A study of bulk current mechanism in P3HT-based thin film transistors and approach for current suppression

(1)

A study of bulk current mechanism in P3HT-based thin ﬁlm transistors

and approach for current suppression

Chia-Hao Chang

a

, Chao-Hsin Chien

a,b,⇑ a

Department of Electronics Engineering and Institute of Electronics, National Chiao Tung University, Hsinchu, Taiwan, ROC

b

National Nano Device Laboratories, Hsinchu, Taiwan, ROC

a r t i c l e

i n f o

Article history:

Received 27 February 2012

Received in revised form 12 June 2012 Accepted 25 June 2012

Available online 2 August 2012 Keywords:

P3HT OTFTs

Polymer semiconductor Bulk current effect Electrodes MWCNTs

a b s t r a c t

In this article, poly(3-hexylthiophene) (P3HT) based thin ﬁlm transistors with Ti capped source and drain electrodes (S/D) was employed to have an insight into the mechanism of bulk current effect, which led to poor subthreshold swing and large off current. Our newly developed PTFTs do show greater characteristics in the aspects of on/off current ratio and subthreshold swing than those with the conventional Au/Ti S/D. In order to explain the results, we propose that the bulk current is composed of two components, i.e., side-wall and top-face injections. We ascribe the improvements to the reduction of top-face injection bulk current due to the larger injection barrier between the P3HT and Ti. As comparing the PTFTs with Ti capped laterally grown multi-wall carbon nanotube (MWCNT) S/D with the PTFTs with MWCNT S/D, we also observe similar tendency of bulk current reduction. From the viewpoint of device operation, better subthreshold swing and smaller off current result in faster device switching and lower power consumption. By this new approach of S/D structure, there are two and 20 times improvements on the sub-threshold swing and off current, respectively, although the reduction of the bulk current also leads to a slight decrease of the on-current.

1. Introduction

Organic semiconductors, such as polymers, are attrac-tive since they have the potential to be easily applied on a large scale ﬂexible substrate due to the fact that the re-quired process temperature is pretty low. Solution process-able, self-organized, regioregular poly(3-hexylthiophene) (P3HT) is very suitable to the fabrication of organic thin ﬁlm transistors (OTFTs) since theses transistors demon-strate relatively high carrier mobility. The hole mobility of OTFTs with a P3HT active layer (i.e., PTFTs) have been demonstrated as high (0.01–0.3 cm2/V s) as amorphous

silicon thin ﬁlm transistors with a reasonable on/off ratio (more than 100 in air and 106 _{in an inert atmosphere)}

[1–5]. However, the performance of PTFTs can be signiﬁ-cantly inﬂuenced by the channel coating method and the macromolecular parameters of channel material [1,3–5]. The electrical properties can be improved by incorporating surface treatment on the surface of gate and electrodes, and post treatments such as O2plasma treatment and ther-mal annealing [3,4,6,7]. Recently, many researchers have proposed that the characteristics of PTFTs were correlative to the thickness of active layer [8–10]. This effect was attributed to the existence of bulk current in the active layer before the formation of accumulation layer, and, usu-ally, higher bulk current would be conducted in a thicker active layer [8,9]. Moreover, poor off-current and sub-threshold swing were observed when the thickness of ac-tive layer was increased [8–10]. This topic has been getting more and more attention, and different methods

http://dx.doi.org/10.1016/j.orgel.2012.06.042

⇑ Corresponding author. Address: Electronics Engineering and Institute of Electronics, National Chiao-Tung University, 1001 University Road, Hsinchu 300, Taiwan, ROC. Tel.: +886 3 5712121x54252; fax: +886 3 5724361.

E-mail address:[email protected](C.-H. Chien).

Contents lists available atSciVerse ScienceDirect

Organic Electronics

(2)

the systematic studies, we have an insight on the origin of the effects coming from the bulk current of PTFTs and provide an approach to suppress this leakage current. Moreover, we also observe similar tendency of bulk cur-rent reduction when comparing the PTFTs with Ti capped laterally grown multi-wall carbon nanotube (MWCNT) S/ D with those with MWCNT S/D.

2. Experimental procedure

Fig. 1depicts the schematic cross section of PTFTs. The speculated components of bulk current injected from the different source electrodes in subthreshold region for PTFTs with conventional and Ti capped Au/Ti S/D are shown inFig. 1a and b, respectively. We will address cur-rent components in detail later. The N++_{doped silicon} sub-strate and thermally grown oxides of 200 nm thickness were used as gate electrodes and insulators, and then a lift-off process patterned Au/Ti (50 nm/2.5 nm) or Ti

over a growth time of 10 min. For Ti capped laterally grown MWCNT S/D formation, a Ti (10 nm) capped Fe/Ti (5 nm/10 nm) was used and went through the same pro-cesses as MWCNT S/D formation, and the MWCNTs were intentionally grown on the side-wall of the electrodes, i.e., laterally grown MWCNTs, from the exposed Fe parti-cles. Some article had demonstrated methods to laterally grow MWCNTs[14,15].Fig. 2a and b depict the schematic cross sections of PTFTs with MWCNT S/D and Ti capped lat-erally grown MWCNT S/D. Prior to the deposition of P3HT as the channel, the substrate was cleaned by acetone, iso-propyl alcohol and deionized (DI) water in an ultrasonic bath for 5 min, respectively, and treated by hexamethyl disilazane (HMDS) passivation in a vacuum oven at 150 °C to create a hydrophobic surface. Then, the P3HT ac-tive layer was spun in yellow bay using a two-step pro-cess—1000 rpm for 10 s in the ﬁrst step and 1500 rpm for 25 s in the second step, followed by baking at 150 °C for 3 min. Chloroform was used as the solvent for P3HT,

Fig. 1. A schematic cross section of PTFTs is depicted, and the components of bulk current injected from different source electrodes under subthreshold region for PTFTs with conventional and Ti capped Au/Ti S/D are shown in (a) and (b), respectively.

(3)

resulting in a weight percentage of solutions for P3HT in chloroform of 0.05 and 0.2 wt.%. All of the P3HT solutions were treated with ultrasonic vibration for 5 min before usage and were percolated by a PTFE membrane syringe ﬁlter with a pore size of 0.2

l

m. The regioregular P3HT used in this study was purchased from FEM Inc. (Mw = 50000), and the provided coupling ratio of head–tail to head–head and tail–tail was approximately 90%. The top views of PTFTs with MWCNT S/D and Ti capped laterally grown MWCNT S/D were observed by using optical micro-scope (OM). The electrical properties of P3HT TFTs were measured by a Hewlett–Packard 4156C semiconductor parameter analyzer, and the devices were evaluated at their saturation region with a drain voltage of 60 V. We extracted mobility by maximum transconductance in the saturation region. All of the devices were processed and characterized in air atmosphere.

3. Results and discussion

The drain current–gate voltage (ID–VG) characteristics of the fabricated PTFTs with conventional and Ti capped S/D

are shown inFig. 3. The employed weight percentages of solutions of P3HT in chloroform were 0.05 and 0.2 wt.%. By comparing the results of the PTFTs with conventional S/D with different channel thicknesses, we ﬁnd the PTFTs with thick active layer coated with 0.2 wt.% solution exhi-bit poorer off current and subthreshold swing than those with thin active layer coated with 0.05 wt.% solution. We think that this result might be closely related to the pres-ence of bulk current[8,9]. This early conduction increases with the raising thickness because of the decrease in the channel resistance. Larger the early conduction, i.e., bulk current, more positive threshold voltage shift is. Moreover, since the channel resistance varies with the gate bias that inducing band bending of active layer before the accumu-lation layer formation, the bulk current will increase with increasing negative gate voltage bias and then also contrib-ute to the on-current. As a result, in order to have a supe-rior switching performance, this bulk current shall be avoided as far as possible. The goal can be attained by using asymmetric electrodes or incorporating an insulating mesa-like structure in between source and drain elec-trodes [11–13]. However, these processes obviously are not consistent with low cost and easy process.

Jung et al. had proposed that the carrier injection into the accumulated channel was via two possible paths, i.e., top-face and side-wall injections[16]. By two dimensional numerical simulation and analytical model, Islam demon-strated that the subthreshold performance of the devices with a stack contact (A electrode structure composed by a low work function metal upper layer and a high work function metal under layer.) could be signiﬁcantly im-proved [10]. Nevertheless, unfortunately, the detail of mechanism of bulk current is still insufﬁcient because the device structure used in his simulation is not similar to the typical device structure in which both the surface of gate insulator and electrodes are covered with the chan-nel layer[10]. Inspired by the proposal of Jung et al. and based on the properties of PTFTs with Au/Ti S/D, we sup-pose that the bulk current might also consist of two injec-tion current paths, i.e., top-face and side-wall injecinjec-tions, as shown in Fig. 1a. In an aim to verify our inference, we

Fig. 2. Schematic cross sections of PTFTs is depicted, and PTFTs with MWCNT and Ti capped laterally grown MWCNT S/D are shown in (a) and (b), respectively.

Fig. 3. Transfer current characteristics of PTFTs with Au/Ti and Ti capped Au/Ti S/D, and different P3HT thicknesses, which are coated by 0.05 and 0.2 wt.% solution.

(4)

D, i.e., without employing Au, exhibit almost no current ﬂow in the accumulation layer, which is not shown here. For the PTFTs with thin active layer, the mobility is 0.008 cm2_{/Vs (0.006 cm}2_{/Vs) for PTFTs with Au/Ti S/D} (PTFTs with Ti capped Au/Ti S/D), comparable to the values shown in the recent literatures[6,7]. In addition, the on/off current ratio and subthreshold swing were improved from 1.1 103 and 9.85 V/decade for PTFTs with Au/Ti S/D to 5.4 103_{and 5.3 V/decade for PTFTs with Ti capped Au/} Ti S/D. The improvement cannot be predicted by the well-known formula for on/off current ratio[17,18]. It also cannot be explained by the space-charge limited current (SCLC) model, which is in common used to explain the bulk current effect[8,10].

The drain current–drain voltage (ID–VD) characteristics under a gate voltage bias of 0 V are shown inFig. 4. The performances of PTFTs with different P3HT thicknesses, which were coated by 0.2 and 0.05 wt.% solution, are shown inFig. 4a and b, respectively. There might be no is-sue on the drain current which is fully composed of the bulk current as the gate voltage is of 0 V, and this current usually is regarded as the leakage current. Obviously, the drain current of PTFTs with Ti capped Au/Ti S/D are lower

injection part by the Ti capping layer is more conspicuous. These results again are consistent to our interpretations.

As shown inFig. 5, we extracted the values of the con-tact and channel resistances of the samples inFig. 3 by transmission line method (TLM). The values of contact resistance of both PTFTs with Au/Ti S/D and with different channel thicknesses are close to 2 109_X_-

_l

_{m. This} repre-sents the coverage of P3HT on the channel and S/D region was still pretty well when thin channel was used. There-fore, the possibility that different P3HT thicknesses lead to different performances, shown inFig. 3, due to the dis-tinct coverage of P3HT can be excluded. As we have men-tioned, the larger side-wall injected bulk current due to the thicker P3HT layer is responsible for the resultant low-er channel resistance. Furthlow-ermore, while the S/D was re-placed by Ti capped Au/Ti, the contact resistance only slightly increases from 2 109 _{to 6 10}9_X_-

_l

_{m or from} 1.5 109to 2.0 109X-

l

m for PTFTs with thin or thick active layer, respectively. The increase is thought to be cor-related to the suppression of top-face injected bulk current and only weight a small portion of contact resistance. As a result, we think our argument that the side-wall injected bulk current is the dominant current component for the

Fig. 4. Drain current–drain voltage (ID–VD) characteristics of PTFTs with Au/Ti and Ti capped Au/Ti S/D under a gate voltage bias of 0 V. The performances of

(5)

PTFTs with thick active layer is rather plausible, and the re-sult is consistent with the conspicuous performance improvement resulted from suppressing top-face injection bulk current for PTFTs with thin active layer. The parame-ters of all PTFTs with thin and thick P3HT layer with Au/Ti and Ti capped Au/Ti S/D are summarized inTable 1.

This effect of reducing bulk current by incorporating a layer of Ti upon the electrodes shall still be observable when the electrode material is changed. We therefore re-placed Au by MWCNTs since we had demonstrated that pentacene and P3HT based TFTs showed superior perfor-mance by employing MWCNTs as the electrodes[19,20]. We obtained the increased on-current and reduced contact resistance. However, it is anticipated that there will be a larger bulk current, i.e., easier early conduction, because of the reduced contact resistance. The drain current–gate voltage (ID–VG) characteristics of the fabricated PTFTs with MWCNT S/D or Ti capped laterally grown MWCNT S/D are shown inFig. 6a. The performance of PTFTs with Au/Ti S/D is also shown as the control. The employed weight percent-age of solution of P3HT in chloroform was 0.2 wt.%. As shown in Fig. 6a, both the PTFTs with MWCNT and Ti capped laterally grown MWCNT S/D have the beneﬁts resulting from employing MWCNT S/D and, as expected,

larger bulk current than the PTFTs with Au/Ti S/D is also observed. However, obviously, the improvements on the subthreshold swing and off-current by incorporating Ti capping layer are still observed. Moreover, the perfor-mance of PTFTs with Ti capped directly on Fe/Ti layer with-out MWCNT growth is shown inFig. 6b. For comparison, the performance of PTFTs with MWCNT S/D and Au/Ti S/ D is also incorporated. Obviously, if there was no MWCNTs grew on the side-wall of electrode, i.e., no laterally grown MWCNTs, no current could conduct in the P3HT layer be-cause of high injection barrier even the gate voltage bias is high enough to form the accumulation layer. This result could exclude the doubt about that the drain current of PTFTs with Ti capped laterally grown MWCNT S/D might not come from laterally grown MWCNTs but the pre-treated Ti capped Fe/Ti.

In addition, there might be concern that the effective channel length was reduced due to the laterally grown

Fig. 5. The results of contact and channel resistance correlated to the samples inFig. 3, which were extracted by transmission line method (TLM).

Table 1

Summation of the parameters for the PTFTs with different thicknesses, and Au/Ti or Ti capped Au/Ti S/D.

Samples, PTFTs with lfet

(cm2 / Vs) S.S. (V/ dec.) Vt (V) Ion/Ioff

0.05 wt.% Channel and Au/ Ti S/D

0.00884 9.85 22 1.1 103

0.2 wt.% Channel and Au/ Ti S/D 0.00803 15.7 3 5.26 102 0.05 wt.% Channel and Ti capped Au/Ti S/D 0.00609 5.30 25 5.39 103 0.2 wt.% Channel and Ti capped Au/Ti S/D 0.00607 10.2 4 7.5 102

Fig. 6. (a) Transfer current characteristics of PTFTs with MWCNT S/D and Ti capped laterally grown MWCNT S/D. (b) Transfer current characteris-tics of PTFTs with vertical MWCNT S/D and no laterally grown MWCNT S/ D. In other words, only pretreatment was performed. P3HT was coated with 0.2 wt.% solution, and the transfer curves of PTFTs with Au/Ti S/D are also incorporated for comparison.

(6)

MWCNTs. Hence to confirm the length of laterally grown MWCNTs is needed. However, we failed to directly confirm the laterally grown MWCNTs by using scanning electron microscope (SEM) because of the fact that there was the magnetic interference induced by Fe. As an alternative, the top views of the PTFTs with different electrodes were observed by using OM as shown inFig. 7. The channel re-gion of the PTFTs with Ti capped laterally grown MWCNT S/D (Fig. 7b) was similar to that of the PTFTs with Ti capped Fe/Ti S/D without MWCNT growth (Fig. 7a). In other words, if the length of the laterally grown MWCNTs is sufficient to influence the channel length, the edges of the Ti capped electrodes would be as black as the vertical grown MWCNTs (Fig. 7c). Therefore, we infer that the length of laterally grown MWCNTs is so short such that the influence on the channel length can be ignored. Supposedly, the short length might be due to the thin catalytic layer (5 nm), resulting in a small exposure area for carbon source and slower MWCNT growth rate. Accordingly, the doubt about the influence of carbon nanotubes on the channel length can be excluded. Moreover, someone may suspect that we have failed to grow the laterally grown MWCNTs. Instead of that, there might be carbide formation and, in turn, lead to the improvement in mobility. This doubt could be excluded because of the fact that the work func-tion of Ti–C and Fe–C are 4 eV and 4.5 eV, respectively

[21,22]. Clearly, relatively large carrier injection barrier will be induced, which does not beneﬁt the performance if the carbides were really formed. Thus, the laterally grown MWCNTs do exist and contribute to the improve-ment in the on-current level.

4. Conclusion

In this paper, we present an insight into the bulk cur-rent effect of PTFTs, and provide an approach to suppress it. The bulk current was composed of side-wall and top-face injection currents. With proper scheme, bulk current could be signiﬁcantly reduced and enhance the perfor-mance of PTFTs. Our newly developed PTFTs with a modi-ﬁed structure by employing a Ti capping Au/Ti S/D did show the improvements on subthreshold swing and on/ off current ratio. We think that these improvements mainly come from the reduction of top-face injected bulk current. For the PTFTs with thin active layer, there is great-er than one ordgreat-er of magnitude of reduction in the off-cur-rent level, as well as an obviously increased on/off curoff-cur-rent ratio, and a much lower subthreshold swing. Furthermore, we fabricated the novel PTFTs which used a Ti capped lat-erally grown MWCNTs as the electrodes to strengthen the foundation of our inference. With the evidence provided by

Fig. 7. Optical micrographs of the PTFTs with different electrodes, (a) Ti capped Fe/Ti S/D without MWNCT growth, (b) Ti capped laterally grown MWCNT S/ D, and (c) vertical grown MWCNT S/D.

(7)

the newly developed transistors, we ﬁnd our approach to suppress the bulk current is effective and useful.

References

[1] H. Sirringhaus, P.J. Brown, R.H. Friend, M.M. Nielsen, K. Bechgaard, B.M.W. Langeveld-Voss, A.J.H. Spiering, R.A.J. Janssen, E.W. Meijer, P. Herwig, D.M. de Leeuw, Nature 401 (1999) 685.

[2] H. Sirringhaus, N. Tessler, R.H. Friend, Science 280 (1998) 1741. [3] G. Wang, J. Swensen, D. Moses, A.J. Heeger, J. Appl. Phys. 93 (2003)

6137.

[4] S. Cho, K. Lee, J. Yuen, G. Wang, D. Moses, A.J. Heeger, M. Surin, R. Lazzaroni, J. Appl. Phys. 100 (2006) 114503.

[5] H. Yang, T.J. Shin, Z. Bao, C.Y. Ryu, J. Polym. Sci. Part B: Polym. Phys. 45 (2007) 1303.

[6] S.K. Park, Y.H. Kim, J.I. Han, D.G. Moon, W.K. Kim, IEEE Trans. Elect. Dev. 49 (2002) 2008.

[7] S.K. Park, Y.H. Kim, J.I. Han, D.G. Moon, W.K. Kim, M.G. Kwak, Synth. Met. 139 (2003) 377.

[8] M.J. Deen, M.H. Kazemeini, Y.M. Haddara, J. Yu, G. Vamvounis, S. Holdcroft, W. Woods, IEEE Trans. Elect. Dev. 51 (2004) 1892.

[9] E. von Hauff, F. Johnen, A.V. Tunc, L. Govor, J. Parisi, J. Appl. Phys. 108 (2010) 063709.

[10] M. Nurul Islam, J. Appl. Phys. 110 (2011) 114906.

[11] M.H. Lim, I.Y. Lee, S.G. Jeong, J. Lee, W.S. Jung, H.Y. Yu, G.H. Kim, Y. Roh, J.H. Park, Org. Electron. 13 (2012) 1056.

[12] W.Y. Chou, S.T. Lin, H.L. Cheng, F.C. Tang, Y.J. Lin, C.F. You, Y.W. Wang, Appl. Phys. Lett. 90 (2007) 222103.

[13] J.Z. Wang, Z.H. Zheng, H. Sirringhaus, Appl. Phys. Lett. 89 (2006) 083513.

[14] Y.S. Han, J.K. Shin, S.T. Kim, J. Appl. Phys. 90 (2001) 5731. [15] B.H. Chen, P.Y. Lo, J.H. Wei, M.J. Tsai, C.L. Hwang, T.S. Chao, H.C. Lin,

T.Y. Huang, Electrochem. Solid-State Lett. 8 (2005) G290. [16] K.D. Jung, Y.C. Kim, H. Shin, B.G. Park, J.D. Lee, E.S. Cho, S.J. Kwon,

Appl. Phys. Lett. 96 (2010) 103305.

[17] A.R. Brown, C.P. Jarrett, D.M. de Leeuw, M. Matters, Synth. Met. 88 (1997) 37.

[18] G. Horowitz, Adv. Mater. 10 (1998) 365.

[19] C.H. Chang, C.H. Chien, J.Y. Yang, Appl. Phys. Lett. 91 (2007) 083502. [20] C.H. Chang, C.H. Chien, IEEE Electron Device Lett. 32 (2011) 1457. [21] F. Santerre, M.A. El Khakani, M. Chaker, J.P. Dodelet, Appl. Surf. Sci.

148 (1999) 24.