Temperature-Dependent Electrical Properties of Graphene Inkjet-Printed on Flexible MaterialsDe Kong,† Linh T. Le,† Yue Li,† James L. Zunino,‡ and Woo Lee*,†

†Department of Chemical Engineering and Materials Science, Stevens Institute of Technology, Hoboken, New Jersey 07030, UnitedStates‡U.S. Army ARDEC, Picatinny Arsenal, New Jersey 07806, United States

ABSTRACT: Graphene electrode was fabricated by inkjet printing, as a new means ofdirectly writing and micropatterning the electrode onto flexible polymeric materials.Graphene oxide sheets were dispersed in water and subsequently reduced using aninfrared heat lamp at a temperature of ∼200 °C in 10 min. Spacing between adjacent inkdroplets and the number of printing layers were used to tailor the electrode’s electricalsheet resistance as low as 0.3 MΩ/□ and optical transparency as high as 86%. Thegraphene electrode was found to be stable under mechanical flexing and behave as anegative temperature coefficient (NTC) material, exhibiting rapid electrical resistancedecrease with temperature increase. Temperature sensitivity of the graphene electrodewas similar to that of conventional NTC materials, but with faster response time by anorder of magnitude. This finding suggests the potential use of the inkjet-printedgraphene electrode as a writable, very thin, mechanically flexible, and transparenttemperature sensor.

Graphene has received significant attention because of itspotential as highly flexible electrically conductive electro-

des for various applications ranging from optoelectronic toenergy storage to biomedical devices.1−3 We recently reportedthat graphene oxide (GO) sheets dispersed in water can beinkjet-printed and thermally reduced at 200 °C in nitrogen(N2) to produce relatively thick graphene electrodes withpromising electrochemical properties for energy storage.4 Thebroader implication of our previous finding is that hydrophilicGO nanosheets could be dispersed up to 0.2 wt % in purewater, as a scalable ink. In contrast, hydrophobic graphenesheets are difficult to inkjet-print because of the difficultyproducing a stable dispersion, even if organic solvents are used.5

Our inkjet printing approach is expected to offer a new,economically viable avenue of producing micropatternablegraphene because of (1) active developments in producinglarge quantities of potentially low-cost GO sheets derived fromgraphite powder;6 (2) direct phase transformation from simple,environmental friendly water-based inks to graphene micro-patterns in an additive, net-shaped manner with minimummaterial use, handling, and waste generation; and (3) rapidtranslation of new discoveries for integration with flexibleelectronics using commercially available inkjet printers rangingfrom desktop to roll-to-roll.As schematically illustrated in Figure 1a, the goal of this

investigation was to evaluate the electrical and opticalproperties of inkjet-printed and infrared (IR) lamp-reducedgraphene electrodes upon optimizing the spacing betweenadjacent ink droplets (D) and the number of printed layers (N)as two major process parameters. Figure 1b shows anelectrically conductive graphene micropattern inkjet-printedon polyethylene terephthalate (PET), which was used as an

example of temperature-sensitive and mechanically flexiblesubstrates. In this investigation, inkjet-printed GO sheets werereduced in room environment using an IR heat lamp from alocal hardware store with the distance between the substratesand the lamp controlled to be 3 cm while monitoring (1)substrate temperature and (2) electrical resistance (R). Asshown in the Figure 1c, the substrate temperature rose to ∼220°C during the 12 min exposure duration. R became measurableat ∼5 min into the exposure, and continuously decreased untilit reached a steady-state value at ∼10 min.Figure 2a shows that GO sheets had various sheet

dimensions and shapes. As summarized in Figure 2b, theaverage lateral dimension was ∼530 nm with ∼35% GO sheetssmaller than 300 nm and ∼30% larger than 1000 nm. Theformation of a coffee ring was observed from the dried-outstructure of a single 10 pL GO ink droplet containing thenominal GO concentration of 2 mg/mL in water (Figure 3a).The coffee-ring structure was similar to what has been observedin various inkjet-printed materials.7,8 As a result of pinning atthe edge of the low contact angle area of the droplet, most GOsheets appeared to stack and form aggregated structures 5−10nm high and 100−200 nm wide at the perimeter (Figure 3b).Interestingly, we consistently observed a “star”-shaped assemblyof nanoscale features at the center region of the droplet (Figure3c), while leaving a significantly lower number of sheetsscattered between the center and the perimeter regions. Theheight and width of these nanoscale features were in the ranges

Received: May 1, 2012Revised: August 21, 2012Published: August 27, 2012

Article

pubs.acs.org/Langmuir

© 2012 American Chemical Society 13467 dx.doi.org/10.1021/la301775d | Langmuir 2012, 28, 13467−13472

of 10−20 nm and 50−200 nm, respectively. As the evaporationfront receded toward the center of the shrinking droplet, itappeared that the droplet became depinned and the GO sheetsbecome entrained, accumulated, and eventually deposited toform the star-shaped assembly at the center region of thedroplet.Figure 3d−f shows that the dried-out structure of GO sheets

printed on silicon substrates became continuous withdecreasing D from 50 to 20 μm at N = 1. Even with increasingN to 5, the structure obtained with D = 50 μm remained largelydiscontinuous (not shown). On the other hand, the structureproduced at D = 40 μm became more interconnected at N = 5.Despite oxygen plasma substrate treatment prior to the printingstep, the effect of D on the formation of discontinuousmorphology was more pronounced on hydrophobic PET andKapton substrates than on hydrophilic silicon and glasssubstrates. Nevertheless, 20 μm was determined to be an

adequate spacing to produce completely continuous morphol-ogy even on Kapton and glass substrates used for opticaltransparency and electrical sheet resistance (Rs) measurements.Note that the printer used for this study was capable ofoperating with 5 μm resolutions in the x- and y-directions.Prior to the IR lamp treatment, characteristic GO peaks were

present in the Fourier transform infrared (FTIR) spectrum(Figure 4a) including the following: (1) CO stretchingvibration at 1735 cm−1, (2) OH stretching at 3428 cm−1, (3)OH deformation vibration at 1411 cm−1, (4) aromatic CCstretching vibration at 1610 cm−1, and (5) alkoxy COstretching vibration at 1041 cm−1.9 After the exposure, the 1411cm−1 and 1041 cm−1 peaks disappeared with the 3428 cm−1

peak significantly decreased, and the small 1735 cm−1 peak stillremained. These changes suggested the significant removal ofOH functional groups from the exposed GO sheets.However, the 1735 cm−1 peak did not disappear, suggesting

Figure 1. Flexible graphene micropatterns produced by inkjet-printing of GO sheets and photothermal reduction using an IR heat lamp in ambientenvironment: (a) illustration of the overall processing concept with the spacing between adjacent ink droplets (D) and the number of printed layers(N) as major printing variables; (b) micropatterns printed on a transparent PET substrate; and (c) electrical resistance and temperature changesmeasured in real-time during the photothermal reduction step of the inkjet-printed graphene produced at D = 30 μm and N = 3.

Figure 2. (a) SEM image and (b) lateral size distribution of GO sheets deposited on Si from the dried-out structure of one ink droplet containing 0.1mg/mL GO.

Langmuir Article

dx.doi.org/10.1021/la301775d | Langmuir 2012, 28, 13467−1347213468

that the CO stretching vibration of six-ring lactones was stillpresent.10 The 1610 cm−1 CC peak was present, indicatingthat the sp2 structure of carbon atoms was retained.11

Two prominent Raman peaks were observed before and afterthe IR lamp reduction step (Figure 4b): (1) G bandcorresponding to the first-order scattering of photons by sp2

carbon atoms and (2) D band arising from small domain-sizedgraphitic regions.12,13 The intensity ratio of the D to G bands

(ID/IG) increased from 0.79 to 0.94 upon reduction. This ratiochange suggested that (1) most of the oxygenated functionalgroups were removed from GO sheets by the reduction stepand (2) sp2 network was established. Upon reduction, the Gband was slightly shifted to 1602 cm−1 from 1607 cm−1.However, the G and D bands of the reduced GO sheets presentat 1602 cm−1 and 1354 cm−1 were considerably higher thanthose of chemically vapor deposited (CVD) graphene typically

Figure 3. Morphology of dried-out structures produced by a single GO ink droplet on Si: (a) SEM and (b,c) AFM images. (d,e,f) SEM imagesshowing the effects of decreasing D on the development of continuous film morphology on Si.

Figure 4. (a) FTIR and (b) Raman spectra of GO sheets before and after IR heat lamp reduction.

Figure 5. Effects of D and N on (a) electrical sheet resistance and (b) optical transparency.

Langmuir Article

dx.doi.org/10.1021/la301775d | Langmuir 2012, 28, 13467−1347213469

observed at 1575 cm−1 and 1350 cm−1. These peak shiftsindicated the relative lack of sp2 character and the remainingpresence of some oxygenated functional groups, consistent withthe FTIR results.The FTIR and Raman results suggested that the IR heat

lamp treatment was effective in reducing printed GO films tographene films to a significant extent, but not completely. TheIR lamp reduction method is expected to be particularly usefulfor printing onto thermally and chemically sensitive materialsand devices. Also, this method is advantageous for easyintegration with roll-to-roll, additive manufacturing since it onlytakes minutes as opposed to hours required for the thermal andchemical methods without the need for controlled reductionenvironments and equipment.As shown in Figure 5a, Rs of the graphene electrodes

fabricated on Kapton decreased with (1) decreasing D and (2)increasing N. At D = 40 μm, the films were not conductive at N= 2, but became conductive with N = 3 at ∼26 MΩ/□ andwith N = 5 at 14 MΩ/□. The high Rs values of these samples

could be explained by (1) the development of noncontinuousmorphology at large D and small N and (2) consequentlyblocking of electron transport paths. At D = 20 μm, Rs

decreased from ∼12 MΩ/□ to ∼0.3MΩ/□ with increasingN from 2 to 5.As shown in Figure 5b, graphene electrodes printed on glass

substrates became less transparent with (1) reducing D and (2)increasing N. At D = 20 μm, transparency rapidly decreasedfrom ∼76% to 45% upon increasing N from 2 to 5. It is well-known that an increase in the stacking of CVD graphene layersdecreases light transparency of 2.3% per graphene sheet.14

Assuming this number for our sample obtained at N = 2, weroughly estimated that ∼10 graphene sheets may be stacked onaverage to result in 76% transparency. This estimation wasconsistent with the average thickness of the dried out structureof each ink droplet being on the order of ∼10 nm as suggestedby the AFM data in Figures 3b and 3c.Based on the above results, D = 20 μm and N = 2 were

determined to be optimum printing parameters for producing

Figure 6. (a) Relative electrical resistance changes upon mechanical bending. (b) Experimental configuration. Error bars represent 3 measurementsmade for each bending angle.

Figure 7. (a) Temperature-dependence on electrical resistance. (b) Linear fit (red) between ln (R) versus T−1. (c) Relative electrical resistanceresponses upon repeated fingertip tapping. (d) Experimental configuration.

Langmuir Article

dx.doi.org/10.1021/la301775d | Langmuir 2012, 28, 13467−1347213470

continuous electrode morphology with Rs = 12 MΩ/□ at 76%transparency. This optoelectrical performance is similar to thatreported by Torrisi et al.5 with Rs = 102 MΩ/□ at 74%transparency for graphene sheets exfoliated by ultrasonicatinggraphite powder, dispersed in an organic solvent, and inkjet-printed. However, in comparison to CVD graphene,15,16 Rs ofour sample was about 7 orders of magnitude higher at a giventransparency of 86%. The lower Rs of the CVD graphene wasexpected since it contains relatively defect-free graphenestructure. Nevertheless, the comparison highlights a significantchallenge associated with the use of inkjet-printed graphene foroptoelectronic applications.Figure 6 shows that R of the electrode printed on Kapton at

D = 20 μm and N = 2 decreased with increasing the degree ofbending (2θ). The overall decrease in R was 5.6% at 2θ = 27.4°.Apparently, local bending stresses increased the effectivemobility of electrons, although the mechanism behind thisbehavior is not clear. Some hysteresis was observed duringrecovery, but the resistance ultimately returned to the initialvalue prior to bending. This recovery behavior implies that themechanical structure of the graphene electrode remained to berelatively stable during the mechanical bending test.Figure 7a shows that R of the graphene electrode decreased

significantly with temperature. The effect of the temperature onthe electrode resistance is similar to what has been recentlyobserved by (1) Sahoo et al.17 for filter-deposited andchemically reduced GO sheets using hydrazine vapor and (2)Zhuge et al.18 with filter-deposited and metal-defused GOsheets. As shown in Figure 7b, the following equation was usedto model the observed temperature dependence as a negativetemperature coefficient (NTC) behavior

=−·

⎛⎝⎜

⎞⎠⎟R R B

T TT T

exp( )

T 00

0

where RT is the electrical resistance as a function of temperature(T), B is the material constant and a measure of temperaturesensitivity, and R0 is the resistance at the reference temperature(T0 = 298 K). From the data fitting, B was determined to be1860 K in the temperature range of 298 to 358 K with therespective resistance changes from 4.4 × 106 to 2.4 × 106 Ω.This B value is close to that of the conventional metal oxideNTC materials, typically in the range of 2000 to 5000 K.19 Thetemperature coefficient of resistance (α) was also used asanother measure of temperature sensitivity where α = R−1·(dR/dT). α for our graphene electrodes was determined to be−0.0148 K−1 at 298 K, which is about 1 order of magnitudelarger than that of the chemically reduced GO sheets17 as wellas that of metal-defused GO sheets.18 Also, the α value of ourgraphene electrode is about 3 orders of magnitude higher thanthat of carbon nanotubes.20

As shown in Figure 7c,d, temperature-sensing function of thegraphene electrode was evaluated by tapping the electrode witha human finger in the ambient room environment. Therepeated taps resulted in the resistance decreases shown in theFigure 7c. In contrast, no change in the resistance was observedwhen the electrode was tapped with other objects that were inthermal equilibrium with the room environment (not shown).This observation also indicated that the effect of slight substrateflexing during tapping on the resistance changes was muchsmaller than that of touching with the finger tip. These resultssuggested that the resistance changes were as a result of heattransfer between the finger tip and the electrode.

The response time to the touching was about 0.5 s, and therecovery time to its initial resistance value upon removing thefinger tip was about 10 s. In comparison, typical response timefor conventional NTC metal oxide materials is more than 10s,21 suggesting an order-of-magnitude faster temperature-sensing function of the inkjet-printed graphene electrode.The observed NTC behavior suggests the inkjet-printedgraphene functions as an intrinsic semiconductor with perhapsthermally activated transfer of electrons between the reduceddomains of the GO sheets as well as between the sheets. Itappears that a major reason for the fast time response of thegraphene electrode is a very small volume of the inkjet-printedelectrode and therefore a significantly lower thermal massinvolved with transient heat transfer.In conclusion, our results suggest that micropatternable

graphene electrodes can be easily fabricated by inkjet printingof GO sheets and subsequent photothermal reduction using theIR heat lamp in ambient environment in about 10 min. D andN were optimized as the major printing parameters to producethe continuous morphology of the graphene electrode foroptimum Rs and transparency. R of the electrode decreasedduring mechanical bending, but returned to its initial valueupon recovery, suggesting the electrode’s structural stabilitywith mechanical flexing. Also, the electrode’s NTC behaviorwith high temperature sensitivity and fast response timesuggests new potential as a writable, very thin, flexible, andtransparent temperature sensor.

■ EXPERIMENTAL SECTIONCommercially available GO sheets (Cheap Tubes, Brattleboro, VT)dispersed in water (2 mg/mL) were used to prepare inks at several GOconcentrations by dilution for some initial experiments. For mostexperiments, 2 mg/mL was used as the nominal concentration of theGO ink. The viscosity, surface tension, and ζ-potential of the nominalGO ink were measured to 1.06 mPa·s, 68 N/m, and −20 mV,respectively.4 Glass slides (1.2 mm thick, Thermo Scientific,Portsmouth, NH), Kapton-HN (DuPont, Wilmington, DE), andPET (3M, St. Paul, MN) films were used as examples of transparentsubstrates. Also, polished Si (University Wafer, Boston, MA) was usedfor characterization purposes. Glass and Si substrates were cleanedusing a piranha solution and deionized water several times, then driedwith nitrogen gas prior to printing. Si, Kapton, and PET were treatedwith O2 plasma for 30 s prior to printing using Plasma Cleaner(Harrick Plasma, Ithaca, NY).

As previously described,4 a Dimatix Material Printer (DMP 2831,Fujifilm Dimatix, Santa Clara, CA) was used to print the GO inksusing cartridges that generate 10 pL droplets. The cartridge height andsubstrate temperature were maintained at 0.5 mm and 25 °C,respectively. GO electrodes were inkjet-printed as 0.8 cm × 0.8 cmsquare patterns. The GO electrodes were reduced with an infrared(IR) heat lamp (250 W, GE, Cleveland, OH). Raman spectroscopy(Spectra Pro 2300i, Princeton Instrument, Trenton, NJ) wasconducted using the excitation line of 632.8 nm. FTIR (TENSORSeries 27 FT-IR Spectrometers, Bruker Optics, Billerica, MA) wasperformed in a transparency mode using 100 μL droplet-cast sampleson silicon before and after reduction. The drop casting method wasused for the FTIR measurements, since the signal from the printedsamples was not strong enough to be measured.

The morphology and pattern formation of the printed GOelectrodes were characterized by optical microscopy (SMZ1500,Nikon, Melville, NJ) and scanning electron microscopy (SEM, CarlZeiss SMT Auriga FIB-SEM workstation, Peabody, MA), and atomicforce microscopy (AFM, Nanoink, Skokie, IL). Transparency wasrecorded at 560 nm using a multimode microplate reader (SynergyHT, BioTek Instruments, Inc., Winooski, VT).

Langmuir Article

dx.doi.org/10.1021/la301775d | Langmuir 2012, 28, 13467−1347213471

Rs was measured using a digital multimeter (Keithley InstrumentsInc., Cleveland, OH) and a custom-made four-point probeconfiguration shown in Figure 5a. The four-point probe was preparedby inkjet printing silver nanoparticles ink (Cabot Corporation, Boston,MA) onto Kapton followed by annealing at 200 °C using a hot plate(Corning, Lowell, MA) in the air. Electrical resistance changes duringthe reduction process were measured by the multimeter with adistance of 2 mm between two probes. Similarly, electrical resistancechanges during mechanical bending were measured with a distance of0.8 mm between two probes. Temperature dependence character-ization was conducted similarly using a tunable hot plate (Corning,Lowell, MA) in the air and a thermocouple attached to the grapheneelectrode. The fingertip tapping experiment was performed with the 4-point probe device by applying a constant voltage of 10 V across thesample and recording the corresponding current change using themultimeter. The graphene electrode surface was covered with Scotchtape, and a plastic glove was worn, as shown in Figure 7d.

■ AUTHOR INFORMATIONCorresponding Author*E-mail: wlee@stevens.edu.NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSThe authors thank the U.S. Army - ARDEC for funding thisproject under the contract of W15QKN-05-D-0011. Thisresearch effort used microscope resources partially funded bythe National Science Foundation through NSF Grant DMR-0922522. We also thank Andrew Ihnen at Stevens and BrianFuchs at ARDEC for various discussions.

■ REFERENCES(1) Eda, G.; Fanchini, G.; Chhowalla, M. Large-area ultrathin films ofreduced graphene oxide as a transparent and flexible electronicmaterial. Nat. Nanotechnol. 2008, 3, 270−274.(2) Cohen-Karni, T.; Qing, Q.; Li, Q.; Fang, Y.; Lieber, C. M.Graphene and nanowire transistors for cellular interfaces and electricalrecording. Nano Lett. 2010, 10, 1098−1102.(3) Shan, C.; Yang, H.; Song, J.; Han, D.; Ivaska, A.; Niu, L. Directelectrochemistry of glucose oxidase and biosensing for glucose basedon graphene. Anal. Chem. 2009, 81, 2378−2382.(4) Le, L. T.; Ervin, M. H.; Qiu, H.; Fuchs, B. E.; Lee, W. Y.Graphene supercapacitor electrodes fabricated by inkjet printing andthermal reduction of graphene oxide. Electrochem. Commun. 2011, 13,355−358.(5) Torrisi, F.; Wu, T. H., W.; Sun, Z.; Lombardo, A.; Kulmala, T.;Hshieh, G. W.; Jung, S. J.; Bonaccorso, F.; Paul, P. J.; Chu, D. P.;Ferrari, A. C. Ink-Jet Printed Graphene Electronics. ACS Nano 2012, 6(4), 2992−3006.(6) Jacoby, M. Graphene Moves Toward Applications. Chem. Eng.News November 21, 2011.(7) Kang, J. S.; Kim, H. S.; Ryu, J.; Hahn, H. T.; Jang, S.; Joung, J. W.Inkjet printed electronics using copper nanoparticle ink. J. Mater. Sci.Mater. Electronics 2010, 21, 1213−1220.(8) Yoshitsugu, K.; Ueno, I. Effect of suspended particles onspreading of volatile droplet on solid substrate. J. Vis. 2011, 14, 285−294.(9) Ren, P. G.; Yan, D. X.; Ji, X.; Chen, T.; Li, Z. M. Temperaturedependence of graphene oxide reduced by hydrazine hydrate.Nanotechnology 2011, 22, 055705.(10) Mei, X. G.; Ouyang, J. Y. Ultrasonication-assisted ultrafastreduction of graphene oxide by zinc powder at room temperature.Carbon 2011, 49, 5389−5397.(11) Chen, W. F.; Yan, L. F.; Bangal, P. R. Preparation of grapheneby the rapid and mild thermal reduction of graphene oxide induced bymicrowaves. Carbon 2010, 48, 1146−1152.

(12) Lin, Z.; Yao, Y.; Li, Z.; Liu, Y.; Li, Z.; Wong, C.-P. Solvent-Assisted Thermal Reduction of Graphite Oxide. J. Phys. Chem. C 2010,114, 14819−14825.(13) Ferrari, A. C.; Robertson, J. Interpretation of Raman spectra ofdisordered and amorphous carbon. Phys. Rev. B 2000, 61, 14095−14107.(14) Nair, R. R.; Blake, P.; Grigorenko, A. N.; Novoselov, K. S.;Booth, T. J.; Stauber, T.; Peres, N. M. R.; Geim, A. K. Fine structureconstant defines visual transparency of graphene. Science 2008, 320,1308.(15) Kim, K. S.; Zhao, Y.; Jang, H.; Lee, S. Y.; Kim, J. M.; Kim, K. S.;Ahn, J.-H.; Kim, P.; Choi, J.-Y.; Hong, B. H. Large-scale patterngrowth of graphene films for stretchable transparent electrodes. Nature2009, 457, 706−710.(16) Bae, S.; Kim, H.; Lee, Y.; Xu, X.; Park, J.-S.; Zheng, Y.;Balakrishnan, J.; Lei, T.; Ri Kim, H.; Song, Y. I.; Kim, Y.-J.; Kim, K. S.;Ozyilmaz, B.; Ahn, J.-H.; Hong, B. H.; Iijima, S. Roll-to-roll productionof 30-in. graphene films for transparent electrodes. Nat. Nanotechnol.2010, 5, 574−578.(17) Sahoo, S.; Barik, S. K.; Sharma, G. L.; Khurana, G.; Scott, J. F.;Katiyar, R. S. Reduced graphene oxide as ultra fast temperature sensor.2012, arXiv:1204.1928v1 [cond-mat.mes-hall].(18) Zhuge, F.; Hu, B.; He, C.; Zhou, X.; Liu, Z.; Li, R. Mechanism ofnonvolatile resistive switching in graphene oxide thin films. Carbon2011, 49, 3796−3802.(19) Kang, J.; Ryu, J.; Han, G.; Choi, J.; Yoon, W.; Hahn, B.; Kim, J.;Ahn, C.; Choi, J.; Park, D. LaNiO3 conducting particle dispersedNiMn2O4 nanocomposite NTC thermistor thick films by aerosoldeposition. J. Alloys Compd. 2012, 534, 70−73.(20) Bartolomeo, A. D.; Sarno, M.; Giubileo, F.; Altavilla, C.; Iemmo,L.; Piano, S.; Bobba, F.; Longobardi, M.; Scarfato, A.; Sannino, D.;Cucolo, A. M.; Ciambelli, P. Multiwalled carbon nanotube films assmall-sized temperature sensors. J. Appl. Phys. 2009, 105, 064518.(21) Kanungo, J.; Saha, H.; Basu, S. Pd sensitized porous siliconhydrogen sensorInfluence of ZnO thin film. Sens. Actuators, B:Chem. 2010, 147, 128−136.

Langmuir Article

dx.doi.org/10.1021/la301775d | Langmuir 2012, 28, 13467−1347213472