Raman Analysis of Inverse Vulcanised Polymers

Inverse vulcanised polymers have received significant research attention on account of their easy, modifiable, and low-cost synthesis. These polymers are synthesized from the industrial by-product, elemental sulfur, resulting in a...


I. General considerations
All chemicals were used as received.All Chemicals were obtained from Sigma Aldrich unless otherwise specified.Ground sulfur sublimed powder reagent grade ≥99.5 % was obtained from Brenntag UK & Ireland.Dicyclopentadiene (stabilised with BHT) [precursor to Cyclopentadiene] >97%, and 1,3-Diisopropenylbenzene (stabilised with TBC) >97% were obtained from Tokyo Chemicals Industry.CHNS combustion microanalyses were performed on an elementar Vario Micro cube, with a first analysis performed to acquire rough data that was then used to calibrate the instrument for a second, more accurate analysis.Differential scanning calorimetry was performed on a TA instruments DSC25 discovery series equipped with an RCS90 and using Tzero aluminium hermetic pans and aluminium lids, in the heating range -90 °C to 150 °C, at a heating rate of 10 °Cmin -1 and a cooling rate of 5 °Cmin -1 .UV/Vis spectra were obtained using a CARY 5000 UV-Vis-NIR spectrophotometer.Fluorescence measurements were performed on an Edinburgh instruments Fluorescence Lifetime Spectrometer 980, with excitation using a xenon lamp and detection using a photon multiplier tube.Raman spectra were obtained using one of the following instruments: for handheld 1064 nm Raman, a Snowy Range Instruments model CBex 1064 was used; for all other 1064 nm Raman, a Metrohm i-Raman EX 1064 was used; for 532 nm and 785 nm Raman, an inVia Reflex Qontor Confocal Raman Microscope was used.

II.A. General Method
Inverse vulcanisation reactions are particularly sensitive to their reaction conditions, so great care was taken to ensure consistency in the reaction method.The same hotplate and thermocouple were used for every reaction.The hotplate was equipped with an aluminium heating pan and block and to protect the reaction vials from the variable conditions in the laboratory, the heating pan, heating block and reaction vials were wrapped tightly in an excess of aluminium foil.Sulfur (3, 5, or 7 g ± 0.0099 g) was melted at a desired temperature in a 40 mL reaction vial, without a lid, with 200 rpm stirring from a 14 mm cross shaped stirrer.The system was left for 10 to 20 mins to allow thermal equilibration.The selected crosslinker (3, 5, or 7 g ± 0.0099 g) was poured into the 40 mL reaction vial, to give a reaction of 10 g scale, and the stirring rate was immediately increased to 900 rpm.The reaction was monitored by dip testing: when an aliquot of the reaction was removed on the end of a spatula, and the aliquot remained a single phase upon cooling, (that is no sulfur precipitated) the reaction solution was poured into a preheated mould and left in the oven at 135 °C overnight to cure.Note that dip testing was not successful for DVB polymers as there was no point at which the aliquot would remain a single phase, but at the same time leave the reaction solution of low enough viscosity to pour.Where necessary, cured polymers were ground to powder.

II.B. Thin Films
To make thin films, rather than pouring the pre-polymer into a pre-heated mould, it was poured onto a preheated quartz slide.This slide was suspended on silicone blocks at its ends, so that the minimum surface area was in contact with another surface: when the quartz slide was in contact with the oven tray, polymer leaked to the underside, fusing the quartz slide to the tray.A second preheated quartz slide was then lay upon the top of the first, flattening the pre-polymer into a thin fil.A gentle pressure was exerted on the top of the assembly by hand using tweezers, in order to squeeze out any air bubbles and thin out the film.This had to be done with care, as if the upper slide shifted over the lower one, the film would be ruined.The assembly was then returned to the oven to cure.This process was difficult to perform, and required perfect timing for the pouring of the polymer: too early, and sulfur would precipitate upon the quartz slide and bubble formation would be promoted by the remaining unreacted crosslinker that could volatilise, too late and the polymer would be too viscous to form a thin film.
Pouring had to be done rapidly to prevent excessive cooling of the reaction solution.The perfect quantity of polymer had to be poured onto the quartz plate: too much and the film could be too thick or polymer would obscure the outer faces of the slide, too little and the film could be too thin, or could promote the formation of bubbles.Even with perfect film making technique, some bubbles were usually present in the film.This was unavoidable, and if the bubbles were dispersed in such a way that no sufficiently large area of the film was left unaffected and suitable for analysis, then the sample had to be discarded.Some polymer usually accumulated as hanging drops on the underside of the quartz plate during curing, which were then solidified in place.These could be removed by placing the film into a freezer to embrittle the polymer, after which it could be chiselled off with the flat end of a spatula.

II.C. Under Atmosphere Reactions
For the cases of DVBα-Sβ-N2 and DVBα-Sβ-Air, while the sulfur was thermally equilibrating, the reaction vial was sealed with a septum and then purged with nitrogen or compressed air for ten minutes, after which time, active gas flow was removed and a gas balloon of either nitrogen or compressed air was added to the septum.The reactions could not be left under active purge while the crosslinker was present as the gas flow removed the volatilised crosslinker, promoting further crosslinker evaporation, quickly removing all crosslinker.The crosslinker was purged with nitrogen or compressed air for two minutes before it was syringe injected into the reaction vial.The stirring was then increased to 900 rpm and the reactions were left to vitrify and then cure on the hotplate overnight.Oven curing was not suitable as a nitrogen atmosphere could not be implemented.

III.A. Spherical Gold Nanoparticles
Spherical gold nanoparticles were synthesized by the citrate synthesis method. 1 Aqueous sodium citrate (1.4 mL, 38.8 mM) was quickly added to aqueous HAuCl 4 (200 mL, 2.53x10 -4 M) which was already under reflux.The refluxing solution was stirred for 30 min until a clear red solution was obtained.ICP analysis of the solution indicated the purity of the solution: it contained only gold metal in exactly the predicted concentration.The UV/Vis spectrum indicated a plasmon band at 534 nm.The nanoparticle solution was concentrated by centrifugation before use in SERS.

III.B. Gold Nanorods
Gold nanorods were synthesized by the method described by Vidgerman and Zubarev 2 .To 10 mL of a HAuCl

VI. Fluorescence Spectra of the Crosslinkers
Figure S9: Fluorescence spectra of a 0.00001 % v/v solution of DVB in chloroform.Note that DVB is a mixture of the para and meta isomers, and is 80 % pure, with most of the impurity being the para and meta isomers of ethyl styrene.This combination of components may explain the complexity of this spectrum relative to that of DIB.

VII. Absorption Spectra and Tauc Plots of the Polymer Thin Films
Firstly, the UV/Vis of the unreacted crosslinkers showed no absorbance in the visible region; expected since they are clear colourless liquids.All however did show absorbance at UV wavelengths (see the supporting information, Section V).From these UV absorptions, fluorescence spectroscopy was performed, the spectra of which are provided in the supporting information, Section VI.DIB and DVB were found to be very effective at fluorescence, whereas DCPD and squalene were very poor at fluorescence, though they did provide some nominal signal.Crucially, all of the crosslinkers only fluoresced when excited with UV wavelengths, and only fluoresced in the UV region of the spectrum, suggesting that leftover unreacted crosslinker is not responsible for the observed fluorescence of the inverse vulcanised polymers, and it is the polymers themselves that fluoresce when under Raman analysis.
With this conclusion, the polymers were studied by UV/Vis and fluorescence spectroscopy.Because blocks and powders of the polymers gave too strong a signal in UV/Vis, thin films of polymer were created (Figure S13, see the supporting information, Section II.B. for details).UV/Vis spectra (Figure S14) were obtained in the region of 235 nm to 1100 nm, to cover the range of Raman excitation laser wavelengths that were accessible in this study.In general, as shown in Figure S14, the polymers showed negligible absorbance at longer wavelengths, but as the wavelength was decreased, the absorbance began to increase, rapidly rising to the detector limit in the UV.
Figure S13: A photograph of inverse vulcanised polymer thin films as well as tabulated assorted spectral data regarding the UV/Vis of the films.The polymer films are adhered between two 2.5 cm by 2.5 cm quartz plates, lain on top of a sheet of paper with the alchemical symbol brimstone printed onto the page, indicating the colour and transparency of the films.See the supporting information, Section VII., for the associated Tauc plots that give the polymer bandgaps.DIB and DVB polymers showed no absorbance at wavelengths longer than 850 nm, and very minor absorbance at 785 nm (0.0015 for DVB50-S50 and 0.0018 for DIB50-S50), suggesting they would only poorly absorb that excitation laser wavelength.Since fluorescence was observed during 785 nm Raman analysis, it may be that these polymers are exceptionally efficient at fluorescence, and this small absorption is sufficient to induce fluorescence capable of swamping out the weaker Raman signal.The UV/Vis data in the table of Figure 3 suggests that these polymers do not absorb at 1064 nm, and this excitation laser would not suffer from fluorescence.
DCPD polymers showed absorbances at wavelengths as long as 1072 nm, explaining their deeper brown colour.Qualitatively, the thin films of DCPD synthesized at 135 °C and 160 °C appear very similar in their profiles, but the films synthesized at the lower temperature have much more significant absorbance at longer wavelengths.Similarly, squalene polymers also showed absorbance at longer wavelengths, though not with any great efficiency at wavelengths longer than 1000 nm.In general, the UV/Vis spectra suggest that DCPD and squalene polymers would absorb 785 nm laser light, giving rise to fluorescence in the Raman spectrum, and would also absorb 1064 nm light to some degree, suggesting that fluoresce may be possible at this wavelength.Therefore, it is expected that DCPD and squalene polymers may be harder to analyse by 1064 nm Raman spectroscopy than DVB or DIB polymers.
Figure S14: UV/Vis spectra of A) DVB, B) DIB, C) DCPD, and D) Squalene inverse vulcanised polymers.The first number in the naming convention is the weight percentage of crosslinker used in the reaction, the second number is the weight percentage of sulfur used in the reaction, and the third number, where present, is the reaction temperature in °C, where two different reaction temperatures were used.
Interestingly, these polymers, even in the form of tens of micrometre thick films, showed tremendous absorbances in the near UV (below 500 nm to the spectrometer limit of 300 nm), suggesting that these polymers could find applications as UV blocking materials, though the degradation seen in the attempts at UV Raman, detailed later on, would need to be studied first.
Using these UV/Vis spectra, the fluorescence spectra of the polymers were obtained at several excitation wavelengths of interest (see the supporting information, Section VIII.).Unfortunately, the limitations of the equipment prevented studies of the fluorescence under 1064 nm excitation light, but the fluorescence was measured under excitation wavelengths of 784 nm, 532 nm, 266 nm and any others where polymers showed peaks or shoulder peaks in their UV/Vis spectra.In brief, the polymers fluoresced at all wavelengths with which they were irradiated.When irradiated with shorter wavelength light, the fluorescent signals were of greater intensity; extremely intense with UV irradiation and very weak with 784 nm irradiation.This further confirms that fluorescence is an obstacle to Raman spectroscopy with conventional excitation lasers, but also suggests that polymer analyses by Raman spectroscopy should be carried out in the dark.This is because the polymers fluoresced at wavelengths far from the excitation laser wavelength, suggesting that stray light, could cause fluorescence that would interfere with Raman signal acquisition.The polymers also showed degradation under deep UV irradiation (266 nm).See the supporting information, Section VIII.for a more detailed discussion of the fluorescence data.

VIII. Absorbance and Fluorescence Spectra of the Polymer Thin Films
The fluorescence spectra were obtained on the same thin films used for the UV/Vis spectrometry.Whereas taking into account the film thickness when comparing the UV/Vis data was simple, for the fluorescence spectrometry such a comparison may be less appropriate.This is because the optimum alignment for each sample cannot be taken into account.Samples were placed into the spectrometer and aligned such that they gave the maximum possible signal.However, this cannot take into account the spot size of the beam on each sample, which may have been different for each set up, skewing any comparison of the results between different samples.It should still be appropriate to compare the intensities of the spectra obtained for the same sample, at different excitation wavelengths, because the alignment of the system was not changed if the sample itself was not changed.In contrast, when comparing the intensities of two different samples' spectra, caution should be taken.Note that the samples did not degrade at any of the tested excitation wavelengths, except 266 nm, thereafter which, the polymer films appeared greyed where they had been irradiated.The 266 nm excitation wavelength spectra were always obtained last, to avoid any potential effects of sample degradation upon other excitation spectra.266 nm excitation spectra were obtained as quickly as possible, using only one scan, in the hopes of obtaining the polymers' spectra and not the spectra of the degradation products, though again, caution should be exercised when observing these spectra.Besides this, all spectra were averaged over five scans, with emission and detection bandwidths of 2.5 nm, and a dwell time of 0.2 s, with the exception of 784 nm excitation spectra which needed longer dwell times of 1 s.These factors have been taken into account in the intensity scales of Figures S20 to S24, so accepting the potential issue of the alignment, all spectra should be comparable to one another.Under 784 nm irradiation, all the polymers gave poor signals that became increasingly weak at emission wavelengths that were further from the excitation wavelength.With 784 nm excitation, the polymers showed nothing more than the background signal at emission wavelengths longer than 850 nm.The polymers showed a shoulder peak at about 825 nm emission.In all cases where 784 nm excitation was applied, the emission intensity was greater at wavelengths closest to the excitation wavelength, which is unfortunate, because this is where the Raman signal occurs in Raman spectroscopy.Even though the fluorescence from the polymers is weak with 784 nm excitation, because it occurs close to the excitation wavelength, it is still capable of obscuring the Raman spectrum, which is itself, a weak signal.This then illustrates why no Raman signal could be obtained with a 784 nm excitation laser.Between all the polymers excited with 784 nm light, it does not seem as though there is a consistent trend between the sulfur content and the fluorescence intensity, though this could be due to the aforementioned issue with the system alignment.Comparing between the polymers, it appears that polymers of DVB and DIB are the most efficient at fluorescing at 784 nm excitation, potentially due to contributions from their aromatic components.Interestingly, DCPD polymers reacted at 160 °C seem to be less effective at fluorescence than those reacted at 135 °C, suggesting that a more complete reaction may reduce the propensity to fluoresce.
The emission of fluorescence under 532 nm excitation often gave fluorescent signals with what appears to be a Raman signal overlapping with it.It appears that when the polymer contains an aromatic unit (DIB or DVB), the fluorescence is most intense with a high loading of crosslinker, and is least intense with a low loading of crosslinker.For the other crosslinkers used, it seems that the opposite is true; fluorescence is most intense with a high loading of crosslinker.Just like when under 784 nm excitation, DIB and DVB polymers gave more intense fluorescence than DCPD or squalene polymers when under 532 nm excitation.Again, it must be remembered that these results could be skewed by the different alignments used between the different polymers.Fluorescence stemming from 532 nm irradiation usually appeared as a decay in intensity, followed by a rise in intensity in the form of a broad peak.However, this broad peak was not always present, and even when it was, it usually had several smaller peaks super imposed upon it.For DVB's 532 nm excitation fluorescence spectra, DVB30-S70 showed essentially no broad peak, revealing several smaller peaks that are likely to be Raman signals.Interestingly, in the spectrum of DVB70-S30, despite the broad peak, one can see overlapping signals that roughly align with the supposed Raman spectrum seen in DVB30-S70's fluorescence spectrum.There are several other cases of this between the different polymer samples, but there appears to be no clear way to predict the appearance of these Raman signals.In contrast to the aforementioned failure of 532 nm Raman spectroscopy to acquire a Raman signal through the fluorescence, this data might suggest that obtaining a Raman spectrum with a 532 nm laser could be possible where inverse vulcanised polymers are concerned.However, there is a crucial flaw: under 532 nm excitation, the sulfur -sulfur band region would be expected to occur at wavelength between 544 nm and 547 nm, and comparing this wavelength range to the fluorescence spectra, it is obvious that no such signal ever appears with any clarity.Thus it can be concluded that analysis of the sulfur -sulfur band would not be possible with a 532 nm excitation laser, as the signal would always be eclipsed by fluorescence.
The 266 nm excitation spectra are somewhat more dubious to analyse because the polymer films degraded under UV irradiation, so it is hard to know what is the fluorescence of the polymer, and what is the fluorescence of the degradation products.Further confusing would be the attenuation of signal intensities as the polymers became increasingly degraded under the progressively increasing exposure times.Regardless, it is safe to say that no sulfur -sulfur Raman band will be observed in these fluorescence spectra, as this region would be expected to occur at 269 nm when under 266 nm excitation, and the fluorescence spectra were all obtained starting from 275 nm to ensure the photo multiplier tube detector was not over exposed and damaged during the experiment.In general, all 266 nm excitation spectra were more intense than spectra obtained with longer wavelength excitation.All the polymers' spectra appeared fairly similar, with a smaller peak centred at 300 nm, a larger peak centred at around 410 nm, and then a broad shoulder peak at around 480 nm.The data acquired at longer emission wavelengths should be treated with the most scepticism, as it is at these wavelengths where the polymer would have been exposed to 266 nm excitation for the longest and therefore would have been the most degraded.Once again, the polymers show no consistent pattern of fluorescence intensity with sulfur loading, and under this excitation wavelength, the polymers of different crosslinkers did not show much difference in their fluorescence intensities: no crosslinker gave a polymer that was particularly more efficient at fluorescence than another.Analogous to the 532 nm excitation spectra, there are several cases in the 266 nm excitation where weak peaks appear that could be Raman signals.
Particularly peaks appear at 327 nm and 349 nm quite consistently between different spectra.It is plausible to consider that the broad peak at 300 nm could also be a Raman signal, but it is difficult prove whether this is true.For 266 nm excitation, it is important to note that the polymers gave nominal signals all across the measured range.This is in fact true for 532 nm excitation as well, though with weaker intensity.The only time the signal ever fell to background levels, was with 784 nm excitation at emission wavelengths longer than 850 nm.This implies that if one was measuring a Raman spectrum using 532 nm excitation, but the polymer was also exposed to other wavelengths of light, these could additionally cause fluorescence.Therefore, it can be concluded that elimination of background light would be beneficial not only for ambient light background reduction in Raman spectroscopy, but also the elimination of additional fluorescence.This may be a particularly important consideration for 1064 nm Raman spectroscopy, which would not be expected to incite fluorescence on its own, but could be affected by fluorescent transitions induced by shorter wavelengths of light from ambient sources.That is, the polymers may have the potential to absorb wavelengths of light shorter than 1064 nm, and then fluoresce at wavelengths longer than 1064 nm, thereby obscuring the Raman signal.

IX. Screening of Different Raman Spectroscopic Techniques
1064 nm Raman spectroscopy is very similar to conventional Raman spectroscopy, with the exception that it uses the less widespread option of a 1064 nm excitation laser.This laser wavelength is usually too low energy to excite an electronic transition, thereby preventing fluorescence.Also a result of the low laser energy, laser burn is uncommon.Unfortunately, the Raman cross section is dependent on λ excite -4 , where λ excite is the wavelength of the excitation laser. 3From this dependency it can be shown that when using a 1064 nm laser, the same signal will fall to 30 % of the intensity it would show when using a 784 nm excitation laser, limiting 1064 nm Raman spectroscopy to samples that are strongly Raman active.Since inverse vulcanised polymers contain a high density of highly Raman active modes, it was predicted that 1064 nm Raman spectroscopy would be capable of providing interpretable signals, and indeed 1064 nm Raman using a handheld instrument was successful in most cases at providing Raman spectra of polymer samples.DVB polymer blocks all gave spectra with the least baseline interference, showing identifiable bands all across the spectral range.DIB also gave interpretable spectra but were more difficult to acquire, and had more substantial baselines.DCPD and squalene showed very substantial baselines that heavily obscured most of the signals, which falls in line with the results of the UV/Vis spectroscopy that indicated these polymers had more substantial absorbances at longer wavelengths.Importantly, it was quite difficult to damage the polymer samples with 1064 nm laser irradiation, a stark advantage over several techniques soon to be discussed.
Fourier transform Raman spectroscopy can largely be considered as an extension of 1064 nm Raman spectroscopy, as they both use the same laser wavelength.However, Fourier transform Raman spectroscopy uses the addition of a Michelson interferometer to allow multiplexing measurements to bring several further advantages, such as shorter acquisition times which minimises the chance of laser burn, better resolution as there are no resolution limiting thin apertures, and an improvement in the background of the signal. 3Samples for Fourier transform Raman spectroscopy were prepared with a hand operated press for preparation of KBr pellets.The well of the 7 mm pellet die was filled with a desired polymer powder with the excess powder being removed.The die set was placed in the hand operated press and the handle was squeezed and held for 15 seconds.The pellet was formed in the center of the die.The die with the pellet was then placed in the probe laser beam chamber of the FT-Raman spectrometer (Bruker Vertex 70 with a 70 a RAM II FT-Raman module).The FT-Raman operated with a Nd-YAG laser of wavelength 1064 nm and a spectral range from 50 to 3600 cm -1 with resolution better than 0.4 cm -1 .Spectra were averaged over 100 scans (130s total acquisition time) with a laser power between 1 and 100mW obtaining an excellent rejection ratio.Similar to 1064 nm Raman spectroscopy, Fourier transform Raman spectroscopy was successful in providing spectra of the polymers, with the same observations that DVB and DIB polymer spectra were easier to obtain than DCPD or squalene spectra.UV Raman spectroscopy is also similar to conventional Raman spectroscopy, with the exception that it uses UV wavelengths to excite Raman transitions.This seems counterintuitive, as a shorter laser wavelength will promote auto-fluorescent transitions that would obscure the Raman spectrum.However, even though the laser does promote auto-fluorescence, the energy of the laser shifts the fluorescence to much higher Raman shifts, leaving the lower end of the Raman spectrum free of a fluorescent background.This is because most fluorescent transitions occur at wavelengths longer than 300 nm, which when placed in comparison with a 266 nm excitation laser, equates to a Raman shift of about 4260 cm -1 .Very few Raman modes occur at such high shifts, and so the spectrum is left free of fluorescence.An additional advantage of UV Raman is that since the Raman cross section is dependent on λ excite -4 , a 266 nm excitation laser gives a signal approximately 75 times stronger than that given by a 784 nm excitation laser. 3However, the shorter laser wavelength of UV Raman spectroscopy promotes laser burn, as it imparts a high amount of thermal energy to the sample, but also carries sufficiently energetic photons to allow photochemical reactions.This proved to be prohibitive to the analysis of inverse vulcanised polymers in UV Raman, as the polymers rapidly degraded and decomposed.Though the spectra showed no signs of fluorescence, they rapidly changed with the duration of laser exposure and therefore the degree of degradation, regardless of methodological optimisations.Thus, UV Raman is not suitable for the analysis of inverse vulcanised polymers and perhaps indicates that these polymers can be sensitive to UV light.Therefore, inverse vulcanised polymers should be stored in the absence of UV light, and possibly in total darkness.Further studies into the degradation of inverse vulcanised polymers in relation to their storage conditions is a research avenue that should receive attention in the future.Given that UV Raman spectroscopy was successful in avoiding the fluorescence of the polymers, further attention was paid to the excitation wavelength itself.Whether there was an excitation wavelength at the fringe of visible light and UV light that could avoid auto-fluorescence, without degrading the polymers before a signal can be obtained, was investigated.366 nm light again, rapidly induced sample degradation, so alongside 266 nm Raman spectroscopy, it is not suitable.On the other hand, a 488 nm laser did show promise.Most samples gave good Raman signals with different sulfur related bands, however not all were free of fluorescence and careful method optimisation was necessary to manage laser absorption and heating damage.
Kerr gated Raman spectroscopy is an advanced Raman spectroscopic technique that allows the separation of Raman signals and fluorescent signals based on the time lag between their emissions.Because Raman scatter occurs through an intermediate virtual state, relaxation from this virtual state is essentially instantaneous.Fluorescence on the other hand, occurs through an intermediate real state.Because the molecule is excited to a real state, it spends a finite amount of time in that state before relaxing and emitting a fluorescent photon.Thus, fluorescent signals are emitted a short amount of time after a Raman signal.Kerr gated Raman spectroscopy uses this short time delay to separate the Raman signal from the fluorescent signal, thereby eliminating fluorescent backgrounds. 4However, Kerr gated Raman spectroscopy is not a widespread technique, and so it is unlikely it will see significant uptake in the field of inverse vulcanised polymers; the analysis performed here was done as a proof of concept.Kerr gated Raman spectroscopy was successful in obtaining polymer spectra for all samples, even the ones where other techniques struggled due to fluorescence.Though the signals are weaker, Kerr gated Raman spectroscopy could be a useful last resort where other techniques fail.The method for preparing samples for Kerr-gated Raman spectroscopy was the same as the method for Fourier transform Raman Spectroscopy.

X. Evidence of Oxidation of Inverse Vulcanised Polymers
DVB polymers provided clear spectra, even without method optimisation, so they were useful candidates for simplistic initial analyses.One such analysis was that, in the 1064 nm Raman spectra, there appeared to be evidence of sulfonic acids and sulfones (1025 -1060 cm -1 and 1050 -1210 cm -1 respectively).If these moieties are present in the structures of inverse vulcanised polymers, then it suggests that oxidation may play a role in the polymerisation process.This conclusion and the assignments themselves were difficult to confirm on their own, so to confirm the presence of oxidised sulfur, the syntheses of DVB30-S70, DVB50-S50 and DVB70-S30 were repeated, but this time an inert atmosphere of nitrogen was maintained throughout the polymerisation and curing processes (the polymers had to be cured upon the hotplate rather than in the oven).Due to the affixed septum and gas balloon on the reaction vial, and the curing upon the hotplate which has been shown to give different results to oven curing previously, these new polymers: DVB30-S70-N2, DVB50-S50-N2 and DVB70-S30-N2, were not directly comparable to the previous ones.Therefore, a second batch of polymers were synthesized in the same way as DVBα-Sβ-N2, but this time, the balloons and reaction vials were filled with air instead of nitrogen, giving DVB30-S70-Air, DVB50-S50-Air and DVB70-S30-Air.All of the reactions for DVBα-Sβ-N2 and DVBα-Sβ-Air took longer than their unsealed reaction vial counterparts: DVBα-Sβ.This is likely because the presence of the septum prevented the loss of volatilised DVB.It has been shown previously that the greater the quantity of crosslinker in comparison to sulfur, the slower the reaction proceeds; likely due to the decreased proportional presence of initiating sulfur radicals. 5Therefore, if DVB evaporates over the course of the reaction, as it has been proven to do previously, the proportion of sulfur in the reaction is raised, leading to a faster rate. 5Sealing the reaction vial would prevent the loss of DVB and therefore keep the proportion of sulfur lower than if the septum was absent, thereby decreasing the reaction rate.The conclusion that the loss of DVB is prevented by the fixture of a septum is supported by the CHNS data for DVBα-Sβ-N2 and DVBα-Sβ-Air, where the percentage of sulfur is lower and the percentage of carbon and hydrogen is higher than those values for DVBα-Sβ.In fact, the CHNS data for DVBα-Sβ-N2 and DVBα-Sβ-Air mirror their predicted values very closely (Table S2), unlike DVBα-Sβ, which show significant evidence of crosslinker evaporation.It should be no surprise then, that with such different sulfur compositions between DVBα-Sβ-N2 and DVBα-Sβ-Air compared to DVBα-Sβ, that their glass transition temperatures are also very different.The data in Table S2 indicates that DVBα-Sβ-N2 and DVBα-Sβ-Air are indeed different, though this difference is quite small between DVB50-S50-N2 and DVB50-S50-Air, wherein the CHNS data values are all within a percent of each other, and the glass transition temperatures are very close.For the other four polymers, the differences are pronounced.The glass transition temperature is lower when the reaction was done under nitrogen, and the CHNS values differ by some margin.For DVB30-S70-N2 and DVB70-S30-N2, the %C and %H values are lower than the expected values, whilst the %S value is higher than expected.Contrastingly, DVB30-S70-Air and DVB70-S30-Air adhere much more closely to their expected values, and therefore, their %C and %H values are higher, and their %S values are lower, than their under-nitrogen counterparts.Though it seems that the atmosphere the reaction is performed under affects the resulting polymers, the reasons for these differences cannot be determined here.It was supposed that the DVBα-Sβ-Air polymers may have incorporated some oxygen atoms into their structure, and that any leftover unaccounted-for mass in the CHNS analysis may be due to oxygen.If this was the case, then the unaccounted-for mass should be higher in the DVBα-Sβ-Air polymers compared to DVBα-Sβ-N2, however this was not consistently the case in the data of Table S2.
Regardless, the reactions of DVBα-Sβ-N2 took longer than the analogous DVBα-Sβ-Air reaction, suggesting that oxidation does play some role in the reaction mechanism.Furthermore, as depicted in Figure S34, the DVBα-Sβ-N2 polymers are visually different to their DVBα-Sβ-Air counterparts, being slightly lighter and more yellow in colour.Unfortunately, 1064 nm Raman spectroscopy could not identify a significant difference between DVBα-Sβ-N2 and DVBα-Sβ-Air, as their spectra appeared the same (Figure S33).The peaks that were initially assigned to sulfonic acids and sulfones appeared in both DVBα-Sβ-N2 and DVBα-Sβ-Air spectra, suggesting these peaks are not related to oxidation products, or that somehow sulfonic acids and sulfones are forming despite the exclusion of oxygen from the reaction; a conclusion that seems highly unlikely.Regardless, even though Raman spectroscopy could not distinguish the polymers formed under nitrogen and under air, the other characterisations suggest that the role of oxidation in the mechanism of inverse vulcanisation merits more dedicated studies.

XIII. Density Functional Theory Calculation Method
To identify a suitable method of predicting Raman spectra, the experimental 1064 nm Raman spectra of six model molecules were obtained: dimethyldisulfide, dimethyltrisulfide, diallyldisulfide, diallyltrisulfide, diphenyldisulfide, and dibenzyldisulfide; all containing a sulfur -sulfur bond.Next, the Raman spectra of these molecules were predicted using density functional theory in the Gaussian 09 code, using different density functionals with the def2-TZVPP basis set.All spectra were calculated assuming 1064 nm excitation at 298.15 K, in the gas state in a vacuum, after an initial energy minimization using MMFF molecular mechanics.The four functionals tested were, BP86, HSE06, ωB97XD, and M062X, (all used in conjunction with the def2-TZVPP basis set, with and without Gimme's B3(BJ) empirical dispersion correction) as it was thought that these functionals would provide a good spread of functionals at different levels of theory: for example ωB97XD is a range separated hybrid generalised gradient approximation functional, whereas M062X is a global hybrid meta generalised gradient approximation.B3LYP and EDF2 were also preliminarily tested with a variety of basis sets, but although EDF2 was better than B3LYP, both were quite poor at replicating the Raman spectra.The predicted Raman spectra from these calculations, were then compared to the experimental Raman spectra, and it was found that the spectra calculated by an initial energy minimisation with MMFF molecular mechanics, followed by geometry optimisation and energy calculation with the BP86 functional and the def2-TZVPP basis, with a D3(BJ) empirical dispersion correction, gave the best fit to the experimental data.Grimme et al. singled out a charged sulphur ring system, S 8 2+ , as having sulfur -sulfur bonds elongated by DFT-D3, which are then corrected with DFT-D3(BJ) which may explain why this empirical dispersion was advantageous in this case. 6See below for the comparison of the experimental and calculated spectra.To further improve the fit of the predicted Raman spectra to the experimental Raman spectra, parameterisation can be carried out, which would entail using the experimental peak data to correct the predicted data, which is detailed in section XV.Figures S40 to S50 all show predicted Raman spectra which were predicted by the optimised method mentioned above (geometry optimisation and energy calculation using BP86+D3(BJ)/def2-TZVPP).For the model compounds, energy minimisation with MMFF molecular mechanics was used before geometry optimisation.For the polymer models, a conformer search was performed instead of energy minimisation, which was also done using MMFF molecular mechanics.Table S3: Selected predicted Raman data for the lowest energy conformers of the polymer models.The column "S-S mode" indicates whether the vibrational mode had a major, minor, or no contribution from the stretching of S-S bonds.Other information such as the reduced mass, force constants of vibration, and IR intensities were also calculated, but there was too much information to represent here.All data presented here is parameterised by the method shown in section XV.

XV. Parameterisation of the Predicted Spectra
In terms of parameterising the Raman shifts, it was found that the best improvement in accuracy was given by simply adding 15 cm -1 to the values, because on average the calculated spectra underestimated the Raman shifts of sulfur region modes by approximately 15 cm -1 .Parameterisation of the intensities of the calculated spectra was a great deal more complicated than the Parameterisation of the Raman shifts, but was also more important.In general, the intensity of Raman modes is much more difficult than the peak Raman shifts to predict accurately through computational chemistry.
The first obstacle to parameterising the intensities was to overcome the fact that the experimental and calculated spectra of the standard molecules have different units that are not easily interchanged.An additional problem is that the experimental spectra are influenced by experimental parameters such as laser intensity and integration time, which the calculated spectra were not.To circumvent this problem, normalisation and a ratiometric approach was employed.Regardless of any parameters and units, the ratio of two different peak intensities should not vary.Because an aromatic mode at approximately 1600 cm -1 exists within the spectra of the DVB polymers as well as two of the standard molecules, diphenyldisulfide and dibenzyldisulfide, the ratio of the intensity of the aromatic mode over the intensity of the sulfur modes could be used to apply the corrections.That is, the ratio in the experimental spectra could be used to correct that ratio from the calculated data.Since this correction centres around the aromatic mode, it was sensible to normalise the experimental and calculated spectra to their aromatic modes, thereby eliminating the issue of differing units.Observing the comparison of the experimental and calculated spectra of the standard molecules, and comparing the experimental DVB polymer spectra to the calculated polymer model spectra, it is clear that the calculated spectra consistently underestimate the intensities of the sulfur -sulfur modes in comparison to the aromatic modes.The next complication to the parameterisation was that the degree to which the sulfur modes were underestimated could be dependent on the Raman shift that the modes occurred at.As such, the parameterisation function had to vary with Raman shift, so that the parametrisation value could vary with the Raman shift.The best intensity parameterisation model was found to be a polynomial function as detailed in Figure S51.Note that the experimental intensity values have been baseline corrected, to eliminate the effects of non-Raman scatter effects, which was found to be crucial.The number of significant figures included in the intensity parameterisation was also found to be crucial.From this parameterisation model, the polymer model calculated spectra can be parametrised by multiplying the peak intensity of a Raman mode by the appropriate B/A value for that Raman shift.

XVI. Step by
Step Guide for the Quantification of Dark Sulfur by Raman Spectroscopy 1. Acquire a polymer spectrum and divide the intensity by the integration time.2. Identify the elemental sulfur signal at 220 cm -1 and then use a linear baseline correction to eliminate the contribution of the fluorescent background.3. Acquire the spectrum of elemental sulfur using the same laser power as was used for the polymer; the integration time can be different in order to prevent issues with saturating the signal detector.Divide the intensity of this spectrum by the integration time.4. Determine the mass percentage of sulfur in the polymer by another method, such as CHNS combustion microanalysis. 5. Multiply the elemental sulfur spectrum intensity by the determined mass percentage of sulfur over 100.6. Calculate the following ratio: intensity of the 220 cm -1 peak in the elemental sulfur spectrum ÷ intensity of the 220 cm -1 peak in the polymer spectrum.Multiply the result by 100 to get the percentage of sulfur in the polymer that is not polymerised.
Even if the mass percentage of sulfur in the polymer cannot be obtained confidently, this method can still be used to qualitatively compare the content of elemental sulfur between different polymers.The aforementioned process should simply be carried out until the point at which the mass percentage of sulfur is needed.Instead of proceeding further, the signal intensity of the 220 cm -1 peak can then be directly compared to the equivalent peak in other polymers.Note that in order for this qualitative method to be valid, all polymer spectra must be obtained at the same laser power, and then the intensities of the spectra must be corrected for any differences in integration time.

XVII. Calculation of Expected Sulfur Rank
If it is assumed that all double bonds react and that there are no sulfur loops in the polymer, then it can be deduced that the number of double bonds equals the number of sulfur bridges.This is because it does not actually matter how the sulfur bridges interconnect the organic units, the number of sulfur bridges is theoretically the same in every scenario.Figure S52  Inside every 'unit cell', regardless of the connectivity of the structure, there are two double bonds and four halves of a sulfur bridge.In other words, there is a one to one ratio of double bonds to sulfur bridges.Similar models can be generated for crosslinkers containing three, four, or more double bond sites.One way to rationalise the justification of why the connectivity does not affect the ratio of double bonds to sulfur bridges is to consider a large cyclic chain of sulfur atoms.This chain can be crosslinked by inserting a carbon skeleton with two double bonds.This would divide the cyclic sulfur chain into two sulfur bridges with two double bond sites inserted into it.If a second crosslinker is added, the sulfur chains are divided again to yield four sulfur bridges, with four double bond sites.Figure S53 illustrates how it does not matter where the crosslinkers are inserted; the result is always the same.With this fact in mind, that for all ideal structures there is one sulfur bridge for every double bond, it is simple to calculate the average expected sulfur rank.The number of sulfur atoms should be calculated from the sulfur mass that was input into the polymer.The number of sulfur atoms constituting elemental sulfur (sulfur that is not part of the polymer network) should be deducted from this total.This number of sulfur atoms should then be divided by the number of double bonds that were successfully polymerised.This number of double bonds can be calculated if the molecular mass of the crosslinker, the number of double bonds in each molecule of crosslinker, and the mass of crosslinker that actually polymerised is known.The latter of these three can be determined from the elemental analysis by the following equation, rearranged for "mass of crosslinker":

XX. DIB Case Study
In order to prove that Raman spectroscopy can have useful applications outside of the proof-of-concept examples already mentioned, Raman analysis was used to analyse inverse vulcanised polymers of DIB.DIB polymers have been observed to have different properties depending on the reaction temperature and curing time.For example, when DIB polymers are synthesized at high temperature (180 °C), they rapidly vitrify into a hard, high T g glass.However, if they are cured for longer, the T g reduces, and the DIB polymers can become more like a viscous, sticky liquid.As such, Raman spectroscopy was employed to explore the reasons for these observations.DIB50-S50 polymers were synthesized by the general method described before, but this time they were reacted at either 135 °C or 180 °C, and then either not cured at all, or cured at 135 °C for 2hr or overnight.This gave the following notation: DIB50-S50-Tγ-Δ, where γ is the temperature of reaction in °C, and Δ is the curing time of either 0 hours, 2 hours, or overnight.These polymers were subsequently analysed by Raman spectroscopy, the results of which can be seen in Figure S54.As can be seen in Figure S54A and Figure 54B, there is a marked difference between the spectra of DIB polymers synthesized at 135 °C and 180 °C.One of the most noticeable differences is that polymers synthesized at 180 °C show much less fluorescence than polymers synthesized at 135 °C.This influenced the integration times used to obtain the spectra, as the polymers synthesized at 180 °C were obtained with much longer integration times, as there was less concern of the signal swamping out the detector with the fluorescent background.A study performed by Onose et al. suggested that the colour of some polymers may result from the formation of 1,2-dithiole-3-thione rings as terminal functionalities.Their work suggested that exo-olefins like DIB can be prone to the formation of these terminal functionalities, and that prolonged heating of such polymers, such as that experienced during the curing step, can result S48 in greater populations of 1,2-dithiole-3-thione termination products. 7Since these ring structures absorb visible light, they could be responsible for polymer fluorescence, and so could explain why DIB polymers that are cured for longer, show greater fluorescent backgrounds.This would also suggest that when DIB polymers are synthesized at high temperature, the reaction pathway that produces 1,2-dithiole-3-thione products, is suppressed or outcompeted.Thus, in the future, the behaviour of 1,2-dithiole-dithiones in relation to this Raman analysis should be investigated through a combination of experimental and computational methods.
Interestingly, for both the polymers synthesized at 135 °C and 180 °C, a two-hour curing time results in a significant reduction in Raman signal through the fluorescent background.This is surprising, as it suggests that fluorescence increases and then decreases as the curing step proceeds, though it is difficult to explain this without detailed understanding of the origin of the fluorescence.The intensity of the aromatic signals in comparison to the sulfur -sulfur band does not change much over the course of the reaction, which suggests that the proportion of crosslinker does not change much as curing proceeds, though this conclusion is complicated by the fact that different sulfur ranks give different intensities to the sulfur -sulfur band, meaning the intensity of the sulfur -sulfur band will change as a function of sulfur rank and not just the crosslinker proportion.For this reason, Raman analysis is not recommended for quantitatively assessing the proportion of crosslinker in a polymer.
Observing Figure S54A the intensity at about 500 cm -1 increases as the curing process proceeds, which indicates an increase in the proportion of rank 2 chains.This conclusion agrees with previous conclusions from the DVB curing experiments.These data suggest that as the curing step proceeds, longer sulfur rank chains are consumed and broken down, and the sulfur atoms are distributed into more shorter sulfur rank chains.Note that the band deconvolution data in the supporting information, Section XIX, does not at first glance, seen to support this conclusion, but this is because the Gaussians are not centred on 500 cm -1 and also have varying FWHM values.This conclusion is also supported by the fact that Figure S54A shows that the intensity at about 380 cm -1 goes up as the curing time increases, and this may be due to an increase in the number of rank 3 chains, in accordance with Figure 9. Overall, the data in Figure S54A suggests that the sulfur rank decreases as expected as curing proceeds.
Figure S54B shows that DIB polymers synthesized at 180 °C are drastically different to analogous polymers synthesized at 135 °C.For instance, band deconvolution of the sulfur -sulfur band indicated that DIB polymers synthesized at 135 °C could have the sulfur -sulfur band accurately fitted by only five Gaussian peaks (excluding those for elemental sulfur) whereas polymers synthesized at 180 °C needed six Gaussian peaks to accurately describe their sulfur -sulfur band (again, excluding the peaks for elemental sulfur).It seems that in Figure S54B, the peaks at about 460 cm -1 are weaker than the 460 cm -1 peaks in Figure S54A, and so the other peaks in Figure S54B seem stronger in comparison to the 460 cm -1 peak, than the analogous peaks do in the Figure S54A spectra.The 460 cm -1 peak may receive contributions from all sulfur ranks, much like Group 4 in Figure 9, and if this is the case, it suggests that there may be a higher proportion of longer sulfur ranks present.Figure 9 helps to explain this justification, as it suggests that most sulfur ranks contribute very similar intensities to Group 4, but progressively longer sulfur ranks contribute progressively greater intensities outside of Group 4. According to Figure 9, the 392 cm -1 peak may correspond to sulfur rank 3, and according to the band deconvolution data in Section XIX. of the supporting information, this peak gains intensity as the curing proceeds, again supporting the previous conclusion that he sulfur rank shortens with increasing curing.Additional evidence to this is the gain in intensity at 498 cm -1 and the gain in intensity at 532 cm -1 as curing proceeds.Thus, it can be concluded that the longer the curing time, the shorter the sulfur rank becomes, and that, the higher the initial reaction temperature, the higher the sulfur rank, which is in line with the well-established theory that elemental sulfur forms longer homopolymer chains at higher temperatures, and since it is these homopolymers that initially connect the crosslinkers together, a higher sulfur rank results.Further computational studies may be required to fully understand why the sulfur -sulfur band of DIB polymers synthesized at low and high temperature are so different.

Figure S1 :
Figure S1: Representative DSC thermograms of the first batch of inverse vulcanised polymers.

Figure S2 :
Figure S2: UV/Vis of the spherical gold nanoparticle solution.
4(aq) (5 mM) and hexadecyl trimethylammonium bromide (100 mM) solution, 460 µL of a NaOH (aq) (10 mM) and NaBH 4(aq) (10 mM) was added with rapid stirring to make the seed solution.To 10 mL of a HAuCl 4(aq) (5 mM) and hexadecyl trimethylammonium bromide (100 mM) solution, 70 µL of an AgNO 3(aq) (100 mM) solution, and 700 µL of a hydroquinone (100 mM) solution was added with rapid stirring.To this solution, 160 µL of the seed solution was added with rapid stirring, which continued overnight.ICP analysis of the solution indicated the purity of the solution: it contained only gold metal in just below the expected concentration.The UV/Vis spectrum indicated a plasmon band at 1044 nm.

Figure S5 :Figure S6 :
Figure S5:Raman spectra of inverse vulcanised polymers obtained with a 532 nm excitation laser at 0.1 % power and a 10 s exposure time.

Figure S7 :
Figure S7: Raman spectra with 532 nm excitation laser, using 0.1 % intensity, with a 10 second exposure time from a 50x objective lens, for different polymers coated with gold nanoparticles.The inset image is of a DVB50-S50 polymer surface coated with gold nanoparticles.Several different laser intensities were tried with different focal points which had different concentrations of gold nanoparticles, all of which failed to give a spectrum.

Figure S8 :
Figure S8: UV/Vis spectra of the crosslinkers, diluted in chloroform to the following v/v concentrations: 0.001 % DVB, 0.001 % DIB, 0.5 % DCPD and 0.01 % squalene.Data below 400 nm may not be reliable due to an unreliable baseline.

Figure S10 :
Figure S10: Fluorescence spectra of a 0.0001 % v/v solution of DIB in chloroform.

Figure S11 :
Figure S11: Fluorescence spectra of a 0.1 % v/v solution of DCPD in chloroform.

Figure S12 :
Figure S12: Fluorescence spectra of a 0.1 % v/v solution of squalene in chloroform.

Figure S20 :
Figure S20: Fluorescence emission scans at different excitation wavelengths for different DVB inverse vulcanised polymers.

Figure S21 :
Figure S21: Fluorescence emission scans at different excitation wavelengths for different DIB inverse vulcanised polymers.

Figure S22 :
Figure S22: Fluorescence emission scans at different excitation wavelengths for different DCPD inverse vulcanised polymers, which were synthesized at 135 °C.

Figure S23 :
Figure S23: Fluorescence emission scans at different excitation wavelengths for different DCPD inverse vulcanised polymers, which were synthesized at 160 °C.

Figure S24 :
Figure S24: Fluorescence emission scans at different excitation wavelengths for different squalene inverse vulcanised polymers.

Figure S33 :
Figure S33: Raman spectra, obtained on a handheld instrument, of DVB polymers synthesized under air and nitrogen atmospheres, A) before spectral processing, and B) after spectral processing.

Figure S36 :
Figure S36: DSC thermograms of the second batch of DIB inverse vulcanised polymers.

Figure S38 :
Figure S38: DSC thermograms of the second batch of squalene inverse vulcanised polymers.

Figure S40 :
Figure S40: Comparison of the experimental and predicted Raman spectra for dimethyldisulfide, as well as a geometry optimised structure for dimethyldisulfide.

Figure S41 :
Figure S41: Comparison of the experimental and predicted Raman spectra for dimethyltrisulfide, as well as a geometry optimised structure for dimethyltrisulfide.

Figure S42 :
Figure S42: Comparison of the experimental and predicted Raman spectra for diallyldisulfide, as well as a geometry optimised structure for diallyldisulfide.

Figure S43 :
Figure S43: Comparison of the experimental and predicted Raman spectra for diallyltrisulfide, as well as a geometry optimised structure for diallyltrisulfide.

Figure S44 :
Figure S44: Comparison of the experimental and predicted Raman spectra for diphenyldisulfide, as well as a geometry optimised structure for diphenyldisulfide.

Figure S45 :
Figure S45: Comparison of the experimental and predicted Raman spectra for dibenzyldisulfide, as well as a geometry optimised structure for dibenzyldisulfide.

Figure S49 :Figure S50 :
Figure S49: Predicted Raman spectra for the lowest energy conformer of the rank 5 polymer model.

Table S2 :
Analyses of inverse vulcanised polymers synthesized under air or nitrogen

DIB Case Study Band Deconvolution DataTable S4 :
where G n,predicted is the intensity of the group calculated from the equation, C n,rm is the intensity that sulfur rank m should contribute to group n (steps 9 and 10), and r m are the proportional populations of sulfur ranks 2, 3, 4, 5, and 6 respectively.This should lead to several simultaneous equations, where the number of equations is equal to the number of groups.The r m value for a given sulfur rank is the same in each equation.14.Adjust the r m values in a trial and error iterative fashion until the values for G n,predicted are as close as possible to the corresponding G n,real values.This process is complete when a minimum difference between the G n,real values and the G n,predicted values is achieved, and none of the r m values are negative, nor nonsensical (for example r 2 > r 3 < r 4 implies that sulfur rank 3 has a lower population than sulfur rank 2 and sulfur rank 4, which makes no logical sense).15.Express the r m values as percentages of the sum of all r m values.The percentages obtained indicate what percentage each sulfur rank makes of the whole population.If desired, the average sulfur rank can easily be calculated from these percentage populations.Band deconvolution data for DIB polymers synthesized under different conditions.Values in orange are for elemental sulfur XIX: