We differentiate the spacer-dependent peak shift in coupled and decoupled super absorbing structures based on magnetic resonance and interference mechanism, respectively, which is experimentally validated by low-cost and large-area structures fabricated using lithography-free processes. The reversible real-time spectral tunability is then demonstrated by incorporating a thermally tunable polymeric spacer layer.

The unprecedented ability of nanoplasmonic/metamaterial structures to concentrate light has attracted significant research interests in recent years.1 Recently developed three-layered metal-dielectric-metal (MDM) metamaterial super absorbers2 provide a promising solution to couple the light into subwavelength dimension efficiently. These super absorbing structures have been developed in different spectral regions for various applications,3 ranging from organic photovoltaics,4,5 organic light-emitting diodes,6,7 thermal energy management,8,9 to surface-enhanced Raman spectroscopy (SERS).10 Particularly, for those energy-harvesting applications, it is essential to realize the spectral tunability of the resonant absorption band to maximize the overlap between the structurally enhanced absorption bands with the intrinsic absorption profile of energy harvesting materials.11 It has been demonstrated that the plasmonic resonance peak of three-layered super/perfect absorbers show strong dependence on the lateral dimension of top metallic nanopatterns.12–15 Although the resonance of periodically patterned structures can be tuned precisely using top-down lithography techniques, they are too expensive to implement in large-area fabrication and develop practical applications. To overcome this cost barrier, recently, we employed random metallic nanoparticles (NPs) to replace periodic nanopatterns to enable fast, low-cost, and large-area fabrication of super-absorbing metamaterial structures.16 However, when the nanopatterns are fabricated, their resonances are fixed. It is very difficult to tune the lateral dimension of these NPs in real time. In this work, we will report another freedom to tune the spectral position of the super absorbing resonance, i.e., the spacer thickness. Coupled and decoupled resonances can be obtained depending on the spacer thickness. Combined with spatial tunable materials, the tunable spacer layer may enable switchable or controllable super absorbing structures/materials for on chip optical applications.17 

Fig. 1(a) shows a conceptual illustration of the three-layered absorber structure, consisting of a dielectric spacer layer (e.g., SiO2) sandwiched by top random silver (Ag) NPs and an optically opaque Ag ground plate. In this experiment, these three layers were deposited by e-beam evaporation. The mass thicknesses of the two metallic layers were set to 12 nm and 200 nm, respectively. The top Ag random NPs were obtained by thermally annealing the top metal film under 200 °C. The morphology of the thermally treated Ag film was characterized by the scanning electronic microscope (SEM) in Fig. 1(b), showing randomly distributed Ag NPs with inter-particle distances between 10 and 30 nm. These NPs with small nanogaps have been reported to assist the extremely concentrated localized field for improved nonlinear optics,18 photoluminescence,19 and SERS signals.20 Fig. 1(c) shows the tilted cross-sectional SEM image of the fabricated structure with the spacer thickness of DT = 35 nm. One can see that the top Ag NPs can be considered as domes approximately. The lateral dimension distribution of these NPs was analyzed statistically based on the top-view SEM image shown in Fig. 1(b). As plotted in the inset of Fig. 1(d), the lateral dimension of dominant Ag NPs (e.g., percentage > 5%) ranges from 20 to 70 nm. The solid red curve is the Gaussian fit for the NP lateral dimension distribution with the peak at 30–40 nm. Large NPs (i.e., size > 80 nm) only occupy 2%, and therefore, with negligible contribution to the overall optical absorption properties. The absorption spectrum of the fabricated three-layered structure was characterized using a Fourier transform infrared spectroscopy (FTIR, Bruker VERTEX 70) with an extended light source covering visible and near-infrared wavelength region, as shown in Fig. 1(d). An absorption peak of 97.6% was observed at the wavelength of 651 nm under the normal incidence. Due to the broad range of the lateral dimensions of top NPs, the absorption profile is relatively broad (e.g., over 80% absorption in a relatively wide spectral range of 537–768 nm).

FIG. 1.

(a) Conceptual illustration of the proposed three-layered metal-dielectric-metal structure consisting of a SiO2 spacer layer sandwiched by top Ag NPs and a Ag ground plate. DT denotes the spacer thickness. (b) The top view SEM image of random Ag NPs with the nominal mass thickness of 14 nm. (c) The cross-sectional SEM image of the three-layered structure with the spacer thickness DT = 35 nm. (d) Absorption spectrum of the MDM structure with DT = 35 nm under normal incidence. Inset: Statistic size distribution of NPs shown in (b). The red curve is a Gaussian fit. (e) Measured absorption spectra of three-layered MDM samples as the function of the spacer layer thickness DT at normal incidence. Black dots indicate the absorption peak position.

FIG. 1.

(a) Conceptual illustration of the proposed three-layered metal-dielectric-metal structure consisting of a SiO2 spacer layer sandwiched by top Ag NPs and a Ag ground plate. DT denotes the spacer thickness. (b) The top view SEM image of random Ag NPs with the nominal mass thickness of 14 nm. (c) The cross-sectional SEM image of the three-layered structure with the spacer thickness DT = 35 nm. (d) Absorption spectrum of the MDM structure with DT = 35 nm under normal incidence. Inset: Statistic size distribution of NPs shown in (b). The red curve is a Gaussian fit. (e) Measured absorption spectra of three-layered MDM samples as the function of the spacer layer thickness DT at normal incidence. Black dots indicate the absorption peak position.

Close modal

To explore the influence of the SiO2 spacer thickness, DT, we then fabricated 24 pieces three-layered metal-dielectric-metal samples with the DT from 7 to 103 nm (with the step size of ∼4 nm). The top Ag NPs were fabricated in the same deposition condition. Fig. 1(e) shows the dependence of the optical absorption spectra on the spacer thickness, where black dots indicate the absorption peak positions. Obviously, one can see two distinct trends of resonant wavelength shift: When DT increases from 7 nm to 19 nm, the absorption peak blue-shifts from 704 nm to 619 nm. When DT continues to increase from 27 nm to 103 nm, the absorption peak red-shifts from 619 to 1000 nm, covering a broad spectral band from visible to near-infrared wavelengths.

To understand the mechanism of these two distinct spectral shifts, we first employed the spectroscopic ellipsometer (J. A. Woollam, VASE) to characterize the effective optical constants of the top nanopatterned layer as shown in Fig. 2(a). These data were then substituted in the three-dimensional (3D) finite element method model to simulate the absorption spectra and spatial distribution of magnetic field (similar processes were also employed in Refs. 21 and 22). As shown in Fig. 2(b), if we consider the NP layer as a uniform effective thin film, the simulated resonant absorption peak will red-shift monotonically with the increasing spacer layer thickness (i.e., the Fabry-Pérot resonant mechanism or decoupled situation23), which, however, only agrees with the experimental observation in the large DT region. Obviously, in the small DT region, the optical behavior cannot be explained by the effective medium theory and interference mechanism accurately,24,25 where each NP will function as an optical antenna coupled with the bottom ground plane, exciting the magnetic resonance supported by metamaterial super absorber/patch antenna.26 

FIG. 2.

(a) Measured effective refractive indices of the top Ag NP layer. (b) Simulated spacer-dependent absorption peak shift (black dots) by considering a uniform effective top layer. Red diamonds are peak positions extracted from experimental data shown in Fig. 1(e) for comparison. (c) Modeled spacer-dependent absorption spectra for a three-layered MDM structure. (d) Modeled spatial distribution of magnetic field at the peak wavelength of 789 nm.

FIG. 2.

(a) Measured effective refractive indices of the top Ag NP layer. (b) Simulated spacer-dependent absorption peak shift (black dots) by considering a uniform effective top layer. Red diamonds are peak positions extracted from experimental data shown in Fig. 1(e) for comparison. (c) Modeled spacer-dependent absorption spectra for a three-layered MDM structure. (d) Modeled spatial distribution of magnetic field at the peak wavelength of 789 nm.

Close modal

To validate this magnetic resonance mechanism, we then model the optical absorption property of a three-layered nanopatterned structure by controlling the spacer thickness. It should be noted that such optical patch antenna is mainly dependent on the geometric parameters of individual patterns rather than the periodic alignment of these patterns, as explained in Ref. 12. In this modeling, periodic boundary conditions were employed in the 3D model since it is easier for numerical simulation. Although using periodic structure to interpret the optical properties of random structures will introduce inevitable inaccuracy, we will show that the two distinct wavelength shift mechanisms can still be revealed. Similar process was also employed in other literature (e.g., Ref. 12 employed periodic modeling to interpret the optical property of randomly distributed optical patch antennas). In this modeling, the lateral dimension of Ag NPs is selected to 30 nm (i.e., the peak dimension in the inset of Fig. 1(d)) and the gap distance between adjacent NPs is 12.5 nm. As shown in Fig. 2(c), a strong blue-shift of the absorption peak was observed as the spacer thickness DT increases to 15 nm. Beyond this thickness, the absorption peak red-shifts, agreeing well with the observed shift in the experiment. Although the coupled-decoupled point is different from the experimental observation of the directly deposited structure with random NPs, the transition from blue-shift resonance to red-shift resonance was reproduced using numerical simulation. To further reveal the physical mechanism of the strongly coupled resonance, the spatial distribution of the magnetic field when DT = 5 nm was plotted in Fig. 2(d) at the peak wavelength of 789 nm. One can see that the resonant mode is confined under the Ag nanopattern due to the strongly coupled magnetic resonance within the optical patch antenna constructed by the top antenna and the ground plane separated by a thin spacer layer.13 

For the decoupled situation, by tuning the phase difference at different interfaces within the three-layered structure, the destructive interference condition was met to achieve strong absorption. This thin-film interference phenomenon received extensive interest in recent years27 for energy harvesting and conversion applications (e.g., Ref. 28). Under an ideal condition, the metal reflecting substrate can be treated as the perfect electron conductor (PEC) with a π phase shift for the reflection component. However, due to the intrinsic optical constants of different metal films, the absorption and phase change at the metal substrate surface are different from the ideal situation (e.g., Refs. 24 and 29). Therefore, the optical properties of the bottom metal film have to be considered in the entire interference mechanism (see more details in a recent comprehensive review article27). In this experiment, Ag back reflector was selected to reduce the penetration of light into the metal, thus resulting in the major absorption in the top effective thin film, which is potentially useful for energy harvesting and conversion applications. Importantly, since the spectral position of the resonance is sensitive to the spacer thickness, a reversibly tunable three-layered super absorbing structure is achievable by introducing stretchable materials like stimuli-responsive hydrogels and polymer/biomolecules,30 as will be demonstrated in the next paragraph.

After revealing the spacer-dependent wavelength shift mechanism, we then explored the potential to realize real-time tunability. According to previous literature, phase change materials like VO2 (Ref. 31) and Ge3Sb2Te6 (GST-326)32 have been employed as the spacer layer based on the large optical constant change. For the proposed mechanism, a spacer layer with tunable thickness is required. Here, we employed a nanofilm consisting of poly(N-isopropylacrylamide) modified polyallylamine hydrochloride (PNIPAM-PAH) as the polycation and polyacrylic acid (PAA) as the polyanion to demonstrate the feasibility33 (see the supplementary material34 for details of sample preparation). PNIPAM-PAH is a temperature-sensitive polymer with a lower critical transition temperature (LCST) at 32 °C, which can be dissolved in water at 25 °C and become water-insoluble at 35 °C due to the collapse of polymer network.35 As shown in Fig. 3(a), the PNIPAM-PAH solution is transparent at 25 °C and milky at 35 °C, respectively. Meanwhile, PNIPAM-PAH is a weakly charged polycation and can form a nanofilm with negatively charged PAA through the layer-by-layer (LbL) assembly method.36 Due to the temperature-sensitive property of the PNIPAM-PAH, the thickness of the resulting film is expected to change as the environmental temperature changes between room temperature and above PNIPAM-PAH's LCST. Because poly(N-isopropylacrylamide) is hydrophilic at 25 °C while hydrophobic at 35 °C, we expected the PNIPAM-PAH/PAA film to lose water molecules above LCST. As a result, the film is thicker at the low temperature while thinner at high temperature, as illustrated in Fig. 3(b). Therefore, the spacer thickness can be tuned by controlling the environmental temperature. To validate this principle, we fabricated a PNIPAM-PAH/PAA multilayer thin film on top of a 60-nm-thick aluminum (Al) film followed by a simple LbL dip coating protocol.37 The initial thickness of the PNIPAM-PAH/PAA film is controlled by the number of LbL cycles (i.e., the number of PNIPAM-PAH/PAA bilayers). To avoid the thermal damage of the polyelectrolyte, an 8-nm-thick Ag film was directly sputtered on top of the polymer film. Since the mass thickness is smaller than the percolation threshold of Ag film on top of these polymer surfaces, NPs can be obtained directly as shown by the SEM image in Fig. 3(c). The inset of Fig. 3(c) is a statistic analysis of the NP size distribution. As shown by the Gaussian fit of the histogram (red curve), the majority NPs are between 15 nm to 25 nm.

FIG. 3.

(a) Photos of PNIPAM-PAH in water at 25 °C and 35 °C. (b) Conceptual illustration of multilayer PNIPAM-PAH/PAA films sandwiched by top Ag NPs and Al ground plates. (c) Top SEM image of the directly deposited Ag NPs with the mass thickness of 8 nm. The inset is the histogram of the NP size distribution. The red line is the Gaussian fit. (d) Relative film thickness of a 10-bilayered PNIPAM-PAH/PAA nanofilm on a silicon substrate with the temperature changed repeatedly between 25 °C (black squares) and 35 °C (blue dots). (e) The absorption spectrum of a three-layered absorber sample with the PNIPAM-PAH/PAA thickness of ∼500 nm. (f) Temperature dependent absorption peak in a PNIPAM-PAH/PAA-based absorber.

FIG. 3.

(a) Photos of PNIPAM-PAH in water at 25 °C and 35 °C. (b) Conceptual illustration of multilayer PNIPAM-PAH/PAA films sandwiched by top Ag NPs and Al ground plates. (c) Top SEM image of the directly deposited Ag NPs with the mass thickness of 8 nm. The inset is the histogram of the NP size distribution. The red line is the Gaussian fit. (d) Relative film thickness of a 10-bilayered PNIPAM-PAH/PAA nanofilm on a silicon substrate with the temperature changed repeatedly between 25 °C (black squares) and 35 °C (blue dots). (e) The absorption spectrum of a three-layered absorber sample with the PNIPAM-PAH/PAA thickness of ∼500 nm. (f) Temperature dependent absorption peak in a PNIPAM-PAH/PAA-based absorber.

Close modal

In addition, we also confirmed that the thickness of PNIPAM-PAH/PAA film can be changed repeatedly by controlling the environmental temperature. Specifically, the optical dielectric constants and thicknesses of these polymer films were characterized using spectroscopic ellipsometer (see the supplementary material34 for characterization details). As shown in Fig. 3(d), the average thickness of the film at 25 °C is ∼22 nm and undergoes approximately 10% decrease as the temperature increases to 35 °C. One can see that multiple heating-cooling cycles did not affect the relative film thickness. Next, temperatures from 25 °C to 40 °C were applied to the three-layered absorber sample with the PNIPAM-PAH/PAA thickness of ∼500 nm to demonstrate the tunability of PNIPAM-PAH/PAA-based absorbers in the interference coupling region. Two strong absorption peaks were obtained at the wavelength of 530 nm and 742 nm, as shown in Fig. 3(e). By controlling the temperature between 25 °C and 40 °C, the round-trip tuning of the absorption peak (e.g., at 742 nm, see the dashed square in Fig. 3(e)) was demonstrated in a PNIPAM-PAH/PAA-based absorber. As shown in Fig. 3(f), when the temperature increased from 25 °C to 40 °C, the polymer spacer layer will shrink, resulting in the blue-shift of the resonant peak from 742 nm to 730 nm. On the other hand, when the temperature decreased from 40 °C back to 25 °C, the resonant absorption peak can be tuned back to 741 nm. Therefore, we demonstrated the reversible tunability of the resonant absorption of a fully fabricated three-layered absorber structure. If spacer materials with larger spatial tunability can be employed in the three-layered structure (e.g., air gap cavity38), a larger spectral tunability will be realizable. It should be noted that the localized heat effect of plasmonic modes can be neglected since the incident light source is a tungsten-halogen lamp (with the power density of ∼100 mW/cm2), which is significantly weaker than those strong laser sources employed in previously reported plasmonic thermal effect studies (e.g., 20 kW/cm2 in Ref. 39, 350 kW/cm2 in Ref. 40). In addition, when we placed neutral density filters to reduce the light source intensity to characterize the optical properties of the sample, the reflection/absorption spectra are identical, indicating the negligible heating effect from the light source. Therefore, the absorption peaks shift can only be attributed to the thickness change of PNIPAM-PAH/PAA film induced by the sample stage temperature, instead of the incident light power.

In conclusion, we explored the spacer thickness dependent mechanism of a three-layered super absorber structure in magnetic resonance and interference regimes. As the spacer thickness increases, two distinct trends of peak shifts were observed, which were attributed to a transition between strongly coupled to decoupled modes. Experimental results revealed real-time tuning of the resonant absorption peak position on the dielectric spacer thickness. A stimuli-responsive polymeric PNIPAM-PAH/PAA multilayer was then employed as the thickness-tunable spacer. As a result, we observed a round-trip and real time absorption peak tuning. This real-time tuning mechanism not only benefits the development of various novel optical devices with flexibly tunable resonance peak positions for color change and display applications, but also can potentially work as a highly compact and portable optical thermometer to efficiently transduce any change in the temperature to the peak wavelength shift.41 

Q. Gan acknowledges funding support from National Science Foundation (Grant Nos. ECCS1507312, CBET1445934, and ECCS1425648). S. Jiang was supported by NSFC (Award No. 51001029). Y. Xu was supported by the NSFC under Grant No. 61177070. W. Li acknowledges funding support from New Faculty Startup Funds from Texas Tech University. N. Zhang and Z. Liu acknowledge the financial support from Chinese Scholarship Council (CSC).

1.
D. K.
Gramotnev
and
S. I.
Bozhevolnyi
,
Nat. Photonics
8
,
13
(
2014
).
2.
C. M.
Watts
,
X.
Liu
, and
W. J.
Padilla
,
Adv. Mater.
24
,
OP98-OP120
(
2012
).
3.
Y.
Cui
,
Y.
He
,
Y.
Jin
,
F.
Ding
,
L.
Yang
,
Y.
Ye
,
S.
Zhong
,
Y.
Lin
, and
S.
He
,
Laser Photonics Rev.
8
,
495
520
(
2014
).
4.
Q.
Gan
,
F. J.
Bartoli
, and
Z. H.
Kafafi
,
Adv. Mater.
25
,
2385
(
2013
).
5.
Q.
Gan
,
F. J.
Bartoli
, and
Z.
Kafafi
,
IEEE Photonics J.
4
,
620
(
2012
).
6.
J.
Tischler
,
M.
Bradley
, and
V.
Bulovic
,
Phys. Rev. Lett.
95
,
036401
(
2005
).
7.
F.
Withers
,
O.
Del Pozo-Zamudio
,
A.
Mishchenko
,
A. P.
Rooney
,
A.
Gholinia
,
K.
Watanabe
,
T.
Taniguchi
,
S. J.
Haigh
,
A. K.
Geim
,
A. I.
Tartakovskii
, and
K. S.
Novoselov
,
Nat. Mater.
14
,
301
306
(
2015
).
8.
Z.
Yu
,
N. P.
Sergeant
,
T.
Skauli
,
G.
Zhang
,
H.
Wang
, and
S.
Fan
,
Nat. Commun.
4
,
1730
(
2013
).
9.
A. P.
Raman
,
M. A.
Anoma
,
L.
Zhu
,
E.
Rephaeli
, and
S.
Fan
,
Nature
515
,
540
544
(
2014
).
10.
N.
Zhang
,
K.
Liu
,
Z.
Liu
,
H.
Song
,
X.
Zeng
,
D.
Ji
,
A.
Cheney
,
S.
Jiang
, and
Q.
Gan
,
Adv. Mater. Interfaces
2
,
1500142
(
2015
).
11.
K.
Liu
,
H.
Hu
,
H.
Song
,
X.
Zeng
,
D.
Ji
,
S.
Jiang
, and
Q.
Gan
,
IEEE Photonics Technol. Lett.
25
,
1266
1269
(
2013
).
12.
A.
Moreau
,
C.
Ciracì
,
J. J.
Mock
,
R. T.
Hill
,
Q.
Wang
,
B. J.
Wiley
,
A.
Chilkoti
, and
D. R.
Smith
,
Nature
492
,
86
89
(
2012
).
13.
C.
Ciracì
,
R. T.
Hill
,
J. J.
Mock
,
Y.
Urzhumov
,
A. I.
Fernández-Domínguez
,
S. A.
Maier
,
J. B.
Pendry
,
A.
Chilkoti
, and
D. R.
Smith
,
Science
337
,
1072
1074
(
2012
).
14.
M. J.
Rozin
,
D. A.
Rosen
,
T. J.
Dill
, and
A. R.
Tao
,
Nat. Commun.
6
,
7325
(
2015
).
15.
G. M.
Akselrod
,
J.
Huang
,
T. B.
Hoang
,
P. T.
Bowen
,
L.
Su
,
D. R.
Smith
, and
M. H.
Mikkelsen
,
Adv. Mater.
27
,
8028
8034
(
2015
).
16.
K.
Liu
,
X.
Zeng
,
S.
Jiang
,
D.
Ji
,
H.
Song
,
N.
Zhang
, and
Q.
Gan
,
Nanoscale
6
,
5599
5605
(
2014
).
17.
R.
Guo
,
M.
Decker
,
F.
Setzpfandt
,
I.
Staude
,
D. N.
Neshev
, and
Y. S.
Kivshar
,
Nano Lett.
15
,
3324
3328
(
2015
).
18.
M.
Kauranen
and
A. V.
Zayats
,
Nat. Photonics
6
,
737
748
(
2012
).
19.
A.
Kinkhabwala
,
Z.
Yu
,
S.
Fan
,
Y.
Avlasevich
,
K.
Müllen
, and
W. E.
Moerner
,
Nat. Photonics
3
,
654
657
(
2009
).
20.
Y. J.
Oh
and
K. H.
Jeong
,
Adv. Mater.
24
,
2234
2237
(
2012
).
21.
N.
Zhang
,
K.
Liu
,
H.
Song
,
Z.
Liu
,
D.
Ji
,
X.
Zeng
,
S.
Jiang
, and
Q.
Gan
,
Appl. Phys. Lett.
104
,
203112
(
2014
).
22.
L.
Malassis
,
P.
Massé
,
M.
Tréguer-Delapierre
,
S.
Mornet
,
P.
Weisbecker
,
P.
Barois
,
C. R.
Simovski
,
V. G.
Kravets
, and
A. N.
Grigorenko
,
Adv. Mater.
26
,
324
(
2014
).
23.
D.
Zhao
,
L.
Meng
,
H.
Gong
,
X.
Chen
,
Y.
Chen
,
M.
Yan
,
Q.
Li
, and
M.
Qiu
,
Appl. Phys. Lett.
104
,
221107
(
2014
).
24.
H.
Song
,
L.
Guo
,
Z.
Liu
,
K.
Liu
,
X.
Zeng
,
D.
Ji
,
N.
Zhang
,
S.
Jiang
, and
Q.
Gan
,
Adv. Mater.
26
,
2737
2743
(
2014
).
25.
H.
Song
,
S.
Jiang
,
D.
Ji
,
X.
Zeng
,
N.
Zhang
,
K.
Liu
,
C.
Wang
,
Y.
Xu
, and
Q.
Gan
,
Opt. Express
23
,
7120
7130
(
2015
).
26.
N.
Liu
,
M.
Mesch
,
T.
Weiss
,
M.
Hentschel
, and
H.
Giessen
,
Nano Lett.
10
,
2342
(
2010
).
27.
M. A.
Kats
,
R.
Blanchard
,
S.
Ramanathan
, and
F.
Capasso
,
Opt. Photonics News
25
,
40
(
2014
).
28.
S.
Linic
,
P.
Christopher
, and
D.
Ingram
,
Nat. Mater.
10
,
911
(
2011
).
29.
M. A.
Kats
,
R.
Blanchard
,
P.
Genevet
, and
F.
Capasso
,
Nat. Mater.
12
,
20
(
2013
).
30.
X.
Huang
,
Y.
Sun
, and
S.
Soh
,
Adv. Mater.
27
,
4062
4068
(
2015
).
31.
H.
Wang
,
Y.
Yang
, and
L.
Wang
,
Appl. Phys. Lett.
105
,
071907
(
2014
).
32.
A.
Tittl
,
A. U.
Michel
,
M.
Schäferling
,
X.
Yin
,
B.
Gholipour
,
L.
Cui
,
M.
Wuttig
,
T.
Taubner
,
F.
Neubrech
, and
H.
Giessen
,
Adv. Mater.
27
,
4597
4603
(
2015
).
33.
34.
See supplementary material at http://dx.doi.org/10.1063/1.4943089 for details.
35.
M. J.
Serpe
,
C. D.
Jones
, and
L. A.
Lyon
,
Langmuir
19
,
8759
8764
(
2003
).
36.
S. S.
Shiratori
and
M. F.
Rubner
,
Macromolecules
33
,
4213
4219
(
2000
).
37.
P. T.
Hammond
,
Mater. Today
15
,
196
206
(
2012
).
38.
N. G.
Boddeti
,
X.
Liu
,
R.
Long
,
J.
Xiao
,
J. S.
Bunch
, and
M. L.
Dunn
,
Nano Lett.
13
,
6216
6221
(
2013
).
39.
J. B.
Herzog
,
M. W.
Knight
, and
D.
Natelson
,
Nano Lett.
14
,
499
(
2014
).
40.
A. S.
Urban
,
M.
Fedoruk
,
M. R.
Horton
,
J. O.
Rädler
,
F. D.
Stefani
, and
J.
Feldmann
,
Nano Lett.
9
,
2903
(
2009
).
41.
T.
Xu
,
Y.
Wu
,
X.
Luo
, and
L.
Jay Guo
,
Nat. Commun.
1
,
59
(
2010
).

Supplementary Material