Two-Dimensional Carbon Nanotube Networks: A Transparent ...
Mater. Res. Soc. Symp. Proc. Vol. 905E © 2006 Materials Research Society
0905-DD06-05.1
Two-Dimensional Carbon Nanotube Networks: A Transparent Electronic Material George Gruner Department of Physics and Astronomy University of California Los Angeles Box 951547 Los Angeles, CA 90095-1547 ABSTRACT The random, two-dimensional network formed of electrically conducting nanoscale wires, called carbon nanotubes, is a transparent electronic material that can be fabricated using room-temperature printing or spraying technologies. Depending on the network density, networks with both metallic- and semiconducting-like attributes can be fabricated. Both display high conductivity, high carrier mobility and optical transparency. The networks also have high mechanical flexibility, robustness and environmental resistance. Application opportunities range from lightweight, transparent conducting films, to electrically conducting fabrics, to active electronic devices and sensors. INTRODUCTION Transparent and conducting materials have found applications in a variety of areas, ranging from energy to electronics. The current choice of materials are transparent conducting oxides, with indium-tin-oxide, ITO as the prime example1,2. In recent years we have seen the emergence of novel materials ranging from conducting plastics to composites. The value proposition in all cases lies not necessarily in increased performance, but in cheap fabrication and additional attributes such as mechanical flexibility. This paper discusses the properties of another “material”, a two-dimensional network of carbon nanotubes (NTs), nano-scale wires with exceptional mechanical and electrical properties. These properties have been demonstrated3,4 through extensive experiments performed on single NTs (throughout this paper, single wall carbon NTs SWNTs are discussed). Manufacturability and system integration is required for the exploitation of the attributes, an objective difficult to achieve if, for example, devices incorporating single tubes are the objective. For this reason, an alternative avenue that exploits large-scale statistical averaging of the properties of individual tubes appears to be more promising for a variety of applications. A random network of such NTs is an obvious – and perhaps the most straightforward – realization of this concept. Such a network is also referred to as a “thin film”, although this nomenclature may be misleading when applied to networks with significantly less that full coverage of a substrate. There is growing interest in single wall carbon NT thin films for applications in the area of macroelectronics and in the general area of optoelectronics where flexible, transparent and conducting coatings together with room-temperature fabrication are required.
0905-DD06-05.2
This paper discusses some of the fabrication issues properties of nanotube networks and the various proof-of-concept realizations of the usefulness of the material in a variety of applications. THE MATERIAL: A TWO DIMENSIONAL RANDOM CARBON NANOTUBE NETWORK What we call a “material’ is a two-dimensional network of nanoscale wires on a substrate3. The building blocks are carbon NTs, the world’s smallest conducting wires4. The diameter of the tubes is less than that of DNA, and the tubes can be grown up to millimeter lengths. They come in metallic and semiconducting forms (currently a mixture of these is available) with higher abundance of semiconducting tubes. The conductivities are comparable to copper and silicon and exceed that of any conducting polymer by orders of magnitude. In air the semiconducting tubes are p-doped, but can be n-doped by surface attachment of electron donors5,6. The tubes are robust and have excellent mechanical and thermal properties. The architecture that leads to a novel material with highly desirable transparency and conductivity is that of a two-dimensional random NT network, that provides a large number of conducting pathways. An appropriate analogy of the architecture is that of an interconnected network of freeways, providing a fast transport medium – potentially significantly faster than a uniform, but lower medium (analogous to surface roads or the terrain itself). The network is flexible, and the carbon tubes that form the network have robust environmental resistance. The network is characterized by two types of disorder. The first is topological: above a certain critical density there is a large number of conducting pathways, connecting, say, a source and drain electrode, and the number of pathways is a strong function of the density. Second, these pathways are electrically conducting, though the individual NTs are separated by barriers, representing the interface between the tubes7. As expected, due to the different relative positiondependent contacts between the individual tubes, there is a distribution of barrier heights – leading to a second source of randomness. These features determine the concentration, temperature, frequency and voltage dependent conductivity of the network. FABRICATION TECHNOLOGIES Various avenues are available for producing SWNTs. Direct growth on a surface8 can produce high quality films, but it requires high temperatures which are not compatible with flexible substrates. In addition, only rather rare networks can be grown at present. Various routes are available for dissolving the nanotubes in an appropriate solvent. We 9,10, and others have used surfactants to separate the NTs in a solution, with Sodium Dodecyl Sulfonate (SDS) or Sodium Dodecyl Benzene Sulfonate (NaDDBS) is the current choice of surfactant. Simple drop casting or spin coating leads to networks where the NTs are “bundled” together 9,11, forming bundles of considerable size and containing hundreds of NTs. A process where a sieve is applied9,10,12 leads to better dispersion – and at the same time better avenues for removing the solubilization agents. The current choice of sieve, porous alumina is not transparent and the network has to be transferred onto a transparent surface, such as glass or PET. A printing method13, using PDMS
0905-DD06-05.3
(poly-dimethylsolixane) leads to a typical NT network as shown in Figure 1a. The bundle size is estimated as approximately 3nm, thus the bundles contain approximately ten tubes. The root mean square roughness of the film is 8 nm as estimated from the AFM image. Examining the edge of the film, we obtain the film thickness of 25 nm, which leads to a conductivity of 1600 S/cm10. The measured conductivity allows us to assess the overall quality of the films, in comparison with films fabricated by methods reported in the literature. Films deposited directly onto various surfaces and having significantly larger bundles do not exceed the conductivity value of 200 S/cm. For films shown in Figure 1. the NT bundle size is approximately 3 nm, compared to the 20 nm bundle and 200 S/cm obtained by us earlier9,12. In contrast, films grown onto a surface using chemical vapor deposition (CVD) lead to wellFigure.1. (a) Carbon nanotube network on a PET separated individual NTs, also with a high substrate. (b) Network conductivity as function of conductivity14. In Fig 1b the conductivity is the nanotube bundle diameter, d. displayed as the function of the inverse of the bundle diameter – including films prepared using CVD, direct deposition and printing - clearly indicating that smaller bundle size is associated with larger conductivity. The reason behind this tendency is that the current flows at the surface of the bundles, with NTs in the interior of the tube not contributing to the conduction process. One should also mention the difference between our results and the results of Wu et al15 may lie in the fact that our networks contain undoped NTs, while in Ref 15, the tubes are prepared by acid reflux, and thus may have been doped (see below). NETWORK PROPERTIES As expected the conductivity is concentration, temperature, frequency and voltage dependent, and such dependences of networks with different densities are displayed on Figures 2 and 3. The concentration dependence, as evaluated at room temperature shown in Figure 2 can be understood in terms of a two dimensional (2D) percolation theory16. For low densities, there is no conduction, until a threshold, given by
Figure 2. Dependence of the network sheet conductance on the network density. The full line is the density dependence expected for a two dimensional percolating network. Ref 9b.
0905-DD06-05.4
l N c = 4.24 (1) is reached. Here l is the length and Nc is the critical density of the NT bundle. From AFM images, l =2µ leading to Nc= 1,43 bundle/µm2, which is compared favorably with the theory, giving Nc= 1,2 bundle/µm2. Above the threshold the conductivity is given by ~ (N-Nc) (2)
with the experiment leading to 1.5, while theory predicts =1.33. The relatively good agreement is surprising, in the light of the fact that there is a distribution of the barrier heights that are associated with the charge transfer from one bundle to the other. This is evidenced by the temperature and frequency dependences, displayed for a few concentrations in Figure 3. One notices that at high densities, well above the percolation threshold the resistivity is only weekly temperature dependent – mimicking the behavior of a “bad” metal. In contrast for smaller network density, in particular near to the percolation threshold the conductivity depends strongly on the temperature, giving evidence that temperature driven excitation processes dominate the conductivity. At the same time, the conductivity is frequency dependent for small densities, even at radio frequencies, this becoming weaker with increasing network density. The temperature dependences have been examined in detail17 and experiment, when compared to theory give evidence for temperature Figure 3. Temperature and frequency dependence driven transitions of electrons (or holes) from tubes of the sheet conductance G of carbon nanotube (or bundles) to tubes (or bundles). The overall networks of different densities. The same amount temperature and concentration dependence, including of nanotubes were dissolved in different volumes of solvent, thus smaller mL in the figure indicates both low density and high density can be, at least larger network density. semi-quantitatively understood based on what has been said before, with one additional factor: at low densities the conducting networks all have at least some semiconducting nanotube networks (NTs), and at higher densities pathways with allmetallic NTs are feasible. This can be described18 with a model that includes both semiconducting and metallic pathways. A significant body of work exists on the elaboration on these notions (and some offering detailed fits to theories), all of which are based on the inherent randomness, and temperature driven transport across random barriers. The frequency dependent conductivity () supports this general picture. () can be described by an expression () = A(i) (3) used in general to describe the conduction process in the presence of randomness19. What is important for the context of applications is that nanotube networks can serve as a semiconducting channel (essential for transistor operation) and also as a ”metallic” interconnect.
0905-DD06-05.5
Networks at different densities lead to different sheet conductances and optical transmittances. What is important to applications is the sheet resistance, given by R sq = 1 dc .d for a film with thickness d. The sheet conductance /sq is the inverse of the sheet resistance. Due to the one-dimensional nature of the wires, the network is highly transparent in the infrared and also in the visible spectral range. Early experiments21 gave evidence that the reflectivity is low in the visible, and thus the transmission is mainly determined by the absorption in the network. The equation below describes the optical transmittance assuming that the film thickness is much less than the wavelength and also assuming that the imaginary part of the conductivity can be neglected20
T=
1 1 = 2 2 (1+ ac d) 2 1+ 2 ac c R sq dc
(4)
Here ac is the conductivity at optical frequencies, and c the speed of light. It has been shown21 that, in contrast to the dc conductivity, the ac conductivity in the visible spectral range is largely proportional to the density of the network, but is not, or is only very weakly influenced by inter-tube resistances, by the NT length, or by doping. Thus higher “quality “ networks – pure and well-dispersed nanotubes for example – will lead to higher conductivity but the same transparency for the same density. Figure 4 illustrates the transparency-sheet conductance Figure 4. Sheet resistance versus optical relation, achieved to date. The full line is Equation 4 transmittance of carbon nanotube networks of with the dc conductivity dc=1600S/cm. Significant varying densities. improvement is expected through nanotube purification, better dispersion of the NTs, and reducing the inter-tube resistance. The tubes are p-doped in air due to the O-covering the tubes. The tubes can be chemically modified, leading to changes of the electronic structure and consequently the transport properties. Such modifications are usually performed by dissolving the NTs in a solution and performing the fictionalization in a wet chemistry environment22. Both further p-doping and ndoping can be achieved by covering the network with electron withdrawing and electron donating species. This has been demonstrated with the chemicals NH3, and NO2 and detailed analysis has been performed of the case of NH3 in water23. Hole doping of the NTs was shown to occur by treatment with electron withdrawing molecules, such as I, Br, K, while electron donating species such as PEI21 lead to an N-type material. Experiments using transistors are able to distinguish between charge transfer to the tubes, and the effects of scattering potential (caused by the dopant) in reducing the mobility. Few experiments24,25 have been analyzed using such notions, much further work remains to be done in this area.
0905-DD06-05.6
APPLICATIONS Application opportunities that are based on highly transparent and conducting films or layers include energy (solar cells, solid state lighting), electronics (passive and active matrix displays, smart windows), appliances (defrosting windows, touch screen) and security (RFID tags, electromagnetic shielding and sensors). The technology can be divided into several areas: transparent conducting layers, transparent/flexible electronic devices and functionalized networks for optoelectronics, photovoltaics and sensors. Flexible and optically transparent conducting layer The current technology is based on high temperature deposition of a brittle, inflexible material, ITO1,2. Measurements on NT network layers, fabricated at UCLA are already close to the parameters of commercial ITO (Figure 5) with significant advances anticipated through materials optimization. In addition, doping and density tailoring can lead to highly application-specific networks. The green squares on Figure 5 represent Figure 5 Sheet resistance of various materials at 80% the current state of the art. Significant transparency at 550nm. The “Single NT limit” refers to a of aligned, infinitely long nanotubes. Some application improvements are expected through network barriers are indicated by arrows. using more purified nanotubes, better dispersion of the tubes and reducing the NT-NT resistances. The ultimate “Single NT limit” refers to a network of aligned, infinitely long tubes. Conducting fabrics. Carbon NT networks deposited on fabric26 lead to highly conducting, wearable materials. The conducting characteristics well exceed those of other approaches such as transparent polymer coatings that address the issue of electrically conducting fabrics. In addition, the network is light-weight, mechanically robust and resistant to environmental factors. An Figure 6 Conducting fabric, arrived at by spraying carbon nanotubes onto a fatigue. Ref 26.
0905-DD06-05.7
illustration of an electrically conducting fatigue is shown on Figure 6. “Nanotube” networks for organic lighting, photovoltaics and smart windows Nanotube networks cab be combined with other, functional layers, layers with electrochromic properties, layers where charge separation occurs upon the application of a dc voltage, or visible light. Such geometrics support opto-electronic or photovoltaic devices. Several of these concepts have been demonstrated to date. A “smart window” configuration with two nanotube27 network electrodes is displayed on Figure 7. For solar cell and OLED applications the work function of the materials is important. Calculations and analysis28,29 of field emission experiments indicate that the work function is approximately 5 eV comparable to ITO. This sets the stage of exploration of such networks as replacement of ITO, and progress has been made in this area.
Figure 7. (a) Smart window configuration using two nanotube networks as transparent electrodes. (b) Change of the transmission with voltage applied between the electrodes. Ref 27.
Flexible/transparent electronic devices
Figure 8. (a) Layout of the FET with carbon nanotube gate and conducting channel. (b) Dependence of the source-drain current on the gate voltage in the low voltage linear Isd(Vsd) region. Ref 34.
Current flexible electronic devices, such as FETs, field effect transistors are mainly based on polymeric materials. The overall performance of devices is limited by the fundamental conducting properties of polymers, and some parameters such as the mobility are not, and probably will never be, comparable to conventional silicon based devices. In spite of significant efforts, the mobility of the devices, while improving still does not exceed 1 cm2/Vsec, limiting the application potential of the devices. Transistors have been fabricated using lowdensity NT network as the conducting channel. The concept of such transistors emerged a few years ago30, and has been realized31,32 by using CVD grown tubes on silicon substrate. Flexible transistors33 together with transparent/flexible transistors34 have also been fabricated as of today. Figure 8 displays such a transistor configuration where both the conducting channel and the gate
0905-DD06-05.8
were fabricated using networks of different densities. The transistor is highly transparent, with the transmittance determined by the density of the NT network that serves as the gate electrode. The mobility of such transistors fabricated using a NT network (blue cross in Figure 9) exceed well all current solutions for high mobility, flexible transistors. Other parameters such as the onoff ratio and hysteresis have still to be improved, although significant progress has been recently achieved35 in the area. Functionalized networks for optoelectronics Devices with carbon nanotube networks can be combined with small molecules and polymers with carbon functionalities. As an example a light sensitive small molecule such as porphyrin can be combined with the network36a leading to a mechanism where light is directly converted into an electronic signal – the fundamental Figure 9. The evolution of the FET device mobility over the last decades. The mobility of carbon nanotube FETs is indicated by the blue cross. Some application barriers are also indicated.
.
process in many optoelectronic and photovoltaic devices. The structure is illustrated on Figure 10, together with the response obtained upon illumination with a light. Examination of the wavelength dependence of the response gives Figure 10. Detection of light using a porphyrinevidence that light-induced electronic transitions of nanotube network complex. Ref 36a. the porphyrin molecules is responsible for the change of the transistor characteristics. Similar observations were made by using the light sensitive polymer PmPv36b. CONCLUSIONS The ultimate usefulness of the material lies in its competitive position with respect to other materials used, and therefore a few comments on this aspect of the work are in order. First, both spraying and printing, as explored to date allow up-scaling of the fabrication of films. Second, the performance characteristics are similar, or exceed that of other materials such as a polymer based transparent conductor, offered by AGFA37, and the performance of ITO on a plastic surface. ITO is readily used in a variety of applications, and we conclude that carbon nanotube films, described in this paper have the technical characteristics that allow applications
0905-DD06-05.9
in areas where flexible coatings are required, such as transparent EM shielding, smart windows, touch screen displays, solar cells, OLEDs and flat panel displays. The third, and quite possibly the most important factor is the cost of the material and of the process being used. The cost of the fabrication process is expected to be less that that for ITO for which high temperatures and vacuum deposition is required. With the abundance of carbon, the price of carbon NTs is expected to decrease with further commercialization, in contrast to the steady increase of the price of ITO. By considering all the attributes discussed above, it is evident that carbon NT films will find their application in a variety of areas that require flexible, transparent and conducting coatings in a patterned or non-patterned form. ACKNOWLEDGEMENTS I appreciate the contributions of my students and postdoctoral associates, L. Hu, Y. Zhou, D. Hecht, , E. Artukovic and M. Briman. This work was supported by NSF grant 0404029. REFERENCES 1. 2. 3. 4. 5. 5a. 6. 7. 8. 9a. 9b.
B.G. Lewis and D.C. Paine, “Applications and Processing of Transparent Conducting Oxides,” MRS Bull. 25 (8), 22 (2000). R.G. Gordon, “Criteria for Choosing Transparent Conductors,” MRS Bull. 25 (8), 52 (2000). T. Durkop, S.A. Getty, Enrique Cobas, and M.S. Fuhrer, “Extraordinary Mobility in Semiconducting Carbon Nanotubes,” Nano Lett 4, 35 (2004). M.S. Dresselhaus and H. Dai, and Guest Editors, “Carbon Nanotubes: Continued Innovations and Challenges,” MRS Bull. 29 (4), 237 (2004). J.A. Shim, N.W. Kam, and H. Dai, “Polymer functionalization for air-stable n-type carbon nanotube field-effect transistors,” J Am Chem Soc. 123, 11512 (2001). C. Zhou, J. Kong, E. Yenilmez, and H. Dai, “Modulated Chemical Doping of Individual Carbon Nanotubes,” Science 290, 1552 (2000). G. de la Torre, W. Blau, and T. Torres, “A survey on the functionalization of single walled nanotubes. The chemical attachment of phthalocyanine moieties,” Nanotechnologies 14, 765 (2003). M. Stadermann, S.J. Papadakis, M.R. Falvo, J. Novak, E. Snow, Q. Fu, J. Liu, Y. Fridman, J.J. Boland, R. Superfine, and S. Washburn, “Nanoscale study of conduction through carbon nanotube networks,” Phys. Rev. B69, 201402(R) (2004). Y. Zhou, A. Gaur, S.-H.Hur, C. Kocabas, M. A. Meitl, M. Shim, J. A Rogers, “pChannel, n-Channel Thin Film Transistors and p-n Diodes Based on Single Wall Carbon Nanotube Networks”, Nano Lett. 4, 2031 (2004) N.P. Armitage, J-C.P. Gabriel, and G. Gruner, “Langmuir-Blodgett nanotube films,” Appl. Phys. Lett. 2003 L. Hu, D.S. Hecht, and G. Grüner, “Percolation in Transparent and Conducting Carbon Nanotube Networks,” Nano Lett. 4 (12), 2513 –2517 (2004)
0905-DD06-05.10
10. 11. 12. 13.
Y Zhou, L. Hu, and G. Grüner, “A Printing Method for carbon nanotube thin films,” (to be published) M. Kaempgen, G.S. Duesberg, and S. Roth “Transparent Carbon Nanotube Coatings,” Appl. Surf. Sci., 252, 425 (2005) The conductivity is calculated by assuming that nanotubes fully fill the space, this assumption is clearly not valid, and leads to an underestimation of the conductivity. M.A. Meitl, Y. Zhou, A. Gaur, S. Jeon, M.L. Usrey, M.S. Strano, and J.A. Rogers, Nano Lett. 4, 1643 (2004)
14.
Hongjie Dai, "Growth and Characterization of Carbon Nanotubes", book chapter in "Topics in Applied Physics", Vol. 80, "Carbon Nanotubes", Edited by M. Dresselhaus, G. Dresselhaus and P. Avouris, Springer Verlag (2000).
15.
Z.C. Wu, Z.H. Chen, X. Du, et al. “Transparent, conductive carbon nanotube films,” Science, 305, 1273 (2004) G. Stauffer, “Introduction to Percolation Theory,” Taylor & Francis: London, 1985 J. Vavro, J.M. Kikkawa, and J.E. Fischer, “Metal-Insulator Transition in Doped Single Wall Carbon Nanotubes,” Physical Review B71, 155410 (2005) E. Bakyarova, M.E. Itkis, N. Cabrera, B. Zhao, A. Yu, J. Gao, and R.C. Haddon, “Electronic Properties of Single-Walled Carbon Nanotube Networks,” J. Am. Chem. Soc., 127, 5990 (2005) A.B. Kaiser, G. Düsberg, and S. Roth, “Heterogeneous model for conduction in carbon nanotubes,” Physical Review B (Condensed Matter and Materials Physics), 57, 1418 (1998) G. Grüner, “Nonlinear and frequency-dependent transport phenomena in low dimensional conductors,” Physica D: Nonlinear Phenomena, 8 , 1 (1983) M. Dressel, and G.Gruner, “Electrodynamics of Solids,” Cambridge University Press, 2002 B. Ruzicka and L. Degiorgi, “Optical and dc conductivity study of potassium-doped single-walled carbon nanotube films,” Phys. Rev. B61, R2468 (2000) Bokrath et al. “Chemical Doping of Individual semiconducting carbon nanotube ropes,” Phys. Rev B61, 10606 (2000) K. Bradley, J.C. Gabriel, M. Briman, A. Star, G. Gruner, “Charge Transfer from Ammonia Physisorbed on Nanotubes,” Phys. Rev Lett. 91 (21), 218301 (2003) G. P. Siddons, D. Merchin, J. H. Back, J. K. Jeong, M. Shim, “Highly Efficient Gating and Doping of Carbon Nanotubes with Polymer Electrolytes,” Nano Lett. 4, 927931(2004) A. Star, T-R Han, J-C.P. Gabriel, K. Bradley, and G. Grüner, “Interaction of AromaticCompounds with Carbon Nanotubes,” Nano Letters, 3 (10), 1421 –1423 (2003) K. Bradley, M. Briman, A. Star, G. Gruner, “Charge Transfer from Adsorbed Proteins”, Nano Lett. 4, 253-256 (2004) D.S. Hecht, L. Hu, and G. Gruner, “Electronic Properties of Carbon Nanotube/Fabric Composites,” Accepted in Current Applied Physics (2005) L. Hu, D. Li, Y. Zhou, R. Kaner and G. Gruner, “Patterned transport carbon nanotube electrodes for electrochromatic device applications,” (to be published) B. Shan and K. Cho “First-Principles study of Work Functions of Single Wall Carbon Nanotubes” Phys. Rev. Lett. 94 (23), 236602 (2005)
16. 17a. 17b. 18. 19. 20. 21. 22. 23. 24. 25a. 25b. 26. 27. 28.
0905-DD06-05.11
29. 30. 31. 32. 33. 34. 35.
36a. 36b. 37.
N. de Jonge, M. Allioux, M. Doytcheva, M. Kaiser, K.B.K. Teo, R.G. Lacerda and W.I. Milne. "Characterization of the field emission properties of individual thin carbon Nanotubes,” Appl. Phys. Lett. 85, 1607 (2004) G. Gruner, Patent Application J.C.P Gabriel, Mater. Res. Symp. Proc. 762, Q.12.7.1 (2003) E.S. Snow, J. Novak, P.M.Campbell, D. Park, “Random networks of carbon nanotubes as an electronic material,” Appl. Phys. Lett. 82 (13), 2145 (2003) K. Bradley, J-C.P. Gabriel, and G. Grüner, “Flexible Nanotube Electronics,” Nano Letters, 3 (10), 1353 (2003) E. Artukovic, M. Kaempgen, D.S. Hecht, S. Roth, and G. Grüner, “Transparent and Flexible Carbon Nanotube Transistors,” Nano Lett. 5 (4), 757 (2005) S.-H. Hur, M.-H. Yoon, A. Gaur, M. Shim, A. Facchetti, T.J. Marks and J.A. Rogers, “Organic Nanodielectrics for Low Voltage Carbon Nanotube Thin Film Transistors and Complementary Logic Gates,” Journal of the American Chemical Society 127(40), 13808 (2005) D.S. Hecht, R.A. Ramirez, E. Artukovic, M. Briman, K. Chichak, J.F. Stoddart, G. Gruner, “Bio-inspired Detection of Light using Porphyrin Sensitized Carbon Nanotube FETs,” Submitted (2005) A. Star, Y. Lu, K. Bradley, G. Gruner, “Nanotube optoelectronic memory devices.” Nano Lett. 4 1587 (2004). Orgacon™ EL – P3000, AGFA
0905-DD06-05.1
Two-Dimensional Carbon Nanotube Networks: A Transparent Electronic Material George Gruner Department of Physics and Astronomy University of California Los Angeles Box 951547 Los Angeles, CA 90095-1547 ABSTRACT The random, two-dimensional network formed of electrically conducting nanoscale wires, called carbon nanotubes, is a transparent electronic material that can be fabricated using room-temperature printing or spraying technologies. Depending on the network density, networks with both metallic- and semiconducting-like attributes can be fabricated. Both display high conductivity, high carrier mobility and optical transparency. The networks also have high mechanical flexibility, robustness and environmental resistance. Application opportunities range from lightweight, transparent conducting films, to electrically conducting fabrics, to active electronic devices and sensors. INTRODUCTION Transparent and conducting materials have found applications in a variety of areas, ranging from energy to electronics. The current choice of materials are transparent conducting oxides, with indium-tin-oxide, ITO as the prime example1,2. In recent years we have seen the emergence of novel materials ranging from conducting plastics to composites. The value proposition in all cases lies not necessarily in increased performance, but in cheap fabrication and additional attributes such as mechanical flexibility. This paper discusses the properties of another “material”, a two-dimensional network of carbon nanotubes (NTs), nano-scale wires with exceptional mechanical and electrical properties. These properties have been demonstrated3,4 through extensive experiments performed on single NTs (throughout this paper, single wall carbon NTs SWNTs are discussed). Manufacturability and system integration is required for the exploitation of the attributes, an objective difficult to achieve if, for example, devices incorporating single tubes are the objective. For this reason, an alternative avenue that exploits large-scale statistical averaging of the properties of individual tubes appears to be more promising for a variety of applications. A random network of such NTs is an obvious – and perhaps the most straightforward – realization of this concept. Such a network is also referred to as a “thin film”, although this nomenclature may be misleading when applied to networks with significantly less that full coverage of a substrate. There is growing interest in single wall carbon NT thin films for applications in the area of macroelectronics and in the general area of optoelectronics where flexible, transparent and conducting coatings together with room-temperature fabrication are required.
0905-DD06-05.2
This paper discusses some of the fabrication issues properties of nanotube networks and the various proof-of-concept realizations of the usefulness of the material in a variety of applications. THE MATERIAL: A TWO DIMENSIONAL RANDOM CARBON NANOTUBE NETWORK What we call a “material’ is a two-dimensional network of nanoscale wires on a substrate3. The building blocks are carbon NTs, the world’s smallest conducting wires4. The diameter of the tubes is less than that of DNA, and the tubes can be grown up to millimeter lengths. They come in metallic and semiconducting forms (currently a mixture of these is available) with higher abundance of semiconducting tubes. The conductivities are comparable to copper and silicon and exceed that of any conducting polymer by orders of magnitude. In air the semiconducting tubes are p-doped, but can be n-doped by surface attachment of electron donors5,6. The tubes are robust and have excellent mechanical and thermal properties. The architecture that leads to a novel material with highly desirable transparency and conductivity is that of a two-dimensional random NT network, that provides a large number of conducting pathways. An appropriate analogy of the architecture is that of an interconnected network of freeways, providing a fast transport medium – potentially significantly faster than a uniform, but lower medium (analogous to surface roads or the terrain itself). The network is flexible, and the carbon tubes that form the network have robust environmental resistance. The network is characterized by two types of disorder. The first is topological: above a certain critical density there is a large number of conducting pathways, connecting, say, a source and drain electrode, and the number of pathways is a strong function of the density. Second, these pathways are electrically conducting, though the individual NTs are separated by barriers, representing the interface between the tubes7. As expected, due to the different relative positiondependent contacts between the individual tubes, there is a distribution of barrier heights – leading to a second source of randomness. These features determine the concentration, temperature, frequency and voltage dependent conductivity of the network. FABRICATION TECHNOLOGIES Various avenues are available for producing SWNTs. Direct growth on a surface8 can produce high quality films, but it requires high temperatures which are not compatible with flexible substrates. In addition, only rather rare networks can be grown at present. Various routes are available for dissolving the nanotubes in an appropriate solvent. We 9,10, and others have used surfactants to separate the NTs in a solution, with Sodium Dodecyl Sulfonate (SDS) or Sodium Dodecyl Benzene Sulfonate (NaDDBS) is the current choice of surfactant. Simple drop casting or spin coating leads to networks where the NTs are “bundled” together 9,11, forming bundles of considerable size and containing hundreds of NTs. A process where a sieve is applied9,10,12 leads to better dispersion – and at the same time better avenues for removing the solubilization agents. The current choice of sieve, porous alumina is not transparent and the network has to be transferred onto a transparent surface, such as glass or PET. A printing method13, using PDMS
0905-DD06-05.3
(poly-dimethylsolixane) leads to a typical NT network as shown in Figure 1a. The bundle size is estimated as approximately 3nm, thus the bundles contain approximately ten tubes. The root mean square roughness of the film is 8 nm as estimated from the AFM image. Examining the edge of the film, we obtain the film thickness of 25 nm, which leads to a conductivity of 1600 S/cm10. The measured conductivity allows us to assess the overall quality of the films, in comparison with films fabricated by methods reported in the literature. Films deposited directly onto various surfaces and having significantly larger bundles do not exceed the conductivity value of 200 S/cm. For films shown in Figure 1. the NT bundle size is approximately 3 nm, compared to the 20 nm bundle and 200 S/cm obtained by us earlier9,12. In contrast, films grown onto a surface using chemical vapor deposition (CVD) lead to wellFigure.1. (a) Carbon nanotube network on a PET separated individual NTs, also with a high substrate. (b) Network conductivity as function of conductivity14. In Fig 1b the conductivity is the nanotube bundle diameter, d. displayed as the function of the inverse of the bundle diameter – including films prepared using CVD, direct deposition and printing - clearly indicating that smaller bundle size is associated with larger conductivity. The reason behind this tendency is that the current flows at the surface of the bundles, with NTs in the interior of the tube not contributing to the conduction process. One should also mention the difference between our results and the results of Wu et al15 may lie in the fact that our networks contain undoped NTs, while in Ref 15, the tubes are prepared by acid reflux, and thus may have been doped (see below). NETWORK PROPERTIES As expected the conductivity is concentration, temperature, frequency and voltage dependent, and such dependences of networks with different densities are displayed on Figures 2 and 3. The concentration dependence, as evaluated at room temperature shown in Figure 2 can be understood in terms of a two dimensional (2D) percolation theory16. For low densities, there is no conduction, until a threshold, given by
Figure 2. Dependence of the network sheet conductance on the network density. The full line is the density dependence expected for a two dimensional percolating network. Ref 9b.
0905-DD06-05.4
l N c = 4.24 (1) is reached. Here l is the length and Nc is the critical density of the NT bundle. From AFM images, l =2µ leading to Nc= 1,43 bundle/µm2, which is compared favorably with the theory, giving Nc= 1,2 bundle/µm2. Above the threshold the conductivity is given by ~ (N-Nc) (2)
with the experiment leading to 1.5, while theory predicts =1.33. The relatively good agreement is surprising, in the light of the fact that there is a distribution of the barrier heights that are associated with the charge transfer from one bundle to the other. This is evidenced by the temperature and frequency dependences, displayed for a few concentrations in Figure 3. One notices that at high densities, well above the percolation threshold the resistivity is only weekly temperature dependent – mimicking the behavior of a “bad” metal. In contrast for smaller network density, in particular near to the percolation threshold the conductivity depends strongly on the temperature, giving evidence that temperature driven excitation processes dominate the conductivity. At the same time, the conductivity is frequency dependent for small densities, even at radio frequencies, this becoming weaker with increasing network density. The temperature dependences have been examined in detail17 and experiment, when compared to theory give evidence for temperature Figure 3. Temperature and frequency dependence driven transitions of electrons (or holes) from tubes of the sheet conductance G of carbon nanotube (or bundles) to tubes (or bundles). The overall networks of different densities. The same amount temperature and concentration dependence, including of nanotubes were dissolved in different volumes of solvent, thus smaller mL in the figure indicates both low density and high density can be, at least larger network density. semi-quantitatively understood based on what has been said before, with one additional factor: at low densities the conducting networks all have at least some semiconducting nanotube networks (NTs), and at higher densities pathways with allmetallic NTs are feasible. This can be described18 with a model that includes both semiconducting and metallic pathways. A significant body of work exists on the elaboration on these notions (and some offering detailed fits to theories), all of which are based on the inherent randomness, and temperature driven transport across random barriers. The frequency dependent conductivity () supports this general picture. () can be described by an expression () = A(i) (3) used in general to describe the conduction process in the presence of randomness19. What is important for the context of applications is that nanotube networks can serve as a semiconducting channel (essential for transistor operation) and also as a ”metallic” interconnect.
0905-DD06-05.5
Networks at different densities lead to different sheet conductances and optical transmittances. What is important to applications is the sheet resistance, given by R sq = 1 dc .d for a film with thickness d. The sheet conductance /sq is the inverse of the sheet resistance. Due to the one-dimensional nature of the wires, the network is highly transparent in the infrared and also in the visible spectral range. Early experiments21 gave evidence that the reflectivity is low in the visible, and thus the transmission is mainly determined by the absorption in the network. The equation below describes the optical transmittance assuming that the film thickness is much less than the wavelength and also assuming that the imaginary part of the conductivity can be neglected20
T=
1 1 = 2 2 (1+ ac d) 2 1+ 2 ac c R sq dc
(4)
Here ac is the conductivity at optical frequencies, and c the speed of light. It has been shown21 that, in contrast to the dc conductivity, the ac conductivity in the visible spectral range is largely proportional to the density of the network, but is not, or is only very weakly influenced by inter-tube resistances, by the NT length, or by doping. Thus higher “quality “ networks – pure and well-dispersed nanotubes for example – will lead to higher conductivity but the same transparency for the same density. Figure 4 illustrates the transparency-sheet conductance Figure 4. Sheet resistance versus optical relation, achieved to date. The full line is Equation 4 transmittance of carbon nanotube networks of with the dc conductivity dc=1600S/cm. Significant varying densities. improvement is expected through nanotube purification, better dispersion of the NTs, and reducing the inter-tube resistance. The tubes are p-doped in air due to the O-covering the tubes. The tubes can be chemically modified, leading to changes of the electronic structure and consequently the transport properties. Such modifications are usually performed by dissolving the NTs in a solution and performing the fictionalization in a wet chemistry environment22. Both further p-doping and ndoping can be achieved by covering the network with electron withdrawing and electron donating species. This has been demonstrated with the chemicals NH3, and NO2 and detailed analysis has been performed of the case of NH3 in water23. Hole doping of the NTs was shown to occur by treatment with electron withdrawing molecules, such as I, Br, K, while electron donating species such as PEI21 lead to an N-type material. Experiments using transistors are able to distinguish between charge transfer to the tubes, and the effects of scattering potential (caused by the dopant) in reducing the mobility. Few experiments24,25 have been analyzed using such notions, much further work remains to be done in this area.
0905-DD06-05.6
APPLICATIONS Application opportunities that are based on highly transparent and conducting films or layers include energy (solar cells, solid state lighting), electronics (passive and active matrix displays, smart windows), appliances (defrosting windows, touch screen) and security (RFID tags, electromagnetic shielding and sensors). The technology can be divided into several areas: transparent conducting layers, transparent/flexible electronic devices and functionalized networks for optoelectronics, photovoltaics and sensors. Flexible and optically transparent conducting layer The current technology is based on high temperature deposition of a brittle, inflexible material, ITO1,2. Measurements on NT network layers, fabricated at UCLA are already close to the parameters of commercial ITO (Figure 5) with significant advances anticipated through materials optimization. In addition, doping and density tailoring can lead to highly application-specific networks. The green squares on Figure 5 represent Figure 5 Sheet resistance of various materials at 80% the current state of the art. Significant transparency at 550nm. The “Single NT limit” refers to a of aligned, infinitely long nanotubes. Some application improvements are expected through network barriers are indicated by arrows. using more purified nanotubes, better dispersion of the tubes and reducing the NT-NT resistances. The ultimate “Single NT limit” refers to a network of aligned, infinitely long tubes. Conducting fabrics. Carbon NT networks deposited on fabric26 lead to highly conducting, wearable materials. The conducting characteristics well exceed those of other approaches such as transparent polymer coatings that address the issue of electrically conducting fabrics. In addition, the network is light-weight, mechanically robust and resistant to environmental factors. An Figure 6 Conducting fabric, arrived at by spraying carbon nanotubes onto a fatigue. Ref 26.
0905-DD06-05.7
illustration of an electrically conducting fatigue is shown on Figure 6. “Nanotube” networks for organic lighting, photovoltaics and smart windows Nanotube networks cab be combined with other, functional layers, layers with electrochromic properties, layers where charge separation occurs upon the application of a dc voltage, or visible light. Such geometrics support opto-electronic or photovoltaic devices. Several of these concepts have been demonstrated to date. A “smart window” configuration with two nanotube27 network electrodes is displayed on Figure 7. For solar cell and OLED applications the work function of the materials is important. Calculations and analysis28,29 of field emission experiments indicate that the work function is approximately 5 eV comparable to ITO. This sets the stage of exploration of such networks as replacement of ITO, and progress has been made in this area.
Figure 7. (a) Smart window configuration using two nanotube networks as transparent electrodes. (b) Change of the transmission with voltage applied between the electrodes. Ref 27.
Flexible/transparent electronic devices
Figure 8. (a) Layout of the FET with carbon nanotube gate and conducting channel. (b) Dependence of the source-drain current on the gate voltage in the low voltage linear Isd(Vsd) region. Ref 34.
Current flexible electronic devices, such as FETs, field effect transistors are mainly based on polymeric materials. The overall performance of devices is limited by the fundamental conducting properties of polymers, and some parameters such as the mobility are not, and probably will never be, comparable to conventional silicon based devices. In spite of significant efforts, the mobility of the devices, while improving still does not exceed 1 cm2/Vsec, limiting the application potential of the devices. Transistors have been fabricated using lowdensity NT network as the conducting channel. The concept of such transistors emerged a few years ago30, and has been realized31,32 by using CVD grown tubes on silicon substrate. Flexible transistors33 together with transparent/flexible transistors34 have also been fabricated as of today. Figure 8 displays such a transistor configuration where both the conducting channel and the gate
0905-DD06-05.8
were fabricated using networks of different densities. The transistor is highly transparent, with the transmittance determined by the density of the NT network that serves as the gate electrode. The mobility of such transistors fabricated using a NT network (blue cross in Figure 9) exceed well all current solutions for high mobility, flexible transistors. Other parameters such as the onoff ratio and hysteresis have still to be improved, although significant progress has been recently achieved35 in the area. Functionalized networks for optoelectronics Devices with carbon nanotube networks can be combined with small molecules and polymers with carbon functionalities. As an example a light sensitive small molecule such as porphyrin can be combined with the network36a leading to a mechanism where light is directly converted into an electronic signal – the fundamental Figure 9. The evolution of the FET device mobility over the last decades. The mobility of carbon nanotube FETs is indicated by the blue cross. Some application barriers are also indicated.
.
process in many optoelectronic and photovoltaic devices. The structure is illustrated on Figure 10, together with the response obtained upon illumination with a light. Examination of the wavelength dependence of the response gives Figure 10. Detection of light using a porphyrinevidence that light-induced electronic transitions of nanotube network complex. Ref 36a. the porphyrin molecules is responsible for the change of the transistor characteristics. Similar observations were made by using the light sensitive polymer PmPv36b. CONCLUSIONS The ultimate usefulness of the material lies in its competitive position with respect to other materials used, and therefore a few comments on this aspect of the work are in order. First, both spraying and printing, as explored to date allow up-scaling of the fabrication of films. Second, the performance characteristics are similar, or exceed that of other materials such as a polymer based transparent conductor, offered by AGFA37, and the performance of ITO on a plastic surface. ITO is readily used in a variety of applications, and we conclude that carbon nanotube films, described in this paper have the technical characteristics that allow applications
0905-DD06-05.9
in areas where flexible coatings are required, such as transparent EM shielding, smart windows, touch screen displays, solar cells, OLEDs and flat panel displays. The third, and quite possibly the most important factor is the cost of the material and of the process being used. The cost of the fabrication process is expected to be less that that for ITO for which high temperatures and vacuum deposition is required. With the abundance of carbon, the price of carbon NTs is expected to decrease with further commercialization, in contrast to the steady increase of the price of ITO. By considering all the attributes discussed above, it is evident that carbon NT films will find their application in a variety of areas that require flexible, transparent and conducting coatings in a patterned or non-patterned form. ACKNOWLEDGEMENTS I appreciate the contributions of my students and postdoctoral associates, L. Hu, Y. Zhou, D. Hecht, , E. Artukovic and M. Briman. This work was supported by NSF grant 0404029. REFERENCES 1. 2. 3. 4. 5. 5a. 6. 7. 8. 9a. 9b.
B.G. Lewis and D.C. Paine, “Applications and Processing of Transparent Conducting Oxides,” MRS Bull. 25 (8), 22 (2000). R.G. Gordon, “Criteria for Choosing Transparent Conductors,” MRS Bull. 25 (8), 52 (2000). T. Durkop, S.A. Getty, Enrique Cobas, and M.S. Fuhrer, “Extraordinary Mobility in Semiconducting Carbon Nanotubes,” Nano Lett 4, 35 (2004). M.S. Dresselhaus and H. Dai, and Guest Editors, “Carbon Nanotubes: Continued Innovations and Challenges,” MRS Bull. 29 (4), 237 (2004). J.A. Shim, N.W. Kam, and H. Dai, “Polymer functionalization for air-stable n-type carbon nanotube field-effect transistors,” J Am Chem Soc. 123, 11512 (2001). C. Zhou, J. Kong, E. Yenilmez, and H. Dai, “Modulated Chemical Doping of Individual Carbon Nanotubes,” Science 290, 1552 (2000). G. de la Torre, W. Blau, and T. Torres, “A survey on the functionalization of single walled nanotubes. The chemical attachment of phthalocyanine moieties,” Nanotechnologies 14, 765 (2003). M. Stadermann, S.J. Papadakis, M.R. Falvo, J. Novak, E. Snow, Q. Fu, J. Liu, Y. Fridman, J.J. Boland, R. Superfine, and S. Washburn, “Nanoscale study of conduction through carbon nanotube networks,” Phys. Rev. B69, 201402(R) (2004). Y. Zhou, A. Gaur, S.-H.Hur, C. Kocabas, M. A. Meitl, M. Shim, J. A Rogers, “pChannel, n-Channel Thin Film Transistors and p-n Diodes Based on Single Wall Carbon Nanotube Networks”, Nano Lett. 4, 2031 (2004) N.P. Armitage, J-C.P. Gabriel, and G. Gruner, “Langmuir-Blodgett nanotube films,” Appl. Phys. Lett. 2003 L. Hu, D.S. Hecht, and G. Grüner, “Percolation in Transparent and Conducting Carbon Nanotube Networks,” Nano Lett. 4 (12), 2513 –2517 (2004)
0905-DD06-05.10
10. 11. 12. 13.
Y Zhou, L. Hu, and G. Grüner, “A Printing Method for carbon nanotube thin films,” (to be published) M. Kaempgen, G.S. Duesberg, and S. Roth “Transparent Carbon Nanotube Coatings,” Appl. Surf. Sci., 252, 425 (2005) The conductivity is calculated by assuming that nanotubes fully fill the space, this assumption is clearly not valid, and leads to an underestimation of the conductivity. M.A. Meitl, Y. Zhou, A. Gaur, S. Jeon, M.L. Usrey, M.S. Strano, and J.A. Rogers, Nano Lett. 4, 1643 (2004)
14.
Hongjie Dai, "Growth and Characterization of Carbon Nanotubes", book chapter in "Topics in Applied Physics", Vol. 80, "Carbon Nanotubes", Edited by M. Dresselhaus, G. Dresselhaus and P. Avouris, Springer Verlag (2000).
15.
Z.C. Wu, Z.H. Chen, X. Du, et al. “Transparent, conductive carbon nanotube films,” Science, 305, 1273 (2004) G. Stauffer, “Introduction to Percolation Theory,” Taylor & Francis: London, 1985 J. Vavro, J.M. Kikkawa, and J.E. Fischer, “Metal-Insulator Transition in Doped Single Wall Carbon Nanotubes,” Physical Review B71, 155410 (2005) E. Bakyarova, M.E. Itkis, N. Cabrera, B. Zhao, A. Yu, J. Gao, and R.C. Haddon, “Electronic Properties of Single-Walled Carbon Nanotube Networks,” J. Am. Chem. Soc., 127, 5990 (2005) A.B. Kaiser, G. Düsberg, and S. Roth, “Heterogeneous model for conduction in carbon nanotubes,” Physical Review B (Condensed Matter and Materials Physics), 57, 1418 (1998) G. Grüner, “Nonlinear and frequency-dependent transport phenomena in low dimensional conductors,” Physica D: Nonlinear Phenomena, 8 , 1 (1983) M. Dressel, and G.Gruner, “Electrodynamics of Solids,” Cambridge University Press, 2002 B. Ruzicka and L. Degiorgi, “Optical and dc conductivity study of potassium-doped single-walled carbon nanotube films,” Phys. Rev. B61, R2468 (2000) Bokrath et al. “Chemical Doping of Individual semiconducting carbon nanotube ropes,” Phys. Rev B61, 10606 (2000) K. Bradley, J.C. Gabriel, M. Briman, A. Star, G. Gruner, “Charge Transfer from Ammonia Physisorbed on Nanotubes,” Phys. Rev Lett. 91 (21), 218301 (2003) G. P. Siddons, D. Merchin, J. H. Back, J. K. Jeong, M. Shim, “Highly Efficient Gating and Doping of Carbon Nanotubes with Polymer Electrolytes,” Nano Lett. 4, 927931(2004) A. Star, T-R Han, J-C.P. Gabriel, K. Bradley, and G. Grüner, “Interaction of AromaticCompounds with Carbon Nanotubes,” Nano Letters, 3 (10), 1421 –1423 (2003) K. Bradley, M. Briman, A. Star, G. Gruner, “Charge Transfer from Adsorbed Proteins”, Nano Lett. 4, 253-256 (2004) D.S. Hecht, L. Hu, and G. Gruner, “Electronic Properties of Carbon Nanotube/Fabric Composites,” Accepted in Current Applied Physics (2005) L. Hu, D. Li, Y. Zhou, R. Kaner and G. Gruner, “Patterned transport carbon nanotube electrodes for electrochromatic device applications,” (to be published) B. Shan and K. Cho “First-Principles study of Work Functions of Single Wall Carbon Nanotubes” Phys. Rev. Lett. 94 (23), 236602 (2005)
16. 17a. 17b. 18. 19. 20. 21. 22. 23. 24. 25a. 25b. 26. 27. 28.
0905-DD06-05.11
29. 30. 31. 32. 33. 34. 35.
36a. 36b. 37.
N. de Jonge, M. Allioux, M. Doytcheva, M. Kaiser, K.B.K. Teo, R.G. Lacerda and W.I. Milne. "Characterization of the field emission properties of individual thin carbon Nanotubes,” Appl. Phys. Lett. 85, 1607 (2004) G. Gruner, Patent Application J.C.P Gabriel, Mater. Res. Symp. Proc. 762, Q.12.7.1 (2003) E.S. Snow, J. Novak, P.M.Campbell, D. Park, “Random networks of carbon nanotubes as an electronic material,” Appl. Phys. Lett. 82 (13), 2145 (2003) K. Bradley, J-C.P. Gabriel, and G. Grüner, “Flexible Nanotube Electronics,” Nano Letters, 3 (10), 1353 (2003) E. Artukovic, M. Kaempgen, D.S. Hecht, S. Roth, and G. Grüner, “Transparent and Flexible Carbon Nanotube Transistors,” Nano Lett. 5 (4), 757 (2005) S.-H. Hur, M.-H. Yoon, A. Gaur, M. Shim, A. Facchetti, T.J. Marks and J.A. Rogers, “Organic Nanodielectrics for Low Voltage Carbon Nanotube Thin Film Transistors and Complementary Logic Gates,” Journal of the American Chemical Society 127(40), 13808 (2005) D.S. Hecht, R.A. Ramirez, E. Artukovic, M. Briman, K. Chichak, J.F. Stoddart, G. Gruner, “Bio-inspired Detection of Light using Porphyrin Sensitized Carbon Nanotube FETs,” Submitted (2005) A. Star, Y. Lu, K. Bradley, G. Gruner, “Nanotube optoelectronic memory devices.” Nano Lett. 4 1587 (2004). Orgacon™ EL – P3000, AGFA
