Chapter 4 Results and Discussions
4.3 The Relationship of the Mobility to the Grains
4.3-1 The Transfer Characteristics without V
TShift
-60 -40 -20 0 20
Figure 4. 5 The transfer characteristics of the OTFTs, which were fabricated at various deposition temperatures with (a) HMDS-treated substrate, (b) PαMS-treated substrate, obtained under VD=-60V.
Compared with Figure 4. 2, Figure 4. 5 illustrates that the VT shift is not clear, and the devices turn on at negative bias, not at positive bias as before. The VT shift as shown in Figure 4. 5 is less than 10V, which would solve a basal problem in estimating mobility. It is a good beginning to study the relationship of mobility to grains in the next sections.
But, why the surface treatments work?
It is notice that the contact angle of bare substrates drops dramatically with decreasing temperature. In contrast, however, the contact angles of the two treated substrates are nearly identical. This is likely the reason that VT shift could be wiped off by surface treatments.
Note that the turn on voltage of PαMS-treated devices center on -5V except for the one prepared at 90℃. It has turned on at a more negative bias than the others. This obvious shift might be due to the glass transition of PαMS, which should be regarded as another modified material.
On the other hand, the devices with surface treatments have turned on at negative bias, whereas the bare ones have actuated at positive bias. This part will be discussed in the section 4.5.
Surface Treatment Substrate Temperature (oC) Contact Angle (H2O)
20 49.8 Bare
90 60.0 20 67.5 HMDS
90 67.0 20 84.0 PαMS
80 82.8
Table 4. 1 Contact angles measured on 3 kinds of substrates at various temperatures.
4.3-2 Inhibition of Phase Transition
0.5Å/sec on a (top) HMDS-treated substrate, (bottom) PαMS-treated substrate at various deposition temperatures.
6 XRD spectra of 60-nm-thick pentacene deposited at a fixed flux
30oC
Coherence of Phases
he rectangle of Figure 4. 6, the crystal progressively evolves into
Molecular Orientation As indicated by t
bulk phase (00l) on the HMDS-treated substrates, whereas it maintains thin-film phase (00l’) with the higher substrate temperature on PαMS-treated substrates. It is believed that the phase evolution is related to the strain relaxation of initially strained thin-film phase. This figure proves that the phase transition has been repressed by the PαMS treatment, probably due to the strain release by polymer chains.
, the pentacene crystal on the treated substrates demonstrates a str
should make the Ehrlich-Schwoebel barri
Intensity
In the XRD patterns
ong Bragg reflection at 19.2°, which is absent in the case of bare substrates. This peak is estimated to be an unresolved doublet arising from (110) and (111) reflections with a d-spacing of ~2.5Å [29]. The appearance of (110) and (111) reflections infer that it exists flat-lying molecules on the substrate surface which is likely to make better adhesion between inorganic/organic interface.
Additionally, the flat-lying molecules
er lower. If the molecules deposit without E-S effect, the film growth is close to layer-by-layer mechanism.
e HMDS-treated patterns, the intensity becomes stronger with increasing temp
For th
erature, which is similar to the untreated ones. For the PαMS-treated patterns, however, it is an opposite case. The film deposited at RT shows better reflections than the ones deposited through substrate heating.
4.3-3 Morphology Evolution
0.5Å/sec on a 200-nm-thick SiO2 substrate with HMDS treatment at various
. (B) T =30℃, R=9.630nm (C) T =50℃, R=12.496nm.
) T =70℃, R=9.325nm. (E)T =90℃, R=8.788nm.
Figure 4. 7 AFM images of 60-nm-thick pentacene deposited at a fixed flux rate
=
deposition temperature.
(A)T =17℃, R=10.672nm (D
g 2 e o which are 5×5μm2.
All the ima es are 3×3μm except th gray (bigger) tw (R: roughness presented by root-mean-square form)
Figure 4. 8 AFM images of 60-nm-thick pentacene deposited at a fixed flux rate
=0.5Å/sec on a 200-nm-thick SiO2 substrate with PαMS treatment at various
. (B) T =30℃, R=8.970nm. (C) T =50℃, R=10.102nm.
) T =70℃, R=9.071nm. (E) T =90℃, R=10.211nm.
Evolution of Grain Size deposition temperature.
(A)T =17℃, R=9.027nm (D
mages of 60-nm-thick pentacene with HMDS-treated substrates, while
ference in grain gaps between
ure favors pentacene to diffuse on the Figure 4. 7 is the first group of AFM i
Figure 4. 8 is the second group with PαMS-treated ones. Similar to bare substrates, HMDS-treated samples also display an obvious enlargement with deposition temperature. But for the PαMS-treated samples, the grain sizes grow in the beginning, slow down at 70℃, and finally become smaller at 90℃.
It also shows a pronounced morphological dif Figure 4.
8(E) and the others.
In common cases, higher substrate temperat
surfa
volution of Grain Shape
ce and to nucleate on fewer sites. But, a large grain should be formed by lower rate of nucleation and higher rate of lateral growth. The glass transition temperature (Tg) at atmosphere of our PαMS is 85℃, but it should be lower than 85℃ under high vacuum. Above the Tg point, molecule chains could obtain more energy which would raise the nucleus number, and diminish grain size. Accordingly, this morphological difference probably results from the violent molecular motion of PαMS at 90℃.
E
Apart from the grain size, it is noticeable that the dendrite-shaped grains disappear at the medium temperature on the PαMS-treated substrates. It also disappears at the high temperature on the bare substrates. The dendrites are independent of the crystal phase, but dependent on deposition temperature.
The model of Diffusion-Limited Aggregation (DLA) is widely used to explain the g
law of heat conduction (Fourier’s law),
rowth of dendrites [30]. DLA is the process whereby particles undergo a random walk (Brownian motion) to push the molecules approach the growing clusters. Once the molecules stick to the cluster, the transient diffusion at the step edges occurs immediately.
From the
T k q=− ∇ ,
the direction of heat flow depends on the temperature gradient at the interface. When the films grow into a supercooled liquid, the interface is unstable. Suppose a small protrusion forms at the interface, the negative temperature gradient in the liquid becomes more negative. Therefore, heat is removed more effectively from the protrusion than from the surrounding regions. That will allow the arms to grow preferentially, and hinder the neighboring arms from forming [19] [31].
Review 4.1 The temperature distribution for the dendrite growth.
[From Chalmers, B., trans. AIME, 200 519 (1954)] [19]
To return to our system, when the pentacene molecules freeze, the latent heat will be released at the vapor/solid interface. If the latent heat is not removed immediately, it will result in the temperature distribution such as Review 4.1. The dendrites are controlled by the rate at which the latent heat can be removed from the vapor/solid interface, and if the vapor temperature reach a critical supercooling. That is why the dendrites disappear at the relatively high deposition temperature, which is influenced by the thermal property of substrates simultaneously.
Evolution of Initial Layers
While the TFT turns on, the carriers concentrate in the initial several layers to form the channel [27] [32]. From the Figure 4. 7 and Figure 4. 8, surface treatment seems to affect the morphology even with a thickness far beyond the first monolayer. Thus, the underlying morphology might deviate a lot from the appearance. It is more significant to see the 8nm-thick (~5ML) pentacene films rather than 60nm-thick ones.
Figure 4. 9 AFM images of 8-nm-thick pentacene deposited on a 200-nm-thick SiO2
substrate without surface treatment. All the images are 3×3μm2.
Figure 4. 10 AFM images of 8-nm-thick pentacene deposited on a 200-nm-thick SiO2
substrate with HMDS treatment. All the images are 3×3μm2.
Figure 4. 11 AFM images of 8-nm-thick pentacene deposited on a 200-nm-thick SiO2
substrate with PαMS treatment. All the images are 3×3μm2.
Compare Figure 4. 9 with Figure 4. 1, we will see the evolution of underlying grains deposited on the bare substrate is analogous to the case of outlying ones. New layers nucleate on the already existing DLA-like monolayer before the latter coalesce
to a complete epitaxial layer. It is reported that the quasi-epitaxial growth of pentacene molecules is disrupted by the upward diffusion bias generated by the Ehrlich-Schwoebel barrier, leading to mound growth [20].
If the Ehrlich-Schwoebel barrier cannot be neglected, there will be an adatom
“uphill current.” Furthermore, as the uncovered areas are much smaller than the islands, shadowing effect will pronounce. They will cause fast upward growth, bringing deep crevices observed, not blurred boundary.
For the PαMS-treated substrates in contrast, it does not agree with the above finding. Unlike Figure 4. 8, the grain boundary is too blurred to recognize how many grains is (Figure 4. 11). It is divergent from the 60-nm-thick appearance.
But why the E-S barrier seems to be trivial in the PαMS-treated system?
The effect of the Ehrlich-Schwoebel barrier might be either strengthened or mitigated by the deposition temperature [20], and surface treatments. By elevating temperature during deposition, the molecules will get more energy so that their probability of jumping down a step edge would be promoted. On the other hand, surface treatments can make the molecules “lie down,” which might change the Ehrlich-Schwoebel barrier. These two factors might help the grains to form layer-by-layer structure. But which one is dominant needs advanced study.
According to both patterns of XRD (Figure 4. 6), the flat-lying pentacene occurs at RT. Moreover, the more blurred boundary, exhibiting lower Ehrlich-Schwoebel barrier of the film, is also shown at low temperature (Figure 4. 11). Consequently, it is deduced that the E-S barrier of the film with flat-lying pentacene is lower than that with oblique-standing one. The film growing on the treated substrate is probably near to the layer-by-layer mechanism, which fills in the deep crevices. To reduce the boundary barrier, layer-by-layer growth is better than the mound growth.
Figure 4. 12 Schematic view of the structure of deposited pentacene on (Left) bare oxide substrates with strong effect of Ehrlich-Schwoebel barrier, exhibiting deep crevices, (Right) surface modified substrates, showing unclear boundaries.
4.3-4 The Comparison of Mobility
10 30 50 70 90
0.2 0.4 0.6 0.8
Temperature(oC)
mobility (cm
2 /Vsec)
PαMS-treated HMDS-treated
Tg of PαMS
Figure 4. 13 The results of the mobility calculation from the “saturation regime” of the modeling of field-effect transistor.
The result of mobility calculation (Figure 4. 13) presents the relationship between the mobility and substrate temperature. For the HMDS-treated devices, the mobility improves with increasing temperature from 0.21 to 0.33 cm2/Vsec, but
reaches a saturation performance at 90℃. In contrast, the mobility of the PαMS-treated device at 70℃ can be as high as 0.72 cm2/Vsec, which is much better than that of untreated or HMDS-treated ones, while it drops to 0.1cm2/Vsec at 90℃. It is believed that the grain and interfacial boundary may reduce the carrier mobility.
Therefore, the higher substrate temperature, resulting in the larger grains, makes better performance.
However, for the two treated devices, the highest mobility occurs while the grain size is intermediate (at 70℃). Large grains, which mean less density of grain boundary, will not help the charge transport. It is suspected that more traps maybe induced at higher substrate temperature [25]. In our system, boundary traps or bulk boundary may result from the phase transition of HMDS, while interfacial traps may cause from the rubbery state of PαMS. Additionally, Figure 4. 8(E) displays deeper grain gaps which may bring high density of grain boundary. These two negative factors hold the mobility of PαMS-treated device back.
1 3 5 7 9
0.1 0.3 0.5 0.7
Mobility(cm2 /Vsec)
Grain Boundary density(/
μm)
HMDS-treated PαMS-treated
90o
C, with serious phase transition 90o
C, above Tg of PαMS
Figure 4. 14 The relationship of mobility to grain boundary density.
The AFM images of 60nm-thick films mentioned in 4.3-4 were dealt by the program1 appendixed at the final of this thesis.
People have already known that the aversion of grain boundaries to the transport properties. Except for the serious phase transition of HMDS-treated device, and the rubber state of PαMS-treated device, the mobility is proportional to the grain boundary density.
Externally, the transport path of PαMS-treated device can bear to meet 10 boundaries per μm, while the case of PαMS-treated one can only bear 6 per μm. It is probably due to the blurred boundary in the initial layer, resulting in the particular better performance.