An alternative bend-testing technique for a flexible indium tin oxide film
Yen-Liang Chen
a,b, Hung-Chih Hsieh
a, Wang-Tsung Wu
a, Bor-Jiunn Wen
b, Wei-Yao Chang
a, Der-Chin Su
a,*a
Department of Photonics and Institute of Electro-Optical Engineering, National Chiao Tung University, 1001 Ta-Hsueh Road, Hsinchu 30010, Taiwan
b
Center for Measurement Standards/Industrial Technology Research Institute (ITRI), Bldg. 16, 321, Kuang Fu Rd. Sec. 2, Hsinchu 30011, Taiwan
a r t i c l e
i n f o
Article history:
Received 28 January 2010
Received in revised form 12 July 2010 Accepted 21 July 2010
Available online 24 July 2010 Keywords:
Indium tin oxide film Bending test Refractive index Electro-optic modulation Heterodyne interferometry
a b s t r a c t
The two-dimensional refractive index distribution of a flexible indium tin oxide film deposited on a PET layer is measured before/after the bend-testing with an alternative technique based on Fresnel equations and the heterodyne interferometry. Their standard deviations are derived and they vary more obviously than the resistance variations measured in the conventional method. Hence the standard deviation of the refractive index can be used as the indicator to justify the durability of a flexible indium tin oxide film. The validity is demonstrated.
Ó 2010 Elsevier B.V. All rights reserved.
1. Introduction
Due to the high optical transmittance and the high conductivity of an indium tin oxide (ITO) film, it is widely used as the electrode for flat panel display devices, solar sells and organic light emitting diodes (OLED)[1]. In practice, the electronic patterns are printed on a thin ITO film deposited on a polyethylene terephthalate (PET) layer with laser patterning processes[2]to fabricate a flexi-ble electronic substrate. The residual stress in part areas on the film caused by the temperature variation during the processes affects its quality and lifetime. The bending test can make the effect of the residual stress to be more obvious[3–7], so the resis-tance variation after definite bending cycles is always used as an indicator to justify its durability. However the resistance measure-ment is less sensitivity and the durability test becomes tedious due to its time-consuming bending cycles. To improve the sensitivity of this test, the two-dimensional microscopic refractive index distri-bution variations are observed instead of the conventional single-value resistance measurement.
Due to its photoelasticity[8], its refractive index is related with the residual stress. An alternative technique for measuring the two-dimensional refractive index distribution is presented based on Fresnel equations and the heterodyne interferometry. In this method, a collimated linearly/circularly polarized heterodyne light beam in turn enters a modified Twyman-Green interferometer, in which an ITO film is located in one arm for test. Two groups of
full-field interference signals are taken by a fast CMOS camera. The sampling intensities recorded at each pixel are fitted to derive a sinusoidal signal, and its associated phase can be calculated. Then, substituting these two groups of phase distribution data into the special equations derived from Fresnel equations, its two-dimensional refractive index distribution and the standard deviation can be estimated. This film is tested before/after some different bending cycles. The standard deviation varies more obviously than the resistance variation measured with the conven-tional method. Hence the standard deviation of the two-dimen-sional refractive index distribution can be used as an indicator to justify its durability. The validity of this technique is demonstrated. 2. Principle
Fig. 1shows a schematic diagram of this technique. For conve-nience, the +z-axis is chosen to be along the light propagation direction and the +y-axis is along the direction pointing out the paper plane. A light beam coming from a heterodyne light source [9] has a frequency difference f between the x- and the y-polarizations, and its Jones vector can be written as[10]
E1¼ 1 ffiffiffi 2 p e ipft eipft : ð1Þ
The light beam is expanded and collimated by a beam expander BE. It enters a modified Twyman-Green interferometer, which con-sists of a beam-splitter BS, a quarter-wave plates Q2with the fast axis at 45° with respect to the y-axis, a reference mirror M, a test sample S, an analyzer AN with the transmission axis at 0° with
0141-9382/$ - see front matter Ó 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.displa.2010.07.003
* Corresponding author.
E-mail address:[email protected](D.-C. Su).
Displays 31 (2010) 191–195
Contents lists available atScienceDirect
Displays
respect to the y-axis, an imaging lens IL, and a CMOS camera C. In this interferometer, two optical paths are (1) BS ? Q2?M ? Q2?BS ? AN ? IL ? C (the reference beam), and (2) BS ? S ? BS ? AN ? IL ? C (the test beam). For convenience, we firstly as-sume that the S is an isotropic material. So, their amplitudes can be derived and expressed as
Er1¼ ðANð0Þ RBS Q2ð45Þ M Q2ð45Þ E1Þ ei/d1 ¼ 1 0 0 0 ei/r=2 0 0 ei/r=2 ! 1 ffiffiffi 2 p 1 i i 1 r m 0 0 rm 1ffiffiffi 2 p 1 i i 1 1 ffiffiffi 2 p e ipft eipft ! ei/d1¼irme i /ð d1pft/r=2Þ ffiffiffi 2 p 1 0 ; ð2Þ and Et1¼ ðANð0Þ S RBS E1Þ ei/d2 ¼ 1 0 0 0 r 0 0 r ei/r=2 0 0 ei/r=2 ! 1 ffiffiffi 2 p e ipft eipft ! ei/d2 ¼ re iðpftþ/d2/r=2Þ ffiffiffi 2 p 1 0 : ð3Þ
Here, RBS, M and S are the reflection matrix of the BS, M and S; rm and r are the reflection coefficients of the M and the S; /d1 and /d2are the phase variations due to the optical path lengths of the reference and test beam, respectively. /ris the phase difference be-tween the x- and y-polarizations coming from the reflection at BS. Thus, the interference signals recorded by the C can be written as
IA¼ jEr1þ Et1j 2 ¼ I01þ
c
1 cosð2pft þ /1Þ ¼1 2fr 2þ r2 m 2rrmcos½2pft þp
2 ð/d1 /d2Þg; ð4Þwhere I01, and
c
1and /1are the mean intensity, the visibility and the phase of the interference signal, respectively. From Eq.(4), we have/1¼
p
2 ð/d1 /d2Þ: ð5Þ
Secondly, the quarter-wave plate Q1with the fast axis at 45° to the y-axis is inserted into the optical setup as shown inFig. 1and the light amplitude becomes
E2¼ Q1ð45Þ E1¼ 1 2 1 i i 1 eipft eipft ! ¼1 i 2 cosðpftÞ sinðpftÞ cosðpftÞ þ sinðpftÞ ¼1 2 1 i eipft þ1 2 i 1 eipft: ð6Þ
From Eq.(6), we can see that there is a frequency difference f between the right- and the left-circular polarizations, and it is a circularly polarized heterodyne light beam.
The amplitudes of the reference beam and the test beam can be derived as above and expressed as
Er2¼ ðANð0Þ RBS Q2ð45Þ M Q2ð45Þ E2Þ ei/d1 ¼ 1 0 0 0 ei/r=2 0 0 ei/r=2 ! 1 ffiffiffi 2 p 1 i i 1 r m 0 0 rm 1ffiffiffi 2 p 1 i i 1 1 i 2 cosðpftÞ sinðpftÞ cosðpftÞ þ sinðpftÞ ei/d1 ¼i þ 1 2 rm½cosðpftÞ þ sinðpftÞe ið/d1/r=2Þ 1 0 ; ð7Þ and Et2¼ ðANð0Þ S RBS E2Þ ei/d2 ¼ 1 0 0 0 r 0 0 r ei/r=2 0 0 ei/r=2 ! 1 i 2 cosð
p
ftÞ sinðp
ftÞ cosðp
ftÞ þ sinðp
ftÞ ei/d2 ¼i 12 r½cosð
p
ftÞ sinðp
ftÞeið/d2/r=2Þ 1 0
: ð8Þ
Here the interference signals measured by the C can be written as
IB¼ jEr2þ Et2j2¼ I02þ
c
2 cosð2pft þ /2Þ¼ A cosð2pftÞ þ B sinð2pftÞ þ C; ð9Þ
Fig. 1. Schematic diagram of this technique. LS: laser light source; EO: electro-optic modulator; FG: function generator; VA: voltage linear amplifier; BE: beam expander; Q: quarter-wave plate; BS: beam-splitter; M: mirror; S: sample; AN: analyzer; IL: imaging lens; MO: microscopic objective; DL: doublet; C: CMOS camera.
where I02,
c
2and /2are the mean intensity, the visibility and the phase of the interference signals, respectively. A, B, and C are real numbers. Here we haveA ¼ rrmsinð/d1 /d2Þ and B ¼ 1 2ðr 2 m r 2 Þ: ð10Þ
From Eq.(10), the phase can be calculated as
/2¼ tan1 B A ¼ cot1 2rrmsinð/d1 /d2Þ ðr2 m r2Þ : ð11Þ
According to Fresnel equations[10], r can be derived by substi-tuting Eq.(5)into Eq.(11)and expressed as
r ¼ cos /1þ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi cos2/ 1þ cot2/2 q cot /2 rm¼ n 1 n þ 1; ð12Þ
where n is the refractive index of S. Eq. (12) can be rewritten as
n ¼cot /2 rmcos /1þ rm ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi cos2/ 1þ cot2/2 q cot /2þ rmcos /1 rm ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi cos2/ 1þ cot2/2 q : ð13Þ
The camera C with the frame frequency fcis used to record m frames at time t1, t2, . . . , tm. Each pixel records a series of interfer-ence intensities I1, I2, . . . Im, which are the sampled intensities of a sinusoidal signal. The data of /1and /2can be derived by using the absolute phase determination method[11] and the least-square sinusoidal fitting algorithm[12]. If these processes are applied to all other pixels, then the associated data /1(x,y) and /2(x,y) can be obtained similarly. Substitute them into Eq.(13), the refractive index distribution n(x,y) can be calculated.
In reality, the S is a photoelastic material, its Jones matrix S in Eq.(3)and Eq.(8)should be modified and represented as
S ¼ rpp rps rsp rss
: ð14Þ
Consequently, /1and /2can be derived again with two simul-taneous equations Eqs.(3) and (4)and Eqs.(8) and (9), and they can be written as /1¼ tan1 rmcosð/d1 /d2Þ þ rpssin /r rmsinð/d1 /d2Þ rpscos /r ; ð15Þ /2¼ cot1 2rppðrmsinð/d1 /d2Þ rpscos /rÞ r2 m r2ppþ r2ps 2rmrpssinð/d1 /d2 /rÞ " # ; ð16Þ
where rpp¼ðnenðnoÞ cos 2eþ1Þðnaoþnþ1Þeno1, rps¼ðnðnoenþ1ÞðneÞ sin 2oþ1Þa, no is the ordinary refractive index, ne is the extraordinary refractive index and
a
is the angle between the principal axis of the stress and the y-axis. Substitute Eq. (15) and Eq. (16) into Eq. (13), the relationship between n and the refractive index variation Dn caused by the residual stress can be obtained, whereDn ¼ ne no is also known as the birefringence. Because a two-dimensional photoelastic model exerted by the forces in its own plane will behave as a general retarder[8], the retardation induced byDn is proportional to the associated stress variation. Under the experimental conditions rm= 99%, /r= 10°, /d1–/d2= 10° and no= 1.8, the relationship between n and Dn can be calculated and depicted in Fig. 2 ata
= 0°, 30°, 60° or 90° by substituting Eqs. (15) and (16) into Eq.(13). From this figure, we can see that the measured data of n is quasi-linear to the stress ifa
–90°. Although the principal axis of the stress at each pixel is different from others, the standard deviation can be measured to show the effect of the residual stress as the resistance measurement in the conventional method.3. Experiments and results
To demonstrate the validity of this technique, an ITO film sam-ple coated on PET layer (CPFilms/OC 100) was tested. The isolation lines with about 0.85 mm pitch were printed on the ITO film with laser patterning processes. It was originally a plane plate as shown inFig. 3a, where the gray, the white and the black parts represent the ITO film, the PET layer and the isolation lines, respectively. Under the bending test, it was bent with a constant bending force F applied on its two sides simultaneously as shown inFig. 3b. Then, the bending force was removed and it became a plane plate again. In our tests, it was bent with a bending machine (ITRI-CMS/FCIS/ 08) under the conditions of 20 mm bending radius, 20°/s bending angular speed, and 8 cycles/min bending frequency. Although it has a 60 mm 60 mm dimension, only a 260
l
m 260l
m area near its center with higher stress caused by the bending test was measured, as shown in Fig. 3a. An He–Ne laser with 632.8 nm wavelength, an electro-optic modulator (New Focus/Model 4002), a CMOS camera (Basler/A504 K) with 8-bit gray levels and 350 350 pixels, a reference mirror with rm= 99% and a 10 im-age lens IL were used. To measure the full-field absolute phase, the saw-tooth voltage signal with the amplitude V being lower than its half-wave voltage Vp was applied to drive the EO, thephase at the break-point position of the periodical sinusoidal seg-ment was used to be the reference phase[11]. Under the condi-tions f = 20 Hz, Vp= 144 V, V = 120 V and fc= 300 frames/s, 300 frames were taken in 1 s each time. For easier reading, the mea-sured results of the two-dimensional refractive index distribution are displayed in color. Fig. 4 shows the results before bending, andFig. 5a and b show the results after bending 1000 cycles and 4000 cycles, respectively. The mean values of the measured area can be calculated and they are 1.855, 1.849 and 1.811, respectively; that is, they have 0.32% and 2.37% variations. Their associated stan-dard deviations can also be derived and they are 0.006, 0.008, and 0.033, respectively. So, they have 33% and 450% variations. 4. Discussion
For comparison, the resistances of this sample were also mea-sured before/after bend-testing by a digital multimeter (Model 2700, Keithley Instrument Inc.). The nodes are located as shown inFig. 3a. The measured resistances of the above three situations are 2.613 kX, 2.623 kX and 2.674 kX, respectively. They have 0.38% and 2.33% variations. For easy understanding, the variations
Fig. 2. The relation curves of n versus Dn at differenta.
of two methods and the standard deviation measured with this method are depicted inFig. 6. Either the resistance or the mean value of the refractive index acts as an average effect inside the
conduct area, so they have consistent variations as shown in
Fig. 6. From this figure, we can infer that the standard deviation of the refractive index may be used as an indicator to justify the durability of the ITO film. It varies more sensitively than the resis-tance does.
To collect the reflected parallel beam from the ITO surface and magnify the object simultaneously, it is better to use the imaging lens IL composed of a microscopic objective (MO) and a doublet (DL). Because the IL is an afocal optical system, the ITO surface needs to be located in the front focal plane of the MO and the im-age plane of the CMOS camera should be correspondingly located in the rear focal plane of the DL. The transverse magnification of the IL in our experiments is 10.
All the lower left circles inFigs. 4, 5a and bhave larger refractive index variations, so it has more residual stress. Comparing the data in these three circles, we can see that the residual stress increases with more bending cycles[3–5]. This tendency can be also found in two other circles inFig. 5a and b.
5. Conclusion
The two-dimensional refractive index distribution of an ITO film deposited on a PET layer has been measured before/after bend-testing with an alternative method based on Fresnel equations and the heterodyne interferometry. The variation of the standard deviation is larger than the resistance variation measured in the
Fig. 4. The two-dimensional refractive index distribution n(x,y) before bending.
Fig. 5. The two-dimensional refractive index distribution n(x,y) after bending (a) 1000 cycles and (b) 4000 cycles.
Fig. 3. The sample is (a) unbent and (b) bent; where gray color and white color represent the ITO film and the PET layer, respectively.
conventional method, so it may be used as the indicator to justify its durability. The validity has been demonstrated.
Acknowledgement
This study was supported in part by the National Science Coun-cil, Taiwan, ROC, under Contract NSC95-2221-E009-236-MY3. References
[1] C. Sujatha, G.M. Rao, S. Uthanna, Characteristics of indium tin oxide films deposited by bias magnetron sputtering, Mater. Sci. Eng. B94 (2002) 106–110. [2] H.Y. Tsai, H. Yangm, C.T. Pan, M.C. Chou, Laser patterning indium tin oxide
(ITO) coated on PET substrate, Proc. SPIE 4230 (2000) 156–163.
[3] S. Grego, J. Lewis, E. Vick, D. Temple, Development and evaluation of bend-testing techniques for flexible-display applications, J. Soc. Inf. Display 13 (2005) 575–581.
[4] S.P. Gorkhali, D.R. Cairns, G.P. Crawford, Reliability of transparent conducting substrates for rollable displays: a cyclic loading investigation, J. Soc. Inf. Display 12 (2004) 45–49.
[5] M.H. Lee, K.Y. Ho, P.C. Chen, C.C. Cheng, S.T. Chang, M. Tang, M.H. Liao, Y.H. Yeh, Promising a-Si:H TFTs with high mechanical reliability for flexible display, Tech. Dig. IEDM (2006) 299–302.
[6] B. J. Wen, T. S. Liu, C. H. Chen, H. Y. Ko, Z. Y. Chung, S. C. Liao, P-72: optical-characteristic measurement of flexible display for reliability test, In: SID Symp. Dig. Tech. Papers, vol. 40, 2009, pp. 1378–1381.
[7] J.R. Lee, D.Y. Lee, D.G. Kim, G.H. Lee, Y.D. Kim, P.K. Song, Characteristics of ITO films deposited on a PET substrate under various deposition conditions, Metal Mater. Int. 14 (2008) 745–751.
[8] K.J. Gåsvik, Optical Metrology, third ed., John Wiley & Sons (2002) 217–245. [9] D.C. Su, M.H. Chiu, C.D. Chen, Simple two frequency laser, Prec. Eng. 18 (1996)
161–163.
[10] E. Hecht, Optics, forth ed., Addison-Wesley, 2002.
[11] Y.L. Chen, D.C. Su, A method for determining full-field absolute phases in the common-path heterodyne interferometer with an electro-optic modulator, Appl. Opt. 47 (2008) 6518–6523.
[12] IEEE, Standard for Terminology and Test Methods for Analog to-Digital Converters, IEEE Std. 1241–2000 (2000) 25–29.
Fig. 6. The variations of the mean value (symbol j), the standard deviation (symbol d) of this method, and the resistance (symbol ) at 1000 and 4000 bending cycles.
