BACKGROUND OF THE INVENTION.

Carbon nanotubes are a novel concept within the art. Carbon nanotubes exhibit unusual mechanical, thermal, optical, electrical, and electronic properties. Procedures according to the prior art for producing nanotubes are known to give mixtures of semiconducting and metallic types of nanotubes. The electronic properties of such single-walled nanotubes, SWNTs are extremely sensitive to minute differences in chirality, i.e. geometrical structure, so that nanotubes with only slightly different form may behave either as metals or as semiconductors with varying bandgaps.

Carbon nanotubes present a special structure of carbon. The nanotubes have a diameter at nanometer scale and a length typically of several micrometers. In the methods existing today for production of SWNTs these methods result in a mixture of two types of SWNTs, with a 1 to 2 ratio for metallic-to-semiconducting nanotubes.

Below is referred to the expression stabilized in regard to the nanotubes. This expression is used in the sense that the stabilizing medium is used to keep the nanotubes or nanotubes ropes separated from each other when in a liquid medium.

Single- walled carbon nanotubes SWNTs have been intensively studied as a potential building block for future nanoelectronics owing to their unique electronic/electrical properties, e.g., semiconducting SWNTs, s-SWNTs, as high-conductivity channels in carbon nanotube field-effect transistors (CNFETs), and metallic SWNTs, m-SWNTs, as high conductivity interconnects in very large scale integrated circuits.

Separation of the two types of nanotubes (NTs) is essential for the fabrication of NT- based electronic devices. Various approaches, physical, chemical, or biochemical, have been pursued for NT separation, e.g. as a physical method, the dielectrophoresis

technique employed by Krupke et al. Science 301, 344 (2003); Nano Lett. 4, 1395 (2004) may be mentioned.

According to the prior art the pristine nanotubes are dispersed in D ₂ O or H ₂ O with the aid of ionic surfactants, e.g. Sodium dodecyl sulphate (SDS) and Sodium dodecyl- benzene sulphonate (SDBS).

Dielectrophoresis is defined as the lateral motion imparted on uncharged particles as a result of polarization induced by non-uniform electric fields. The AC-dielectrophoresis in this description is a movement resulting from the same phenomenon but here an alternating electric field is applied.

BRIEF SUMMARY OF THE INVENTION.

The invention concerns the separation of semiconducting single-walled carbon nanotubes s-SWNTs from metallic m-SWNTs using a combination of surface functionalization of SWNTs with non-ionic surfactant/-s to the nanotubes and AC- dielectrophoresis technique. This is to solve the problem in the prior art regarding the separation and to improve the separation to a level which has not been attained before.

According to the invention it has been shown that the mentioned electric double layer should be minimized or avoided when separating the nanotubes into metallic and semiconducting species. This is accomplished according to the invention by stabilizing the pristine nanotubes in a liquid medium comprising also various, one or more, non- ionic surfactants during the dielectrophoresis.

Without the electric double layer, the metallic nanotubes still move to the direction in which the strength of the electric field is the highest (which is conventionally referred to as a positive dielectrophoresis response), i.e. to the electrodes, whereas the semiconducting nanotubes move in the opposite direction where the strength of the electric field is minimum (which is conventionally referred to as a negative dielectrophoresis response), i.e. away from the electrodes. The different direction in

which the metallic and semiconducting nanotubes move is caused by the fact that the dielectric constant of metallic nanotubes is much larger than that of the medium (H ₂ O, D ₂ O, and other media having the same type of characteristics as to dielectric constant etc.), while the dielectric constant of the semiconducting nanotubes is smaller than that of the medium. Various non-ionic surfactants and polymers, e.g. Trition XlOO, Tween series, PVP, etc. can be used for the stabilization of nanotubes in various liquid media. The stabilization of nanotubes is still realized by the presence of polar chains surrounding the nanotubes.

The invention also relates to production of CNFETs through a combination of surface functionalization of semiconducting SWNTs using ionic surfactant/-s with AC dielectrophoresis technique. Also DC dielectrophoresis may be used. Here one of the problems is to have purified, i.e. separated, semiconducting SWNTs and also to remove any detrimental surfactant which remains on devices produced in order to give electrically good characteristics to the devices.

Once a source of nanotubes that only comprises semiconducting nanotubes is obtained they may be used according to the invention in producing e.g. CNFETs according to the invention. Now, use is made of the electric double layer produced according to the invention. There are thus only semiconducting nanotubes suspended in a liquid medium. In this case, as opposed to the process for separation of the pristine SWNTs, ionic surfactants, polymers, or charged DNA etc. are added in the medium to stabilize the semi-conducting nanotubes. In this way, the desired electric double layer is produced around the nanotubes, which consist of a charged nanotube-surfactant (polymer, DNA etc) complex and its counterions freely moving about in the medium.

Under the dielectrophoresis, the polarization of the electric double layer makes the effective dielectric constant of the semiconducting nanotube-surfactant complex larger than that of the medium. This results in the semiconducting nanotubes moving to the direction in which the electric field is the highest (i.e. the electrodes). This fact is used

according to the invention to provide a method of placing semiconducting nanotubes between pre-defined electrodes by means of polarization of the electric double layer.

The invention also relates to mass production of CNFETs.

The invention also relates to an improvement of the electrical contact using sequential rinsing in combination with an all-around contact geometry for CNFETs fabricated using AC/DC dielectrophoresis technique..

These objects are attained according to the invention in accordance with the appended claims in which:

A method for separation of semiconducting single-walled carbon nanotubes (SWNTs) from a mixture of semiconducting single-walled carbon nanotubes (s- SWNTs) and metallic single-walled carbon nanotubes (m-SWNTs) using a solution comprising a non-ionic surfactant for stabilizing the nanotubes.

Where the method also comprises the steps of: a) forming said suspension by dispersing the mixture of semiconducting single- walled carbon nanotubes and metallic single-walled carbon nanotubes in a solution where the non-ionic surfactant has a concentration larger than a critical micelle concentration; b) treating by agitation the thus formed suspension using ultrasonic radiation.

Said method may further comprise the steps of: c) decanting at least part of the supernatant suspension; d) treating the decanted suspension using AC dielectrophoresis; whereat in this last step the semiconducting and metallic SWNTs are separated.

In the method the formed suspension may be centrifuged in a step ahead of the AC dielectrophoresis .

In the method the solution is an aqueous solution and the steps of using ultrasonic agitation, centrifugation and AC dielectrophoresis are repeated until the purity of semiconducting SWNTs reaches a pre-defined percentage.

According to the invention semiconducting SWNTs are separated from metallic SWNTs using the method.

According to the invention use is made of the semiconducting SWNTs separated according to the invention for use in semiconductor devices.

According the invention a method for producing CNFETs comprising the steps of: a) dispersing s-SWNTs in an aqueous solution containing ionic surfactant having a concentration larger than a critical micelle concentration; b) treating the thus formed suspension using ultrasonic agitation and thereafter decanting at least a part of the supernatant liquid forming an s-SWNT-containing suspension; c) depositing of the semiconducting SWNT-containing suspension to predefined electrode parts, and d) aligning of the SWNTs between pre-defined electrode parts using AC/DC dielectrophoresis.

In the method ultrasonic agitation may be followed by a centrifugation of the suspension.

In said method according to the invention the semiconducting SWNT-containing suspension, having a controlled SWNT density is deposited over one or more chips having pre-defined electrode parts whereat AC/DC dielectrophoresis is applied to align the SWNTs in parallel between the electrode parts. The invention also provides for CNFETs manufactured using the above method.

The invention also provides a method for producing CNFETs exhibiting improvements in the electrical contact comprising the steps of: a) providing electrode pairs on a Si wafer covered by an insulating layer using lithography; b) fabricating CNFETs using AC/DC dielectrophoresis according to the described method. c) rinsing of the fabricated CNFETs for removal of surfactant residues; and repetition of the lithography step, using the same pattern as for the pre-defined electrodes, to deposit another metal layer on top of the contact parts of the semiconducting SWNTs.

In this method the Si wafer, which has an insulating layer is a SiO ₂ /Si substrate. However other combination may be used such as instead of Si -SiC, InP ₅ GaAs as examples. The SiO ₂ may be substituted by other equal dielectrics such as Si ₃ N ₄ , HfO ₂ , Al ₂ O ₃ and other permittivity dielectrics.

In the above method the liquid used for rinsing sequentially comprises deionized water and ethanol or other polar detergents that may be combined with water. Also part of the invention is CNFETs manufactured using the described method according to the invention.

DRAWINGS

The invention will now be further described with reference to the drawings in which:

Fig. 1 illustrates schematically the suspension comprising the pristine SWNTs according to one embodiment of the invention. Fig. 2 illustrates schematically the separation of the metallic and semiconducting

SWNTs being separated by AC/DC-dielectrophoresis according to one embodiment of the invention. Fig. 3 illustrates a further embodiment according to the invention of how the separation shown in Fig. 2 may be performed.

Fig. 4 illustrates schematically the production process for making carbon nanotube field-effect transistor according to the invention.

Fig. 5 a shows an AFM image of a CNFET with an individual s-SWNT. Fig. 5b shows the corresponding I _d -V _g characteristics in log linear scale. Fig. 6 illustrates the diffuse electric double layer formed by the ionic surfactant used in the method of making devices according to the invention. Fig. 7 illustrates the effect on the electric double layer by the electric field. Fig. 8a-d illustrates the verification of the existence of the electric double layer on

SWNTs suspended in different solutions. Fig. 9a,b illustrates the effect of rinsing of the manufactured devices according to the invention.

Fig. 10 also illustrates the effect of rinsing. Fig. 11 illustrates the process from functionalization to the finished CNFETs

DETAILED DESCRIPTION OF THE INVENTION.

In Fig. 1 is shown how the solution comprising the pristine SWNTs metallic 2 and semiconducting 3 in a container 1, where the solution comprises a non-ionic surfactant for stabilizing the nanotubes according to the above description. The suspension is formed by dispersing the mixture of semiconducting 3 and metallic 2 single-walled carbon nanotubes in the solution where the non-ionic surfactant has a concentration larger than a critical micelle concentration.

Thereafter the suspension is treated by agitation of the thus formed suspension using ultrasonic radiation in order to separate and disperse the SWNTs, of the two types still co-existing.

The suspension may thereafter be centrifuged in order to remove any bigger particles and part of the thus centrifuged solution is thereafter decanted in order to undergo AC dielectrophoresis.

In Fig. 2 is shown schematically the principle of the separation according to the invention. Two electrodes 4 are arranged in the solution. An AC-current is applied between the two and here is shown how the metallic SWNTs 2 are attracted to the electrodes while the semiconducting SWNTs 3 move away from the field between the electrodes. The man skilled in the art will know that a field of this type will show gradients inducing movements. The SWNTs which have a length typically in the order of micrometers, no electric double layer is formed in this process on account of the non-ionic surfactant, move as described above and the distance between the electrodes is preferably larger than the length of the SWNTs. Thus the metallic SWNTs may be removed from the solution.

In Fig. 3 is schematically illustrated how this process may be arranged in a continuous manner by arranging tube-like sections one after the other, where the circumference of the sections represents one electrode 4 and the second electrode 4 may be arranged in the form of a concentrically arranged rod. The process may be continuous; the flow 5 into the device at the arrow may contain metallic and semiconducting SWNTs in a ratio of approximately 1 :2 and at the flow 6 emerging from the device the ratio may be of the order 1 :100.

In Fig.4 is illustrated schematically, in cross-section, the production process for making nanotube field effect transistors according to the invention. Different types of devices such as diodes, FETs, etc may be produced in the same manner. In the figure is shown a Si-substrate 8, commonly called wafer, and thereon a layer 7 of SiO ₂ . Two metallic electrodes 4 are arranged on top of the SiO ₂ -layer through any suitable process within the art. The distance between the electrodes may be in the order of nms to μms. It should be noted that the electrodes become a part of the produced device.

The electrodes may be made from Pd, Rd, Ru, Pt, Au, Ag, Ni, Co, or their alloys or any other metal suicides, carbides, etc either both of the electrodes made from the same metal for FETs or of different metals for diodes, FETs etc.

A drop of the solution comprising only semiconducting CNTs, preferably separated according to the process according to the invention as described above, is placed in contact with the electrodes. In the solution the stabilization surfactant is of the ionic type and the liquid may be of various types. The purpose is thus here to form an electric double layer around the semiconducting nanotubes. Such ionic molecules may be ionic surfactants, polymers, charged DNA, etc. as described above.

Next the AC/DC-dielectrophoresis is accomplished using the metal electrodes 4 on the wafer. The theory according to the invention is discussed below in connection with description of experimental results made by the present inventors.

EXPERIMENTAL

According to the method of producing CNFETs etc according to the invention it has been shown that individual s-SWNTs suspended in an aqueous solution of Sodium dodecyl sulphate SDS display a positive dielectrophoresis response, i.e., move to the region of the highest field intensity, even at 100 MHz. The deposition of SWNTs was studied using standard electrical transport measurements of the SWNTs abridged on electrode pairs. These findings present a viable approach for fabrication of CNFETs based on AC/DC-dielectrophoresis, given that pre-purified s-SWNTs are available.

The SWNT soots prepared by using high-pressure conversion of CO (HiPco) was ultrasonically suspended in deionized water containing 1 weight % SDS [a concentration, which is greater than the critical micellar concentration (CMC)]. Additives, 0.35 weight % EDTA (ethylenediaminetetraacetic acid) and 4 volume % Tris-HCl buffer (pH=7.9), were included in the solution for complexing residue transition-metal ions and stabilizing the pH value, respectively. Immediately after sonication, the sample was centrifuged at 16,000 g for 6 hours. The upper 30 % of the supernatant was carefully decanted, with the micelle coated SWNTs in the supernatant. The average diameter of the SWNTs was determined to be l.l±O.l nm using Raman spectroscopy and optical absorption. The length of the SWNTs was 5 to 10 μm.

Heavily Sb doped Si (100) wafers were used for electronic device fabrication. After thermal oxidation to grow a 150 nm thick layer of SiO ₂ , microelectrode pairs of 20 nm thick Pd were prepared by combining a standard photo-lithography with standard liftoff technique. The distance between the electrode pairs varied from 1 to 5 μm. The width of the electrode pairs was 1 or 2 μm. Any typ of lithography suitable for the purpose could of course be used.

For AC dielectrophoresis of the SWNTs a drop of 2 μl of the SWNT bearing suspension was deposited on the device area where a bias of 5 or 100 MHz with a peak-to-peak voltage of 5 V was applied to the electrode pair. After one minute the bias was switch off and the solution was washed by deionized water and blow-dried with N ₂ . The alignment of the SWNTs was then characterized by means of standard electrical transport measurements on an HP 4165 A precision semiconductor analyzer as well as using atomic force microscopy (AFM) operated in a tapping mode. When making electrical transport measurements the Si substrate may be heavily doped and used as the gate electrode and the two electrodes respectively as source and drain.

Obtained were 160 devices with SWNTs connecting their electrode pairs when the dielectorphoresis was operated at 5 MHz according to electrical measurements. Of those 16% showed a p-type transistor behavior with a large on/off -state current ratio of 10 ⁶ to 10 ⁸ . According to AMF, these CNFETS comprised either individual s- SWNTs or small diameter NT-bundles (diameter < 6 nm). Among them 4 CNFETs only had one individual s-SWNT since the diameter of the NTs was below 1.4 nm according to AFM-profiling.

The AFM image of such a device with one individual s-SWNT is shown in Fig. 5a. The CNFET exhibites an individual s-SWNT having a diameter of 1.1 nm and it was deposited by AC-dielectrophoresis at 5 MHz.

In Fig. 5b is shown the corresponding I _d -V _g characteristics in log linear scale before and after RTP ( rapid thermal processing) at 500 ⁰ C for 30 seconds in Ar. The gate

modulation of this particular CNFET reached 10 ⁸ with a negligible off-current I _off of 10 ^'14 A. The on-current I _0n was 300 nA most likely due to a poor Pd/-SWNT contact. The on-current I _0n increased to 1 μA after the rapid thermal processing. This indicates that an appropriate post treatment is essential according to the invention.

When the dielectrophoresis was operated at 100 MHz ₅ a positive response of the SWNTs was still obtained and a CNFET with a 1.2 nm thick individual s-SWNT displayed a gate modulation as large as 10 ⁷ . Tests were made to make certain that the result of the experiments was not dependent on the EDTA and Tris-HCL buffer. Individual s-SWNTs were again deposited. Thus these additives were not essential to the positive electrophoresis response of the s-SWNTs in the SDS aqueous solution.

The relative high probability of aligning s-SWNTs as well as the good alignment relative the microelectrode pairs demonstrate a high site-selectivity of the deposition by AC/DC-dielectrophoresis, ruling out the possibility of random depositon of the s- SWNTs from the suspension.

In Fig. 6 is schematically illustrated a diffuse electric double layer formed by the ionic surfactant which is essential according to the invention. Generally the dielectro- phoretic force is negative for a s-SWNT, which is taken advantage of in the first part of the invention. However, when adding ionic surfactants, e.g. SDS, to the solution of the SWNTs these will be coated with SDS micelles, with the hydrophobic end of the SDS being chemically adsorbed on the surface of the SWNTs, whereas the hydrophilic end, i.e. the sulphate polar head with negative charge being oriented toward the water environment. Consequently the adsorption of SDS molecules yields highly charged negative SWNTs. This in turn leads to the formation of the electric double layer.

In the figure the reference number denotes the surface 61 of a SWNT on which the hydrophobic end, i.e. the lipid chain 65, of the SDS molecule being chemically adsorbed on the surface of the SWNT. As shown in the picture the molecules are arranged essentially with the direction of their chain in a direction perpendicular to the

SWNT surface. The hydrophilic polar heads 66 essentially forming a stable layer IHP, denoted 64 around the SWNT and outside this layer is a layer of positively charged counterions from the SDS-molecules arranged forming a layer OHP, denoted 62 and outside this area there is a diffuse layer in which some of the counterions 67 are freely moving.

When a system like that in Fig. 6 is subjected to an external electric field, the ion concentration in the diffuse region of the electric double layer is so redistributed that the electric double layer as a whole reassembles a dipole as shown in Fig. 7.

In Fig. 7 is shown schematically the effect on the electric double layer by the electric field. Fig. 7 is a cross-sectional illustration of the polarization of an electric double layer induced by the distortion of the diffuse region by the electric field E. The border of the diffuse layer is shown with long dashed lines 78 and the circumference of the SWTN shown as solid line 71 that is coated by an SDS micelle of negative charge, shown as short dashed lines 74. The radius of the SWNT is denoted as 79.

In order to verify the existence of the electric double layer the inventors performed AC-dielectrophoresis on SWNTs suspended in different solutions. This is illustrated in Fig. 8 a-b where the difference between AC dielectrophoresis at 5 MHz in a solution where SDS is added and 8 c-d when the deposition is performed in ethanol.

In 8a and 8c are presented AMF image of the SWNTs deposition and 8b and 8d shows the corresponding schematic illustration of the motion of an SWNT under the influence of electric force.

Micro-electrode pairs with a separation of 5 μm were used. For SWNTs in the SDS- solubilized medium, Fig. 8a shows that the SWNTs accumulated tend to have their central points preferentially located at the edge of the electrodes where the electric field strength is the highest. This may be explained by the motion of the SWNTs towards the edge of the electrodes driven by the positive dielectrophoresis force

perpendicular to the surface of the SWNTs due to the radial component of the electric double layer, Fig. 8b. The axial component of the electric double layer, along with the bulk SWNT as a dipole, leads to the alignment of several SWNTs in the middle going straight across the microelectrode pair. In comparison, the SWNTs dispersed in ethanol (ε=25.3), although as ropes of 10 to 15 nm in diameter due to their Van der Waals interactions, show a clear tendency to extend along the electric field with one end located on the edge of the electrode as shown i Fig. 8c. Without the electric double layer, the SWNT ropes in the external electric field acquire dipole moments mainly along their axis, i.e. length. The alignment of the SWNT ropes is then a result of their motion towards the edge of the electrodes with their axis parallel to the applied field Fig. 8d.

However the inventors have also noted that the presence of ionic surfactants like SDS adsorbed on semiconducting SWNTs degrades the electrical properties of the device as it has been fabricated and further experimental work was undertaken. Electrical characterization in combination with x-ray spectroscopy (XPS) was used for monitoring the adsorption and removal of SDS residuals.

An s-SWNT was deposited through AC/DC-dielectrophoresis across two Pd electrodes, the resulting device shown in Fig. 9a is an AFM image. The produced device was rinsed in deionized water DIW for various lengths of time in order to investigate if any SDS-residuals could be removed. The DIW rinse was also combined with an ethanol rinse.

In Fig. 9b are shown the corresponding electrical characteristics of these CNFET before and after surface cleaning by a combination of DIW rinse and ethanol immersion. The inset shows the I ₀ - V ₀ characteristic in a log-linear scale, yielding a gate modulation of 10 ⁷ for this device after the surface cleaning.

The effect may also be seen in Fig. 10 which shows partial XPS spectra taken on Pd films for (A) a sample with a high density of SWNTs deposited from the aqueous SDS

solution after rinse for 5 s; (B) a reference sample without SWNT and after a DIW rinse for 5 s; (C) the same sample as for (A) but after a DIW rinse for 60 min followed by an ethanol immersion for another 60 min. The observation of Na and S peaks indicates the presence of SDS residuals. Evidently the rinse action makes a big difference.

Also noted was that when the CNFETs were immersed in DIW and then in ethanol, each for 60 minutes and then baked at 150 ⁰ C for 10 min a good improvement was attained. This was the case for CNFETs with either individual s-SWNTs or small- diameter s-SWNT-ropes.

In Fig. 11 is shown schematically the process of making a CNFET. The same reference numbers are used throughout the figure. In Fig. 1 Ia is shown a functionalized s-SWNT 113, which has been functionalized by providing e.g. SDS- molecules 112 (the surfactant) on the tube 111 while it is in a solution of e.g. water and SDS.

In Fig. 1 Ib is shown a positive AC-dielectrophoresis response of an s-SWNT. The arrow indicates the line of travel towards the metallic electrodes 117 for the SWNTs. Also seen is the P+ doped Si wafer 115 on which a layer 114 of SiO ₂ has been grown. Reference number 116 signifies the drop of the s-SWNT containing suspension.

In Fig. 1 Ic is shown the SiO ₂ -layer 114, and the aligned s-SWNT 113 between the predefined electrodes 117.

In Fig. 1 Id is illustrated how a drop of rinse solution 118 is used for rinsing away the surfactant. In reality this represents a scheme of rinses with one or more solutions and also drying. The clean SWNT 111 is illustrated Fig. l ie, which is a partial enlarged view of Fig. Hd.

Having in mind the different behavior of the s-SWNTs in relation to the m-SWNTs in forming CNFETs it is appreciated that using the method of separation of metallic and semiconducting SWNTs according to the first part of the invention is a necessity when making device according to the second part of the invention.