Influence of mechanical strain on the electrical properties of flexible organic thin-film transistors

This content has been downloaded from IOPscience. Please scroll down to see the full text.

Download details:

IP Address: 140.113.38.11

This content was downloaded on 25/04/2014 at 00:45

Please note that terms and conditions apply.

Influence of mechanical strain on the electrical properties of flexible organic thin-film

transistors

View the table of contents for this issue, or go to the journal homepage for more 2011 Semicond. Sci. Technol. 26 034005

(http://iopscience.iop.org/0268-1242/26/3/034005)

Semicond. Sci. Technol. 26 (2011) 034005 (5pp) doi:10.1088/0268-1242/26/3/034005

Influence of mechanical strain on the

electrical properties of flexible organic

thin-film transistors

Fang-Chung Chen

, Tzung-Da Chen, Bing-Ruei Zeng and Ya-Wei Chung

Received 30 September 2010, in final form 16 November 2010 Published 14 February 2011

Online atstacks.iop.org/SST/26/034005 Abstract

In this study, we systematically investigated the bending effect on the electrical properties of flexible organic thin-film transistors (OTFTs) fabricated on stainless steel substrates. We found that the compressive strain resulted in an increased mobility, while the tensile strain degraded the electrical performance. We further used a transfer line method to extract the channel and parasitic resistances under either compressive or tensile strain. The results indicated that the parasitic resistance increased apparently under the tensile bending condition, which probably could be attributed to the damage of the source/drain contacts. Additionally, we deduced that mechanical strains influence the energy barrier height between the grains of pentacene thin films, thereby resulting in the variation of channel resistances. Overall, the flexible OTFTs fabricated on the metal foils exhibited high mechanical flexibility and stability. (Some figures in this article are in colour only in the electronic version)

1. Introduction

Organic thin-film transistors (OTFTs) have received much attention because of their potential applications on low-cost, light-weight, and large-area electronic devices, such as smart cards, paper-like displays, and radio-frequency identification tags [1–5]. In particular, the mechanical flexibility of organic materials facilitates the preparation of organic devices using a high-throughput technique, such as roll-to-roll processing on flexible substrates. At present, flexible OTFTs have been widely reported [6–12]. Among these flexible devices, most of them were fabricated on plastic substrates, such as polyethylene terephthalate and polyethylenenaphthalate films [6–8]. On the other hand, metal foils, such as stainless steel (SS) substrates, are also gaining great interest in the field of flexible electronics due to their excellent mechanical properties, thermal stability and chemical robustness. The low oxygen and moisture permeability can provide effective protection from possible oxidation reactions [9–12]. OTFTs made on SS substrates, however, have been seldom reported. Further, a thorough analysis of the electrical properties of these OTFTs under bending conditions still remains rare. In

1 _{Author to whom any correspondence should be addressed.}

this work, with the use of pentacene as the semiconducting material, and poly(vinyl cinnamate) (PVCN) [13] as the insulator, we fabricated flexible OTFTs on SS substrates. The bending properties of these devices were systematically investigated. We found that compressive strain resulted in an increased mobility, while tensile strain degraded the electrical performance. From the extracted resistance, we deduced that the compressive strain forced the pentacene grains to become more compact, thereby facilitating the charge transport in the grain boundaries. On the other hand, the tensile strain caused an increase in the spacing between the grains, thereby increasing the energy barrier height for charge hopping. Overall, the flexible OTFTs fabricated on the metal foils exhibited high mechanical flexibility and stability.

2. Experimental details

The device structure is illustrated in figure 1. To fabricate flexible OTFTs on SS substrates, a thick photoresist (EOC-130B) layer was first spin-coated to planarize the surface of the SS foils [12]. Subsequently, a layer of polyimide was also fabricated onto the photoresist layer to modify

Semicond. Sci. Technol. 26 (2011) 034005 F-C Chen et al

(a)

(b) (c)

Figure 1.(a) The device structure and materials used in this study. The schematic presentation of the bending test apparatus:

(b) compressive and (c) tensile bending. The arrow indicates the direction of the charge transport.

the surface energy for improving the adhesion of the upper layers on the substrate. Then, Al was thermally evaporated through a shadow mask to function as the gate electrode. For the dielectric layer, PVCN was firstly dissolved in dichlorobenzene (15 wt%) and was spin coated on the substrates. After a drying process at 80 ◦C for 10 min, the sample was irradiated with UV light (λ = 264 nm) to crosslink the PVCN layer [13]. The substrate was then etched by propylene glycol monomethyl ether acetate to define the insulating area. After the etching process, the substrate was dried at 80 ◦C for 10 min. The thickness of the resulting PVCN layer was∼400 nm. The measured capacitance was 6.12 × 10−9 F cm−2 [14]. For the semiconducting layer, pentacene was thermally deposited (80 nm) on top of the PVCN layer. Finally, Au electrodes were evaporated to serve as the source/drain electrodes. We used a shadow mask to define channel length (L) and width (W ). The W /L ratio was kept at 20. The electrical characteristics were measure by Keithley 4200 in ambient air and at room temperature. For the bending tests, the devices were stressed with various bending radii (figures1(b) and (c)). The bending direction was parallel to that of charge transport.

3. Results and discussion

Figure2(a) shows the device characterization obtained under different mechanical bending conditions. Before the bending test, the device exhibited typical p-channel field effect behavior [14,15]; the extracted mobility (μ) following the conventional field effect transistor model in the saturation regime was 0.17 cm2 _V−1 _s−1_{. The threshold voltage (V}

t) and on–

off ratio were −12.0 V and 104_{, respectively. Under either}

compressive or tensile strain with a bending radius of 10 mm, we only observed minor current change in the transfer curves (figure2(a)). The threshold voltage slightly shifted positively

(a)

(b)

Figure 2.(a) The transfer characteristics of the devices measured under different bending conditions. (b) The output characteristics of the devices measured under different bending conditions.

by 0.30 V under compressive strain. On the other hand, the Vt shifted toward the negative direction by 0.45 V under

tensile strain, suggesting that a higher driving voltage was required to turn on the channel conduction. From the output characteristics (figure2(b)), we could more clearly tell that the drain current increased under compressive strain and decreased under the tensile condition. Further, the current differences increased with the gate voltage (VG).

The strain on the substrate surface (εsurface) at a certain

bending curvature radius (R) can be expressed using the equation [16] εsurface=  df+ ds 2R  (1 + 2η + χ η2₎ (1 + η)(1 + χ η), (1)

where df is the film thickness, ds is the substrate thickness, η= df/ds, χ= Yf /Ys, and Yf and Ys are Young’s moduli

of the film and the substrate, respectively. Because df and ds were ∼1.5 and 150 μm, respectively, we estimated η to be ∼0.01. Further, Young’s modulus of organic materials usually ranges from 0.1 to 10 GPa [17], and the Ys of SS

substrates is∼200 GPa [18]. Hence, χ should be quite small as well. Therefore, the strain (εsurface) can be approximately

determined as ds/2R. Figure 3(a) displayed the normalized mobility with respect to the value obtained without bending as a function of the calculated strain. We can see that the mobility increased with increasing compressive strain and decreased with increasing tensile strain. More interestingly, we were able to fit a linear curve well to this set of data [19]. The slope of the fitting curve is 0.078 (%−1), suggesting that 1% of compressive or tensile strain will lead to an increase

-1.0 -0.5 0.0 0.5 1.0 0.92 0.94 0.96 0.98 1.00 1.02 1.04 1.06 No rm al iz ed M o b ili ty ( μ / μ0 ) Strain (%) Compressive Tension

Figure 3.Mobility of the OTFTs as a function of strain. The initial values of mobility were normalized for ready comparison.

or a decrease of ∼8% in mobility, respectively. Assuming the charge transport is thermal activated and governed by an Arrhenius-type equation, the mobility can be described as

μ∼ μ0exp(−E/kBT), where E is the energy barrier for the

charge hopping process [19,20]. Therefore, we can estimate the effect of mechanical strain on E in the flexible OTFTs as E∝ −kBT ln(1+0.078ε), where strain (ε) is expressed in

percent [19].

In order to further clarify the effect of mechanical bending, a transfer line method (TLM) [21] was adapted to analyze the resistance responses of the devices to the applied strain. In the linear regime, the total device turn-on resistance (Ron) mainly

comes from two contributions: the channel resistance (Rch)

and the parasitic resistance (Rp) according to the following

equation: Ron=  ∂VDS ∂ID VG=const. VDS→0 = Rch+ Rp = L W μiCi(VG− VT) + Rp. (2)

Ron could be calculated from the linear regime of the output

curves (figure2(b)). Because RPis independent of the channel

length, it can be extracted by plotting Ron as a function of

channel length (L) (figure4). The intercepts of the curves with the y-axis indicated the parasitic resistances. After knowing

Rp, the channel resistance could be also deduced. From

figure 4, the calculated RP and Rch before bending were

0.89 M and 13.7 k μm−1, respectively. Under compressive strain, we can clearly see that the parasitic resistance only decreased slightly to 0.86 M; the channel resistance decreased to 13.1 k μm−1. However, both RP and Rch

under tensile strain increased to 1.01 M and 14.6 k μm−1, respectively. In particular, we observed a large increase (∼12%) in the parasitic resistance under tensile strain.

To understand the origin of the increased RP, we

performed a so-called sticky tape test under either compressive or tensile strain with a bending radius of 10 mm [22]. As shown in figures 5(a) and (b), we found that the top Au source/drain electrodes of the devices could be much more easily removed under the tensile condition. Because Young’s modulus of Au (∼80 GPa) is much larger than that of pentacene (∼15 GPa) [23], the strain of Au films was smaller than that of the pentacene layer under a similar shear stress. Therefore, as

0 30 60 90 120 150 180 0 1x106 2x106 3x106 4x106 Tot al re sistance ( Ω ) Channel length (μm) Flatten Compress Tensile

Figure 4.The device resistances as a function of channel length under different bending conditions. According to the transfer line method, the intercept of the fitting curve indicates the parasitic resistance.

shown figure5(c), we deduce that the source/drain electrodes tended to be delaminated, resulting in a poor contact. As a result, the contact resistance was increased under tensile strain. In contrast, the contact resistance reminded almost unchanged under the compressive bending condition, which was probably due to the opposite direction of the stress on the device layers.

Figure6shows the surface morphology of the pentacene layer, which exhibited typical polycrystalline morphology. For a polycrystalline TFT, the mobility is usually governed by the grain boundary mobility [24, 25]. In other words, the hopping process is usually limited by the activation energy barrier between the grains. Following the above assumption, we propose a model to explain the above observation which is as follows. Under a tensile strain (figure 6(c)), because the spacing between the gains probably becomes longer, the energy barrier between the grains is increased, thereby reducing the hopping rate of holes. On the other hand, when the substrates are under compressive strain, the grains come closer and, subsequently, the energy barrier is decreased (figure6(d)). Therefore, we observed an increased mobility under an inward (compressive) bending condition.

Finally, we performed a cyclic bending test at a radius of 10 mm to examine the mechanical stability. During the bending test, the device was subject to either tensile or compressive strain. After bending, the electrical properties were measured in a flat state. Figure7reveals the normalized mobility as a function of bending times. For the device tested under compressive strain, the mobility gradually increased with bending cycles. On the other hand, the mobility of the device under tensile strain increased first and then decreased after ∼200 times of bending cycles. Further, after the applied bias was released, the mobility of the compressive device somehow recovered back to its origin value gradually, implying that the variation of the mobility was probably due to the polarization of the dielectric layer and/or adsorption of H2O and O2from the atmosphere [26,27]. On the other hand,

for the tensile device, the damage of the source/drain contact might overwhelm the bias-stress effect in the beginning of the stress test, resulting in a decreased mobility. The two factors then probably became balanced and, therefore, the mobility

Semicond. Sci. Technol. 26 (2011) 034005 F-C Chen et al

(a)

(c)

(b)

Figure 5.The pictures of the sticky tapes lifted-off from the device either under (a) compressive strain or (b) tensile strain. The bending radius was 10 mm. Note that more Au metal was removed under tensile strain. (c) The schematic representation of the cross-section of the device under the tensile strain condition.

(a)

(b)

(c) (d)

Figure 6.(a) The surface morphology of the pentacene layer. The schematic representation of the gains of the pentacene layer on a (b) flat substrate; (c) substrate under tensile strain; (d) substrate under compressive strain. 0 200 400 600 800 1000 0.94 0.96 0.98 1.00 1.02 1.04 1.06 1.08 1.10 Norm aliz ed mobil ity ( μ / μ0 )

Bending cycles (times) Compression

Tension

Figure 7.Variation of the normalized mobility as a function of bending cycles under either compressive or tensile strain.

became almost unchanged after ∼400 times of bending cycles.

4. Conclusion

We have investigated the effect of mechanical strain on pentacene thin-film transistors. The devices were fabricated on flexible stainless steel substrates and tested under different bending conditions. We observed that compressive strain resulted in an increased mobility, while tensile strain led to degraded electrical performance. We further used TLM to extract the channel and parasitic resistances under either compressive or tensile strain with a bending radius of 10 mm. The results indicated that the parasitic resistance increased by∼12% under the tensile bending condition, which probably could be attributed to the damage of the source/drain contacts. Additionally, we deduced that mechanical strains influence the energy barrier height between the grains of pentacene thin films, thereby resulting in the variation of channel resistances. Finally, the fatigue test revealed that the OTFTs in this study exhibited supreme flexibility, especially under compressive bending conditions.

Acknowledgments

The authors thank the National Science Council (NSC 99-2221-E-009-181 and NSC 98-2218-E-009-002) and the Ministry of Education (though the ATU program) for financial support.

References

[1] Zaumseil J and Sirringhaus H 2007 Chem. Rev.1071296

[2] Sung C F, Kekuda D, Chu L F, Lee Y Z, Chen F C, Wu M C and Chu C W 2009 Adv. Mat.214845

[3] Facchetti A, Yoon M H and Marks T J 2005 Adv. Mat.171705

[4] Subramanian V, Chang P C, Lee J B, Molesa S E and Volkman S K 2005 IEEE Trans. Compon. Packag. Technol.

28741

[5] Chung C S, Cheng J A, Huang Y J, Chang H F, Chen F C and Shieh H P D 2008 Appl. Phys. Lett.93053305

[6] Kubozono Y, Haas S, Kalb W L, Joris P, Meng F, Fujiwara A and Batlogg B 2008 Appl. Phys. Lett.93033316

[7] Tan H S, Kulkarni S R, Cahyadi T, Lee P S, Mhaisalkar S G, Kasim J, Shen Z X and Zhu F R 2008 Appl. Phys. Lett.

93183503

[8] Jedaa A and Halik M 2009 Appl. Phys. Lett.95103309

[9] Wu C C, Theiss S D, Gu G, Lu M H, Sturm J C, Wagner S and Forrest S R 1997 IEEE Electron Device Lett.18609

[10] Afentakis T, Hatalis M, Voutsas A T and Hartzell J 2006 IEEE

Trans. Electron Devices53815

[11] Chuang T K, Troccoli M, Kuo P C, Jamshidi-Roudbari A, Hatalis M K, Biaggio I and Voutsas A T 2007 Appl. Phys.

Lett.90151114

[12] Chen F C, Wu J L, Lee C L, Huang W C, Chen H M P and Chen W C 2009 IEEE Electron Device Lett.30727

[13] Jang J, Kim S H, Hwang J, Nam S, Yang C, Chung D S and Park C E 2009 Appl. Phys. Lett.95073302

[14] Chen F C, Chung C S, Lin Y S, Kung L J, Chen T H and Shieh H P D 2006 Org. Electron.7435

[15] Chen F C, Chen Y P, Huang Y J and Chien S C 2010 J. Phys.

D: Appl Phys.43405103

[16] Won S H, Chung J K, Lee C B, Nam H C, Hur J H and Jin J 2004 J. Electrochem. Soc.151G167

[17] Stuart B H 2002 Polymer Analysis (England: Wiley) [18] Peng I H, Liu P T and Wu T B 2009 Appl. Phys. Lett.

95041909

[19] Sekitani T, Kato Y, Iba S, Shinaoka H, Someya T, Sakurai T and Takagi S 2005 Appl. Phys. Lett.86073511

[20] Horowitz G 2003 Adv. Funct. Mater.1353

[21] Necliudov P V, Shur M S, Gundlach D J and Jackson T N 2003 Solid-State Electron.47259

[22] Ma W, Yang C, Gong X, Lee K and Heeger A J 2005 Adv.

Funct. Mater.151617

[23] Tahk D, Lee H H and Khang D Y 2009 Macromolecules

427079

[24] Horowitz G, Hajlaoui M E and Hajlaoui R 2000 J. Appl. Phys.

874456

[25] Annibale P, Albonetti C, Stoliar P and Biscarini F 2007

J. Phys. Chem. A11112854

[26] Jung T, Dodabalapur A, Wenz R and Mohapatra S 2005 Appl.

Phys. Lett.87182109