomeric polyurethane:
viscoelastic experimental characterizations and constitutive modelling with
nonlinear viscosity functions, Int. J. Non Linear Mech. 126 (2020), 103546.

10