Article pubs.acs.org/Langmuir
Dose-Controlled, Floating Evaporative Self-assembly and Alignment of Semiconducting Carbon Nanotubes from Organic Solvents Yongho Joo, Gerald J. Brady, Michael S. Arnold, and Padma Gopalan* Department of Materials Science and Engineering, University of Wisconsin, Madison, Wisconsin 53706, United States S Supporting Information *
ABSTRACT: Arrays of aligned semiconducting single-walled carbon nanotubes (s-SWCNTs) with exceptional electronic-type purity were deposited at high deposition velocity of 5 mm min−1 by a novel “dose-controlled, floating evaporative self-assembly” process with excellent control over the placement of stripes and quantity of s-SWCNTs deposited. This approach uses the diffusion of organic solvent on the water−air interface to deposit aligned s-SWCNT (99.9%) tubes on a partially submerged hydrophobic substrate, which is withdrawn vertically from the surface of water. By decoupling the s-SWCNT stripe formation from the evaporation of the bulk solution and by iteratively applying the s-SWCNTs in controlled “doses”, we show through polarized Raman studies that the s-SWCNTs are aligned within ±14°, are packed at a density of ∼50 s-SWCNTs μm−1, and constitute primarily a well-ordered monodispersed layer. The resulting field-effect transistor devices show high performance with a mobility of 38 cm2 V−1 s−1 and on/off ratio of 2.2 × 106 at 9 μm channel length.
■
INTRODUCTION Single-walled carbon nanotubes (SWCNTs) have been one of the key building blocks for nanoscale science and technology over the past 2 decades because of their interesting physical and chemical properties. SWCNTs are particularly promising for high-speed and low-power semiconductor electronics.1,2 A challenge, however, is the hierarchical organization of these building blocks into organized assemblies and, ultimately, useful devices. Ordered structures are necessary as random network SWCNTs thin films result in suboptimal electronic properties including reduced channel conductance and mobility. Numerous techniques for aligning SWCNTs have been explored to solve this shortcoming and achieve higher conductance and mobility.3−6 These approaches can be divided into two main categories: (a) direct growth via chemical vapor deposition and arc discharge and (b) postsynthetic assembly. In the case of direct growth, both metallic and semiconducting SWCNTs are produced. In this case, the performance of SWCNT field-effect transistors (FETs) is limited by metallic SWCNTs (mSWCNTs), thus motivating attempts to purify semiconducting SWCNT (s-SWCNT) samples with homogeneous electronic properties. A variety of postsynthetic sorting methods have been developed to separate m- and s-SWCNTs according to their specific physical and electronic structure, which are usually implemented in aqueous or organic solutions.7 To take © 2014 American Chemical Society
advantage of the high purity of s-SWCNTs that can be produced by these solution-based sorting approaches in semiconductor electronic devices, solution-based methods for assembling and aligning s-SWCNTs such as evaporation-driven self-assembly,5,8 blown-bubble assembly,6 gas flow selfassembly,9 spin coating,10 Langmuir−Blodgett and −Shafer methods,11,12 contact-printing assembly,13 and alternating current (ac) electrophoresis14 have been developed. While each of these methods has its strengths, new methods are still needed to improve the fidelity of s-SWCNT assembly and alignment. Specifically, better control is needed on the wafer scale, of the placement of s-SWCNTs, the type and functionalization of SWCNTs deposited, and their degree of alignment, density, and purity, to enable the fabrication of practical s-SWCNT electronic devices. Here, we report a novel yet simple method to deposit aligned s-SWCNTs on substrates via dose-controlled, floating evaporative self-assembly (DFES). One unique advantage of this method is that it allows for the deposition of s-SWCNTs with exceptional electronic-type purity that are prepared using conjugated polymers as semiconductor-selective agents in organic solvents. Received: January 13, 2014 Revised: February 27, 2014 Published: March 2, 2014 3460
dx.doi.org/10.1021/la500162x | Langmuir 2014, 30, 3460−3466
Langmuir
Article
Figure 1. Schematic illustration of the iterative process used to fabricate aligned nanotubes-SWCNTs driven by the spreading and evaporation of controlled doses of organic solvent at the air/water interface.
octylfluorenyl-2,7-diyl)] (PFO). PFO-BPy and PFO were chosen as the selective polymers as each demonstrates optimized semiconductor selectivity for wrapping arc-discharge s-SWCNTs. Arc Discharge. The mixtures of arc-discharge SWCNT powders (2 mg mL−1) and PFO-BPy (American Dye Source, Inc., Baie D’Urfé, Quebec, Canada; #AD153-UV, 2 mg mL−1) were sonicated (Fisher Scientific, Waltham, MA; Sonic dismembrator 500) for 30 min in toluene (30 mL). The solution was centrifuged in a swing bucket rotor at 50000 g for 5 min, and again at 50000 for 1 h. The supernatant was collected and filtered through a syringe filter. A distillation removed toluene over a 30 min duration. The resulting residue of PFO-BPy and s-SWCNTs was redispersed in tetrahydrofuran (THF). The s-SWCNT solution in THF was centrifuged at 15 °C for 12 h. The supernatant (containing excess PFO-BPy) was discarded and the pellet was redispersed into THF. After removal of the THF, the residue was dispersed in chloroform to a concentration of 10 μg mL−1. Solutions were bath-sonicated until immediately before use for DFES to prevent bundling and aggregation of SWCNTs. HiPco. The initial dispersion of HiPco SWCNTs were prepared using 2 mg mL−1 of HiPco powder and 2 mg mL−1 of PFO (American Dye Source, #AD329-BE) in toluene. The same sonication, centrifugation, and distillation procedures as the arc-discharge SWCNTs were used for the dispersion of sSWCNTs, separation of unwanted material, and removal of excess polymer. Langmuir−Blodgett (LB) Trough and Substrate. DFES was conducted in a LB trough (KSV NIMA Medium size KN 2002) without compression of the LB barriers, to improve vibrational isolation and to accurately control the substrate withdrawal speed. The s-SWCNTs were spread at 23 °C and a Wilhelmy balance (Platinum plate) was used. Milli Q water (resistively ca. 18.2 MΩ-cm) was used as the water subphase. The Si/SiO2 substrates were cleaned with a Piranha solution (H2O2 (33%)/concentrated H2SO4 (67%)) for 20 min and rinsed with deionized water. After Piranha treatment, the substrates were exposed to hexamethyldisilizane (HMDS) vapors to create a hydrophobic self-assembling monolayer (vapor deposition). DFES Process. Figure 1 shows a schematic illustration of our method. A 2 μL dose of >99.9% purified arc-discharge sSWCNT ink in chloroform (concentration = 10 μg mL−1) was dropped by a glass syringe (50 μL, Model 1705 AD SYR) onto the water surface 0.5 cm distance away from the vertically oriented substrate, which was withdrawn with a certain speed controlled by a LB dipping mechanism (vertical dipping clamp) as shown in Figure 1i.16 The dose covered the water surface by spreading at the air/water interface and reached the substrate quickly as a result of surface tension effects (Figure 1ii). This
Unlike s-SWCNTs sorted using anionic surfactants in aqueous solution, conjugated polymers like polyfluorene derivatives are advantageous because they sensitively and selectively “pick out” semiconducting species directly during dispersion from raw sSWCNT powders, thereby avoiding the need for subsequent postdispersion sorting. The dose-controlled, floating evaporative self-assembly method is a substantial evolution over standard evaporation-driven self-assembly, which has been used to deposit surfactant-encapsulated s-SWCNTs from aqueous solution. In the standard approach, s-SWCNTs are uniformly dispersed in aqueous solution and the slow evaporation of the solution leads to stripes of aligned s-SWCNTs on a hydrophilic receiving substrate. As the solution evaporates, the air/water interface moves along the substrate according to a “slip-stick” motion, which defines where the stripes are deposited. In contrast, in dose-controlled, floating evaporative self-assembly, we gain complete control over where the stripes are deposited and the quantity of s-SWCNTs deposited. These advantages are achieved by decoupling the s-SWCNT stripe formation from the evaporation of the bulk solution and by iteratively applying the s-SWCNTs in controlled “doses”. In more detail, in the technique a receiving substrate is withdrawn from a water trough with control over position and velocity. Single doses of s-SWCNT “organic inks” are delivered to the substrates in controlled aliquots of ∼2 μL droplets. Each aliquot of ink was dropped onto the trough surface in the vicinity of the substrate which then spreads across the surface to the substrate to nearly instantaneously form an aligned sSWCNT stripe that spans the entire width of the substrate. The width and density of the s-SWCNTs within each stripe are controlled by the velocity of the substrate. Overall, the result is the rapid deposition of s-SWCNTs into aligned arrays that can be adapted for scalable integration into a wide variety of microelectronic applications. Here, we demonstrate the method; characterize the degree of alignment and s-SWCNT density using polarized Raman spectroscopy, atomic force microscopy (AFM), and scanning electron microscopy (SEM); and demonstrate the potential of the aligned s-SWCNT stripes in FETs for high-performance microelectronics.
■
EXPERIMENTAL SECTION Preparation of Semiconducting SWCNTs. Two different types of s-SWCNT inks were examined. The first type of ink was processed from arc-discharge SWCNT powders (NanoLab, Inc., Waltham, MA). In this case, we employed the polyfluorene derivative poly[(9,9-dioctylfluorenyl-2,7-diyl)-alt-co-(6,6′-{2,2′bipyridine})] (PFO-BPy), which has been shown to selectively wrap highly semiconducting SWCNT species.15 The second type of ink was processed from high-pressure carbon monoxide (HiPco) produced powers (NanoIntegris Inc., Skokie, IL). In this case we used the polyfluorene derivative poly[(9,9-di-n3461
dx.doi.org/10.1021/la500162x | Langmuir 2014, 30, 3460−3466
Langmuir
Article
Figure 2. (a) Absorption spectra of PFO-BPy sorted and unsorted SWCNT solutions. The unsorted spectra contain peaks from metallic SWCNTs at the M11 transition. S22 and S33 peaks are the second and third energy transitions of semiconducting SWCNTs that are broadened due to an overlap of a wide diameter distribution. The M11 peaks were absent in the sorted spectra because of significant removal of metallic impurities. (b) Absorbance spectra after removal of polymer by repeated washing cycles. PFO-BPy wrapped s-SWCNTs were dispersed in tetrahydrofuran and sedimented repeatedly to wash away the excess polymer.
poly[(9,9-dioctylfluorenyl-2,7-diyl)-alt-co-(6,6′-{2,2′-bipyridine})] (PFO-BPy), which has been shown to selectively wrap highly semiconducting SWCNT species with large diameters present in arc-discharge soot ranging from 1.3 to 1.7 nm.17 Corresponding solution spectra of the sorted and unsorted SWCNT solutions are shown in Figure 2a for comparison. We measured the concentration of the SWCNTs in toluene prior to and following sonication to calculate a yield of ∼0.6%. It should be noted that the primary focus of the study was to achieve the highest purity s-SWCNTs possible. Thus, the sonication and centrifugation steps were optimized to separate as much of the impurities as possible, resulting in yields that are lower than what has previously been achieved using the same SWCNT and polyfluorene starting materials.17 The metallic peaks present in the unsorted spectra around 700 nm were absent after sorting with PFO-BPy. Following the initial sorting process, excess polymer chains were removed by repeated dispersion and centrifugation of the SWCNTs in tetrahydrofuran (Figure 2b). The second type of ink was processed from high-pressure carbon monoxide (HiPco) produced powers (NanoIntegris Inc.). In this case we used the polyfluorene derivative poly[(9,9di-n-octylfluorenyl-2,7-diyl)] (PFO), with improved selectivity of small SWCNT diameters ranging from 0.8 to 1.1 nm.15 The difference in SWCNT selectivity seen here is due to the difference in conjugation of PFO-BPy in comparison to PFO. More specifically, PFO-BPy has a stronger affinity for largediameter semiconducting SWCNTs contained within the assynthesized arc-discharge starting material with little to no preference toward specific chirality types.15 This is contrary to PFO, which has a stronger affinity for the smaller diameter near armchair SWCNTs contained within HiPCO starting material.15 In both cases we observed a significant preference for semiconducting over metallic electronic-type SWCNTs in the absorption spectra. Characterization of DFES-Deposited SWCNT Stripes. In previous studies on evaporative self-assembly from aqueous solutions of SWCNTs low pressure was required to speed up the evaporation of water and hence the assembly process. Contrarily, use of high vapor pressure organic solvents allowed us to do rapid assembly under ambient conditions. For example, whereas here we successfully demonstrate a deposition velocity of 5 mm min−1, the deposition velocity using standard evaporative self-assembly from aqueous solution is much slower, only 0.02 and 0.001 mm min−1 at 70 and 760 Torr, respectively.5 As the organic ink spreads, it comes in
phenomenon is generally used in Langmuir−Blodgett techniques to spread amphiphilic molecules dissolved in organic solvent at the air/water interface. Continuous stripes were deposited on the vertically oriented substrate with repeated doses of the ink on the water at the same spot while withdrawing the substrate at a set speed (Figure 1iii). These stripes tend to have regular width and orientation as shown in Figure 1iv. We chose chloroform as a suitable solvent for this study as it spreads and evaporates rapidly across the water surface (determined by measurements of solvent evaporation rates in Figure S1 of the Supporting Information). Imaging. SEM images were collected with an LEO-1530 field-emission scanning electron microscope. The surface morphology of the s-SWCNTs was imaged using a Nanoscope III Multimode atomic force microscope (Digital Instruments). Tapping mode was utilized for the AFM measurement. A triangular cantilever with an integral pyramidal Si3N4 tip was used. The typical imaging force was of the order of 10−9 N. Raman Spectroscopy Characterization. Raman characterization was measured in a confocal Raman microscope with laser excitation wavelength of 532 nm (Aramis Horiba Jobin Yvon Confocal Raman Microscope.). The device was equipped with a linear polarizing filter between the sample and the incident beam laser to allow polarization-dependent measurements. FET Fabrication. First, stripes of arc-discharge s-SWCNTs were deposited on a highly doped Si substrate with a 90 nm thermally grown SiO2, which served as the backgate electrode and dielectric, respectively. Electron beam lithography was then used to pattern the stripes so that they had well-defined widths of 4 μm. Samples were then annealed in a mixture of ≥99.99% Ar (95%): H2 (5%) to partially degrade the PFO-BPy polymer, followed by annealing in vacuum at 1 × 10−7 Torr and 400 °C for 20 min. A second electron beam lithography step was used to define the top-contacted electrodes. Thermal deposition of Pd (40 nm) was used to create source and drain contacts to the s-SWCNT stripe. Finally, the devices were annealed in an argon atmosphere at 225 °C.
■
RESULTS AND DISCUSSION Characterization of SWCNT Solutions. Two different types of SWCNTs inks were explored to investigate the SWCNT diameter dependence of DFES. The first type of ink was processed from arc-discharge SWCNT powders (NanoLab, Inc.). In this case, we employed the polyfluorene derivative 3462
dx.doi.org/10.1021/la500162x | Langmuir 2014, 30, 3460−3466
Langmuir
Article
Figure 3. Microscopic characterizations of s-SWCNT stripes. SEM images of aligned s-SWCNTs as a function of the substrate elevation velocity at 1 (a), 5 (b), and 9 mm min−1 (c). The line width (LW) was controlled by the substrate elevation velocity. Optical microscope image (d) of stripes of sSWCNTs, along with a high-resolution SEM image (e), and AFM images (f), (g).
contact with the partially submerged substrate. Subsequent rapid evaporation of the chloroform (Figures 1ii and S2iii) results in the formation of a stripe of aligned s-SWCNTs on the vertically submerged substrate. As the solvent level rapidly decreases during evaporation, the s-SWCNTs tend to orient perpendicularly to the evaporation front, which is sterically more favored. It is also likely that there is less of a steric penalty for the s-SWCNTs to turn sideways (shown as Figure S2ii), rather than for them to stand up out of the solvent into the air while the solvent front recedes (Figure S2i).18,19 Through the DFES process we can form continuous stripes of aligned sSWCNTs (Figure 1iii,iv) control three pivotal factors: (1) the width of the stripes, (2) the density of SWCNTs within each stripe, and (3) the spacing between the stripes. We have demonstrated control over (1) the width of the stripes and (2) density of the stripes by varying the substrate elevation velocity (controlled by vertically withdrawing the substrate using a LB dipping apparatus) in Figure 3a−c. For a dose concentration of 10 μg mL−1, at a high velocity of 9 mm min−1 the SWCNTs became randomly disordered while at 1 mm min−1 they began to aggregate into large bundles or ropes. At an optimized velocity of 5 mm min−1, the s-SWCNTs in the resulting stripes were well isolated from one another and well aligned. Figure 3d shows an optical micrograph of aligned sSWCNT stripes with widths of 20 (±2.5) μm fabricated under these optimized conditions. By optimizing the elevation velocity, we were able to dictate the width of the stripes. Another crucial factor for scalable electronics will be controlling
(3) the stripe spacing. To demonstrate periodic stripe spacing, we set a constant substrate elevation velocity of 5 mm min−1 and applied one dose per 1.2 s to achieve a stripe periodicity of 100 μm (Figure 3d). With this method it is possible to fabricate aligned SWCNT arrays with control over the stripe width, stripe periodicity, and SWCNT density, in a continuous manner, which makes this appealing for high-throughput microeletronic applications. Unlike the studies on evaporative self-assembly from aqueous solutions of SWCNTs which worked well on hydrophilic substrates, our method from organic solutions gave best results on HMDS-treated substrates. HMDS-treated substrates are probably advantageous for thin film transistor devices as they lead to lower charged impurity concentrations. In the higher resolution SEM and AFM images in Figure 3e− g, the degree of alignment of the arc-discharge s-SWCNTs was notably higher than Langmuir−Blodgett and spin-casting methods,10,11 which is quantified in more detail via Raman spectroscopy, below. We used SEM images to quantify an approximate linear packing density of 40−50 tubes μm−1. The packing density achieved here is between the relatively lower densities achieved from aqueous self-assembly (∼20 tubes μm−1), and higher values achieved using Langmuir−Blodgett and Langmuir−Schaefer methods (>100 tubes μm−1).5,11,20 In addition, the AFM height of the stripes was
