Journal of the Korean Physical Society, Vol. 53, No. 3, September 2008, pp.

14641467

Eect of Electrode Thickness on Bottom-Contact

Pentacene Thin-Film Transistors
Chaeho Kim and D. Jeon
Department of Physics Education and Nano Systems Institute, Seoul National University, Seoul 151-748

(Received 12 December 2007)

The usual way of fabricating bottom-contact pentacene eld eect transistors (FETs) is to
vacuum-evaporate pentacene molecules onto a substrate with patterned electrodes. We observed
that the heights of the source and the drain electrodes aected the drain current of a pentacene
FET made in this way. Atomic force microscopy measurements of the patterned substrate coated
with pentacene revealed a region in the channel where pentacene did not adsorb along the electrodes
edge. This pentacene-denuded zone was caused by the sticking coecient of pentacene molecules
to the side wall of the metal electrode being higher than that of pentacene molecules to other nu-
cleation sites on the Si-oxide channel. The denuded zone was eventually lled with pentacene by
prolonged deposition, but when and how it was lled depended on the height of the electrode. Our
study suggests that dierent heights of the electrodes aect the drain current via the formation of
a pentacene-denuded zone.
PACS numbers: 85.30.Tv, 68.37.Ps, 68.55.-a, 81.10.Bk
Keywords: Pentacene, Organic FET, Electrode thickness, Denuded zone

I. INTRODUCTION source and the drain electrodes aected the growth of

a pentacene lm and, thus, the drain current of a pen-
tacene FET. Our in-situ I-V and ex-situ atomic force
Organic electronic devices are a subject of intensive microscopy measurements suggested that the pentacene
research due to their exibility, pro-environmental pro- growth in the channel near the edges of the source and
cess and low cost. The organic thin lm transistors are the drain electrodes is strongly in uenced by the height
an example application and for this, pentacene is consid- of the electrodes, which subsequently aects the drain
ered a good candidate due to its relatively high mobility. current of the completed pentacene FET.
Pentacene transistor research has focused on many dif-
ferent aspects, such as the formation and the treatment
of electrodes, the formation of a gate insulating layer, the
growth of pentacene and so on. For example, the per- II. EXPERIMENT

formance of a pentacene eld eect transistor (FET) in

relation with the structure of electrodes [1{3], the elec- The bottom-contact electrodes were patterned using
trode materials [4{8] and the treatment of electrodes sur- a conventional photolithography technique on a 300-
face [9{12] has been studied. Here, we report our study nm-thick silicon oxide on a highly-doped n -type Si(100)
on how the height of the source and the drain electrodes wafer that served as a gate. The channel was 10-m
aect the drain current of a pentacene bottom-contact long and 4-mm wide. Source and drain electrodes were
FET. Compared to the top-contact FET, fabrication of formed by thermal deposition of gold on a thin chrome
the bottom-contact FET is easier because the electrodes layer serving as an adhesion layer. This patterned sub-
can be formed by using conventional photolithography strate was then cleaned with ethyl alcohol and acetone
and because pentacene can be vacuum-evaporated di- for 10 min each in an ultrasonic cleaning bath. After
rectly onto the patterned substrate. How a pentacene blowing o the solvent by using nitrogen gas, organic
lm grows is strongly aected by the properties of the residue was removed in an ultraviolet-ozone cleaner for
substrate surface. In other words, how the pentacene 10 min. We used 97 % pure pentacene powder, as-
grows on a at silicon oxide surface and on a patterned purchased from Sigma-Aldrich. Pentacene was thermally
electrode substrate will be very dierent. In particular, evaporated from a boron-nitride crucible heated by using
we were interested in studying how the heights of the a tungsten lament in a 10 7 Torr vacuum chamber. To
stabilize the deposition rate, we preheated the crucible
 E-mail: [email protected]; Fax: +82-2-873-7881 for 1 hour before opening the shutter. The substrate was
-1464-
Eect of Electrode Thickness on Bottom-Contact


{ Chaeho Kim and D. Jeon -1465-

Fig. 2. AFM images of pentacene lms grown on patterned

substrates. The thicknesses of the electrodes were (a) 1.5 nm
and (b) 15 nm. The same amount of pentacene was deposited
simultaneously on the two substrates. A pentacene-denuded
zone is formed in the channel along the edge of the electrodes
for the case of the 15-nm electrode. (c) AFM image obtained
after a prolonged pentacene deposition. The images look sim-
ilar, regardless of the electrode height.
Fig. 1. Drain current vs thickness of pentacene (VDS =
VGS = {20 V) measured in-situ. The drain current did not
ow for the 10-nm-thick electrode. In the cases of 28-nm and
93-nm electrodes, the drain current increased with the start ence of the two images is clear. On the channel of 1.5-
of pentacene deposition, saturated and then increased again. nm-electrode substrate, the pentacene lm is formed by
the connection of pentacene islands expanding from each
nucleation sites. Some islands have reached the edge of
kept at room temperature during deposition and the de- the source and the drain electrodes; thus, a channel for
position rate was around 10  A/min. the drain current is formed. On the other hand, in Fig-
The drain current was measured while pentacene was ure 2(b), one can see a stripe in the channel where pen-
being deposited onto the patterned substrates with 10- tacene is not deposited along the edges of the source and
nm, 28-nm and 93-nm-high electrode to study the eect the drain electrodes. When pentacene was deposited for
of electrode height on the drain current. Both the drain- a prolonged time, a polycrystalline pentacene lm was
source voltage and the gate-source voltages were {20 V. observed regardless of the electrode height, as depicted
in Figure 2(c). The pentacene grains in the channel are
larger than those on the Au electrodes because pentacene
nucleates more easily on Au than on Si oxide [13].
III. RESULTS AND DISCUSSION
The reason for the dierent morphologies in Figures
2(a) and (b) and for the pentacene-denuded zone along
Figure 1 shows the drain current vs. pentacene thick- the electrode edge in Figure 2(b) is not due to the shadow
ness. In the case of a 10-nm-high electrode, the drain eect of the electrodes because the height dierence of
current did not ow at all. In the case of the 28-nm-high the two electrodes is negligible when compared to the
electrode, the current started to increase with the depo- channel width (5 m). We think the pentacene-denuded
sition of pentacene and was almost saturated when the zone in Figure 2(b) is due to the higher sticking rate
pentacene lm was about 15-nm thick. When the lm of pentacene on Au than on other sites on the Si-oxide
became about 30-nm thick, however, the current started channel [13]. A pentacene molecule landing on the Si-
to increase again, although the rate was very small. The oxide channel diuses around until it attaches to a nu-
overall behavior of the drain current for the case of 93- cleation site, a pentacene island, or the sidewall of the
nm-high electrode appeared similar to that for the case Au electrode. Since the sticking rate onto the Au sur-
of 28-nm electrode. The drain current started to increase face is higher, the electrode sidewall acts as a sink for
with a little time delay when pentacene was deposited, impinging pentacene molecules, producing the denuded
saturated when the lm became about 20 nm and in- zone along the sidewall.
creased again when the lm became about 60 nm. To be more precise, we show, in Figure 3(a) and (b),
To nd possible reasons for the observed behavior of the height proles of the dotted lines marked A and B
drain current against the pentacene lm thickness, we in the images of Figures 3(a) and (b), respectively. The
used atomic force microscopy (AFM) to obtain images of average height of pentacene on the oxide channel is the
the pentacene-covered patterned substrates. Figure 2(a) same for Figures 3(a) and (b) and there is no pentacene
shows AFM image of pentacene deposited on the channel adsorbed in the denuded zone. Because of the sidewall
and 1.5-nm-high electrodes and Figure 2(b) shown that adsorption, the pentacene lm grows in two directions
on the channel and 15-nm-high electrodes. The pen- near the electrode edge: from the nucleation sites on the
tacene deposition time was the same for both substrates oxide channel toward the electrode sidewall and from the
and the average thickness of pentacene judged from the electrode sidewall toward the oxide channel.
AFM images is about 2 nm. Although the amount of When the electrode is thin, we observed some pen-
deposited pentacene is the same, the morphology dier- tacene islands formed at the electrode edge, covering
-1466- Journal of the Korean Physical Society, Vol. 53, No. 3, September 2008

Fig. 4. AFM image of the 10-nm-thick electrode after ap-

plying a drain-source bias voltage. The electrode is destroyed
due to electromigration.
Fig. 3. (a) and (b) are height proles along the dashed
lines, A and B, in Figures 2(a) and (b), respectively. Ch.
Pen denotes pentacene growing on the Si-oxide channel, side
pen denotes pentacene growing from the sidewall of the Au or 20-nm thick, as shown in Figure 1. Our experimental
electrode and top pen denotes pentacene growing on the top results suggest that, as long as the current is not limited
surfaces of the Au electrodes. The channel pentacene and the by the resistance of the thin electrode, an electrode just
sidewall pentacene grow in opposite directions. The higher as high as the height of the rst few pentacene layers
the electrode is, the farther from the sidewall the two pen- should be the best. If the electrode is too thin, however,
tacene growth fronts meet, as illustrated in (c) for a thin the current does not ow at all, as indicated by the case
electrode and in (d) for a thick electrode. of a 10-nm-high electrode in Figure 1. This is due to the
electrical breakdown or electromigration of the electrode.
Figure 4 is an AFM image of the broken electrode after
both the oxide channel and the electrode, so it seems applying a source-drain voltage of {20 V to the 10-nm-
that pentacene molecules can diuse up and down the high electrode to measure the drain current. Figure 1
sidewall. This, in turn, means that if the electrode is shows that the current starts to increase again when the
higher, the sidewall area is larger, so more pentacene pentacene lm became 30-nm thick for the 28-nm elec-
molecules can adsorb on the sidewall. Even regarding trode and 60-nm thick for the 93-nm electrode. We think
the convolution eect of the AFM probe tip, Figures 3(a) this is due to the Ohmic current owing between the top
and (b) show that there is more pentacene adsorbed on surfaces of the source and the drain via a pentacene lm
the higher sidewall. The delay in the start of current ow that has grown higher than the electrodes.
at the early stage of pentacene deposition in the case of
the 93-nm-high electrode is also caused by the sidewall
adsorption: It takes more time for a higher electrode for IV. SUMMARY
the pentacene molecules to ll the denuded zone to build
a current path.
Because a higher sidewall can accommodate more pen- We studied how the drain current of a bottom-contact
tacene molecules, the meeting position of the two pen- pentacene FET was aected by the heights of the source
tacene lms growing in opposite directions is aected by and the drain electrodes. The drain current vs pentacene
the height of an electrode. The higher the electrode is, thickness for dierent electrode heights measured in-situ
the farther the meeting position is located from the side- suggested that an electrode which was as high as the
wall, as illustrated in Figures 3(c) and (d). The electrical height of the rst few pentacene layers was most favor-
properties of a pentacene lm growing on the channel, able. To understand the reason, we used AFM to image
i.e., Si oxide and that growing on the electrode sidewall, the morphology of pentacene lms grown the patterned
i.e., Au, are expected to be dierent, just as the mor- substrates and observed that a pentacene-denuded zone
phologies of the two lms are dierent [13]. The drain formed in the channel along the edges of the source and
current for the 28-nm electrode being higher than that the drain electrodes. We think the denuded zone arose
for the 93-nm electrode shown in Figure 1 suggests that from the higher sticking rate of pentacene to the sidewall
one get a higher drain current when the pentacene lm of the Au electrode. If an electrode is higher, more pen-
adsorbed on the sidewall is thinner, i.e., when the elec- tacene molecules will adsorb on the Au sidewall before
trode is thinner. The drain current is known to ow only the denuded zone is lled. Our current measurement
through the rst few pentacene layers close to the chan- showing that a thinner electrode is favorable indicates
nel surface [14,15]. In our case, this is indicated by the that it is better to have fewer pentacene molecules ad-
saturation of current when the pentacene lm became 15- sorbed on the Au sidewall. If the electrode is too thin,
Eect of Electrode Thickness on Bottom-Contact


{ Chaeho Kim and D. Jeon -1467-

the current can be limited by the electrode resistance or, and Y. Aoyagi, Appl. Phys. Lett. 91, 203508 (2007).
in the worse case, the electrode can be destroyed due to [5] C.-H. Chang, C.-H. Chiena and J.-Y. Yang, Appl. Phys.
electromigration. Lett. 91, 083502 (2007).
[6] H. S. Kang, J. W. Lee, M. K. Kim, J. Joo, J. M. Ko and
J. Y. Lee, J. Appl. Phys. 100, 064508 (2006).
ACKNOWLEDGMENTS
[7] M. Ichikawa, L. You, S. Tatemichi, T. Koyama and Y.
Taniguchi, Jpn. J. Appl. Phys. 45, L1171 (2006).
[8] K. Lee and T. H. Kim, J. Korean Phys. Soc. 49, L2226
This work was supported by the Korea Science and (2006).
Engineering Foundation through the Nano Systems [9] S. H. Jee, S. H. Kim, J. H. Ko and Y. S. Yoon, J. Korean
Institute-National Core Research Center, Seoul National Phys. Soc. 49, 2034 (2006).
University. [10] N. Kawasaki, Y. Ohta, Y. Kubozono and A. Fujiwara,
Appl. Phys. Lett. 91, 123518 (2007).
[11] W.-K. Kim, K. Hong and J.-L. Lee, Appl. Phys. Lett.
89, 142117 (2006).
REFERENCES
[12] S. H. Jee, S. H. Kim, K. H. Lee, J. H. Ko and Y. S. Yoon,
J. Korean Phys. Soc. 50, 915 (2007).
[1] Y.-H. Kim, S.-M. Han, W. Lee, M.-K. Han, Y. U. Lee [13] I. Kymissis, C. D. Dimitrakopoulos and S. Pu-
and J.-I. Han, Appl. Phys. Lett. 91, 042113 (2007). rushothaman, IEEE Trans. Electron Devices 48, 1060
[2] D. Fine, L. Wang, D. Sharma and A. Dodabalapur, Appl. (2001).
Phys. Lett. 89, 203118 (2006). [14] M. Kiguchi, M. Nakayama, K. Fujiwara, K. Ueno, T.
[3] M. Yoshida, S. Uemura, S. Hoshino, N. Takada, T. Shimada and K. Saiki, Jpn. J. Appl. Phys. 42, L1408
Kodzasa and T. Kamata, Jpn. J. Appl. Phys. 44, 3715 (2003).
(2005). [15] F. Dinelli, M. Murgia, P. Levy, M. Cavallini, F. Biscarini
[4] S. D. Wang, T. Minari, T. Miyadera, K. Tsukagoshi and D. M. de Leeuw, Phys. Rev. Lett. 92, 116802 (2004).