Microelectronics Reliability 55 (2015) 538–546

Contents lists available at ScienceDirect

Microelectronics Reliability

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

Reliability of thin films: Experimental study on mechanical and thermalbehavior of indium tin oxide and poly(3,4-ethylenedioxythiophene)

http://dx.doi.org/10.1016/j.microrel.2015.01.0130026-2714/� 2015 Elsevier Ltd. All rights reserved.

⇑ Corresponding author.

Atif Alkhazaili a,b, Mohammad M. Hamasha c,⇑, Gihoon Choi a, Susan Lu a, Charles R. Westgate aa State University of New York at Binghamton, Binghamton, NY, USAb Department of Industrial Engineering, Hashemite University, Zarqa, Jordanc Department of Industrial Engineering, University of Business and Technology, Jeddah, Saudi Arabia

a r t i c l e i n f o

Article history:Received 5 August 2014Received in revised form 11 December 2014Accepted 26 January 2015Available online 7 February 2015

Keywords:Bending fatigueDamp heatPCERITOPEDOTReliability

a b s t r a c t

In order to improve the performance of various flexible electronic devices, the research on the transpar-ent conductive thin films becomes very intensive in the recent years. In this work, we studied mechanicaland thermal behaviors of two types of transparent conductive thin films, Poly(3,4-ethylenedioxythio-phene) (PEDOT) as an example on polymer conductive thin film, and indium tin oxide (ITO) as an exam-ple on transparent conductive oxide thin film. Both films are deposited on polyethyleneterephthalate(PET) substrate. Two sheet resistances for PEDOT (i.e., 150 X/sq and 225 X/sq) and one sheet resistancefor ITO (i.e., 60 X/sq) were involved in the study. PEDOT showed good mechanical properties with a smallelectrical resistance change and no clear cracks or deformation on the film surface under the condition ofcyclic bending. However, the resistance of ITO significantly increased with the cyclic bending and crackswere seen initiated in the center of sample and propagated toward the edges. Further, design of experi-ment approach was used to study the effect of different cyclic bending parameters, such as bending diam-eter and frequency. Additionally, damp heat experiment on similar samples was conducted by applying85 �C temperatures and 85% relative humidity on them for 1000 h. The electrical resistance was dramat-ically increased for both films. Scanning electronic microscopy (SIM), Energy Dispersive Spectrometry(EDX) and transmission tests were also used to determine the change of films’ compositions andtransparencies.

� 2015 Elsevier Ltd. All rights reserved.

1. Introduction

Flexible electronic displays have unique features, such as lightweight, low cost, and flexibility. They are used in many applica-tions such as, solar cell, organic light emitting diode (OLED), andelectronic sensors [1]. There were many efforts on improving con-ductive films to increase the solar cell efficiency. Two of the mostcommon conductive transparent thin films are polyethylene dioxy-thiophene (PEDOT) and tin indium oxide (ITO). They are very com-mon in liquid crystal displays (LCDs), sensors, and emissionsdevices applications [2,3].

ITO is a brittle transparent conductive oxide, while PEDOT is aductile polymer. Many researchers paid attention to show theresponse of the ITO films to mechanical and thermal stresses. Forexample, Leterrier et al. [4] noticed that process-induced internalstresses were systematically compressive, and the tensile cracksin the ITO layer always start from surface defects. Further, theynoticed that transition from stable to unstable crack growth starts

at crack length of several hundred times of the coating thickness.Furthermore, they found that polarization of hard coats providesa good quality surface for polymer substrate and enhances cohe-sive specifications of ITO thin film.

Many researchers studied mechanical properties of differentthin film materials that were exposed to cycling bending and ther-mal stresses [5]. Alzoubi et al. [6] investigated the effect of highand low cycle bending on electrical properties of copper thin filmcoated flexible substrate. In their study, they observed that the fati-gue failure was developed faster at strain of 2.6% or higher. Also,Hamasha et al. [7] carried out an experiment to study the effectof cyclic bending fatigue on the surface integrity of ITO under harshenvironmental condition. They conducted, however, the sameexperiment on similar samples, but without involving any bendingto make a comparison and figure out the effect of bending fatigueitself. They found that the combination of high temperature andhumidity under bending fatigue condition is critical on the conduc-tivity and the surface integrity of the thin film.

Park et al. [8] conducted a bending test on the single crystallinesilicon coated plastic substrate. They observed three types of

http://crossmark.crossref.org/dialog/?doi=10.1016/j.microrel.2015.01.013&domain=pdfhttp://dx.doi.org/10.1016/j.microrel.2015.01.013http://dx.doi.org/10.1016/j.microrel.2015.01.013http://www.sciencedirect.com/science/journal/00262714http://www.elsevier.com/locate/microrel

A. Alkhazaili et al. / Microelectronics Reliability 55 (2015) 538–546 539

failure modes when crystalline silicon exposed to different strainrates: slipping, cracking, and delamination. Cairns and Crawford[9] studied the effect of cyclic bending and thermal stresses onITO and PEDOT thin films coated PET. They found that the ITO thinfilm was susceptible to damage more than the PEDOT thin filmunder tensile stress. Moreover, it was observed that cracks startto appear in the ITO film at low rate of strain (less than 2%), whilePEDOT thin film had great mechanical properties. However, thePEDOT film suffers with a severe degradation under high temper-ature conditions. The cracks propagation in thin films depositedon flexible substrate causes electrical failure, while the failure inthe bulk materials is caused by a single sudden crack [4]. Strainlevel, number of cycles, material properties and thickness of thinfilm affect the intensity and size of cracks in the films [10,11]. Ste-phenson et al. [12] used the photographic technology in manufac-turing flexible displays to minimize residual stresses. In their work,two experiments were conducted: cyclic bending and dampedheating experiments to investigate behavior ITO and PEDOT coatedflexible plastic substrate under mechanical and thermal stresses.Grego et al. [13] used the dry-etch approach to generate cracksat relaxed position. Some research focused on the change in elec-tric resistance of PEDOT and ITO coated flexible substrate underbending fatigue. In this research, reciprocating dies were used tocreate cyclic bending of the structure of thin film and substrate[9]. Moreover, some studies showed the benefit of PEDOT onorganic photovoltaic (OPV) performance. For example, it is usedas a buffer layer between ITO layer and photoactive layer (PAL)leading to increase open circuit voltage (VOC) and power conver-sion efficiency [14–16].

High temperature generates thermal stresses leading to a weak-ness in the cohesion between thin film and substrate, and it affectsthe structural and the physical properties of thin film. Hamashaet al. [17] investigated the behavior of ITO coated flexible PETexposed to fluctuating temperature for a certain period of time.In their study, surface and structure of material were inspectedby SEM, and they observed clear thermal damages on the surface.In this paper, cycle bending test was conducted to show effect ofbending diameter, sheet resistance, frequency and number ofcycles on electrical resistance ITO and PEDOT. Further, damp heatexperiment was conducted on the same films to investigate theeffect of the constant high temperature and humidity on the films.Furthermore, Christou and Paradee investigated the fatigue behav-ior and effect of crack propagation in lead free solder in microelec-tronic packaging [18].

2. Experiments

The thin films used in this work were PEDOT and ITO coatedPET. Two different tests were conducted, bending fatigue, anddamp heating, to investigate electrical and mechanical propertiesof these films under bending fatigue, and damp heating conditions.

2.1. Bending fatigue test

Two different PEDOT sheet resistances (150 X/sq and 225 X/sq)were selected for the test, while only one ITO sheet resistance(60 X/sq) was selected. All films were deposited on 127 lm thickpolyethylene terephthalate (PET). Each sheet was cut into

Fig. 1. Sample shape for both PEDOT and ITO coated PET (t: thickness of substrate,H: height of conductive layer, L: length, and w: width).

rectangular stripes with 9.5 cm in length and 1.3 cm in width, asshown in Fig. 1. Both ends of the sample was fixed with a paperstrips from the substrate side to make a loop of sample and paperstrip. The sample ends from the thin film side were connected withthe ohmmeter terminals using copper tapes. The loop of the sampleand strip was secured in on the cycle bending device as shown inFig. 2. A weight of 75 g was clipped on the loop from the ground sideto ensure continuous connection between the die and the sampleduring the bending. The cycle bending device is illustrated inFig. 2. It consists of electrical motor that provides rotation move-ment, which is converted using a connector into cyclic vertical dis-placement or, in other words, reciprocating motion of the die. Thedevice designed to hold a single (see Fig. 2a–c) or double dies (seeFig. 2d–f). In the single die setup, the sample must be bent cyclicallyaround the die during the upward stroke and kept in the flat posi-tion during the downward stroke. In the double die setup, the sam-ple must be bent cyclically around the lower die during the upwardstroke and around the upper die during the downward strokes.

At upward stroke for any setup (Fig. 2b or e), the upper surface ofthe sample (thin film side) is exposed to tension stress, while thelower surface (substrate side) is exposed to compressed stress. Ten-sion and compression stresses are formed in the vertical cross sec-tional direction under bending loads. The maximum tension isformed at the upper surface of the sample (thin film side), whilethe maximum compression stress is developed at the lower surface.Since the relationship between the stress magnitude and its perpen-dicular position is linear, a plane of no stress (i.e., neutral axis) on thelinear path between the two opposing maxima is formed. The ten-sion or compression stress of a certain point is determined by theclassical formula r = My/Ix, where M is the force moment aboutthe neutral axis, y is the perpendicular distance between the pointand the neural axis (y is zero at the neutral axis, positive in theupward direction, and negative in the downward direction) and Ixis the second moment of area. According to the previous formula,the maximum tension is developed at the thin film and the maxi-mum compression is developed at the lower surface of the substrate.However, in the downward stoke in the double die setup, the uppersurface of the thin film side is exposed to compression stress, whilethe substrate side is exposed to tension stress. The same way of anal-ysis is applied in this case. Thus, the positive direction of y becomesdownward so the maximum tension is developed on the lower sur-face of the substrate and the maximum compression is developed onthe upper surface where the thin film is located. However, both sur-faces will be at neural stress in the single die setup. If the film is brit-tle (e.g., ITO), the cyclic tensile stress of the thin film leads to fatigue,and then the cracks are initiated in the area of maximum stress (mid-section of the thin film). Then, the cracks will propagate at the line ofless energy (perpendicularly to the direction of load toward theedges). In ductile films (e.g., PEDOT), the fatigue and cracks are notexpected to be noticed. However, other type of failure like delamina-tion, or damage may be seen.

DOE is considered as an essential key for design improvementand development in manufacturing sector [19]. It is mainly usedby production quality engineers to determine significant factorsthat affect products’ defects. Also, electronic packaging scientistsuse this tool extensively in analyzing parameters and factors thataffect electronics resistance aiming to build design with high reli-ability [20]. For example, Dasgupta et al. [21] studied the factorsthat affect adhesion compound between liquid crystal polymerand the die using DOE analysis. In addition, nanoparticle conduct-ing films were investigated for resistivity to heat transfer throughstudying the main effect parameters using DOE analysis [22]. Theexamples on using DOE procedures to investigate the significantimpact of certain parameters on a certain response in the field ofmaterial science and thin films technology are so many (e.g.,[23–30]).

Fig. 2. Schematic of the bending tester: (a) sample at neutral position in the single die setup, (b) sample bent around the single die during the upward stroke in the single diesetup, (c) sample at neutral position during the downward stroke in the single die setup, (d) sample at neutral position in the double die setup, (e) sample bent around thelower mandrel during the upward stroke in the double die setup, and (f) sample bent around the upper mandrel during the downward stroke in the double die setup [19].

540 A. Alkhazaili et al. / Microelectronics Reliability 55 (2015) 538–546

In the current study, DOE was used to investigate the signifi-cance of potential factors on the percentage change in electricalresistance. Particularly, a general mixed factorial design wasdesigned to investigate the effect of bending diameter, way ofbending (one way or two ways bending), sheet resistance, numberof cycles, and bending frequency on the percentage change in elec-trical resistance (PCER) of the PEDOT film. Moreover, another gen-eral mixed factorial design was designed to study the effect ofbending diameter, way of bending (one way or two ways bending),number of cycles, and bending frequency on the PCER of ITO films.The selected factors and levels for both PEDOT and ITO areexplained in Table 1.

The total number of the experimental run was 144 for the PEDOTdesign and 72 for the ITO design. The orders of experimental runswere totally random to ensure independence and quality of theresults. Electrical resistance was measured initially, and then the

Table 1Factors of bending fatigue life and their levels.

Factor Level 1 Level 2 Level 3

PEDOT experimentNumber of cycle (NOC) 0 250 500Diameter of bending (DOB) (mm) 8.89 5.08Frequency of bending (FOB) (c/min) 107 30Sheet resistance (X/sq) 150 225Way of bending (WOB) 1wb 2wb

ITO experimentNumber of cycle (NOC) 0 250 500Diameter of bending (DOB) (mm) 8.89 5.08Frequency of bending (FOB) (c/min) 107 30Way of bending (WOB) 1wb 2wb

sample is placed on the bending tool to apply the cyclic bendingas explained above, and finally the electrical resistance was mea-sured again after bending application. For every experimental run,the percentage change in the electrical resistance (PCER) was esti-mated according to the following formula: PCER = [DR/Ro] ⁄ 100%(Ro is initial resistance, and DR is difference between initial andfinal electrical resistance). Analysis of variance (ANOVA) was con-ducted to investigate the significance of the considered factors.Moreover, the main effects and interaction between parameterswere investigated to demonstrate the relationship trends.

2.2. Damp heat experiment

The samples were divided into 3 groups (4 samples were PEDOTwith 225 X/sq, 4 samples were PEDOT with 150 X/sq, and 4 sam-ples were ITO with 60 X/sq). All samples were fixed on a wood

Level 4 Level 5 Level 6 Level 7 Level 8 Level 9

750 1000 1250 1500 1750 2000

750 1000 1250 1500 1750 2000

(d) 2000 cycle (b) 500 cycle (A) 0 cycle (c) 1000 cycle

Fig. 3. Cracks developing after specific number of cycle for ITO coated PET.

(A) (b) 500 cycle (c) 1000 cycle (d) 2000 cycle 0 cycle

Fig. 4. Cracks developing after specific number of cycle for PEDOT coated PET.

A. Alkhazaili et al. / Microelectronics Reliability 55 (2015) 538–546 541

plate to prevent them from moving. Then they were placed inside atemperature and humidity control chamber at angle of 45�. Thetemperature and relative humidity were controlled at 85 �C and85%, respectively. The period of samples exposure to this harshcondition was 1000 h. In order to estimate the PCER, the electricalresistance was measured for all samples before and after the expo-sure to the harsh condition. Scanning electron microscopy (SEM)images were taken to study the effect on the surface morphology.Furthermore, X-ray energy-dispersive X-ray spectroscopy (EDX)test was performed before and after the harsh condition exposureto study the response in the film composition. Finally, transmit-tance test of the PEDOT and ITO thin films was carried out at thecenter of sample before and after harsh condition exposure forall damp heat samples using spectroscopic reflectometer system(Angstrom Sun Technologies Inc., USA). The transmission in %was plotted against wave length in nm for all samples.

99.999

90

50

10

10.1

Fig. 5. Validation figure

3. Results and discussion

In the following two subsections, the experimental results willbe presented besides discussing the science behind every result.

3.1. Bending fatigue results and discussion

As reported by many researchers, the increasing of the electricalresistance of thin films with the cyclic bending is due to cracks[6,8,9]. The cracks are most likely to start from the area with ahighest stress, which is located at the middle in the samples, asdetailed by [6,8]. In this research, many optical microscopic imageswere taken for the surface of the thin films after selected experi-mental runs. The cracks were initially observed at ITO surface after250 bending cycles. However, it the cracks were observed

s of PEDOT design.

1.00.50.0-0.5-1.0

99.999

90

50

10

10.1

Residual

Per

cent

86420

1.0

0.5

0.0

-0.5

-1.0

Fitted Value

Res

idua

l0.80.40.0-0.4-0.8

20

15

10

5

0

Residual

Freq

uenc

y

1 5 10 15 20 25 30 35 40 45 50 55 60 65 70

1.0

0.5

0.0

-0.5

-1.0

Observation Order

Res

idua

l

Normal Probability Plot Versus Fits

Histogram Versus Order

Residual Plots for PCER

Fig. 6. Validation figures of PEDOT design.

Table 2Analysis of variance (ANOVA) for ITO coated PET design.

Source DF Seq SS Adj SS Adj MS F P

NOC 8 1621.821 1621.821 202.728 107.63 0.000BD 1 141.310 141.310 141.310 75.02 0.000F 1 77.285 77.285 77.285 41.03 0.000WOB 1 69.780 69.780 69.780 37.05 0.000NOC ⁄ BD 8 170.429 170.429 21.304 11.31 0.000NOC ⁄ F 8 27.198 27.198 3.400 1.80 0.150NOC ⁄WOB 8 16.235 16.235 2.029 1.08 0.425BD ⁄ F 1 231.616 231.616 231.616 122.97 0.000BD ⁄WOB 1 409.868 409.868 409.868 217.60 0.000F ⁄WOB 1 15.293 15.293 15.293 8.12 0.012NOC ⁄ BD ⁄ F 8 53.186 53.186 6.648 3.53 0.015NOC ⁄ BD ⁄WOB 8 110.166 110.166 13.771 7.31 0.000BD ⁄ F ⁄WOB 1 0.032 0.032 0.032 0.02 0.899Error 16 30.137 30.137 1.884

Total 71 2974.357

S = 1.37243, R-Sq = 98.99% and R-Sq(adj) = 95.50%.

Table 3Analysis of variance (ANOVA) for PEDOT coated PET.

Source DF Seq SS Adj SS Adj MS F P

NOC 8 646.901 646.901 80.863 86.44 0.000BD 1 47.238 47.238 47.238 50.49 0.000F 1 55.311 55.311 55.311 59.12 0.000SR 1 0.982 0.982 0.982 1.05 0.309WOB 1 6.029 6.029 6.029 6.44 0.013NOC ⁄ BD 8 11.864 11.864 1.483 1.59 0.146NOC ⁄ F 8 10.471 10.471 1.309 1.40 0.213NOC ⁄ SR 8 2.986 2.986 0.373 0.40 0.917NOC ⁄WOB 8 5.821 5.821 0.728 0.78 0.624BD ⁄ F 1 0.403 0.403 0.403 0.43 0.514BD ⁄ SR 1 0.139 0.139 0.139 0.15 0.701BD ⁄WOB 1 1.232 1.232 1.232 1.32 0.255F ⁄ SR 1 0.596 0.596 0.596 0.64 0.428F ⁄WOB 1 3.946 3.946 3.946 4.22 0.044SR ⁄WOB 1 53.687 53.687 53.687 57.39 0.000NOC ⁄ BD ⁄ F 8 1.398 1.398 0.175 0.19 0.992NOC ⁄ BD ⁄ SR 8 8.647 8.647 1.081 1.16 0.339NOC ⁄ BD ⁄WOB 8 1.723 1.723 0.215 0.23 0.984BD ⁄ F ⁄ SR 1 1.287 1.287 1.287 1.38 0.245BD ⁄ F ⁄WOB 1 0.094 0.094 0.094 0.10 0.753F ⁄ SR ⁄WOB 1 6.032 6.032 6.032 6.45 0.013Error 66 61.745 61.745 0.936

Total 143 928.529

S = 0.967226, R-Sq = 93.35% and R-Sq(adj) = 85.59%.

542 A. Alkhazaili et al. / Microelectronics Reliability 55 (2015) 538–546

propagated and joined together forming larger cracks with thecycling. Fig. 3 shows the cracks at different number of bendingcycle for the ITO samples. The experimental conditions in the fig-ure were two-way bending, BD: 200 mm and frequency: 107 c/min. However, in the case of PEDOT, the cracks were not observedafter any experimental run at both sheet resistances even at themost considered strained condition (i.e., number of cycle of 100,bending diameter of 200 mil, bending frequency of 107 c/min).See Fig. 4. This disappearance of the cracks is possibly due to highductility of the film.

As phenomenological model of the failure mechanisms of ITOfilm, the cracks initiate as a micro crack from the surface imper-fects [19] at the middle of thin film and propagate perpendicularlywith the direction of bending toward the edges. A research by Alzo-ubi et al. [5] shows that the highest stress is located middle of thethin film using the finite element analysis.

As mentioned before, the DOE was conducted on PEDOT and ITOfilms separately. The first step of this design is to validate the nor-

mality assumption of the error and randomness of standardizedresidual with the observations. Figs. 5 and 6 proof the validity ofPEDOT and ITO DOE designs, respectively. In both figures, the nor-mality assumption of the error is tested by constructing a normalprobability plot of standardized residuals, which is very close tolinear (normal). In order to test the time trends or the run order-related effects, standardized residuals were plotted vs. the runorder of the observations. It was observed that the standardizedresiduals are totally randomly distributed around the zero linewith no pattern appeared within the data. For further validation,the frequency values were plotted against standardized residuals,

20001750150012500 250 500 750 1000

6.0

4.5

3.0

1.5

0.0350200

10730

6.0

4.5

3.0

1.5

0.0

2WB1WB

NOC

Mea

n

BD

F WOB

Main Effects Plot for PCERFitted Means

Fig. 7. Main effect plot for PEDOT coated PET design.

20001750150012500 052 005 057 0001

30

20

10

0350200

10730

30

20

10

02WB1WB

NOC

Mea

n

BD

F WOB

Main Effects Plot for PCERData Means

Fig. 8. Main effect plot for ITO coated PET designs.

0

5

10

15

20

25

30

0 500 1000 1500 2000

PCER

(Mai

n Eff

ect)

Number of Cycles

PEDOT

ITO

Fig. 9. Comparison between the main effect of number of cycles on the PCER forboth PEDOT and ITO.

A. Alkhazaili et al. / Microelectronics Reliability 55 (2015) 538–546 543

and the standardized residuals were plotted against fitted values.The distribution of frequency vs. standardized residuals isappeared normal in both cases and randomness are appeared inthe standardized residuals vs. fitted values plot in both cases. Fromthe discussed above, we can conclude that the ANOVA results ofboth designs are valid.

According to the statistical rules, factors with P-value less thanthe significance level (a) are significant. In this paper, a is selectedto be 0.05. ANOVA results for the ITO experimental design is illus-trated in Table 2. All factors are significant in term of affecting thePCER. However, only three of the two-factor interactions are signif-icant (i.e., BD–F, BD–WOB and F–WOB). All the three-factor inter-actions were significant except BD–F–WOB. Similarly, ANOVAresults for the PEDOT experimental design is illustrated in Table 3.All of NOC, BD, F and WOB are significant factors, while the sheetresistant is not. However, the interaction of this non-significantfactor with way of bending is significant, and its three factor inter-action with way of bending and frequency is significant as well.

350200 10730 2WB1WB30

15

030

15

030

15

0

NOC

BD

F

WOB

0250500750

100012501500

17502000

NOC

200350

BD

30107

F

Interaction Plot for PCERData Means

Fig. 10. Two-factor interaction plot for ITO coated PET design.

350200 10730 225150 2WB1WB8

4

08

4

08

4

08

4

0

NOC

BD

F

SR

WOB

0

250

500

7501000

1250

1500

1750

2000

NOC

200

350

BD

30

107

F

150

225

SR

Interaction Plot for PCERFitted Means

Fig. 11. Two-factor interaction plot for PEDOT coated PET design.

544 A. Alkhazaili et al. / Microelectronics Reliability 55 (2015) 538–546

The last significant interaction was between the frequency and theway of bending.

In order to check the trends in the relationship between thePCER and the significant parameters, the main effect plots for bothPEDOT and ITO designs were constructed, as illustrated in Figs. 7and 8. As expected, the PCER increases with the number of cyclesdue to the fatigue and cracks in the films. Additionally, the PCERdecreases with the bending die diameter due to the tighter curva-ture in the sample at lower bending diameter. Tighter curvatureproduces more peak tension and compression that is applied onthe thin film with the cycling. Moreover, there is direct relationshipbetween the PCER and number of cycle, as depicted in the figures.Finally, as expected, the relationship between bending frequency

and PCER is direct. This relationship is due the increasing of fatiga-bility of thin films with the speed of bending. Finally, the two-wayof bending produces more PCER due to high amplitude in the cyclicstress that is applied on thin film. For example, the film’s stress inthe two-way bending cycles from negative and positive (tensionand compression), while it cycles from zero to positive (neutralto tension) in the case of one-way bending. This makes the stressamplitude at two-way bending double of one-way bending. Theeffect of sheet resistance was not included in Fig. 8 due to insignif-icant effect, as appears in the ANOVA results.

In Fig. 9, a comparison between effect of number of cycles onthe PCER for both PEDOT and ITO is illustrated. PCER is higher forthe ITO at all levels. According to this result, we expect that the

Table 4Sheet resistance for all samples before and after damp heat application.

Thin filmtype

Sheet resistance(manufacturer data) (X/sq)

Initialresistance X/sq

Finalresistance X/sq

PEDOT 150 149 86,000PEDOT 150 148.5 55,000PEDOT 150 150.5 46,000PEDOT 150 150 50,000PEDOT 225 225 180,000PEDOT 225 225 920,000PEDOT 225 225.5 DISPEDOT 225 224.5 DISITO 60 60.5 DISITO 60 61 DISITO 60 59.5 DISITO 60 59 DIS

Table 5EDX analysis for all samples before and after damp heat application.

Sheet resistance Peak Initial composition Final composition

wt% at% wt% at%

PEDOT150 O K 0.941572 0.97009 0.991893 0.99576

S K 0.058428 0.02991 0.008107 0.00424225 O K 0.957097 0.978138 0.991893 0.99576

S K 0.042903 0.021862 0.008107 0.00424

ITO60 O K 0.303791 0.758082 0.430462 0.848642

InL 0.682862 0.237482 0 0SnL 0.013348 0.004437 0.569538 0.151358

0

10

20

30

40

50

60

70

80

90

100

0 200 400 600 800 1000 1200

TRAN

SMIS

SIO

N%

WAVE LENGTH nm

PEDOT-225BHPEDOT-225AHPEDOT-150BHPEDOT-150AHITO-BHITO-AH

Fig. 12. Transmittance plot for PEDOT (150 X/sq, and 225 X/sq) and ITO coated PET(before damp heat denoted BH and after damp heat denoted AH).

A. Alkhazaili et al. / Microelectronics Reliability 55 (2015) 538–546 545

PEDOT film performs better than ITO film in term of reliabilityunder cyclic bending condition in the usage (e.g., flexible display).

In Figs. 10 and 11, the two-factor interaction plots where con-structed for both designs. The two-factor interaction will be signif-icant if the differences in the response between two factors at theupper and lower levels are not the same. In order to conclude a notsignificant interaction, the unparalleled curves should be noticedon the curve. The significant two-factor interactions were mentionon the explanation of ANOVA results.

3.2. Damp heat results and discussion

In order to make sure that the initial resistance in term of X/sqfor every sample after cutting fits the original data, it was mea-sured in our lab using 4-pint probe Ohmmeter before the test,and the readings were very close the original data. Moreover, theresistance after the test was measured using the same way. Afterthe test, the electrical resistance was found dramatically increasedfor all PEDOT and ITO samples, as shown in Table 4. It is notablethat the all PEDOT samples with 150 X/sq show electrical continu-ity after the test, while two samples (out of four) with 225 X/sqshow electrical discontinuity after the test. Therefore, it is expectedthat the damp heat has more significant impact on higher sheetresistance PEDOT films (thinner films). All ITO samples were founddisconnected after the test. In order to understand what happen tothe thin films after the application of damp heat, further scanningelectronic microscopic analysis and EDX analysis were carried out.The structure of thin films looks a little different according to theSEM image. However, films’ compositions were changed noticeablyafter the damp heat. Table 5 shows the initial and final composi-tion of the films in term of weight and atomic weight. In the PEDOTfilms, the content of oxygen increased, while the content of sulfurdecreased. According to this noticed, it is expected that the sulfurevaporates with the heat and/or humidity, leads to worsen the

films electrical characteristics. In the case of ITO film, the contentof indium disappeared, which illustrates why the ITO film becamedisconnected.

The transmittance (%) of the film against wavelength (0–1100 nm) is plotted for the fresh and the tested samples of PEDOTat the two different sheet resistances and ITO, as shown in Fig. 12.It is absolutely notable that when the PEDOT and ITO exposed todamp heat, the transmittance percentage decreases. However,the gap between the transmission percentage between fresh andtested 150 X/sq PEDOT film is higher than the 225 X/sq PEDOTfilm. 150 X/sq PEDOT film is a thicker, and suffers of more damageacross the thicker film under damp heat leading to more interrup-tion of the light passing. The effect of the damp heat on transmis-sion of ITO is not considerably high as shown in the figure.

4. Conclusion

Reliability tests on thin films coated flexible substrates playessential role in the design development of flexible displays. Cyclicbending fatigue and thermal aging experiments were conducted onPEDOT and ITO thin films. The cyclic bending fatigue was con-ducted according to DOE procedure and recommendation. Theused factorial design is mixed full factorial with factors of bendingdiameter, number of cycles, bending frequency and way of bend-ing. The used response is PCER. ANOVA results reveal that bendingdiameter, number of cycles, frequency, and way of bending are sig-nificant factors for both designs (i.e., PEDOT and ITO). PCERincrease for the ITO is significantly higher than PEDOT. Moreover,no cracks or deformation were found on the surface for PEDOT film.However, the cracks begun to appear in the center of ITO samplewhere the stress is highly concentrated then propagated towardthe ends of samples. It is shown that the PCER is increased dramat-ically when PEDOT and ITO exposed to harsh environment condi-tions of 85 �C temperature and 85% humidity. After thisexposure, the indium is totally absent from the ITO film and thesulfur decreased to more than 50% in the PEDOT.

References

[1] Krasnov AN. High-contrast organic light-emitting diodes on flexible substrates.Appl Phys Lett 2002;80(20):3853–5.

[2] Chopra KL, Major S, Pandya DK. Transparent conductors – a status review. ThinSolid Films 1983;102(1):1–46.

[3] Mohamed SH, El-Hossary FM, Gamal GA, Kahlid MM. Properties of indium tinoxide thin films deposited on polymer substrates. Acta Phys Pol, A2009;115(3):704–8.

http://refhub.elsevier.com/S0026-2714(15)00020-7/h0005http://refhub.elsevier.com/S0026-2714(15)00020-7/h0005http://refhub.elsevier.com/S0026-2714(15)00020-7/h0010http://refhub.elsevier.com/S0026-2714(15)00020-7/h0010http://refhub.elsevier.com/S0026-2714(15)00020-7/h0015http://refhub.elsevier.com/S0026-2714(15)00020-7/h0015http://refhub.elsevier.com/S0026-2714(15)00020-7/h0015

546 A. Alkhazaili et al. / Microelectronics Reliability 55 (2015) 538–546

[4] Leterrier Y, Medico L, Demarco F, Manson JAE, Betz U, Escola MF, et al.Mechanical integrity of transparent conductive oxide films for flexiblepolymer-based displays. Thin Solid Films 2004;460(1–2):156–66.

[5] Alzoubi K, Hamasha M, Lu S, Sammakia B. Bending fatigue study of sputteredITO on flexible substrate. IEEE J Disp Technol 2011;7(11):593–600.

[6] Alzoubi K, Lu S, Sammakia B, Poliks M. Experimental study of the high cyclefatigue of thin film metal on polyethylene terephthalate for flexible electronicsapplications. IEEE Trans Compon, Pack Manuf 2011;1(1):43–51.

[7] Hamasha M, Alzoubi K, Lu S, Desue S. Durability study on sputtered indium tinoxide thin film on poly ethylene terephthalate substrate. Thin Solid Films2011;519(18):6033–9.

[8] Park S, Ahn JH, Feng X, Wang S, Huang Y, Rogers JA. Theoretical andexperimental studies of bending of inorganic electronic materials on plasticsubstrates. Adv Funct Mater 2008;18(1):2673–84.

[9] Cairns DR, Crawford GP. Electromechanical properties of transparentconducting substrates for flexible electronic displays. Proc IEEE2005;93(11):1451–8.

[10] Crawford GP. Flexible flat panel displays. Chichester (UK): Wiley; 2005.[11] Chen BF, Hwang J, Yu GP, Huang JH. In situ observation of the cracking

behavior of TiN coating on 304 stainless steel subjected to tensile strain. ThinSolid Films 1999;352(1–2):173–8.

[12] Stephenson SW, Johnson DM, Kilburn JI, Mi X-D, Rankin CM, Capurso RG.Development of a flexible electronic display using photographic technology.In: Soc inf display dig tech, Papers XXXV; 2004. p. 774–7.

[13] Grego S, Lewis J, Vick E, Temple D. A method to evaluate mechanicalperformance of thin transparent films for flexible displays. Thin Solid Films2007;515(11):4745–52.

[14] Manceau M, Bundgaard E, Carlé JE, Hagemann O, Helgesen M, Søndergaard R,et al. Photochemical stability of p-conjugated polymers for polymer solarcells: a rule of thumb. J Mater Chem 2011;21(12):4132–41.

[15] Nüesch F, Tornare G, Zuppiroli L, Meng KX, Chen H, Tian H. Interfacemodification to optimize charge separation in cyanine heterojunctionphotovoltaic devices. Sol Energy Mater Sol Cells 2005;87(1–4):817–24.

[16] Krebs FC, Carlé JE, Cruys-Bagger N, Andersen M, Lilliedal MR, Hammond MA,et al. Lifetimes of organic photovoltaics: photochemistry, atmosphere effectsand barrier layers in ITO-MEHPPV:PCBM-aluminium devices. Sol Energy MaterSol Cells 2005;86:499–516.

[17] Hamasha MM, Dhakal T, Alzoubi K, Albahri S, Qasaimeh A, Lu S, et al. Stabilityof ITO thin film on flexible substrate under thermal aging and thermal cyclingconditions. IEEE J Disp Technol 2012;8(7):383–8.

[18] Christou A, Paradee G, Fatigue behavior and effect of crack propagation in leadfree solder in microelectronic packaging. In: International semiconductordevice research symposium, 9–11 December 2009, College Park, MD, USA.

[19] Hamasha MM, Alzoubi K, Switzer JC, Lu S, Desu SB, Poliks MD. A study on crackpropagation and electrical resistance change of sputtered aluminum thin filmon poly ethylene terephthalate substrate under stretching. Thin Solid Films2011;519(22):7918–24.

[20] Montgomery D. Design and analysis of experiments. 5th ed. Hoboken(NJ): Wiley; 2000.

[21] Dasgupta A, Pecht MG, Mathieu B. Design-of-experiment methods forcomputational parametric studies in electronic packaging. Finite Elem AnalDes 1998;30(1–2):125–46.

[22] Holland B, Mcpherson R, Zhang T, Houl Z, Dean R, Johnson R. Ultra-thin,flexible electronics. In: 2008 Electron. compon. technol. conf.; 2008. p. 1110.

[23] Hamasha MM, Dhakal T, Vasekar P, Alzoubi K, Lu S, Vanhart D, et al. Reliabilityof sputtered deposited aluminum-doped zinc oxide under harshenvironmental conditions. J Sol Energy 2013;2013(89):54–61.

[24] Peng C, Hamasha MM, VanHart D, Lu S, Westgate CR. Electrical and opticaldegradation studies on aluminum-doped zinc oxide transparent conductivethin films under cyclic bending conditions. IEEE Trans Device Mater Reliab2012;13(1):236–44.

[25] Alzoubi K, Hamasha MM, Wang L, Zhang H, Yin J, Luo J, et al. Stability ofinterdigitated microelectrodes of flexible chemiresistor sensor arrays. J DispTechnol 2012;8(7):377–83.

[26] Dhakal T, Hamasha MM, Nandur A, Lu S, Vanhart D, Westgate CR. Moistureinduced surface corrosion in AZO thin films formed by atomic layer deposition.IEEE Trans Device Mater Reliab 2012;12(2):347–56.

[27] Mayyas A, Hamasha MM, Alrashdan A, Hassan AM, Hayajneh MT. Effect ofcopper and silicon carbide content on the corrosion resistance of Al–Mg alloysin acidic and alkaline solutions. J Miner Mater Charact Eng 2012;11(4):435–52.

[28] Hamasha MM, Mayyas A, Hayajneh MT, Hassan AM. The effect of time, percentof copper and nickel on naturally aged Al–Cu–Ni cast alloys. J Miner MaterCharact Eng 2012;11(2):117–31.

[29] Hamasha MM, Mayyas A, Hayajneh MT, Hassan AM. The effect of time, percentof copper and nickel on the natural precipitation hardness of Al–Cu–Ni powdermetallurgy alloys using design of experiments. J Miner Mater Charact Eng2011;10(6):479–92.

[30] Hamasha MM, Alzoubi K, Lu S. Behavior of ITO thin film on PET substrate understretching. J Disp Technol 2011;7(8):426–33.

http://refhub.elsevier.com/S0026-2714(15)00020-7/h0020http://refhub.elsevier.com/S0026-2714(15)00020-7/h0020http://refhub.elsevier.com/S0026-2714(15)00020-7/h0020http://refhub.elsevier.com/S0026-2714(15)00020-7/h0025http://refhub.elsevier.com/S0026-2714(15)00020-7/h0025http://refhub.elsevier.com/S0026-2714(15)00020-7/h0030http://refhub.elsevier.com/S0026-2714(15)00020-7/h0030http://refhub.elsevier.com/S0026-2714(15)00020-7/h0030http://refhub.elsevier.com/S0026-2714(15)00020-7/h0035http://refhub.elsevier.com/S0026-2714(15)00020-7/h0035http://refhub.elsevier.com/S0026-2714(15)00020-7/h0035http://refhub.elsevier.com/S0026-2714(15)00020-7/h0040http://refhub.elsevier.com/S0026-2714(15)00020-7/h0040http://refhub.elsevier.com/S0026-2714(15)00020-7/h0040http://refhub.elsevier.com/S0026-2714(15)00020-7/h0045http://refhub.elsevier.com/S0026-2714(15)00020-7/h0045http://refhub.elsevier.com/S0026-2714(15)00020-7/h0045http://refhub.elsevier.com/S0026-2714(15)00020-7/h0050http://refhub.elsevier.com/S0026-2714(15)00020-7/h0055http://refhub.elsevier.com/S0026-2714(15)00020-7/h0055http://refhub.elsevier.com/S0026-2714(15)00020-7/h0055http://refhub.elsevier.com/S0026-2714(15)00020-7/h0065http://refhub.elsevier.com/S0026-2714(15)00020-7/h0065http://refhub.elsevier.com/S0026-2714(15)00020-7/h0065http://refhub.elsevier.com/S0026-2714(15)00020-7/h0070http://refhub.elsevier.com/S0026-2714(15)00020-7/h0070http://refhub.elsevier.com/S0026-2714(15)00020-7/h0070http://refhub.elsevier.com/S0026-2714(15)00020-7/h0075http://refhub.elsevier.com/S0026-2714(15)00020-7/h0075http://refhub.elsevier.com/S0026-2714(15)00020-7/h0075http://refhub.elsevier.com/S0026-2714(15)00020-7/h0080http://refhub.elsevier.com/S0026-2714(15)00020-7/h0080http://refhub.elsevier.com/S0026-2714(15)00020-7/h0080http://refhub.elsevier.com/S0026-2714(15)00020-7/h0080http://refhub.elsevier.com/S0026-2714(15)00020-7/h0085http://refhub.elsevier.com/S0026-2714(15)00020-7/h0085http://refhub.elsevier.com/S0026-2714(15)00020-7/h0085http://refhub.elsevier.com/S0026-2714(15)00020-7/h0095http://refhub.elsevier.com/S0026-2714(15)00020-7/h0095http://refhub.elsevier.com/S0026-2714(15)00020-7/h0095http://refhub.elsevier.com/S0026-2714(15)00020-7/h0095http://refhub.elsevier.com/S0026-2714(15)00020-7/h0100http://refhub.elsevier.com/S0026-2714(15)00020-7/h0100http://refhub.elsevier.com/S0026-2714(15)00020-7/h0105http://refhub.elsevier.com/S0026-2714(15)00020-7/h0105http://refhub.elsevier.com/S0026-2714(15)00020-7/h0105http://refhub.elsevier.com/S0026-2714(15)00020-7/h0115http://refhub.elsevier.com/S0026-2714(15)00020-7/h0115http://refhub.elsevier.com/S0026-2714(15)00020-7/h0115http://refhub.elsevier.com/S0026-2714(15)00020-7/h0120http://refhub.elsevier.com/S0026-2714(15)00020-7/h0120http://refhub.elsevier.com/S0026-2714(15)00020-7/h0120http://refhub.elsevier.com/S0026-2714(15)00020-7/h0120http://refhub.elsevier.com/S0026-2714(15)00020-7/h0125http://refhub.elsevier.com/S0026-2714(15)00020-7/h0125http://refhub.elsevier.com/S0026-2714(15)00020-7/h0125http://refhub.elsevier.com/S0026-2714(15)00020-7/h0130http://refhub.elsevier.com/S0026-2714(15)00020-7/h0130http://refhub.elsevier.com/S0026-2714(15)00020-7/h0130http://refhub.elsevier.com/S0026-2714(15)00020-7/h0135http://refhub.elsevier.com/S0026-2714(15)00020-7/h0135http://refhub.elsevier.com/S0026-2714(15)00020-7/h0135http://refhub.elsevier.com/S0026-2714(15)00020-7/h0140http://refhub.elsevier.com/S0026-2714(15)00020-7/h0140http://refhub.elsevier.com/S0026-2714(15)00020-7/h0140http://refhub.elsevier.com/S0026-2714(15)00020-7/h0145http://refhub.elsevier.com/S0026-2714(15)00020-7/h0145http://refhub.elsevier.com/S0026-2714(15)00020-7/h0145http://refhub.elsevier.com/S0026-2714(15)00020-7/h0145http://refhub.elsevier.com/S0026-2714(15)00020-7/h0150http://refhub.elsevier.com/S0026-2714(15)00020-7/h0150

Reliability of thin films: Experimental study on mechanical and thermal behavior of indium tin oxide and poly(3,4-ethylenedioxythiophene)1 Introduction2 Experiments2.1 Bending fatigue test2.2 Damp heat experiment

3 Results and discussion3.1 Bending fatigue results and discussion3.2 Damp heat results and discussion

4 ConclusionReferences