We introduced a nucleic acid component guanine as a trap-neutralizing layer between silicon dioxide gate dielectric and a pentacene semiconducting layer to obtain increased field-effect mobility in organic field-effect transistors (OFETs). A tripling of the field-effect mobility, from 0.13 to 0.42 cm2/V s, was achieved by introducing a 2 nm guanine layer. By characterizing the surface morphology of pentacene films grown on guanine, we found that the effect of guanine layer on the topography of pentacene film was not responsible for the mobility enhancement of the OFETs. The increased field-effect mobility was mainly attributed to the hydrogen bonding capacity of otherwise unassociated guanine molecules, which enabled them to neutralize trapping sites on the silicon dioxide surface.
Organic field-effect transistors (OFETs) have been intensively researched due to their light weight, low cost, and possible applications in sensors, memory, and large-area displays.1–5 Significant improvements in charge carrier mobility have been obtained through various means, where the field-effect mobility of single-crystal OFETs has already approached 10 cm2/V s.6 The key functional layer of an OFET is the semiconducting layer, and several attempts focusing on the semiconductor have been made to achieve high mobility OFETs.7–9 However, the dielectric surface of an OFET also plays a crucial role in the device performance, since the efficient current channel lies in the first few molecular layers of the semiconductor resting on the dielectric layer.10–13
As is known, the most traditional dielectric layer for OFETs is silicon dioxide, since it possesses a well-characterized dielectric constant, high breakdown voltage, and low surface roughness.14 However, although silicon dioxide has superior insulating performance, its surface defects, which have been reported to affect the generation of the charge traps in the films, are inevitable.15–17 Hence, the surface modification of silicon dioxide is necessary to obtain high performance OFETs. Plasma cleaning, including hydrogen, oxygen, and argon plasmas, can effectively clean the surface of silicon dioxide by removing adventitious organic contamination.18 Meanwhile, surface hydroxyl (OH groups) are unavoidable in pristine silicon dioxide.19,20 Surface treatment with surface-active silanes is thus required after the plasma cleaning. Until now, the most commonly used silanes are hexamethyldisilazane (HMDS) and octadecyltrichlorosilane (OTS), which can effectively eliminate the OH hydrogen atoms on the surface of silicon dioxide.21,22
However, complex processes and long processing times are still strong limitations of silanes.23 The normally adopted deposition of the silanes typically requires a ∼12 h processing time. Although good quality HMDS can also result from spin-coating, this requires more material compared to the self-assembly process and still requires a post-baking. It is desirable to introduce an accessible material with the superior attributes of chemical abundance, biodegradability, and low cost. As the organic electronics field is becoming more and more interdisciplinary, biomaterials become natural candidates to meet these requirements.24,25 Among a variety of biomaterials, the functional base pairs in deoxyribonucleic acid (DNA) are attractive naturally occurring small molecules, due to their compatibility with thermal evaporation.26,27 Such nucleotide bases are electrically insulating and have already been employed as the dielectric layer in OFETs.28–30 Among these bases, guanine is the most stably stacked molecule, and shows a great potential as a surface modifier upon the dielectric layer.31
In this work, we report the utilization of thermally evaporated guanine as a trap-neutralizing layer on the silicon oxide dielectric. Three-fold enhancement of carrier field-effect mobility, from 0.13 to 0.42 cm2/V s, was obtained in pentacene deposited on ambient-temperature substrates. Such performance enhancement was mainly attributed to the guanine molecule exhibiting capability of hydrogen-bonding with the surface OH groups of silicon dioxide, thereby neutralizing traps.
We illustrate the structures of OFETs without and with guanine layers used in this work (Figs. 1(a) and 1(b)). Channel length (L) and width (W) were 250 μm and 1 cm, respectively. Highly n-doped ⟨100⟩ silicon wafers with 100 nm thermally grown oxide were diced into 1 in. × 1 in. substrates, cleaned with piranha solution (Caution: corrosive!), and sonicated in acetone and isopropanol for 15 min each. Guanine (Sigma Aldrich) with a series of thicknesses of 0.5, 1, 2, 3, and 8 nm was thermally evaporated at a rate of 0.1 Å/s. Since the guanine layer thickness was 1%–10% of the oxide thickness, we ignored its effect on the device capacitance. Subsequently, pentacene (Tokyo Chemical Industry Co., Ltd.) with a film thickness of 6 nm was thermally evaporated at a rate of 0.3 Å/s. Finally, 30 nm Au was thermally evaporated through a metal mask at the rate of 0.3 Å/s. The deposition chamber pressure was <2 × 10−6 Torr. The surface morphology of pentacene was investigated via atomic force microscope (AFM) (Dimension FastScan, Brucker) in the tapping mode. The electronic properties were tested using a Semiconductor Parameter Analyzer (4155C, Agilent) in the atmosphere. We also attempted x-ray diffraction characterization, but the pentacene layers were too thin to derive structural information.
Device architecture of the OFET and representation of the location of hydrogen atoms in (a) OFET without guanine layer and (b) OFET with guanine layer; (c) schematic energy-level diagrams of pentacene and guanine base pair with vacuum-level (Evac) alignment; (d) field-effect mobilities of OFETs with different thicknesses of guanine layer.
Device architecture of the OFET and representation of the location of hydrogen atoms in (a) OFET without guanine layer and (b) OFET with guanine layer; (c) schematic energy-level diagrams of pentacene and guanine base pair with vacuum-level (Evac) alignment; (d) field-effect mobilities of OFETs with different thicknesses of guanine layer.
We show energy diagrams that include the gap between the highest occupied molecular orbital and the lowest unoccupied molecular orbital (HOMO-LUMO), electron affinities, and the ionization energies of the pentacene and guanine base pair (Fig. 1(c)).29,30 The ionization energies of pentacene and guanine are 5.10 and 6.31 eV, respectively. The higher ionization energy of guanine indicates that it will not trap the holes in the pentacene film, making guanine suitable to be adopted in this work. As stated above, among the four kinds of DNA base pairs, guanine is part of the most stably stacked base pair. Meanwhile, guanine possesses the highest melting point of 360 °C and high density of 2.2 g/cm3, which increases the stability of the molecule layer.29,31 Moreover, guanine is an excellent dielectric with low dielectric losses of 10−2 at 10 mHz and high breakdown strengths between 1.5 and 3.5 MV cm−1.29 These considerations led to our choice of guanine for this work.
The field-effect mobilities of these OFETs (10 samples each) were extracted from the transfer curves in the saturation region (Fig. 1(d)). We show that the mobility increases by two-fold from 0.13 to 0.27 cm2/V s with the increase in the guanine layer thickness from 0.5 to 1 nm. Upon further increasing the thickness of the guanine to 2 nm, the maximum charge mobility of 0.42 cm2/V s is obtained, over three times higher than that of the control device. For thicknesses >3 nm, the device performance begins to deteriorate below the control level. For example, at 8 nm, the charge mobility sharply decreases to 0.004 cm2/V s.
We illustrate the typical transfer and output characteristics of OFETs of the control and devices with the guanine layer of thinnest thickness of 0.5 nm, best thickness of 2 nm, and the thickest thickness of 8 nm (Fig. 2). When 0.5 nm of guanine is deposited, the saturation current of the OFET deteriorates compared to the bare device. We assumed that this lower performance was due to the possible discontinuity of the guanine film. When the thickness of the guanine layer is increased to 2 nm, the saturation current of the OFET is significantly increased. Further increasing the thickness to 8 nm results in a decrease of the OFET performance again. Here, it is a compromise between the layer uniformity and acceptable substrate roughness. When 2 nm thick layer is formed, such passivation layer is uniform and is able to screen the energetic disorder induced by charges, impurities, and hydroxyl groups at the silicon oxide surface. However, for thicker layer, the high rugosity of the film does hinder the optimal growth of pentacene. As a result, the electrical performance becomes worse. This could also be ascribed to the higher polarity of guanine compared to silicon dioxide, which could introduce a new source of traps.31 The detailed performances of all the devices (10 samples each) are summarized in Table I.
Transfer and output characteristics of OFETs (a), (e) without guanine layer; (b), (f) with 0.5 nm; (c), (g) with 2 nm; and (d), (h) with 8 nm guanine layer.
Transfer and output characteristics of OFETs (a), (e) without guanine layer; (b), (f) with 0.5 nm; (c), (g) with 2 nm; and (d), (h) with 8 nm guanine layer.
Characteristics of the OFETs with variable thickness of the guanine layer.
| Thickness of guanine (nm) | IDS (μA) | μ (cm2/V s) | Vth (V) | SS (V/dec) | NSS × 1013 (cm−2 eV−1) |
|---|---|---|---|---|---|
| 0 nm | −86 ± 11 | 0.130 ± 0.016 | −6 ± 2 | −5 ± 0.3 | 1.99 ± 0.12 |
| 0.5 nm | −36 ± 5 | 0.080 ± 0.010 | −10 ± 1 | −4 ± 0.2 | 1.58 ± 0.08 |
| 1 nm | −164 ± 25 | 0.271 ± 0.018 | −6 ± 3 | −4 ± 0.2 | 1.58 ± 0.08 |
| 2 nm | −254 ± 41 | 0.420 ± 0.031 | −6 ± 3 | −2 ± 0.2 | 0.78 ± 0.08 |
| 3 nm | −98 ± 16 | 0.189 ± 0.011 | −6 ± 2 | −3 ± 0.3 | 1.18 ± 0.12 |
| 8 nm | −2.2 ± 0.5 | 0.004 ± 0.001 | −10 ± 1 | −4 ± 0.2 | 1.58 ± 0.08 |
| Thickness of guanine (nm) | IDS (μA) | μ (cm2/V s) | Vth (V) | SS (V/dec) | NSS × 1013 (cm−2 eV−1) |
|---|---|---|---|---|---|
| 0 nm | −86 ± 11 | 0.130 ± 0.016 | −6 ± 2 | −5 ± 0.3 | 1.99 ± 0.12 |
| 0.5 nm | −36 ± 5 | 0.080 ± 0.010 | −10 ± 1 | −4 ± 0.2 | 1.58 ± 0.08 |
| 1 nm | −164 ± 25 | 0.271 ± 0.018 | −6 ± 3 | −4 ± 0.2 | 1.58 ± 0.08 |
| 2 nm | −254 ± 41 | 0.420 ± 0.031 | −6 ± 3 | −2 ± 0.2 | 0.78 ± 0.08 |
| 3 nm | −98 ± 16 | 0.189 ± 0.011 | −6 ± 2 | −3 ± 0.3 | 1.18 ± 0.12 |
| 8 nm | −2.2 ± 0.5 | 0.004 ± 0.001 | −10 ± 1 | −4 ± 0.2 | 1.58 ± 0.08 |
To further scrutinize the interface between the guanine and the pentacene, we characterized the surface topography by AFM analysis. The relatively high value of the Z-scale compared to the film thickness is due to the inevitable dust and the bottom/top-peak of the film. The thickness of the pentacene films in Figs. 3(a)–3(c) is 6 nm, which is identical to that of the OFET device. We show that the original pentacene film displays polycrystalline microstructure characterized by several highly textured dendritic grains in the AFM image (Fig. 3(a)).32 When the pentacene is grown on the 2 nm guanine layer, the surface morphology of the pentacene film stays almost the same as the pristine pentacene film (Fig. 3(b)). This indicated that the reason for the performance enhancement of the device with the 2 nm guanine layer did not result from the topography improvement of the pentacene layer. In order to further confirm this assumption, the topographies of the pentacene grown on the 0.5 nm guanine layer as well as the morphologies of the 2 nm and 0.5 nm guanine layers were characterized (Figs. 3(c)–3(e)). We show that the topography of the pentacene film grown on the 0.5 nm guanine is slightly changed to smaller crystals (Fig. 3(c)). This further confirmed the assumption that the guanine layer did not have a dramatic effect on the topography of the pentacene film. Meanwhile, the 2 nm guanine layer exhibits a uniform morphology (Fig. 3(d)), whereas the 0.5 nm guanine shows an island structure (Fig. 3(e)). This observation was consistent with the inferior performance of the OFET with a 0.5 nm guanine layer.
Tapping mode AFM images in 5 μm × 5 μm of (a) pentacene film, (b) pentacene film on 2 nm guanine layer, (c) pentacene film on 0.5 nm guanine layer, (d) 2 nm guanine layer, and (e) 0.5 nm guanine layer. (f) Schematic presentation of the molecular structure of guanine and cytosine pair, dashed lines indicate the hydrogen bonds.
Tapping mode AFM images in 5 μm × 5 μm of (a) pentacene film, (b) pentacene film on 2 nm guanine layer, (c) pentacene film on 0.5 nm guanine layer, (d) 2 nm guanine layer, and (e) 0.5 nm guanine layer. (f) Schematic presentation of the molecular structure of guanine and cytosine pair, dashed lines indicate the hydrogen bonds.
As the morphology is not significantly affected by introducing the guanine layer, the interface between pentacene and guanine, compared to pentacene and silicon dioxide, should play an important role in the performance discrepancy. As stated above, surface OH groups are the common defects in silicon dioxide dielectrics. Active hydrogen is inevitably introduced into the SiO2 films during the silicon oxidation process, and fabrication of overlayer films, such as polycrystalline silicon. These hydrogen atoms are mostly found to form Si-OH and Si-H bonds.15,33 This results in a high density of interface traps, which makes the device performance inferior.34–36 In nature, guanine binds to cytosine in a base pair through hydrogen bonding as shown in Fig. 3(f), where some of the hydrogen and oxygen bridging atoms of guanine need to be paired with other atoms of the cytosine molecule.30 When guanine is introduced into the OFET device, the hydrogen bonding capability leads to the interaction with the silicon surface. In addition, guanine is a basic material in its chemical nature and hydrogen atoms usually deliver a positive charge as H+, so guanine molecules may then easily capture H+ due to this basic property.37,38 Therefore, the introduced guanine layer tends to neutralize hydrogen ions in the device channel and silicon dioxide as shown in Fig. 1(b), which leads to the observed device performance enhancement. However, pentacene is mainly a hole conductor, which could also be trapped with the H+ neutralization process. The rationalization of the performance enhancement with the hole capture ability of guanine can be ascribed to two aspects: (1) from Fig. 3(e), the cluster-like morphology shows that there is also a molecule-molecule affinity of guanine. This affinity could simultaneously inhibit the trap neutralization and the hole capture properties of guanine, which helps guanine to obtain a best compromise; (2) the thickness of the guanine layer is also crucial to the device performance, the optimization of which can enable trap cancellation while most holes in the pentacene are maintained.
In order to confirm the trap-neutralization ability of guanine, we calculated the trap density (NSS) at the interface between the organic semiconducting layer and the dielectric layer. The NSS is proportional to subthreshold slope (SS) as shown in Eq. (1)39
where e is the natural constant, kB is Boltzmann's constant, T is the absolute temperature, and q is the electronic charge. As summarized in Table I, the calculated NSS of OFET without the guanine proton trapping layer is 1.99 × 1013 cm−2 eV−1. NSS gets reduced when a critical guanine layer thickness is introduced. The smallest NSS obtained is 0.78 × 1013 cm−2 eV−1 with a 2 nm guanine layer applied, which indicates that the trapping density is effectively reduced by the guanine layer.
In order to further observe the effect of guanine on the surface trap density, we characterized the OFET with and without a 2 nm guanine layer under multiple gate voltage sweeps. From Fig. 4, the saturation current of the OFET with guanine is much higher than that of the OFET without guanine, which is in consistent with our results. Meanwhile, the saturation current of the OFET without the guanine layer increases monotonically (Fig. 4(a)). After five cycles of paired sweeps, the saturation current exhibits a strong increase of 159%. The increase of the current is due to the filling of carrier traps at the dielectric surface after the sweep. On the contrary, we show that when 2 nm guanine is introduced, the OFET presents a much more stable performance of only 42% increase after five sweep cycles (Fig. 4(b)). This indicated that the density of interface trapping sites significantly decreased with the guanine layer.
Transfer characteristics of OFETs (a) without and (b) with 2 nm guanine layer under multiple gate voltage sweeps of 5 times (VDS = −40 V).
Transfer characteristics of OFETs (a) without and (b) with 2 nm guanine layer under multiple gate voltage sweeps of 5 times (VDS = −40 V).
On the other hand, we analyzed the hysteresis of OFETs as shown in Fig. 4. We note that the hysteresis is clockwise, which means the off-to-on sweep current is higher than the on-to-off sweep current. This phenomenon is common in various OFETs which can be explained two ways.40 If the traps at the pentacene/dielectric are electron-capture dominated, when sweeping rapidly from off to on, the negatively charged traps could induce excess holes causing higher IDS in the forward sweep. If the traps are hole-capture dominated, the hole traps that fill fast and empty slowly can also result in lower back sweep current. No matter whether the traps are electron or hole-dominated, they are located close to the channel since the higher back sweep current hysteresis is usually caused by mobile ions in the dielectric or by ferroelectric polarization of the dielectric. After five sweep cycles, the hysteresis voltage of the OFET without the guanine layer is about −3 V, whereas that of the OFET with the guanine layer is almost 0 V. This further confirmed the trap neutralizing ability of the guanine layer in the OFET device.
In conclusion, we demonstrated a higher mobility OFET by depositing the DNA base guanine on the silicon dioxide gate dielectric as a trap-neutralizing layer. We showed that the field-effect mobility of OFET could be increased with the guanine layer addition from 0.13 cm2/V s to 0.42 cm2/V s. Through introducing the guanine layer, the trap density was significantly reduced, which resulted in the enhancement of the charge field-effect mobility. This work represents an early introduction of a simple biomaterial into an organic electronic device.
This research was funded by the National Natural Science Foundation of China (NSFC) (Grant No. 61675041), the Fundamental Research Funds for the Central Universities (Grant No. ZYGX2010Z004), the Foundation of Innovation Groups of NSFC (Grant No. 61421002), the China Scholarship Council (File No. 201506070085 and No. 201506070069), and Science and Technology Department of Sichuan Province (Grant No. 2016HH0027). We acknowledged the Environmental Protection Agency SEARCH grant for funding at Johns Hopkins University. The Yale Institute for Nanoscience and Quantum Engineering (YINQE) and NSF MRSEC DMR 1119826 (CRISP) provided facility support. We also acknowledged Yuhang Yang in our group for helping us with some experiment preparations.
