Langmuir 2007, 23, 4749-4752
4749
Electrical Transport Characteristics of Surface-Conductance-Controlled, Dielectrophoretically Separated Single-Walled Carbon Nanotubes Seunghyun Hong,† Sehun Jung,‡ Jaeboong Choi,‡ Youngjin Kim,†,‡ and Seunghyun Baik*,†,‡ SKKU AdVanced Institute of Nanotechnology (SAINT) and School of Mechanical Engineering, Sungkyunkwan UniVersity, Suwon, KyungGi-Do 440-746, Korea ReceiVed January 23, 2007. In Final Form: March 20, 2007 Alternating current dielectrophoresis has attracted considerable attention as a possible candidate to separate singlewalled carbon nanotubes according to electronic types. Recently, the significant effect of surface charge on the polarizability of semiconducting nanotubes was demonstrated using comparative Raman spectroscopic studies. Here we present electrical transport characteristics of surface-charge-controlled, dielectrophoretically deposited nanotube arrays. The surface charge was controlled using cationic/anionic surfactant mixtures. Complete separation between metallic and semiconducting species was achieved at the electric field frequency of 10 MHz only when the surface charge of nanotubes was neutralized, which is consistent with previous Raman investigation. A theoretical analysis, using zeta potential information as input, further supported the experimental observation.
Introduction Single-walled carbon nanotubes (SWNTs) have unique properties as components of future nanoscale electronic applications.1-3 However, the realization of nanotube-based electronic devices has been hindered by the inability to separate according to electronic structure.4-6 As-produced SWNTs from the HiPco process have polydisperse mixtures of metallic, semimetallic, and semiconducting fractions with more than 50 individual chiral types.7,8 Over the past few years, the separation of nanotubes using alternating current dielectrophoresis has become an active field of research.5-11 An induced dipole moment is developed for a nanotube in a dielectric fluid when an electric field is applied.6,12-13 The Clausius-Mossotti factor (fCM) represents the effective polarizability of the nanotube. Nanotubes move toward regions of high electric field strength (positive DEP) if the polarizability of the tube is greater than that of the suspending medium (Re[fCM] > 0) whereas nanotubes move in the opposite direction (negative DEP) when the polarizability of the tube is less than that of the suspending medium (Re[fCM] < 0).12 Recently, the effect of surface conductance on the polarizability of semiconducting nanotubes was investigated using Raman * Corresponding author. E-mail:
[email protected]. † SKKU Advanced Institute of Nanotechnology (SAINT). ‡ School of Mechanical Engineering. (1) McEuen, P. Phys. World 2000, 13, 31. (2) Avouris, Ph. Acc. Chem. Res. 2002, 35, 1026. (3) Tans, S. J.; Verschueren, A. R. M.; Dekker, C. Nature 1998, 393, 49. (4) Strano, M. S.; Dyke, C. A.; Usrey, M. L.; Barone, P. W.; Allen, M. J.; Shan, H.; Kittrell, C.; Hauge, R. H.; Tour, J. M.; Smalley, R. E. Science 2003, 301, 1519. (5) Kim, Y. J.; Hong, S.; Jung, S. H.; Strano, M. S.; Choi, J. B.; Baik, S. J. Phys. Chem. B 2006, 110, 1541. (6) Krupke, R.; Hennrich, F.; Lohneysen, H.; Kappes, M. Science 2003, 301, 344 (7) Peng, H.; Alvarez, N. T.; Kittrell, C.; Hauge, R. H.; Schmidt, H. K. J. Am. Chem. Soc. 2006, 128, 8396. (8) Baik, S.; Usrey, M.; Rotkina, L.; Strano, M. S. J. Phys. Chem. B 2004, 108, 15560. (9) Krupke, R.; Hennrich, F.; Kappes, M.; Lohneysen, H. Nano Lett. 2004, 4, 1395. (10) Lee, S. W.; Lee, D. S.; Yu, H. Y.; Campbell, E. E. B.; Park, Y. W. Appl. Phys. A 2004, 78, 283. (11) Krupke, R.; Linden, S.; Rapp, M.; Hennrich, F. AdV. Mater. 2006, 18, 1468. (12) Morgan, H. J.; Green, N. G. AC Electrokinetics: Colloids and Nanoparticles; Research Studies Press: Philadelphia, PA, 2003; Chapters 6 and 7. (13) Green, N. G.; Morgan, H. J. Phys. Chem. B 1999, 103, 41.
spectroscopic measurements.5 Positive DEP was always observed for metallic SWNTs whereas semiconducting SWNTs displayed either positive or negative DEP depending on the electric field frequency and the tube surface conductance. In this letter, we present electrical transport characteristics of surface-conductancecontrolled, dielectrophoretically deposited single-walled carbon nanotubes. The surface conductance was controlled by cationic/ anionic surfactant mixtures because the surface conductance is directly proportional to the surface charge.12 The transport measurements showed complete separation between metallic and semiconducting species only when the surface charge was neutralized, which is consistent with the previous Raman investigation.5 A detailed theoretical analysis about the influence of the electrical double layer on nanotube electrokinetics further supported the experimental observation. Raw HiPco SWNTs purchased from Carbon Nanotechnologies, Inc. were suspended in deionized water under two different surface charge conditions. First, nanotubes were dispersed in anionic surfactant sodium dodecyl sulfate (SDS-SWNT solution) according to a previously published protocol.14 The adsorbed anionic surfactant would increase the surface charge of nanotubes.15 The second nanotube suspension (SDS-CTAB-SWNT solution) was designed to electroneutralize the surface charge of nanotubes.5 Nanotubes were suspended in hexadecyltrimtehylammonium dodecyl sulfate (CTADS), which was formed by an equimolar mixture of anionic SDS and cationic cetyltrimethylammonium bromide (CTAB). This catanionic surfactant is an amphiphilic compound that contains both cations and anions and can be considered to be an ion-paired complex.15-18 Detailed sample preparation methods are provided in Supporting Information. Figure 1 shows dielectrophoretically deposited nanotubes on the electrode from the SDS-CTAB-SWNT solution. The morphologies of dielectrophoretically deposited nanotubes from (14) O’Connell, M. J.; Bachilo, S.; Huffman, C.; Moore, V. C.; Strano, M. S.; Haroz, E.; Rialon, K.; Boul, B.; Noon, W.; Kitrell, C.; Ma, J.; Hauge, R. H.; Weisman, R.; Smalley, R. E. Science 2002, 297, 593. (15) Tomasˇic´, V.; Sˇ tefanic´, I.; Filipovic´-Vincekovic´, N. Colloid Polym. Sci. 1999, 277, 153. (16) Wang, C.; Lucy, C. A. Electrophoresis 2004, 25, 825. (17) Tondre, C.; Caillet, C. AdV. Colloid Interface Sci. 2001, 93, 115. (18) Marques, E. F.; Regev, O.; Khan, A.; Lindman, B. AdV. Colloid Interface Sci. 2003, 100, 83.
10.1021/la070206e CCC: $37.00 © 2007 American Chemical Society Published on Web 03/31/2007
4750 Langmuir, Vol. 23, No. 9, 2007
Letters
The surface conductance consists of two components12
λs ) λs,d + λs,s
(4)
where λs,d is the diffuse layer conductance and λs,s is the Stern layer conductance. The diffuse layer conductance was determined using the zeta potential (ζ) and information about the electrolyte. In general, the zeta potential is often equated with the potential of the diffuse layer.12 The following expression includes contributions from both electroosmosis and ionic conduction12,24 Figure 1. SEM image showing a matted sheet of nanotubes deposited at the 2 µm gap of the electrode from the SDS-CTAB-SWNT suspension by ac dielectrophoresis.
both types of suspensions (SDS-SWNT and SDS-CTABSWNT) were similar (Supporting Information). An electrode with a gap size of 2 µm was fabricated using photolithography. The 5 nm titanium and 100 nm gold contacts were sputtered on a SiO2-Si substrate, and the thickness of the SiO2 layer was 300 nm. A drop of HiPco-SWNT was placed on the electrode after applying alternating current at a frequency of 10 MHz and a peak-to-peak voltage (Vp-p) of 10 V. The droplet was blown off using nitrogen gas after a delay of 10 min, and the electric field was turned off. Electrical transport measurements of the deposited nanotubes were carried out using a semiconductor analyzer (Agilent, E5262A) and a probe station (CASCADE RF-1) in air at room temperature. The samples were characterized after rinsing with ethanol to remove surfactants on the electrode. A dielectric particle in a dielectric medium experiences a force in proportion to the Clausius-Mossotti (fCM) factor.12-13,19 Equation 1 describes the fCM factor derived for a long rod with the major axis aligned with the electric field5,9,19
fCM )
p - m m + (p - m)L
Kp p ) p - i ω
Km m ) m - i ω
(1)
(2)
where is the permittivity, K is the conductivity, and ω is the frequency of the electric field. Subscripts m and p refer to the suspending medium and particle, respectively. The depolarization factor L is approximated by L ) d2/l2[ln(2l/d) - 1]. d and l denote the diameter and length of a tube, respectively. The permittivity of SWNTs is inversely proportional to the square of the band gap energy.6,20 Therefore, the permittivity of metallic SWNTs should be exceedingly large, whereas that of semiconducting SWNTs was found to be less than 5.6,21 The conductivity of a spherical particle in an electrolyte can be expressed as22
Kp ) Kint +
2λs a
(3)
where Kint is the internal particle conductivity, λs is the surface conductance, and a is the radius of a spherical particle. The surface conductance, λs, arises from the movement of counterions in the electrical double layer and is directly proportional to the surface charge density caused by adsorbed ionic surfactants.12,23 (19) Jones, T. B. Electromechanisms of Particles; Cambridge University Press: Cambridge, England, 1995; Chapter 3. (20) Benedict, L. X.; Louie, S. G.; Cohen, M. L. Phys. ReV. B 1995, 52, 8541. (21) Pichler, T.; Knupfer, M.; Golden, M. S.; Fink, J. Phys. ReV. Lett. 1998, 80, 4729. (22) O’Lonski, C. T. J. Phys. Chem. 1960, 64, 605. (23) Green, N. G.; Morgan, H. J. Phys. D: Appl. Phys. 1997, 30, L41.
λs,d )
[
(
)
4q2noz2 3m+ D+(e-zqζ/2kBT - 1) 1 + 2 + kBTκ z
(
D-(ezqζ/2kBT - 1) 1 +
m( )
z2
(5)
kBT q
(6)
2z2q2no kBT
(7)
D( ) µ(
κ)
)]
3m-
x
( )
2 kBT 3 D(η q
2
(8)
where q is the charge of an electron, no is the bulk concentration of an ion, z is the valence of an ion, kB is the Boltzmann constant, T is the temperature, D is the diffusion constant, κ is the reciprocal Debye length, m describes the contribution from the electroosmotic transport, µ is the mobility of an ion, is the permittivity, η is the viscosity of the solution, and ζ is the zeta potential. The Stern layer conductance was estimated using the typical ratio of the Stern layer conductance to the diffuse layer conductance. A value of 0.56, which was suggested by Ermolina et al., was used in this study.25
λs,s ) 0.56 λs,d
(9)
The above equations need to be solved for cylindrical geometry in the future to describe the surfactant-suspended nanotube more accurately because the system appears to be a 1D charged rod. Figure 2 shows the calculated fCM factor at a frequency of 10 MHz as a function of two parameters: R ) p/m and β ) Kp/Km. The depolarization factor L was assumed to be 10-5 (d ) 1 nm, l ) 800 nm). The conductivity of the solution was measured using a conductivity meter (Hanna, HI2300). Km was 0.0814 S/m for the SDS-SWNT solution and 0.1440 S/m for the SDSCTAB-SWNT solution. Different values of Km did not lead to a significant change in the graph because the behavior is governed by the ratio Kp/Km. The lines in Figure 2 were calculated using values of m ) 80 and Km ) 0.0814 S/m. p ) 5 was used for semiconducting SWNTs, and a large value, p ) 4000, was assumed for metallic SWNTs.6,21 Metallic SWNTs with large R values always experienced positive DEP whereas semiconducting SWNTs with R values smaller than 0.0625 displayed either positive or negative DEP depending on the β value (i.e., conductivity of the nanotubes5). As an approximation, the previously reported ζ potential values of SDS and CTAB mixtures were used to calculate the surface (24) Lyklema, J. Fundamentals of Interface and Colloid Science; Academic Press: London, 1995; Vol. 2, Chapter 4. (25) Ermolina, I.; Morgan, H. J. Colloid Interface Sci. 2005, 285, 419.
Letters
Figure 2. Theoretical plot showing the variation in the real part of the Clausius-Mossotti (fCM) factor as a function of two parameters, R ) p/m and β ) Kp/Km. The real part of fCM was zero at β ) 1.28. (a) β ) 17.88 was calculated for the SDS-suspended SWNTs (ζ ) 50 mV). (b) β ) 0.29 was calculated for the CTADS-suspended SWNTs (ζ ) 5 mV).
conductivity of semiconducting nanotubes.15 The following parameters were used to calculate the diffuse layer conductance of SDS-suspended nanotubes using eq 5:12,26 q ) 1.60 × 10-19 C, no)1.65 × 1025 m-3, z ) 1, kB ) 1.38 × 10-23 J K-1, T ) 297 K, mobility of Na+ ion at infinite dilution µ ) 5.19 × 10-8 m2 V-1 s-1, viscosity of water η ) 0.890 × 10-3 K g m-1 s-1, permittivity of water ) 800, and zeta potential ζ ) 50 mV. The stern layer conductance was calculated using eq 9. Finally, the surface conductivity of nanotubes was obtained using eq 3. Although the geometry is not spherical, a ) 2.7 nm was assumed to be an approximation because the double layer forms in the radial direction of nanotubes. O’Connell et al. calculated the density profiles of SDS-suspended nanotubes.14 The radius was chosen at the point where most of the sulfate head groups exist because most of the counterions were shown to accumulate at the outer region where r g 2.7 nm. The fixed charge on the surface is defined as a charge with negligible surface mobility,12,24 and the sulfate head group was assumed to be the fixed charge. The internal conductivity of semiconducting nanotubes should be small without the activation of gate bias or doping agents. Therefore, it was regarded to be zero as an approximation. The calculated β value for the SDS-suspended nanotubes was 17.88 as shown by the circle at point (a) in Figure 2. The increased
Langmuir, Vol. 23, No. 9, 2007 4751
surface conductivity (i.e., large ζ potential), caused by the movement of counterions (Na+) in the electrical double layer, resulted in positive dielectrophoresis of the semiconducting nanotubes. The ζ potential was decreased to 5 mV for the equimolar mixture of SDS and CTAB.15 As shown by the circle at point (b), the calculated β value was 0.29 for the SDS-CTABSWNT solution, leading to the negative dielectrophoresis of semiconducting nanotubes. It is important to note that carbon nanotubes can have charged surface groups, especially after oxidative treatments.27,28 In such cases, the added surface charge should be considered. Raw HiPco SWNTs were used in this study without further purification to minimize the effects from charged surface groups. Figure 3a shows electrical transport data of nanotubes deposited by dielectrophoresis from the SDS-SWNT suspension. The source-drain current (ISD) was measured at five different sourcedrain biases (VSD ) 0.4, 0.8, 1.2, 1.6, and 2.0 V) as a function of the back gate voltage (VG). The matted sheet of nanotubes did not show modulation with changing VG, indicating dominant metallic pathways. The gap size of the electrode (2 µm) was longer than the typical length of HiPco-SWNTs. Also, a large number of nanotubes were deposited between the gap, forming a network. Therefore, it is possible that metallic nanotubes short circuited (or bridged) the electrode gap, exhibiting dominant metallic behavior, although both metallic and semiconducting species could be deposited. The current through metallic pathways should be greater than that through semiconducting species. To investigate this possibility, preferential electrical breakdown of metallic species was performed by applying VG ) 30 V and VSD ) 12 V. Preferential electrical breakdown is a simple method of selectively eliminating metallic nanotubes in a mixture of semiconducting and metallic types. This can be achieved by applying a sufficiently high source-drain bias while simultaneously depleting the carriers of semiconducting nanotubes with an appropriate gate voltage, generating current that passes through metallic nanotubes only.29-31 This mechanism was related to self-heating and thermal oxidation.29-31 Transport data after the electrical breakdown process are shown in the lower part of Figure 3a. The order of the current was decreased from 10-4 to 10-6 A after the breakdown procedure. The nanotube array displayed pure p-type semiconducting behavior, and the on/off ratio was on the order of 106. These results indicated that both metallic and semiconducting species were deposited by dielectrophoresis from the SDS-SWNT suspension.
Figure 3. Electrical transport data before and after the preferential electrical breakdown procedure. The source-drain current was measured at five different source-drain biases (1:2, 2:1.6, 3:1.2, 4:0.8, and 5:0.4 V) as a function of the back gate voltage. (a) Dielectrophoretically deposited nanotubes from the SDS-SWNT suspension. (b) Dielectrophoretically deposited nanotubes from the SDS-CTAB-SWNT suspension.
4752 Langmuir, Vol. 23, No. 9, 2007
Figure 3b shows electrical transport data of dielectrophoretically deposited nanotubes from the SDS-CTAB-SWNT suspension before and after the breakdown process. In this case, all of the deposited nanotubes were destroyed by an identical electrical breakdown procedure (Supporting Information). Therefore, the breakdown procedure was stopped when the sourcedrain current was decreased to the order of 10-6 A to compare the data with the transport characteristics shown in Figure 3a. The nanotube network displayed no gate dependence before and after the preferential electrical breakdown process, indicating that only metallic nanotubes were deposited by dielectrophoresis from the SDS-CTAB-SWNT suspension. Semiconducting nanotubes could also be damaged by the electrical breakdown procedure.30 In particular, Joule heat generated in metallic nanotubes can be locally transferred to semiconducting nanotubes in the network of nanotubes, resulting in the destruction of semiconducting species. However, the two different types of data shown in parts a and b of Figure 3 were obtained by an almost identical breakdown procedure, and the distinctive trends were reproducible. Besides, the current electrical transport characteristics are consistent with the previous Raman spectroscopic observation.5 It is also important to note that metallic (26) Lide, D. R., Ed. CRC Handbook of Chemistry and Physics, 85th ed.; CRC Press: London, 2004. (27) Zhao, W.; Song, C. H.; Pehrsson, P. E. J. Am. Chem. Soc.2002, 124, 12418. (28) Dukovic, G.; White, B. E.; Zhou, Z.; Wang, F.; Jockusch, S.; Steigerwald, M. L.; Heinz, T. F.; Friesner, R. A.; Turro, N. J.; Brus, L. E. J. Am. Chem. Soc.2004, 126, 15269. (29) Collins, P. G.; Arnold, M. S.; Avouris, Ph. Science 2001, 292, 706. (30) Seidel, R. V.; Graham, A. P.; Rajasekharan, B.; Unger, E.; Liebau, M.; Duesberg, G. S.; Kreupl, F.; Hoenlein, W. J. Appl. Phys. 2004, 96, 6694. (31) Radosavljevic´, M.; Lefebvre, J.; Johnson, A. T. Phys. ReV. B 2001, 64, 241307.
Letters
SWNTs are expected to be enriched, although not perfectly separated, in the dielectrophoretically deposited sample from the SDS-SWNT suspension because the magnitude of the fCM factor of metallic SWNTs is greater than that of semiconducting SWNTs (Figure 2).5 In summary, we investigated the effects of induced surface charge on the sign of the dieletrophoretic force of semiconducting nanotubes. The surface charge was controlled by cationic/anionic surfactant mixtures. The theory predicted positive dielectrophoresis of SDS-suspended semiconducting nanotubes. The sign of the dielectrophoretic force of semiconducting nanotubes could be changed to negative when the surface charge was neutralized by an equimolar mixture of SDS and CTAB. Positive dielectrophoresis was always observed for metallic nanotubes regardless of the different types of surfactants used. This theoretical analysis was supported by the electrical transport characteristics of surfacecharge-controlled, dielectrophoretically deposited SWNT arrays. Also, the electrical transport data are consistent with the previous Raman observation.5 Acknowledgment. This work was supported by a grant (05K1401-00410) from the Center for Nanoscale Mechatronics & Manufacturing, one of the 21st Century Frontier Research Programs supported by the Ministry of Science and Technology, Korea. Supporting Information Available: Detailed sample preparation methods and electrical transport characteristics of the dielectrophoretically deposited SWNT arrays before and after the electrical breakdown procedure. This material is available free of charge via the Internet at http://pubs.acs.org. LA070206E
