Looking good.... a first attempt to develop conductive carbon nanotube/epoxy adhesives

TWI Bulletin, July/August 2007

Such materials find applications in conductive coatings, thermal management and electronic board assembly, particularly since the advent of lead free processing legislation by the EU.

Amir Bahrami is a senior project leader in the Advanced Materials and Processes group at TWI. He gained an honours degree in materials science and engineering from Queen Mary University of London. After spending time in the business consultancy world and then in the medical materials community he joined TWI in 2005. He has managed a range of small and large projects. He is doing a PhD research degree at the University of Cambridge in the field of carbon nanotubes and is a professional member of the Institute of Nanotechnology and a fellow of the Royal Commission. Amir has now taken on the task of developing and road mapping TWI's activities and business in the field of nanotechnology.

As Amir Bahrami reports the use of conductive materials that are easily applied and cured, particularly at lower temperatures than the current solder replacements, are highly desirable. Many polymeric materials, particularly those used in production of flexible substrates have very low temperature tolerance. The current adhesives available for board assembly are loaded with a minimum of 60-90% silver particulates or flakes. Adhesives seem to have reached their limit in terms of electrical conductivity. More importantly the maximum thermal conductivity achieved by these fillers reach values of a few W/mK. Considering the trend of the electronic industry towards more system integration and smaller components, multifunctional materials that can have satisfactory electrical and thermal conductivity are invaluable.

Carbon nanotubes have the potential to create a conductive network at very low loadings, and with thermal conductivities reaching that of diamond, are promising candidates for the demanding electronic industry.

Carbon nanotubes can be described as cylinders with diameters ranging from one to 100nm depending on their structure, and lengths up to several millimetres. Such cylinders are made of hexagonal lattices of carbon atoms and can be described as a single layer of graphite known as graphene rolled up into a tubular construction. Figure 1 shows a schematic of a single wall carbon nanotube (SWNT) and a graphene sheet.

The carbon atoms in a carbon nanotube are SP ² hybridised similar to graphite. Therefore the electronic structure of a carbon nanotube can be described based on its parent material. The electronic s σ bands are responsible for the strong in-plane covalent bonds within the 2D graphene sheets, while the Π bands are responsible for weak Van der Waals interactions between such graphene sheets in 3D graphite, or a multiwall carbon nanotube. In contrast to the σ bands, the Π bands are close to the Fermi level, so that electrons can be excited from the valence ( Π*) to the conduction ( Π*) band.

However despite their two-dimensional geometry carbon nanotubes, particularly those with small diameters, have one-dimensional electronic structures. This is because the number of allowed electron states in the axial direction is large and in circumferential direction very limited (depending on their diameter). The quantum confinement of the 1D electronic state must be taken into account when analysing carbon nanotubes electronic properties.

The Π, and Π* bands degenerate at high symmetry points close to the corners of the Brillouin zone. (A minimum cell corresponds to a single lattice point of a structure in the reciprocal lattice with translational symmetry). These are called the K points, where the valence and conduction bands touch each other.

For such tubes the density of state at the Fermi level is finite (zero band gap) and they are therefore metallic. If the cutting lines do not pass through K points then the tube is semi-conducting with finite band gap between valence and conduction bands.

Semi-conducting or metallic behaviour depends on the twist in the tube, the orientation of the hexagonal lattices along the length of the tube and the tube diameter. The twist in the tubes is determined by the chiral vector which corresponds to a section of the tube perpendicular to its axis. Considering a single wall tube three structures can exist, armchair, zigzag and chiral ( Fig.2).

The armchair tubes are all metallic as well as some of the zigzag and chiral tubes. Conduction through well separated discrete electron states is known as ballistic conduction as the electrons experience no scattering from impurities or phonons. So they dissipate no energy.

The high aspect ratio of carbon nanotubes (length/diameter typically 10 ³ to 10 ⁴ ) dictates that they can form a conductive network of tubes at low concentrations when embedded in a matrix and impart their excellent electrical and thermal properties in the resultant composite.

Objective

To produce electrically conductive materials from insulating resins at very low loadings of carbon nanotubes.

Materials and method

Carbon nanotube production

Highly aligned multi-walled carbon nanotubes (MWNT) were grown using a chemical vapour deposition (CVD) as carpets on the walls of a quartz tube placed in a furnace. After several hours of growth the quartz tube was allowed to cool down to room temperature before the MWNTs were harvested from the walls.

Carbon nanotube characterisation

The carbon nanotubes produced by the CVD method were characterised using a JEOL JSM6340F high resolution SEM. Their Raman spectra were collected using a Renishaw 1000 Ramascope spectrometer with a 514 nm excitation laser.

Carbon nanotube/epoxy composite preparation

Two types of epoxies were used to prepare carbon nanotube composite.

• Bisphenol-A epoxy from Huntsman (Araldite LY 556), this epoxy was chosen as it is the base for many silver loaded conductive adhesives which are common electronic industry standards such as Epotek H20E from Epoxy Technologies. A 0.05% mix was made using 150g of epoxy and 0.075g of carbon nanotubes.
• Two part epoxy, Araldite 2014. A 0.5% mix was prepared by mixing 130g of Araldite 2014 with 0.65g of carbon nanotubes.

In both cases the appropriate amount of carbon nanotubes were weighed and mixed with a corresponding amount of resin under high shear.

Specimen preparation

Specimens of the epoxy composites were prepared as follows:

• A small amount of the mix was transferred onto a microscope glass slide.
• An amine hardener (second part of Araldite 2014) was added to the glass slide in the manufacturer's recommended ratio of one to four.
• The epoxy/nanotube mix and the hardener were stirred and mixed together.
• After thorough mixing the material was smeared into a thin film (typically several tens of microns) using a glass microscope slide.
• In the case of the Bisphenol-A the films were heated on a hotplate at 110°C for seven hours to cure the epoxy.
• The 2014 samples were cured at room temperature for 24 hours.
• The amount of material used on each specimen was weighed and recorded.
• In total 10 specimens of each mix were made in addition to one sample of pure unmodified resin as control. As well as the control samples, the five best specimens in terms of smoothness and homogeneity were selected for trials. Out of the five samples one was slightly abraded on the surface where the electric contact points would sit, to expose more of the carbon nanotubes to the contacts in case they are masked on the surface by layers of resin.

It is significant that the very first set of ten samples made, were poor and formed an irregular deformed pattern. The second set of samples however were better and good enough to conduct the trials, but were by no means perfect.

Electrical conductivity measurement

Ablestik ATM-0020 test method, Volume Resistivity of Conductive Adhesives was used. Conductivity of the samples was measured using a four point probe method. The samples were connected at four locations ( Fig.3 and 4) and current was forced trough the outer pins (one and four) while the voltage drop was measured over the inner pins (two and three) using a Keithly 2400 Series SourceMeter. Using this method the current flowing through two and three is nearly zero and therefore the individual additional contact resistances do not play a role.

The coating thickness was measured at three points and the average results were recorded. The width of the sample as well as the inner probe distance between the pins two and three, were also measured and recorded.

Results and discussion

SEM

Some of the images of the CVD grown carbon nanotubes are shown in Figure 5.

The SEM pictures showed aligned carpets of carbon nanotubes ( Fig.5a and 5b). In general two types of impurities were seen in particulate form which could be amorphous carbon, and some contamination from the SEM sample preparation process. The average diameters of the multiwall tubes of the aligned carpet were measured to be in the order of 70nm to 80nm.

Raman spectroscopy results

Figure 6 shows a 514nm laser Raman spectrum of a small sample of the aligned carpet of multiwall nanotubes.

The Raman active modes in MWNT are classical tangential modes which are based on the first order Raman spectra of ideal graphite with a characteristic single peak at 1580 cm ^-1 (G-peak) and two main second order scattering namely D peak at about 1350cm ^-1 and G' peak at about 2700cm ^-1 .

In our sample the G' band can be seen at around 2600 cm ^-1 . The G band and D' bands are also present. The ratio of the intensity of the D peak to G peak is a qualitative measure of the impurities and imperfection in the samples as it is thought that the D peak arises from presence of amorphous carbon and lattice defects in carbon nanotubes. This ratio is 0.5 for the tubes in this study.

Epoxy/carbon nanotube composite samples

The Bisphenol-A epoxy had a clear appearance with a slight hint of yellow. After mixing with the 0.05% carbon nanotubes the resultant paste was greyish and transparent, as opposed to the 2014 films ( Fig.7). The 2014 was a yellowish colour paste and addition of the 0.5% carbon nanotubes resulted in an opaque black paste even as a thin film.

b) Opaque 2014 specimen

The Bisphenol-A mixture was generally smoother and the carbon nanotubes seemed to be more evenly dispersed, however there were light patches in the film samples which can be regions deficient in carbon nanotubes compared to the bulk of the sample.

The 2014 mixture was less smooth, but there seemed to be no light patches in the thin film. However there was agglomeration of tubes in different regions which gave rise to the rough texture of the film specimens.

The difference between the texture and appearance of the two samples are due to the filler content. Smaller amounts of carbon nanotubes in the Bisphenol-A epoxy would dictate that there are less agglomerated regions than that in the 2014.

In tandem with the filler content is the mixing time. The Bisphenol-A composite was mixed in high shear for 24 hours which meant the tubes were better dispersed compared to the 2014 which was only mixed for seven hours. On the other hand the tubes in the 2014 would possibly be of longer length on average compared to the ones in Bisphenol-A as they have undergone a shorter period of chopping and agitation in the mixer.

Resistivity measurements

Table 1 shows the results of the resistivity measurements. The volume resistant was calculated using the following formula:

Where r is the volume resistivity (W-Cm), R is the measured resistance (W), w is the sample width (Cm), t is the sample thickness (Cm) and L is the distance between the inner probes (Cm). The widths of the samples were measured to be 2.61cm with an inner probe size of 2.5cm. The current was constantly flickering between 0.1 to 1 mA and the compliance voltage had a correlating fluctuation at a very high frequency from 2.1 to 21 V and back.

As can be seen in Table 1, the Bisphenol-A mix had resistance higher than the 200MW limit of the equipment used. However compared with a pure Bisphenol-A specimen without the addition of carbon nanotube it was observed that the current started flowing, and the equipment gave an initial reading which very rapidly increased to more than the machine's limitation. The pure sample showed no reading at all.

Table 1 Electrical resistance measurement results

This shows that the very small amount of carbon nanotubes (0.075g) added to the Bisphenol-A does create a conductive path, however not sufficient to have a significant effect on the volume resistivity of the pure resin.

In the case of the 2014, there was a conductive path created due to the addition of 0.65g of the carbon nanotubes and this conductive path was sufficient to convert the epoxy from an electrically insulating to an electrically conductive material. The average resistivity was reduced to 3000 W-cm with the abraded sample showing resistivity of about the same value.

The results suggest that in the case of Bisphenol-A composite, if the carbon nanotube content was higher it is more than likely that the resistivity would decrease dramatically and enough to give a reading. This is true in the case of both resin types. The filler content was very small, in fact smaller than any industry used conductive adhesive where there are silver loadings of a minimum 60%, and up to 90%.

Therefore it is possible that slightly increasing the nanotube content will radically decrease the resistivity due to the very high aspect ratio of these tubes and their intrinsic electrical properties. Another issue which was highlighted during the experiments was inconsistency in the readings as well as fluctuations in the compliance voltage and the passing current. These can be attributed to two causes:

• The width of the samples was larger than the length of the contacts and therefore much of the current goes through the tree-branch like network of the carbon nanotube and may not reach the contacts at all.
  
  
• The other reason is that, even the tube network which is spread in between the contacts will not produce a continuous straight path from one point to another. It will produce an irregular network of conductive nanotubes which will provide a tortuous path for the current and therefore reduce the efficiency of conduction.

Conclusions and recommendations

The addition of carbon nanotubes will impart electrical conductivity properties to an insulating material. As the results show this can happen at loadings as small as 0.05wt%. However the resistivity reduction may be too low to measure with standard equipment. The effect was more prominent at higher loadings of 0.5% as can be expected. At such loadings, however, the resistivity was not in line with that expected from current conductive adhesives (~0.001 W-Cm) used in industry. These are usually loaded with 60-90 wt% silver flakes. The results from this feasibility trial, however, are good enough for many applications such as antistatic coatings.

What makes this material attractive is the need for very small quantities to produce a conductive network, as their high aspect ratio dictates that they can find one another end to end and have a very low percolation threshold.

As the concentration of this network and its shape governs the transport properties of the resultant resin, it can be concluded that changing the amount of filler and the shape of the network can result in material with different properties. These can be used for applications as diverse as antistatic (possibly transparent) coatings, to bulk conductive isotropic or anisotropic adhesives. Furthermore carbon nanotubes can impart other functionalities to the resin system for example high thermal conductivity.

As neither the materials nor the processes were optimised, many aspects could have had a negative impact on the results in this study. If these are optimised then better values and performance can be achieved. Some such improvements are:

• Use of purified carbon nanotubes
• Longer shear-mixing duration.
• Use of higher carbon nanotube loadings.
• Optimised sample preparation.

To learn more contact Amir Bahrami at TWI.