Discussion

This discussion focuses on three things. The first point concerns the major composi- tional, structural, mechanical, and resistive differences between the AA1100 sample and the other three GC samples. The second compares the minor differences observed between the high and low temperature Sigradur samples. Finally, the validity of the “Type 1” and “Type 2” GC classifications is discussed.

All of the techniques used in this chapter to characterise the composition, struc- ture, mechanical response, and electrical properties of the GC samples have clearly shown that the AA1100 sample is substantially different from the other three. This sample contains significant amounts of nitrogen, oxygen, and hydrogen. This is a very important fact to consider as a common use of GC is to make crucibles for annealing other materials. It is not known if these impurities are stable in the GC matrix at these temperatures but it may be an important consideration.

It is possible that these impurities originate from the original polymer precursor or from some type of hexamine-like reactant which should have been fully expelled during synthesis [59]. It is also possible that the AA1100 structure has a large amor- phous component with some uniformly dispersed graphitic nanostructures. If this is true, the lack of large graphene sheets could mean that the AA1100 sample is not im-

permeable to nitrogen, oxygen, and especially hydrogen, and that these contaminants have penetrated the material long after the heat treatment process was complete.

It is also possible that the high temperature wasn’t reached uniformly throughout the bulk, or that this sample was synthesised in a contaminated environment. These scenarios are relatively unlikely, as the contaminant elements are uniformly dispersed throughout the entire bulk of the material.

Another important thing to consider is how the impurities and structural differ- ences in the AA1100 sample make it more difficult to compress than the other three GC samples. A model to explain this has been devised here, which incorporates the results from this chapter. These results show that the AA1100 sample is comprised of smaller graphene sheets, and that this is likely to be a result of the non-carbon elements impeding their extended growth. These smaller sheets will greatly effect the compressibility, as it is assumed that the superelastic behaviour of GC originates from the ability of the graphene sheets to slide over each other without breaking any bonds within the sheets, just the bonds connecting the sheets. This mechanism allows for bulk GC to deform without breaking any sheets, and only temporarily adjusting interconnecting bonds and temporarily compressing voids. The results of this chapter show that the AA1100 sample has smaller sheets and hence smaller voids, and also

∼10% non-carbon atoms. These factors mean that to deform the bulk of the material more bonds (that are not in graphene sheets) must be broken to allow the sheets to slide over each other, and they will be limited in how far they can slide. This could explain the higher compressibility measured for the AA1100 sample. The fact that AA1100 sample has smaller graphitic sheets and a significant portion of non-carbon atoms can be used to explain the higher resistivity. Smaller graphitic sheets effec- tively suggest that there will be more grain boundaries for the current to cross, and the non-carbon contaminant atoms will act as trap states which will act to impede net current flow.

Interestingly, none of the techniques used in this chapter show significant differ- ences between the Sigradur-K and Sigradur-G GC samples, even though they have been heat treated to 1000◦C and 2500◦C, respectively.

There are some small differences observed between the two structures. The D and G-peaks in the Raman spectra are slightly narrower for the Sigradur-G sample, and correspond to a slightly larger L[a]. The data also show that the {002} peak in the XRD spectrum is slightly narrower for the Sigradur-G sample, indicating a smaller distribution with regards to the layer spacing within layered graphitic nanostructures. This result is supported by the slightly smaller layer separations measured from the

layered structures in the HRTEM images, which are 3.60 ± 0.02 ˚A, relative to 3.65

± 0.02 ˚A for the Sigradur-K sample. Another, is that the graphene sheets are able to relax slightly more, allowing for neater stacking. The second is that the higher temperature is enough to mobilise (or remove) defects within individual graphene sheets, effectively making larger sections of unbroken sheets. These small structural differences mentioned could easily account for the different values measured for E*, indicating that the Sigradur-G sample is slightly easier to compress than the Sigradur- K, and also for the lower resistivity of Sigradur-G. The G(r) spectra generated from neutron diffraction data show no differences out to a distance of ∼10 ˚A. This corre- sponds to a distance spanning across 4-5 six-membered hexagonal rings. With this result the assumption can be made that both materials are effectively the same over this very small distance. However these are arguably subtle differences and it is surprising that larger differences were not observed given the difference in the heat treatment temperature between the two samples. For example, the characterisation methods used here show no evidence of any clearly defined structural difference as to whether or not the bulk of the materials are comprised of curved and folded graphene sheets [60], or layered fullerenes [104].

Interestingly, when compared to a study conducted by Jenkins et al. [59], the graphitic layer spacing of ∼3.64 ± 0.02 ˚A observed (via HRTEM and XRD) in the Sigradur-K sample estimate a maximum heat treatment temperature of ∼1800◦C. Thus, the results show that the phenolic polymer resin used to create the Sigradur samples does not appear to change when annealed beyond ∼1000◦C up to ∼2500◦C. Moreover, this is quite different when comparing the high and low temperature Alfa Aesar GC samples, which were synthesised at maximum temperatures of 1100◦C and 2200◦C and have been shown here to be quite different. One further possible cause for the difference between samples produced by the different manufacturers could be that the Sigradur samples were annealed at their maximum temperatures for much longer than the Alfa Aesar sample. This thought suggests that the length of time that the sample is held at its maximum temperature during synthesis is as important as the actual maximum temperature itself. This suggests that the stated maximum heat treatment temperature is not the only important factor to consider when classifying a GC material, but other factors such as the original carbonaceous precursor and annealing conditions may also effect the properties of the final material.

The Type 1 and Type 2 GC classifications are underpinned by the concept that GC materials synthesised in specific temperature ranges will contain drastically differing inherent structural properties. The results presented here show that these categories

may not provide any helpful correlation with the compositional, structural, mechan- ical, or electrical properties of GC.