Carbon nanotube films and methods of forming films of carbon nanotubes by dispersing in a superacid

Abstract

A novel method of forming thin films of carbon nanotubes (CNTs) is described. In this method, carbon nanotubes are dispersed in a superacid solution and laid down on a substrate to form a conductive and transparent CNT network film. The superacid, in its deprotonated state, is an anion that has a permanent dipole moment. The superacid solution may be a pure superacid or have additional solvent. Preferably, the superacid solution does not contain an oxidizing agent. Novel, highly conductive and transparent CNT network films are also described.

Claims

1. A CNT network film composition, comprising: a substrate, and a CNT network film on the substrate; wherein the CNT network film has a transparency of at least 50%, a sheet resistance of 5000 ohms/square or less, and a G/D ratio based on the average of Raman excitation bands at 532 and 633 wave numbers of at least 15; and wherein the CNT film has a bulk conductivity exceeding 11,000 S/cm.

2. The CNT network film composition of claim 1 having a transmittance of at least 80% measured at 550 nm.

3. The CNT network film composition of claim 2 wherein the film has an area in the range of 1 cm.sup.2 to 1000 cm.sup.2.

4. The CNT network film composition of claim 1 having a G/D ratio based on the average of Raman excitation bands at 532 and 633 wave numbers of 25 or more.

5. The CNT network film composition of claim 1 wherein the film has a transmittance of at least 90% measured at 550 nm.

6. The CNT network film composition of claim 1 wherein the film has an area of at least 1 cm1 cm.

7. The CNT network film composition of claim 1 wherein the film has a thickness of 10 to 45 nm.

8. The CNT network film composition of claim 7 having a G/D ratio based on the average of Raman excitation bands at 532 and 633 wave numbers of at least 25.

9. The CNT network film composition of claim 1 wherein the CNT network film exhibits a sheet resistance in the range of 200 to 3000 ohms/square.

10. The CNT network film composition of claim 1 wherein virtually no band in the Raman spectrum is observed at 267 cm.sup.1.

11. A CNT network film composition, comprising: a substrate, and a CNT network film on the substrate; wherein the CNT network film has a transparency of at least 90% measured at 550 nm, a sheet resistance of 5000 ohms/square or less, and a G/D ratio based on the average of Raman excitation bands at 532 and 633 wave numbers of at least 15; and wherein the CNT film has a bulk conductivity exceeding 11,000 S/cm.

12. The CNT network film composition of claim 11 having a G/D ratio based on the average of Raman excitation bands at 532 and 633 wave numbers of 25 or more.

13. The CNT network film composition of claim 11 wherein the film has a thickness of 10 to 30 nm.

14. The CNT network film composition of claim 11 wherein the film has an area of at least 1 cm1 cm.

15. The CNT network film composition of claim 11 wherein the film has an area in the range of 1 cm.sup.2 to 1000 cm.sup.2.

16. The CNT network film composition of claim 11 wherein the film has a thickness of 10 to 45 nm.

17. The CNT network film composition of claim 11 wherein the CNT network film exhibits a sheet resistance in the range of 200 to 3000 ohms/square.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) FIG. 1 is a plot of transmittance versus sheet resistance for HiPco SWNT network samples prepared according to this invention.

(2) FIG. 2 is an SEM image of as received SWNT-HiPco-SP CNT fibers.

(3) FIG. 3 is an SEM image of a CNT network film of SWNT-HiPco-SP CNT fibers prepared according to this invention.

(4) FIG. 4 is a Raman spectrum of SWNT-HiPco-SP obtained at 785 nm excitation.

(5) FIG. 5 is a Raman spectrum of as received SWNT-HiPco at 532 nm.

(6) FIG. 6 is a Raman spectrum of SWNT-HiPco at 532 nm after treatment with chlorosulfonic acid.

(7) FIG. 7 is a Raman spectrum of as received SWNT-HiPco at 633 nm before treatment with chlorosulfonic acid

(8) FIG. 8 is a Raman spectrum of SWNT-HiPco at 633 nm after treatment with chlorosulfonic acid.

(9) FIG. 9 is a Raman spectrum of SWNT-HiPco at 532 nm after treatment by prior art dispersion method (Comparative Example 3)

(10) FIG. 10 is a Raman spectrum of SWNT-HiPco at 633 nm after treatment by prior art dispersion method (Comparative Example 3)

(11) FIG. 11 is a derivative of weight loss curve from TGA of SWNT-HiPco that was dispersed in chlorosulfonic acid.

(12) FIG. 12 is a derivative of weight loss curve from TGA of SWNT-HiPco Control that was sonicated in aqueous surfactant solution (Comparative Example 3).

(13) FIG. 13 is a Dark-field STEM image of CNT film prepared by prior art SDBS surfactant method.

(14) FIG. 14 is a Dark-field STEM image of CNT film prepared by the methods of this invention.

(15) FIG. 15 is a Bright-field TEM image of CNT film prepared by prior art SDBS surfactant method.

(16) FIG. 16 is a Bright-field TEM image of CNT film prepared by the methods of this invention.

DETAILED DESCRIPTION OF THE INVENTION

(17) The morphology and structure of a CNT network are described by several parameters. CNTs can be described by length and diameter. Prior art dispersion methods that use sonication, oxidation, and/or covalent modification tend to damage the CNT structure. The dispersion process employed in this invention, superacids, does not change the CNT structure. Analysis of a network film described by this invention will show that the sidewall perfection and the length of the CNTs is high. The length of the individual CNTs within the network can be determined by TEM analysis. The mass fractions of CNTs in a sample can be measured by SEM, TEM, or other appropriate analytical technique.

(18) CNTs can also be described by bundle size. Rebundling is prevented in this invention due to the CNT-acid charge transfer complex, which established a large diffuse electric double layer. As the liquid, such as superacid or superacid/solvent mixture, is removed during the initial CNT network formation, the diffuse portions of the double layers interpenetrate, giving rise to a repulsive interaction between CNTs. The distance over which this overlap occurs depends on the thickness of the double layer and the surface potential. A high dielectric constant maintains the debundled network structure. Analysis by SEM, TEM, AFM, or Raman spectroscopy provides characterization of the bundle size.

(19) A third level of structure is described by the bundle characteristics. In general, CNTs are composed of a mixture of semiconducting and metallic tubes. Bundles will contain a mixture of both according to the statistics of the system. It is possible to make bundles have a more metallic-like character through these use of dopant; generally more metallic-like behavior is desirable for electrical conductivity. Dopants, such as p-dopants, are often added in a second step. In this invention, the dopants are the acids and part of the dispersing process. Evidence of p-doping can be determined spectroscopically, for example, by examining the optical absorbance spectrum. Depletion of filled states by an electron acceptor results in bleaching of the van Hove transitions, and evidence of p-doping by the subject coating. Evidence for p-doping is observed by measuring the spectrum, and then comparing the spectrum after treatment with an n-dopant like hydrazine, which will show evidence of any bleached transitions.

(20) The final level of structure is described by the organization and interactions between bundles. The interactions between bundles provide the cohesion within the film. By analogy with polymer films, the interbundle interactions are the crosslinks between bundles. Preferred structures for good film formation and high electrical conductivity have strong interbundle interactions at the junctions between bundles. In this invention, the interbundle interactions can be enhanced by using a non-solvent to coagulate the structure. The interbundle interactions are controlled by controlling the thermodynamics and the kinetics of the phase separation process that occurs upon addition of a non-solvent to the dispersion or to the initial CNT network. The non-solvent should be miscible with the acid, but a non-solvent for the CNTs. The non-solvent is chosen such that the interaction parameter between non-solvent/CNT is much weaker than between either CNT/acid or acid/non-solvent. This promotes rapid aggregation in the debundled network state.

(21) In the prior art, CNT superacids dispersions have been used with water or dilute sulfuric acid as coagulating agents. We found that the use of water as a non-solvent was ineffective in producing cohesive, debundled CNT films. This is due to the strong interaction between water and CNTs. Water and other molecules that strongly interact with CNTs, adsorb to the surface of the CNTs and decrease the interaction strength between CNTs. To our surprise, the addition of poorly interacting solvents such as diethyl ether to initial CNT networks yielded cohesive, debundled CNT films. The strong, interbundle interactions in the films produced in this invention are exemplified by the shrinkage in the network structure that occurs upon addition to the appropriate solvent. The shrinkage results in higher density networks with outstanding electrical and mechanical properties, as described in the Examples.

(22) The inventive methods preferably avoid the use of water and, preferably, use a nonsolvent to coagulate the CNTs (this step can occur prior to, simultaneously with, or after dispersing the CNTs on a substrate. Non-solvents which are not suited include non-solvents with high capacity for hydrogen bonding, such as water and alcohols. These can be determined from the hydrogen-bond component of their Hansen solubility parameter. Simple ethers (diethyl ether, dipropyl ether, dibutyl ether) can be suitable, but ethers with multiple oxygens such as diglyme have too much hydrogen bonding. One criterion for determining the suitability, or lack thereof, for a potential non-solvent is to examine its value of the hydrogen-bonding component of the Hansen solubility parameter. We have found that .sub.h should have a value below 15 (MPa).sup.1/2 for the for the material to be a potential non-solvent for this invention.

(23) The reactivity of CNTs towards acids depends on the diameter of the nanotube. The purity and the initial morphology also play a role in reactivity. In its broader aspects, the invention applies generally to all CNTs. The CNTs that may be dispersed by this method include straight, bent, or branched single wall carbon nanotubes (SWNTs), straight, bent, or branched multi wall carbon nanotubes (MWNTs), straight, branched, or bent few wall nanotubes (FWNTs), and their mixtures. CNTs with average diameter from 0.5 nm to 60 nm may be used. Given the differences in reactivity as a function of sidewall curvature, selection of an appropriate acid depends on the CNT feedstock. CNTs may also include ultra-long CNTs (>1 cm) and ultra-pure CNTs. The class of peapods, substitutionally doped CNTs, and filled CNTs may be used.

(24) A Bronsted superacid is defined as an acid stronger than 100% sulfuric acid. A Lewis superacid is defined as an acid stronger than anhydrous aluminum trichloride. Suitable acids include Bronsted superacids such as halosulfuric acids (for example, chlorosulfonic acid and fluorosulfuric acid) and perfluoroalkanesulfonic acids (e.g., trifluoromethanesulfonic acid), Lewis superacids such as antimony pentafluoride, arsenic pentafluoride, tantalum pentafluoride, niobium pentafluoride, conjugate Bronsted-Lewis superacids such as tetra(hydrogen sulfato)boric acid sulfuric acid, and carborane acids such as CHB.sub.11R.sub.5Cl.sub.6, CH B.sub.11R.sub.5Br.sub.6, and CH B.sub.11R.sub.5I.sub.6, where R is H or CH3. Preferred acids include chlorosulfonic acid and trifluoromethanesulfonic acid. The invention also includes mixtures of any of these superacids in any permutation. Strong oxidizing acid combinations that degrade CNTs such as nitric acid/sulfuric acid should be avoided. Oleum is also not a suitable acid combination for this invention.

(25) Without being held to a specific theory, we theorize that there are two important factors that determine the suitability of an acid or acid combination for this invention. One is that the acid must not strongly oxidize the nanotubes, to prevent the introduction of defects, which lowers the intrinsic conductivity of the individual nanotubes. Secondly, the acid should have a permanent dipole moment. For protic acids, determination of the dipole moment should be when in the deprotonated state. When the acid solvates the individual nanotubes, the permanent dipole moment indicates that the acid will have a preferred orientation relative to the nanotube. In deprotonated chlorosulfonic acid, the dipole lies along the SCl bond. If the preferred orientation of CSA is, for example, with this dipole normal to the nanotubes surface, with the Cl atom away from the nanotube, when two solvated nanotubes come close together, there will be a strong dipole repulsion helping keep the nanotubes separated. This interaction will be absent with an acid such as sulfuric acid or in an acid mixture such as oleum.

(26) Again, not being bound by any specific theory, we believe that the desired exhibited by compounds suitable for the present invention can in part be expressed in terms of relative dispersion, polar, and hydrogen bonding components of Hansen type solubility parameters, and relative octanol-water partion coefficients (Log P). For example, suitable compounds such as chlorosulfonic acid and triflic acid have overall Hansen solubility parameters of 30-40, with polar fractional components of 20-30, compared to unsuitable compounds such as sulfuric acid, which has a much higher solubility parameter 70, with a higher value of the polar component: 60. These solubility parameter manifest themselves in such properties as surface tension, where chlorosulfonic acid and triflic acid may be in the range of 100-200 dyne-cm, while sulfuric acid is only 30 dyne-cm, and Log P where CSA and TFA have estimated values in the range of positive (+) 0.1-0.6, more soluble in the organic phase, while sulfuric acid is 10 more soluble in the aqueous phase Log P=negative (1). In preferred embodiments of the present invention, the superacid preferably has a Hansen solubility parameter of 40 or less, in some embodiments in the range of 30-40.

(27) The CNT charge transfer complex must be dissolved in a dispersion medium, such as a solvent. Suitable solvents for the CNT/acid charge transfer complex include the neat superacid itself or mixtures of superacid. In some preferred embodiments, the superacid is chosen from the family of volatile superacids such as chlorosulfonic acid trifluoromethanesulfonic acid, or antimony pentafluoride. Preferred volatile superacids have a vapor pressure of at least 0.1 atm at 100 C., more preferably at least 0.5 atm at 80 C.

(28) Alternatively the CNT/acid charge transfer complex may be dissolved in solvents including low-nucleophilicity solvents such as liquid SO.sub.2, SO.sub.2ClF, and SO.sub.2F.sub.2; typical Friedel-Crafts solvents such as diethyl ether, nitromethane, and tetrahydrofuran; or deactivated aromatic solvents such as nitrobenzene. The superacid/solvent combination must have sufficiently high dielectric constant to maintain the large diffuse electric double layer. In preferred embodiments, the solvent has low nucleophilicity and high dielectric constant. Preferably, the dielectric constant is greater than 35, more preferably greater than 60.

(29) Dispersions are prepared by mixing the CNTs with a superacid, and optionally diluting with solvent, under an inert environment at preferably at temperatures ranging from 78 C. to 120 C., depending on the reactivity of the CNT. At higher temperatures the screening length is expected to increase due to an increase in the ionic strength, which will decrease the effective distance of the repulsive interaction. The temperature stability of the superacid must also be considered. The temperature and time of the mixing process should be controlled to insure that sidewall damage to the CNTs is not incurred. The extent of sidewall damage can be monitored by precipitating the dispersion into a non-solvent and analyzing the CNTs by Raman spectroscopy, NIR spectroscopy, or TGA, as described in Measurement Issues in Single Wall Carbon Nanotubes. NIST Special Publication 960-19.

(30) Additives may be incorporated for improved gloss, durability, or other physical and mechanical properties. Preferably, these additives are avoided to maintain the highest conductivity.

(31) Initial CNT networks can be formed from the dispersion using known methods to produce CNT networks on substrates. For example, the dispersion can be passed through a porous membrane filter to form a CNT mat. The membrane filter must be compatible with the solvent combination. Suitable membrane filters include those made of alumina, glass, PVDF, PTFE, PEEK, and polypropylene.

(32) If volatile superacids are used, films can be formed by direct deposition methods such as spin coating, dip coating, and screen printing. Substrates must be selected that are compatible with acid. Suitable substrates include glass, PTFE, PVDF, polypropylene, Kevlar, PEEK, ceramic, and others that would be recognized as being compatible with the acid.

(33) The formation of cohesive CNT networks can be carried out by adding a non-solvent to the CNT film. Preferred non-solvents are miscible with the superacid and have poor interaction with the CNT. Volatile solvents with boiling point less than 80 C. (at 1 atm) are preferred. Non-solvents can include ethers such as diethyl ether or THF, nitrogen-containing compounds such as nitromethane, or ketones such as acetone and methyl ethyl ketone. The non-solvents preferably do not contain water (they are non-aqueous).

(34) The cohesive, CNT network films, once formed, are strong and can be transferred in a second step, if desired to other substrates such as glass, polyester, silicone, polyethylene, etc using methods known in the art. Examples include, but are not limited to floating, direct pressure transfer, and pad printing.

(35) In some embodiments, the CNT network has good cohesion, average bundle sizes less than 10 nm, and is p-doped. In some preferred embodiments, the sample contains 0.01 wt % or less of S.

(36) In preferred embodiments, water and other competing bases or reducing agents should be avoided during dispersion and film formation, in order to prevent nanotube aggregation. Adventitious moisture in the air does not dramatically cause bundle formation on the time scale of the film formation process, but film formation under nitrogen, argon, or other inert gases is preferable. Decomposition of the CNT cation to permanently oxidized species can be avoided by maintaining temperatures below 120 C. during dispersion of the CNTs and, also, preferably, during the film-forming process. Preferably, no sonication is used in the inventive methods.

(37) The following examples describe this invention in greater detail, and demonstrate its advantages over the prior art. One important feature of the disclosed method is that the network morphology created is unique in that it is composed of long CNTs (greater than 500 nm), organized into very small bundles (preferably 20 nm or less, more preferably 10 nm or less, and still more preferably 5 nm or less), and with strong interbundle cohesion. The inventive process promotes film formation of CNT networks. This is demonstrated by comparing examples using different non-solvents. The unique morphology yields films with outstanding electrical conductivity and low sheet resistance for films thinner than 5 nm. This is demonstrated by measuring the sheet resistance and optical transmission of thin films deposited on polyester substrates. The use of superacid disperses and debundles the CNTs without damaging them, cutting them, or fractionating them. This is demonstrated by comparing the Raman spectra and thermal gravimetric analysis data of SWNTs before treatment and after treatment. To better differentiate the advantages inherent in this invention, an example of a control sample prepared by a standard film formation method of sonication in aqueous surfactant is provided. A second advantage of this invention is that superior electrical properties of the CNT film can be achieved at very low loadings of CNTs.

EXAMPLES

(38) Sample Characterization

(39) The sheet resistance of CNTs films was measured by the Jandel Universal Probe and Corrware software. The Universal Probe consists of four 100-um tip radius pins, spaced 1 mm apart, aligned linearly. The films are placed in contact with the pins under 25 g of spring pressure. Current was increased from 0.1 mA to 0.1 mA at a rate of 0.01 mA/sec. All samples exhibited ohmic response. The sheet resistance was determined by: Rs=4.532(V/I)GFC, where V/I is the resistance, and the correction factor (GFC) is based upon film thickness vs. pin spacing. Given the thickness of these films, the correction factor is typically 0.99. Alternatively, the sheet resistance can be measured by painting two parallel electrodes created with silver paint onto the CNT film along opposite edges of 11 square. A digital multimeter can be used to determine the sheet resistance.

(40) Near infrared (NIR) and visible spectra of nanotube films were acquired using a Varian Cary 5000 UV/Vis/NIR spectrophotometer over the range of 3000 nm (3333 cm.sup.1) to 400 nm. The spectra were acquired at a scan rate of 600 nm/minute with a spectral band width of 2 nm with an averaging time of 0.1 seconds. In general, the samples were run against a background of air. A blank substrate, not containing CNTs, as also run against air. The percent transmission was determined at 550 nm by comparing the ratio of the transmission from the sample to the ratio of the transmission from the blank.

(41) The Raman spectra of CNT films were acquired on a Horiba-JY Aramis Raman Confocal Microspectrometer operated in the backscattering mode at 633 nm excitation. The sample was brought into optical focus with a 20 microscope objective. The Raman signal was optimized using the Real Time Display feature with further adjustment of the Z-direction distance. To insure that no sample heating effects were observed, the laser power was minimized to 5 mW at the sample. The stability of the position of the G-band can be used to analyze for heating effects. Data acquisition employed a 1-second acquisition time and 4 cm.sup.1 resolution; 16 spectra were signal averaged. Spectra were acquired at several places on the samples to obtain an overall average of the intensity of specific Raman bands. The G/D ratio was to determined by determining the integrated intensity for the D mode (1322 cm-1 at 532 nm excitation) and the G mode (1580 cm-1 at 532 nm excitation). We also compared the ratio of the maximum intensity and found the result to be the same.

(42) Thermogravimetric analysis was carried out on a Perkin Elmer Pyris 1 TGA. The temperature is calibrated using the Curie point transition of Perkalloy and Alumel, the balance is calibrated using a 100 mg weight and the furnace is calibrated from 50-900 C. by the analysis software. Approximately 7 mg was loaded into Pt pan. The heating rate was 10 C. per min in air.

(43) Film Formation

Example 1. Formation of Cohesive, Debundled CNT FilmsMethod A

(44) Purified-grade SWNTs prepared by the HiPco process (SWNT-HiPco) were obtained from Carbon Nanotechnologies, Inc. The SWNTs are reported to have an ash content less than 15 wt %. A dispersion was prepared by adding 10 mg SWNT-HiPco to 50 mL of chlorosulfonic acid and stirring (using a bar magnet over a stir plate) at room temperature for 12 hours. A known volume of the dispersion was passed through a 0.2 m Anodisc membrane, with vacuum assistance, to create the initial CNT network of desired thickness. In some cases, the dispersion was diluted with more acid to allow preparation of very thin films. Approximately, 50 mL of diethyl ether was added as a non-solvent, while maintaining vacuum. The color of the SWNT network changed from yellow-brown to black upon washing. The film may be observed to release at the interface between the film and the Anodisc, which arises due to shrinkage caused from good non-solvent choice and the formation of cohesive interbundle interactions. The SWNT network film was released from the membrane filter by dipping the coated membrane filter into a water bath. The SWNT network film floated to the surface of the water and was picked up with a substrate such as polyester or glass. Visual inspection of a film shows that the film is homogeneous.

(45) The transmission of CNT coatings prepared by Example 1 and Method A were determined at 550 nm. The optical transmission is directly related to the volume fraction of nanotubes in the coating and the thickness. The transmission versus sheet resistance of these samples is shown in FIG. 1. As expected based on a percolation model, the sheet resistance decreases with increased transmission. Near the percolation threshold, it is expected that the sheet resistance will dramatically increase. The volume fraction at which percolation is achieved depends on the aspect ratio of the nanotubes. Since the transmission versus sheet resistance is fairly linear in the range tested, the loading is evidently well above the percolation threshold, consistent with the presence of long nanotubes that were not shortened by the dispersion process.

(46) It should be noted that the results of further examples will be compared to the data presented here. The results given in FIG. 1 can be roughly represented as lying along a straight line. Films which have superior performance will lie above or to the left of this line, while films with poorer performance will lie below or to the right of this line.

Example 2. Formation of Cohesive, Debundled CNT FilmsMethod B

(47) A dispersion was prepared, as described in Example 1. A known volume of the dispersion was passed through a PVDF membrane filter, with vacuum assistance, and dried by passing a stream of dry air over the membrane to create the initial film. Diethyl ether was added as a non-solvent, while maintaining vacuum. The dried SWNT network film was placed in contact with a polymeric substrate such as Mylar and transferred to the Mylar by applying pressure. Visual inspection of a film shows that the film is homogeneous.

Example 3. Formation of Cohesive, Debundled CNT FilmsMethod C

(48) Direct deposition methods such as spin-coating and dip-coating may be used to prepare thin films on compatible substrates such as glass. As an example, thin films were prepared by spin coating. Purified-grade SWNTs prepared by the HiPco process (SWNT-HiPco) were obtained from Carbon Nanotechnologies, Inc. The SWNTs are reported to have an ash content less than 15 wt %. A dispersion was prepared by adding 50 mg SWNT-HiPco to 10 g of chlorosulfonic acid and stirring at room temperature for 120 h. The resulting viscous dispersion was applied to a glass substrate by spin coating at speeds from 2,500 to 5000 rpm for 2 minutes under Ar atmosphere. After drying, the coated glass slides were subsequently washed with diethyl ether. The sheet resistance of these samples was 176 /square and 110 /square, respectively, and the transmission was greater than 60% at 550 nm. Spin-coating uses less solvent than other processing methods and produces optical quality films with low RMS roughness. Visual inspection of a film shows that the film is homogeneous.

Comparative Example 1. Use of Water as the Non-Solvent

(49) Smalley et al. have described neat superacid systems for producing macroscopic fibers (U.S. Published Patent Application No. 2003/0170166) and alewive structures (U.S. Published Patent Application No. 2003/0133865). In the latter publication, water was used to precipitate aggregates of aligned SWNTs. A dispersion was prepared, as described in Example 1. A known volume of the dispersion was passed through a 0.2 m Anodisc membrane to create the initial CNT network. 50 mL water was added as the non-solvent, while maintaining vacuum. Upon immersion into water to float the sample, the films shredded and tore, rather than releasing, indicating poor interbundle cohesion and poor film forming properties. Water cannot be used as a coagulant to prepare cohesive, debundled CNT films.

Comparative Example 2. Use of Oleum

(50) A dispersion was prepared, as described in Example 1, except substituting oleum for chlorosulfonic acid. A known volume of the dispersion was passed through a 0.2 m Anodisc membrane, with vacuum assistance, to create the initial CNT network of desired thickness. Approximately, 50 mL of diethyl ether was added as a non-solvent, while maintaining vacuum. The SWNT network film was released from the membrane filter by dipping the coated membrane filter into a water bath. The SWNT network film floated to the surface of the water and was picked up with a substrate such as polyester or glass.

(51) Visual inspection of a film shows that the film contains black specks. The black specks arise due to phase separation of the CNTs that occurred during addition of the non-solvent. Cohesive, debundled networks films cannot be prepared from oleum.

(52) The sheet resistance of films prepared from oleum were higher then films prepared by Example 1. For example, a film prepared by the method Comparative Example 2 had a sheet resistance of 592 /square and a percent transmission of 80.1%; as observed by comparison with FIG. 1, this value is to the right of the line, exhibiting poorer performance.

Example 3. Use of Different Non-Solvents

(53) Films were prepared as in Example 1 with Method A, using different non-solvents, for diethyl ether

(54) TABLE-US-00001 Solvent Result Ether Excellent film THF Good film Acetone Good film Methanol Poor film Ethanol Poor film Isopropanol Poor film Diglyme Poor film Chloroform Poor film Nitromethane Excellent film Nitrobenzene Poor film

(55) The suitability of a solvent for this invention is determined by multiple factors, including among others its interaction with the CNT, interaction with acid, rate of evaporation, and hydrophobicity.

(56) Unique Morphology of Films of this Invention

Example 4. SEM and Raman of Super-Purified HiPco SWNT

(57) The morphology of resulting CNT networks can be characterized by a combination of SEM and Raman spectroscopy. Super-purified grade nanotubes (SWNT-HiPco-SP) were chosen to demonstrate the high efficiency of this method for producing debundled CNT networks, even from high purity CNT sources. High purity nanotubes containing low levels of residual amorphous carbon often present the greatest challenge from a standpoint of debundling and dispersion.

(58) Super-purified grade SWNTs prepared by the HiPco process (SWNT-HiPco-SP) were obtained from Carbon Nanotechnologies Inc. The SWNTs are reported to have an ash content less than 5 wt %. A dispersion was prepared by adding 0.3 mg SWNT-HiPco-SP to 140 mL of chlorosulfonic acid and stirring for 3 h at room temperature and 48 h at 80 C. Coatings were deposited on a polyester substrate using Method A of Example 1.

(59) An SEM image of the SWNT-HiPco-SP before dispersion is shown in FIG. 2. An SEM image of a 100 nm thick CNT network prepared from chlorosulfonic acid dispersion is shown in FIG. 3. The powder is composed of dense aggregates composed of entangled super-ropes. After dispersion and film formation, SEM shows that both the aggregate and super-rope structures have been disrupted. The sample is composed of <10 nm bundles. Large quantities of spherical-like impurities are present in both samples. These can be removed by centrifugation of the dispersion prior to film formation.

(60) The radial breathing modes observed in the Raman spectrum are another indicator of the degree of debundling. The electronic dispersion of an individual SWNT can be altered by aggregation. Development of electronic dispersion orthogonal to the usual 1D dispersion can cause bands to go out of resonance or go into resonance. For example, bands at 267 cm.sup.1 and 204 cm.sup.1 arising from (10,2) and (13,2) type nanotubes come into resonance at 785 nm when the nanotubes are bundled. The Raman spectrum of the coating prepared from highly purified HiPco SWNT is shown in FIG. 4. Virtually no band is observed at 267 cm.sup.1, consistent with a highly debundled SWNT sample.

(61) Structure of CNTs in Films Produced by this Invention

Example 5. Treatment of SWNT-HiPco with Chlorosulfonic Acid

(62) Purified-grade SWNTs prepared by the HiPco process (SWNT-HiPco) were obtained from Carbon Nanotechnologies, Inc. The SWNTs are reported to have an ash content less than 15 wt %. A dispersion was prepared by adding 10 mg SWNT-HiPco to 50 mL of chlorosulfonic acid and stirring at room temperature for 12 hours. The treated SWNTs were collected by passing the dispersion through a 0.2 m Anodisc membrane and then washing two times with 150 mL diethyl ether. The powder was dried for 12 hours at 110 C. and then for 24 hours in vacuum at 160 C. The sample was subsequently characterized by Raman spectroscopy and thermal gravimetric analysis.

Comparative Example 3. Treatment of SWNT-HiPco by Sonication (Control)

(63) Purified-grade SWNTs prepared by the HiPco process were obtained from Carbon Nanotechnologies, Inc. The SWNTs are reported to have an ash content less than 15 wt %. A dispersion was prepared by adding 50 mg SWNT-HiPco to 50 mL of 0.62 wt % sodium dodecylbenzene sulfonate in water and sonicating with a tiphorn (20 kHz and 225 Watts) for 30 minutes. The SWNTs were collected by passing the dispersion through a 0.2 m Anodisc membrane and then washing with copious water until no surfactant bubbles were observed. The powder was dried for 12 hours at 110 C. and then for 24 hours in vacuum at 160 C. The sample was subsequently characterized by Raman spectroscopy and thermal gravimetric analysis.

(64) The Raman spectra of SWNT-HiPco before and after film preparation as described in Example 5 are shown in FIGS. 5-8. The Raman spectra of SWNT-HiPco after film preparation as described in Comparative Example 3 are shown in FIGS. 9 and 10. FIGS. 5, 6 and 9 are Raman spectra at 532 nm excitation. FIGS. 7, 8 and 10 are Raman spectra at 633 nm excitation. Raman analysis is a useful indicator of the relative purity of the nanotubes. The relative ratio of the integrated intensity of the G mode near 1600 cm.sup.1 to the D mode near 1325 cm.sup.1 decreases with decreasing purity. Damage to CNTs is accompanied by an increase in the intensity of the D mode. The integrated intensity of the D and G bands for the samples at different excitation wavelength are shown in Table 1.

(65) As shown in the Figures and the Table, the SWNT-HiPco before film preparation exhibit a relatively small D band associated with the purity of the as-received sample. The ratio of the G/D mode intensities is approximately 24, depending on the laser excitation wavelength. After treatment, the ratio of the G/D mode intensity is 23. More scatter is observed as a function of laser excitation wavelength, but this is within experimental error of the method. The results suggest that the treatment and process of this invention do not alter the chemical structure of the CNTs. The difference in the G/D ratio before and after treatment is much less than 30%.

(66) The prior art dispersion treatment (e.g. Comparative Example 3) shows a significant decrease in the G/D. The average is G/D is 20. Comparison of the spectra at 633 nm excitation shows evidence for the advantage of the subject dispersion method over prior art methods. The contribution of the D band is much greater when sonication is used. The spectra at 532 are more difficult to interpret due to the lineshapes caused by the metallic nanotubes.

(67) TABLE-US-00002 TABLE 1 Results from Raman analysis of G and D modes of SWNT-HiPco before and after treatment. Excitation Intensity Intensity G Sample Wavelength D Band Band Ratio G/D Before 532 42271 917987 21.7 Treatment 633 26994 727593 26.9 After Treatment 532 27749 807388 29.1 by Example 5 633 154716 2653570 17.1 After Treatment 532 75197 2083629 27.7 by Comparative 633 148228 1733828 11.7 Example 3

(68) Thermogravimetric analysis is also an indicator of the level of damage that is present in CNT samples. TGA (air at 10 C./min) of powders before and after treatment both show an onset of decomposition near 420 C. The results suggest that little or nor damage was caused by the dispersion treatment. The first derivative of the weight loss curves for the samples after treatment with chlorosulfonic acid is shown in FIG. 11. For comparison purposes, the first derivative curve for a SWNT-HiPco sample prepared by prior art aqueous dispersion methods (e.g. Comparative Example 3) is shown in FIG. 12. This sample shows an onset of decomposition of 358 C., suggesting that significant damage has occurred to the sample.

(69) FIGS. 13 and 15 show STEM and TEM images for the CNT film prepared as in Comparative Example 3. FIGS. 14 and 16 show STEM and TEM images for films prepared by the dispersion method of this invention. Comparison of these images reveals that the comparative treated film is much denser than that treated with acid, that much thinner fibers/bundles are observed in the films of this invention, and that in the films of this invention, the larger bundles are inter-connected by many thinner fibers or bundles.

Example 6. Effect of Centrifugation on CNT Film Properties

(70) Although centrifugation is not required to obtain debundling, it may be optionally used to improve the purity of CNT sources. Spherical impurities can be removed from rod-like impurities using mild centrifugation. This is likely due to differences in the diffusion coefficients of spheres versus rods.

(71) Raw SWNTs prepared by Pulsed Laser Vaporization (SWNT-PLV) were obtained from Oakridge National Laboratory. A dispersion was prepared by adding 51.5 mg SWNT-PLV to 10 mL chlorosulfonic acid and stirred for 96 h. The resulting viscous dispersion was centrifuged at 4000 rpm for 1 h. The pellet and supernatant were individually collected, passed through a 0.2 m Anodisc membrane filter, washed with acetone, and the powders dried under vacuum at 120 C. for 24 h. The powders were analyzed by Raman spectroscopy and thermal gravimetric analysis. Alternatively, the supernatant was diluted and used to prepare a film on Mylar using Method A of Example 1.

(72) Raman analysis of D and G modes of SWNT-PLV samples are shown in Table 2. Analysis of the G-mode area to D-mode area indicates that the supernatant is of higher purity than the initial sample and the pellet, suggesting that centrifugation is effective at removing carbonaceous impurities from raw PLV tubes.

(73) TABLE-US-00003 TABLE 2 Summary of Raman data to assess purity 532 G Mode D Mode G/D Sample Area Area Ratio PLV-Raw 310107 22531 14 PLV-Supernatant 192113 8748 22 PLV-Pellet 200664 36082 6

(74) The SWNT-PLV supernatant was re-dispersed, and transparent membranes were prepared and transferred to polyester substrates. The sheet resistance was evaluated after drying in air for 48 hours, after drying in vacuum oven for 24 h at 100 C., and again after exposing to air for 48 hours. The transmittance was evaluated and compared to blank substrate. The results are shown in Table 3. The sheet resistance and % transmission of this example are similar to those of the results presented in Example 5.

(75) TABLE-US-00004 TABLE 3 Results obtained for transparent conductive coatings prepared from raw SWNT-PLV supernatant. Sheet Resistance (/) After Air After Vac After Air % T at 550 nm Sample Dry Dry Expose (relative to PE) PLV-Supernatant 460 582 570 90.8

Comparative Example 4. Comparison to Prior Art Acid Dispersion

(76) Mixtures of acid, including superacids, with an oxidizing agent such as nitric acid are often used for purification and light covalent modification of nanotubes. Such dispersions have significantly lower FOM than those of the subject invention. In this example, we prepared a CNT film on Anodisc using the method described in Example 1 with the exception that nitric acid was used in place of the chlorosulfonic acid. Raman spectroscopy showed that the G/D ratio of these films was less than 10, indicating that significant damage had occurred to the sample.

(77) Preferred films of the present invention have a G/D ratio of 15 or greater, preferably 25 or greater. In some embodiments, the inventive films have a G/D ratio similar to that shown in the Figures.

(78) Properties with Different Types of CNTs

Example 8. CNT Films with Varied Types of Nanotubes

(79) The invention is suitable for a wide range of nanotube types and purities. Due to the differences in diameter, the conversion to CNT cations for different nanotube sources requires higher temperatures, longer dispersion times, or stronger acids. The conditions must be optimized for each CNT source. Examples are shown below for several sources; however the examples are not optimized cases, but shown simply to demonstrate the applicability of the method to a wide range of nanotube diameters and types.

(80) SWNT-CVD

(81) Purified, optical electronic grade SWNTs prepared by CVD were obtained from Unidym. Films were prepared according to Example 1 and Method A. Coatings were deposited on PET.

(82) SWNT-Arc:

(83) Purified/Low-functionality grade SWNTs prepared by the Arc Discharge process (SWNT-Arc) were obtained from Carbon Solutions. The SWNTs are reported to consist of 70-90% carbonaceous material and 7-10 wt % metal impurities. A concentrate was prepared by adding 19.5 mg SWNT-Arc to 15 mL of chlorosulfonic acid and stirring for 96 h at room temperature. Prior to use, the concentrate was passed through a plug of glass wool and diluted with chlorosulfonic acid by a factor of three. Alternatively, a coating was deposited on a polyester substrate using Method A of Example 1. A coating was deposited on a polyester substrate using Method B of Example 2.

(84) MWNT-CVD<10 nm:

(85) MWNTs prepared by a CVD process (MWNT-CVD<10 nm) were obtained from Helix Material Solutions. The MWNTs are reported to have an average diameter less than 10 nm, a length from 0.5-40 m, and a purity of greater than 95 wt %. A dispersion was prepared from 0.1 mg of MWNT-CVD<10 nm and 10 mL of ClSO3H and stirred at room temperature for 24 h. Coatings were deposited on a polyester substrate using Method A of Example 1.

(86) MWNT-CVD-35 nm:

(87) MWNTs prepared by a CVD process (MWNT-CVD-35 nm) were obtained from Materials and Electrochemical Research Corporation. The MWNTs are reported to have an average diameter of 35 nm, 30 m length, and purity>90 wt % MWNT. A dispersion was prepared from 0.1 mg of MWNT-CVD-35 nm and 10 mL of ClSO3H and stirred at room temperature for 96 h. The dispersion was passed through a plug of glass wool prior to use. Coatings were deposited on a polyester substrate using Method A of Example 1.

(88) The sheet resistance and transmission for CNT films prepared from different sources is shown in Table 5.

(89) TABLE-US-00005 TABLE 5 Results for different CNT sources Film Formation Sheet Resistance % T at 550 nm Source Method (/) (relative to glass) SWNT- A 1243 98.4 CVD SWNT- A 60 90.9 CVD SWNT- A 20 71.6 CVD SWNT-Arc A 170 70.5 SWNT-Arc B 170 46.7 MWNT- A 1650 79.0 CVD <10 nm MWNT- A 3750 69.92 CVD-35 nm

(90) The results in Table 5 indicate the outstanding utility of the method to produce films. Films with extremely low sheet resistance can be prepared with very high visible transmission. Correlation of the percent transmission to film thickness indicates that SWNT with percent transmission of 90% have film thickness of approximately 10 to 15 nm. The results indicate that the bulk conductivity of films produced by this invention can exceed 11,000 S/cm and even 15,000 S/cm.

(91) Doped Structure

Example 9. Doping of Films Formed with Prior Art Dispersion Methods

(92) It is well known that films formed by prior art dispersion methods require a separate doping step, which serves to enhance the conductivity of the film. HiPco SWNT (CNI, Purified Grade) were dispersed in 0.62 wt % sodium dodecylbenzene sulfonate in D2O by tiphorn sonication. The concentration of nanotubes in the dispersion was 5.5 mg/L. The dispersion (6 g) was passed through a 0.02 m mixed cellulose ester filter and washed with water. The resulting CNT mat was transferred to PET by: placing the CNT mat in contact with a sheet of PET, applying heat (80 C.) and pressure (<10,000 psi) for 15 minutes, removing the mixed cellulose ester from the PET by carefully peeling or swelling with acetone, and then washing the coated PET with acetone.

(93) Samples were then immersed in a container of liquid thionyl chloride for 2 minutes, maintained under moisture-free conditions. Samples were removed from the liquid, washed with methylene chloride, and then dried with a stream of air. The film had 74% transmission at 550 nm. The sheet resistance before doping was measured to be 532/ and after was 133/. Comparison to the data of FIG. 1 shows that even when doped, the prior art films have a much lower FOM than films of this invention

Example 10. Doping Films of this Invention

(94) In this example, two films were prepared as in Example 1, with the exception that after the wash step, one sample was immersed in a container of liquid thionyl chloride for 2 minutes, maintained under moisture-free conditions. The sample was removed from the liquid, washed with methylene chloride, and then dried with a stream of air. The sheet resistance of the two films was measured, and showed little difference between the two films. This supports the assumption that using the film formation methods of this invention enhance the conductivity of the films via two mechanisms. One is the film formation methods of this invention create a CNT film that is more debundled and is has a morphology which promotes higher conductivity. The second is that the film formation methods of this invention can lead to a film which is intrinsically doped during the film formation process, and requires no second doping step.