Organic Electronics 12 (2011) 1098–1102

Contents lists available at ScienceDirect

Organic Electronics

journal homepage: www.elsevier .com/locate /orgel

Effects of single walled carbon nanotubes on the electroluminescentperformance of organic light-emitting diodes

Ming Shao a, Matthew P. Garrett b, Xinjun Xu c, Ilia N. Ivanov d,⇑, Stanislaus S. Wong e, Bin Hu a,⇑a Department of Materials Science and Engineering, University of Tennessee, Knoxville, TN 37996, USAb Wu Han National Laboratory for Optoelectronics, Huazhong University of Science and Technology, Wu Han 430074, PR Chinac School of Materials Science and Engineering, University of Science and Technology Beijing, Beijing 100083, PR Chinad Center for Nanophase Materials Sciences, Oak Ridge National Laboratory, Oak Ridge, TN 37831, USAe Department of Chemistry, State University of New York at Stony Brook, New York 11794-3400, USA

a r t i c l e i n f o

Article history:Received 5 February 2011Received in revised form 2 March 2011Accepted 6 March 2011Available online 27 March 2011

Keywords:Single walled carbon nanotubes (SWNTs)Organic light-emitting diodes (OLEDs)ElectroluminescenceCarrier transport

1566-1199/$ - see front matter � 2011 Elsevier B.Vdoi:10.1016/j.orgel.2011.03.003

⇑ Corresponding authors.E-mail addresses: ivanovin@ornl.gov (I.N. Iv

(B. Hu).

a b s t r a c t

Effects of single walled carbon nanotubes (SWNTs) on the electroluminescent performanceof organic light-emitting diodes (OLEDs) have been investigated by mixing them in a hole-conducting layer and in a light-emitting layer in OLEDs. We found that SWNTs playdifferent roles when used as polymer:SWNT composites in OLEDs. When used in a hole-conducting layer, SWNTs facilitate the charge transport in the transport layer and on theother hand they also act as the exciton quenching centers at the transporting/emittinginterface provided their concentration is high enough. When used in a light-emitting layer,SWNTs act as an n-type dopant to increase electron transport in p-type electroluminescentfilm and subsequently improve the balancing degree of bipolar injection, leading to anenhancement in the electroluminescence efficiency.

� 2011 Elsevier B.V. All rights reserved.

1. Introduction

Recently, much attention has been paid to use nanoma-terials for improving the electroluminescent (EL) perfor-mance of organic light-emitting diodes (OLEDs) [1–4].Among those nanomaterials, single walled carbon nano-tubes (SWNTs) are expected to be useful in improvingOLEDs due to their well-established synthesis process,excellent mechanical properties, and good carrier-transporting ability. Although SWNTs have been reportedto improve the EL efficiency when used as polymer:SWNTcomposites, their role in OLEDs is not clear yet. Severalpossible explanations have been proposed, including (a)improving hole injection and transport [5], (b) improvingthe conductivity of polymer films [6], (c) hole-blocking inthe polymer composite [7], and (d) hole trapping of SWNTsin a hole-conducting polymer [8]. However, these explana-

. All rights reserved.

anov), bhu@utk.edu

tions are not suitable for explaining all the phenomena re-ported in the literature.

In this paper, we investigated the roles of SWNTs on theEL performance of OLEDs when mixing them with a hole-conducting material and a light-emitting material. Ourfindings provide a clear understanding of the role ofSWNTs in OLEDs and will be helpful for improving OLEDperformance by using SWNTs.

2. Experimental

The SWNTs were synthesized by a laser ablation meth-od [9], and was purified by hydrothermal and chemicaltreatments. The SWNTs contain two-third semiconductingand one-third metallic nanotubes [10]. As a result, theSWNTs can exhibit both charge transport and electronwithdrawing abilities. The hole injection materialpoly(3,4-ethylenedioxythiophene) doped with poly(sty-rene sulfonate) (Baytron P 4083) (PEDOT) was acquiredfrom the Bayer Company. The hole-transporting material(PVK) and the electron-transporting material (PBD) were

0 50 100 1500

50

100

150

200

0.02% SWNTs

0% SWNTs

0.001% SWNTs

0.01% SWNTs

EL

Inte

nsity

(a.

u.)

Current Density (mA/cm2)

(b)

0 4 8 12 16

0

50

100

150

0.02% SWNTs

0.01% SWNTs

0.001% SWNTs

0% SWCNTs

Cur

rent

(m

A/c

m2 )

Voltage (V)

(a)

Fig. 1. The J–V curves (a) and the EL intensity–current density curves (b)of devices with the structure of ITO/PEDOT:SWNTs (x wt.%, 40 nm)/PFO(80 nm)/Ca (20 nm)/Al (80 nm).

M. Shao et al. / Organic Electronics 12 (2011) 1098–1102 1099

purchased from Sigma–Aldrich Company and used as re-ceived. The light-emitting conjugated polymer polyfluo-rene (PFO, purchased from H.W. Sand Company) andSuper-Yellow (purchased from Merck Company) were usedas received. The electrophosphorescent material Iridium(III) tris(2-(4-totyl)pyridinato-N,C2) [Ir(mppy)3] was pur-chased from American Dye Source Inc. In order to investi-gate the role of SWNTs on the performance of the OLEDs,three types of device architecture were fabricated. The firsttype of devices have the structure of ITO/PEDOT:SWNT (xwt.%, 40 nm)/PFO (80 nm)/Ca (20 nm)/Al (80 nm), wherex is 0, 0.001, 0.01, and 0.02, relative to solid PEDOT content.The second one is ITO/PEDOT:SWNT (0.01 wt.% relative tosolid PEDOT content, 40 nm)/PVK:PBD:Ir(mppy)3

(69 wt.%:30 wt.%:1 wt.%, 60 nm)/Ca (20 nm)/Al (80 nm),which use Ir(mppy)3 as a phosphorescent emissive mate-rial. The third one has a structure of ITO/PEDOT (40 nm)/Super-Yellow:SWNTs (x wt.%, 80 nm)/Ca (20 nm)/Al(80 nm), where x is 0 and 0.005.

To get the uniform dispersion of SWNTs in PEDOT, wesuspended the SWNTs functionalized by with crown ether,making them dispersible in deionized water [11]. Theywere then ultrasonicated the SWNTs solution 30 min, andthen we mixed the SWNT solution with PEDOT solutionand deionized water to form the composites at designeddoping concentration. For the Super-Yellow:SWNTs com-posite preparation, SWNTs dispersed in chloroform wereadded into Super-Yellow chloroform solution to give thedesired doping concentrations. The PEDOT:SWNTs com-posites were spin-coated onto ITO-coated glass substrateand the thickness of the film was 40 nm. The light-emittinglayers for the three types of devices were all formed byspin coating from their chloroform solutions. Finally, thecalcium (Ca) and aluminum (Al) electrode were depositedsequentially by thermal evaporation under the vacuum of2 � 10�6 Torr, and the deposition rate was typically about1 Å/s. The thicknesses and the morphology of the filmsand were measured by a Veeco atomic force microscopy(AFM) using a tapping mode. The resistivity and conductiv-ity of PEDOT:SWNTs composite films on glass substrateswere measured by a four-probe method with a Keithleysource meter. The current and voltage characteristics ofthe OLEDs were measured with Keithley 2400 source me-ter. The EL spectra of OLEDs were recorded by a HORIBA Jo-bin Yvon spectrometer. The luminance and efficiency ofOLEDs were measured by an OLED testing system cali-brated by the National Institute of Standards andTechnology.

Table 1The resistivity (q) and conductivity (C) of PEDOT:SWNT composite filmswith various SWNT concentration.

SWNT concentration (%) q (X cm) C (S/cm)

0 44,902 2.23 � 10�5

0.0004 40,778 2.45 � 10�5

0.0042 31,709 3.15 � 10�5

0.0010 29,775 3.36 � 10�5

0.6622 23,666 4.23 � 10�5

3. Results and discussion

We investigated the roles of SWNTs in a hole conduct-ing layer on the performance of OLEDs at first. Fig. 1ashows the current density–voltage (J–V) characteristics ofOLEDs employing PEDOT:SWNTs composites with variousdoping concentrations under a forward bias. With increas-ing the SWNTs concentration in PEDOT, the current densityof devices becomes higher under the same voltage, whichmeans the hole injection and/or transport are improvedby introducing SWNTs in the PEDOT layer of devices. How-

ever, the EL intensity did not show a monotonous increasewhen increasing the SWNTs concentration in PEDOT underthe same current density. From Fig. 1b we can see thatwhen the SWNTs concentration in PEDOT is no higher than0.01 wt.%, then the higher the doping concentration, thestronger the EL intensity. But when the doping concentra-tion of SWNTs reaches 0.02 wt.%, the EL performance of thedevices deteriorates, and the EL intensity becomes lowerthan that of the device without SWNTs doping. Our resultshave some differences compared to those reported by Wooet al. [8]. In their case, the doping concentration of SWNTsin PEDOT is no less than 0.05 wt.%. Although they also ob-served the enhancement of current density when increas-ing the SWNTs doping concentration in PEDOT, they onlyfound that EL brightness in devices employing PED-OT:SWNTs composites was reduced relative to the devicewithout using SWNTs. As a result, their explanation ofSWNTs as hole traps in PEDOT film should be reconsidered,since there would be no EL intensity enhancement in ourcase if the injected holes were initially trapped by SWNTsin PEDOT film.

We investigated the conductivity of PEDOT:SWNTscomposite films with different doping concentration ofSWNTs. It is shown in Table 1 that when the SWNTs

1100 M. Shao et al. / Organic Electronics 12 (2011) 1098–1102

concentration changed from 0 wt.% to a high value of0.66 wt.%, the conductivity of the composite film increasedmonotonously. This can be ascribed to the increase in par-tial conductive paths when introducing SWNTs into thepolymer film [12], since the metallic components inSWNTs have a better conductivity than PEDOT. As a result,the enhancement of EL intensity shown in Fig. 1b whenincreasing the SWNTs concentration in PEDOT from 0.001to 0.01 wt.% can be ascribed to the increase of the conduc-tivity in PEDOT film with SWNTs doping, and thus leadingto the improvement of hole transport in devices. It shouldbe noted that although SWNTs can be beneficial in improv-ing the carrier transport, they also can act as trap sites forquenching the excitons [13,14]. It is well known that whenexcitons are formed in the emissive layer, they can migrateduring their lifetime and the singlet excitons usually havea migration distance of �20 nm [15]. Thus if the excitonsare formed near the interface of PEDOT:SWNTs/PFO inter-face, they can easily migrate to the SWNTs trapping sitesand thus be quenched provided that the SWNTs concentra-tion is high enough near the interface.

We notice that PEDOT has a highest occupied molecularorbital (HOMO) energy level of �5.2 eV while PFO has aHOMO energy level of �5.8 eV [16], so there is an energy

0 2 4 6 80

500

1000

1500

2000

2500

Lum

inan

ce (

cd/m

2 )

Voltage (V)

(a)

Without SWNTs

With SWNTs

0 200 400 600 8000.0

0.1

0.2

0.3

With SWNTs

Without SWNTs

Lum

inou

s E

ffici

ency

(cd

/A)

Current Density (mA/cm2)

(b)

0 200 400 600 8000.0

0.1

0.2

Without SWNTs

With SWNTs

EQ

E (

%)

Current Density (mA/cm2)

(c)

Fig. 2. The L–V curves (a), the luminous efficiency (b), and the externalquantum efficiency (c) characteristics of devices without and with SWNTsdoping. The device structure is ITO/PEDOT (40 nm)/PFO (80 nm)/Ca(20 nm)/Al (80 nm) and ITO/PEDOT:SWNTs (0.01 wt.%, 40 nm)/PFO(80 nm)/Ca (20 nm)/Al (80 nm), respectively.

barrier of 0.6 eV for holes to overcome. In our devices withthe structure of ITO/PEDOT:SWNTs/PFO/Ca/Al, the lowestunoccupied molecular orbital (LUMO) energy level of thelight-emitting material PFO aligns with the Fermi level ofthe Ca cathode. Consequently, the electron transport fromthe cathode into the light-emitting layer is not hindered.However, since an energy barrier between the HOMO lev-els prevents holes from entering the light emitting layer,the exciton recombination zone in the PFO layer must beclose to the interface of PEDOT:SWNTs/PFO interface. Asa result, the excitons formed in the PFO layer can migrateto the PEDOT:SWNTs layer and be quenched by the SWNTsif the SWNTs concentration is high enough in the PED-OT:SWNTs film. That is the reason why we see reducedEL intensity in OLEDs when the SWNTs concentration isas high as 0.02 wt.% in the PEDOT:SWNTs film comparedwith pure PEDOT film.

Fig. 2a shows the luminance–voltage curves of OLEDsusing pure PEDOT and PEDOT:SWNTs (0.01 wt.%) compos-ite. The maximum luminance of the device with SWNTs inPEDOT is 2150 cd/m2, which is as 1.4 times high as the onewithout SWNTs. Fig. 2b and c illustrate the EL efficiency vs.current density curves of device with and without SWNTsdoping. We can see the maximum luminous efficiency and

0 5 10 15 20 25

0

100

200

300

400

Without SWNTs

With SWNTs

Cur

rent

den

sity

(m

A/c

m2 )

Voltage (V)

(a)

0 50 100 150 2000

5

10

Without SWNTs

With SWNTs

Lum

inou

s E

ffici

ency

(cd

/A)

Current Density (mA/cm2)

(b)

0 50 100 150 2000

1

2

3

4

With SWNTs

Without SWNTs

EQ

E (

%)

Current density (mA/cm2)

(c)

Fig. 3. The J–V curves (a), the luminous efficiency (b), and the externalquantum efficiency (c) characteristics of the electrophosphorescentdevices without and with SWNT doping. The device structure is ITO/PEDOT (40 nm)/PVK:PBD:Ir(mppy)3 (69 wt.%:30 wt.%:1 wt.%, 60 nm)/Ca(20 nm)/Al (80 nm) and ITO/PEDOT:SWNTs (0.01 wt.%, 40 nm)/PVK:PBD:Ir(mppy)3 (69 wt.%:30 wt.%:1 wt.%, 60 nm)/Ca (20 nm)/Al(80 nm), respectively.

Fig. 4. The AFM images of the pure PEDOT film (a) and the PEDOT:SWNTs(0.01 wt.%) composite film (b), measured by tapping mode.

M. Shao et al. / Organic Electronics 12 (2011) 1098–1102 1101

the maximum external quantum efficiency (EQE) of the de-vice with SWNTs in PEDOT are 0.32 cd/A (1.2 times thanthat without SWNTs) and 0.22% (1.4 times than that with-out SWNTs), respectively. The enhancement of EL effi-ciency in devices employing SWNTs in PEDOT can beascribed to the improvement of hole transport, thus lead-

400 500 600 700 8000.0

0.2

0.4

0.6

0.8

1.0

Nor

mal

ized

EL

Inte

nsity

(a.

u.)

Wavelength (nm)

Without SWNT With 0.005 % SWNT

(e)

0 200 400 6000

2

4

6

8

10With 0.005% SWNTs

Without SWNTs

Lum

inou

s E

ffici

ency

(cd

/A)

Current Density (mA/cm2)

(c)

0 2 4 6 8 10 12 14

0

200

400

600

With 0.005% SWNTs

Without SWNTs

Cur

rent

Den

sity

(m

A/c

m2 )

Voltage (V)

(a)

1

1

2

Lum

inan

ce (

cd/m

2)

Fig. 5. The J–V curves (a), the L–V curves (b), the luminous efficiency (c), the exteSWNT doping. The device structure is ITO/PEDOT (40 nm)/Super-Yellow (80 n(0.005 wt.%, 80 nm)/Ca (20 nm)/Al (80 nm), respectively.

ing to more balanced holes and electrons in the light-emitting layer.

As we mentioned above, the improvement of EL perfor-mance in devices with the SWNTs concentration no higherthan 0.01 wt.% lies in their role of increasing hole transportin the conducting PEDOT film. Consequently, for OLEDswith holes as the minor carriers in the light-emitting layer(such as our PFO case), the EL performance can be im-proved by doping a very low amount of SWNTs into thehole conducting polymer. To investigate the impact ofthe mixture of the hole conducting polymer and SWNTson the EL performance of OLEDs with balanced electronsand holes in the light-emitting layer, we have fabricateddevices with the structure of ITO/PEDOT:SWNT(0.01 wt.%)/PVK:PBD:Ir(mppy)3/Ca/Al. It has been reportedthat when the mixture of PVK, PBD and Ir(mppy)3 is usedas a light-emitting layer in OLEDs, highly efficient elec-trophosphorescence is achieved due to balanced electronsand holes in the light-emitting layer [17].

Fig. 3a shows the J–V curves of the electrophosphores-cent devices with and without SWNTs doped in the PEDOTfilms. As can be seen from the figure, the J–V curve of thedevice with PEDOT:SWNTs composite shifts to a lowervoltage compared with the one without SWNTs, which issimilar with the PFO case. However, the maximum EL effi-ciency in the device with SWNTs is reduced relative to theone without SWNTs (shown in Fig. 3a and b). As we

0 200 400 6000

1

2

3

Without SWNT

With 0.005 % SWNT

EQ

E (

%)

Current Density (mA/cm2)

(d)

0 2 4 6 8 10 12 14

0

5000

0000

5000

0000

With 0.005% SWNTs

Without SWNTs

(b)

Voltage (V)

rnal quantum efficiency (d), and EL spectra (e) of devices without and withm)/Ca (20 nm)/Al (80 nm) and ITO/PEDOT (40 nm)/Super-Yellow:SWNTs

1102 M. Shao et al. / Organic Electronics 12 (2011) 1098–1102

expected, a low concentration of SWNTs in PEDOT film willenhance the hole transport in devices and shift the J–Vcurve to a lower voltage. Nevertheless, in OLEDs with bal-anced holes and electrons, introducing more holes to thelight-emitting layer will upset the balance thus reducingthe EL efficiency. Fig. 4 shows AFM images of the PEDOTfilm with and without SWNT dopants. In the case of purePEDOT film, the islands have big sizes. But in the PED-OT:SWNTs (0.01 wt.%) composite film, the islands havesmall sizes, which means the contact at the interface be-tween the PEDOT:SWNTs and the light-emitting layers willbe improved. This morphological result suggests that usingSWNTs in PEDOT layer can enhance hole injection into thelight-emitting layer in the ITO/PEDOT:SWNTs/PFO/CaOLED.

Besides acting as dopants in a hole conducting polymerin OLEDs, SWNTs can also be used in the light-emittingpolymer layer. To investigate the SWNT’s effect on the ELperformance when it is in the light-emitting layer, we havefabricated devices with the structure of ITO/PEDOT/Super-Yellow:SWNTs/Ca/Al. Fig. 5a shows the J–V characters ofdevices with and without SWNTs in the light-emittinglayer. Introducing SWNTs into the Super-Yellow layershifts the J–V curve to the lower voltage compared withthe pure Super-Yellow layer. Since Super-Yellow is apoly(p-phenylene vinylene) (PPV) derivate and it cantransport holes more easily than electrons [18], OLEDsbased on Super-Yellow are hole-dominated devices andelectrons are minority carriers. It is well known thatSWNTs have an electron-withdrawing ability [19,20], sothey can act as an n-type material to improve the electrontransport when doped into a polymer host, and can im-prove charge injection at interfaces [21]. Consequently,the J–V curve of the device with the Super-Yellow:SWNTscomposite shifts to a lower voltage due to improved elec-tron transport in the Super-Yellow layer. It can also be seenin Fig. 5b that the Super-Yellow:SWNTs OLED exhibits alower threshold voltage in electroluminescence as com-pared to Super-Yellow only OLED. This result further sug-gests that the minority charge carrier (electron) transportis enhanced in Super-Yellow:SWNTs OLED. As can be seenfrom Fig. 5c and d, the EL efficiency has also been improvedin devices with the Super-Yellow:SWNTs composite. Themaximum luminous efficiency and maximum EQE reaches9.8 cd/A and 2.9%, which is 1.6 times and 1.7 times thanthe one with pure Super-Yellow layer, respectively. Suchenhancement in EL efficiency is also the result of theimprovement of electron transport in the Super-Yellowlayer due to the SWNTs acting as n-type dopants, thusleading to more balanced electrons and holes in thelight-emitting layer. Fig. 5e shows the EL spectra of deviceswith and without SWNTs in the light-emitting layer andwe can see that mixing the Super-Yellow with SWNTs doesnot change the EL emission spectra.

4. Conclusion

Here we elucidate that SWNTs play different roles whenused as polymer:SWNTs composites in OLEDs. When

SWNTs are mixed in a hole-conducting layer, they can havea helpful or a hurtful influence on the EL performance ofOLEDs, which depends on the SWNTs concentration. Ifthe SWNTs concentration is low enough (in our case60.01 wt.%), they can play a role in improving the conduc-tivity thus enhancing the hole transport in the hole-conducting layer. Also, they improve contact at interfacesby inducing much smaller surface islands than the oneson the surface of purely hole-conducting layers. But if theSWNTs concentration is too high (in our case higher than0.02 wt.%), they can act as centers to quench the excitonsthat have migrated to the interface between the hole-con-ducting layer and the light-emitting layer. When SWNTsare incorporated into an emitting polymer layer, they im-prove the electron transport since SWNTs can act as n-typedopants in the light-emitting layer, which profits from theelectron-withdrawing nature of SWNTs.

Acknowledgements

This research was supported by Center for MaterialsProcessing at the University of Tennessee, Solid State Light-ing Program of the Energy Efficiency and Renewable En-ergy, US Department of Energy. A portion of this researchwas conducted at the Center for Nanophase Materials Sci-ences through a user project (CNMS2009-055), which issponsored at Oak Ridge National Laboratory by the BasicEnergy Sciences, US Department of Energy. Authors ex-press gratitude to Prof. Stanislaus S. Wong (Departmentof Chemistry, SUNY) for providing water-soluble SWNTs.

References

[1] F. Massuyeau, J.L. Duvail, H. Athalin, J.M. Lorcy, S. Lefrant, J. Wery, E.Faulques, Nanotechnology 20 (2009) 155701.

[2] L. Qian, T. Feng, X. Zheng, S.Y. Quan, D.A. Liu, Y.M. Wang, Y.S. Wang,X.R. Xu, Spectrosc. Spect. Anal. 26 (2006) 601.

[3] B. Hu, Y. Wu, Z.T. Zhang, S. Dai, J. Shen, Appl. Phys. Lett. 88 (2006)022114.

[4] Z. Tan, F. Zhang, T. Zhu, J. Xu, A.Y. Wang, J.D. Dixon, L. Li, Q. Zhang,S.E. Mohney, J. Ruzyllo, Nano Lett. 7 (2007) 3803.

[5] Y.G. Ha, E.A. You, B.J. Kim, J.H. Choi, Synth. Met. 153 (2005) 205.[6] S.A. Curran, P.M. Ajayan, W.J. Blau, D.L. Carroll, J.N. Coleman, A.B.

Dalton, A.P. Davey, A. Drury, B. McCarthy, S. Maier, A. Strevens, Adv.Mater. 10 (1998) 1091.

[7] H.S. Woo, R. Czerw, S. Webster, D.L. Carroll, J. Ballato, A.E. Strevens,D. O’Brien, W.J. Blau, Appl. Phys. Lett. 77 (2000) 1393.

[8] H.S. Woo, R. Czerw, S. Webster, D.L. Carroll, J.W. Park, J.H. Lee, Synth.Met. 116 (2001) 369.

[9] J.Y. Kim, E.S. Kim, J.H. Choi, J. Appl. Phys. 91 (2002) 1944.[10] T.W. Odem, J.L. Huang, P. Kim, C.M. Liber, Nature 391 (1998) 62.[11] M.G.C. Kahn, S. Banerjee, S.S. Wong, Nano Lett. 2 (2002) 1215.[12] J.Y. Kim, M. Kim, J.-H. Choi, Synth. Met. 139 (2003) 565.[13] C. Yang, M. Wohlgenannt, Z.V. Vardeny, W.J. Blau, A.B. Dalton, R.

Baughman, A.A. Zakhidov, Physica B 338 (2003) 366.[14] E. Kymakis, G.A.J. Amaratung, Appl. Phys. Lett. 80 (2002) 112.[15] A.L. Burin, M.A. Ratner, J. Phys. Chem. A 104 (2000) 4704.[16] A.J. Campell, D.D.C. Bradley, H. Antoniadis, J. Appl. Phys. 89 (2000)

3343.[17] X. Yang, D. Neher, D. Hertel, T.K. Daubler, Adv. Mater. 16 (2004) 161.[18] H. Spreitzer, H. Becker, E. Kluge, W. Kreuder, H. Schenk, R. Demandt,

H. Schoo, Adv. Mater. 10 (1998) 1340.[19] A. Kongkanand, P.V. Kamat, ACS Nano 1 (2007) 13.[20] J.M. Holt, A.J. Ferguson, N. Kopidakis, B.A. Larsen, J. Bult, G. Rumbles,

J.L. Blackburn, Nano Lett. 10 (2010) 4627.[21] D. Liu, M. Fina, J. Guo, X. Chen, G. Liu, S.G. Johnson, S.S. Mao, App.

Phys. Lett. 94 (2009) 013110.