Adapted from: Ortuso, R. D.; Ricardi, N.; Bürgi, T.; Wesolowski, T. A.; Sugihara, K., The deconvolution analysis of ATR-FTIR spectra of diacetylene during UV exposure. Spectrochim. Acta A Mol. Biomol. Spectrosc.
2019, 219, 23-32.
Introduction to chapter
After decades of studies, the structural change during the polymerisation and the mechanism of the blue-to-red transition are still controversial. Infrared spectroscopy (IR) has been frequently employed for investigating the structure of PDAs due to its ability to keep track on the bond change during the polymerisation and the blue-to-red transitions, besides Raman spectroscopy183, carbon-13 nuclear magnetic resonance (NMR)184,185 and X-ray crystallography58,101,186. To date, numerous Fourier-transform IR (FTIR) studies have focused on the structural changes of PDA, when they are exposed to light at different wavelengths187, pH188, temperatures189,190,191,192 and pressure perturbation193. Density functional theory has also been used to better understand the spectral changes.194,195 . However, the deconvolution analysis of diacetylene IR peaks has been under studied, thus the maximum information may have not been extracted from IR spectroscopy in PDA systems.
In this work, we performed a detailed deconvolution analysis of ATR-FTIR peaks of a common amphiphilic diacetylene, 10,12-tricosadiynoic acid (TRCDA) during the polymerisation and the blue-to-red transition. Our ATR-FTIR setup is equipped with UV light source, which realises in-situ polymerisation. The obtained IR spectra were compared with UV-VIS spectra for their detailed interpretation.
We carried out the deconvolution analysis of ATR-FTIR C≡C, C=C, C=O, CH2 symmetric stretching, and CH2 in-plane bending peaks with TRCDA during UV exposure. Based on the
92 | 212
analysis and the solvent dependence on the IR signals, we found that the triple peaks from C≡C stretching mode that have been previously suspected as a consequence of Fermi resonance194,195 are rather associated with the macromolecular assembly of TRCDA. Besides these C≡C triple peaks, we found that the background in the region increased during the UV exposure due to the C≡C signals from polymers. In addition, the anisotropic compression during polymerisation was also detected, which support the proposed interpretation of X-ray data reported previously. These results are the benefits from the deconvolution analysis.
93 | 212 Results and discussion
Figure 18: UV-VIS spectra. A) UV-VIS spectra of TRCDA after different UV areal doses. B) Position of the red and the blue peak as a function of the UV doses. C) Blue to red ratio, calculated with the height of the peaks. The vertical solid line highlights the point, in which the blue and the red polymers are at a 1:1 ratio (blue to red ratio 100%). D) Variation of the Area under the absorption spectra between 480 nm and 710 nm over the amount of the UV dose. The vertical dashed line identifies the point, where 10% PDA was degraded.
The polymer states (blue/red/degraded) were correlated with the UV irradiation dose
Our initial goal is to correlate the colour of the sample and the UV irradiation dose.
We will use this result as a reference to interpret the ATR-FTIR spectra later. To enable direct comparison between UV-VIS and ATR-FTIR spectra, exactly the same sample holder and the
94 | 212
UV source were used during these experiments. Figure 18A shows the UV-VIS absorption spectra of TRCDA at different areal irradiation doses. Two characteristic peaks, the blue state peak at around 650 nm and the red state peak at around 600 nm were observed, both of which shift during the UV irradiation (Figure 18B). Note that both blue and the red peaks split into a double peak structure196, yet only the most intense peaks were plotted in Figure 18B.
These transitions of the PDA electronic states have been extensively studied previously2,109,116,196. The ratio between the height of these blue and red peaks was plotted in Figure 18C. Note that the extinction coefficients of blue and red PDA are in the same order of magnitude (2.52 x 1010 M-1cm-1nm4 and 3.27 x 1010 M-1cm-1nm4 respectively)197. Ratio 100%
suggests the point, where the amount of the blue and the red polymer are roughly equal.
Ratio > 100% indicates that the amount of the blue PDA is larger than that of the red PDA, while ratio < 100% suggests the opposite. From these results, we correlate the state of the polymer (blue or red) with the irradiation dose as shown in Figure 18C: irradiation dose 0 – 90 J/cm2 is for the monomer to blue state polymerisation (indicated as “blue state”), 90 – 170 J/cm2 is for the blue-to-red transition (indicated as “red state”). Note that these assignments are for the reference, and in reality the monomer, blue, and the red polymers are always expected to be mixed in the sample. When the samples are irradiated more than 170 J/cm2, the ratio further decreases (Figure 18C). To understand this region between 170 – 250 J/cm2, we plotted the area below the absorption spectra between 480 nm and 710 nm as a function of the irradiation dose, normalised by that of the monomer (0 J/cm2 irradiation) in Figure 18D.
The area under the spectra indicates the total amount of the blue and the red polymer in the sample as monomers or degraded polymers that lost the conjugated backbone present no clear absorption peaks198. The initial increase, followed by a plateau until around 170 J/cm2, is in good agreement with the assignment of the polymer states defined by the peak ratio in Figure 18C, because during the monomer to blue state polymerisation the amount of the conjugated polymer should increase, while during the blue-to-red transition the total amount of the blue and the red polymers should stay constant. After 170 J/cm2, the area decreases rapidly, which implies degradation of the polymer. At 250 J/cm2, around 70% of PDAs were degraded, evidenced by the loss of the area by 70%. For the following, these assignments of
95 | 212
“blue (monomer-to-blue state)”, “red (blue-to-red)”, “degraded states” will be used to interpret the ATR-FTIR data.
Figure 19: ATR-FTIR spectra of TRCDA at 0 J/cm2 (black), blue (blue), red (red), and degraded (grey) states. The blank ATR-FTIR ZnSe element was used as reference spectra and the baseline was subtracted. All the peaks that are identified and will be analysed later are indicated by wavenumbers. Note that C≡C peaks (2181, 2167, 2139 cm-1) are not visible in the Figure since their intensities are small. See the zoom-in Figure 20 for the detail.
Table 6: Spectral interpretation of ATR-FTIR signal in terms of the functional group.
Bond Mode Type CH3 Stretching Asymmetric 2970 - 2948 Termination Tail
199,200,201,
202
CH2 Stretching Asymmetric 2948 - 2880 Tail CH3 Stretching Symmetric 2877 - 2867 Termination Tail CH2 Stretching Symmetric 2867 - 2830 Tail
96 | 212
ATR-FTIR spectra were obtained and the peaks were assigned
Next, we obtained the ATR-FTIR spectra of TRCDA during the UV irradiation (Figure 19). The assigned peaks are summarised in Table 6. Overall the IR spectra change during the UV exposure is rather subtle. Nevertheless, these zoom-out spectra are shown to confirm e.g.
the absence of signals from free OH group in the 3500 cm-1 region, which suggests that most of the carboxyl head groups are hydrogen bonded203,204. In the following, we will perform the detailed deconvolution analysis for C≡C (1980 – 2275 cm-1), C=C (1580 - 1500 cm-1), C=O stretching (1800 – 1600 cm-1), CH2 symmetric stretching (3600 – 2380 cm-1) and in-plane CH2
bending (1490 – 1310 cm-1) peaks.
97 | 212
98 | 212
Figure 20: C≡C stretching peaks. A.I) ATR-FTIR spectra for C≡C stretching at different UV areal irradiation doses, where the results from the deconvolution are superposed. The regions within the two vertical dashed lines were used for the deconvolution. The characteristic triple peaks from C≡C stretching that have been reported before are highlighted by violet, green and yellow. A.II) The normalised area underneath the peaks and A.III) Δ peak position versus areal irradiation doses. B) The characteristic triple peaks changed their intensities and positions, depending on their solvent environments.
Triple (C≡C) and double (C=C) carbon-carbon bonds reveal the crosslinking process
As the schemes in Figure 3 and Figure 4 show, an expected major change in the chemical structure during the polymerisation is the reduction of the triple carbon-carbon bond (C≡C), accompanied by the formation of a double carbon-carbon bond (C=C), which is initially absent in monomers.
In the monomer state, we detected multiple peaks around the triple carbon-carbon bond (C≡C) region, including the characteristic triple peaks in TRCDA that have been reported before205 (highlighted by violet, green and yellow deconvolutions in Figure 20A.I). Note that the rest of the deconvoluted peaks are shown in grey. Previously, double peaks in the C≡C region in diacetylene monomers have been interpreted as a consequence of Fermi resonance205, but the origin of the third peak in TRCDA is left elusive. First, we thought that these are signals from spontaneously-polymerised short-chain polymers mixed in monomers because C=C signals indicate the presence of polymer as we will explain later. However, this idea was rejected as the height ratio of these three peaks is almost identical for different batches and storage time of monomers (Figure App.C1), where it is unlikely that the population of different short-chain polymers are exactly the same. These triple peaks seem to be sensitive to solvent (Figure 20B and Figure 20C, where the possible assignment of violet, green and yellow peaks and their peak position as a function of the solvent polarity are shown). Amphiphilic molecules self-assemble into different macromolecular structures in solvent with different polarities206. For example, the intermediate polarity of chloroform (0.26) allow amphiphilic molecules to be dispersed as a monomer, whereas an extreme polar (e.g. water: 1) or non-polar (e.g. pentane: 0.01) solvent induce the assembly of bilayers/micelles or inverted bilayers/micelles, respectively. The fact that monomer C≡C
99 | 212
signal has a relatively simple peak in chloroform, whereas it splits into multiple peaks in higher or lower polar solvent suggests that these multiple peaks come from various packing environments of the monomer. The deconvolution analysis, of the polymerisation, shows that the areas below these triple peaks all decrease during the UV exposure as the population of the triple bonds in monomers diminishes (Figure 20A.II) and more polymer is formed. The slight positive peak shifts during the monomer-to-blue polymerisation (Figure 20A.III) imply the compression of the diacetylene upon the conformational change during the polymerisation. In contrast, the sum of the rest of the peaks within the region (1980 – 2275 cm-1, indicated as grey deconvoluted peaks) increased during the UV exposure (the grey plot in Figure 20A.II). This background increase probably originates from many different C≡C peaks signals from polymers with different chain lengths. IR peak simulation from short polymer chains appear around in this region with different peak shifts (Figure App.A1), supporting this idea. In addition, their larger conformational freedom such as slight torsions or kinking of the polymer backbone could have contributed in splitting the peaks.
The peak from C=C was detected even before the UV irradiation (the deconvoluted light blue peak in 0 J/cm2 in Figure 21A.I). This indicates the presence of polymers dispersed in the TRCDA suspension even before the active UV irradiation. In fact it is well-known that diacetylene self-polymerises when stored in chloroform207. As the sample is irradiated with UV light, the area under the C=C peak increases only by around three folds during the monomer-to-blue polymerisation (Figure 21A.II), indicating that the amount of the double carbon-carbon bond tripled. The area keeps a plateau during the red state (Figure 21A.II), which is in agreement with the fact that little or no bond change is expected during the blue-to-red polymer conformation change. Note that even though the polymerisation reached the saturation, some monomers may have still remained in the samples unpolymerised because we observed some monomers in amorphous structures that are not in a favourable alignment for polymerisation (Figure App.C1C). The slight decrease during the degradation can be due to the opening of the double bonds and formation of degraded species. The stability of the
100 | 212
peak position (Figure 21A.III) indicates that the double carbon-carbon bonds (C=C) are not significantly influenced by the environment during these processes.
101 | 212
Figure 21: C=C and C=O stretching modes. A.I) ATR-FTIR spectra for C=C and C=O stretching at different UV areal irradiation doses, where the results from the deconvolution are superposed. A.II) The normalised area underneath the peaks and A.III) Δ peak position versus areal irradiation doses for characteristic peaks at 1539, 1692, 1698, 1736 cm-1. B) Simulated IR spectra from C=O stretching mode, where hydrogen bonds exist intra or inter layers.
102 | 212
C=O stretching mode reveals different types of hydrogen bonds at the head group
The diacetylene monomer used in this work, TRCDA, has a carboxyl head group. IR peaks from carbon-oxygen double bond stretching (C=O) are commonly employed to study the hydrogen bonding of the head group208. In our samples, a clear C=O stretching peak was observed as expected (Figure 21A.I). The deconvolution revealed three main peaks in the monomer state TRCDA (violet, green and yellow) and broad background peaks (grey) that appear upon UV exposure (Figure 21A.I). The origin of these multiple peaks has been previously explained by the number of the hydrogen bonds per C=O, where the more hydrogen bonds are present, the lower the wavenumber is190,191,204. In our samples, a lack of the characteristic peak for the free OH group around 3500 cm-1 (Figure 19) suggests that most of the carboxyl head groups are hydrogen bonded203,204. Another interpretation hinges on the conformation of the hydrogen bonds rather than on their number such as inter- or intra-layer bonding. Simulation shows that the presence of intra- (simulated with a tetramer and a heptamer, where hydrogen bonds are assumed in parallel to the conjugation) or inter-layer hydrogen bonds (simulated with two monomers, where a head-to-head hydrogen bond is assumed) causes splitting in the C=O stretching peaks (Figure 21B). This implicates that the simulation does not reject the second explanation.
All three peaks present a slight areal decrease upon UV exposure (violet, green and yellow in Figure 21A.II), which suggest the breaking of the hydrogen bonds and the declined interactions between head groups during the UV exposure. Among these three peaks, the 1692 cm-1 peak (yellow) shifted its position slightly in the initial UV irradiation and levelled off immediately during the blue state (Figure 21A.III). In contrast, both 1698 cm-1 (green) and 1736 cm-1 (violet) peaks kept shifting towards higher wavenumbers also during the red state.
This is because the conformational changes during the polymerisation and the blue-to-red transition are different, where different types of hydrogen bonds are broken either during the polymerisation or blue-to-red transition or both. These results are consistent with the picture coming from the alkyl chain peaks (CH2 stretching and CH2 in-plane bending in Figure
103 | 212
22 and Figure 23) and the previous reports58,101,190,203,204, where diacetylene monomers form ordered hexagonal lattice, while the polymerisation distorts the lattice and induce a higher disorder as will be discussed later. Previously a similar change in the C=O intensity or its peak position has been observed with other amphiphilic molecules during the variation of temperature190,191,204,209. As amphiphilic molecules undergo the phase transition induced by temperature, the reduction of the structural order caused a decrease in the C=O stretching signal and a shift in the wavenumber towards higher positions, indicating a higher degree of freedom of the C=O group190,191,204,209. The effect of solvents with different polarity on the hydrogen bonds between carboxyl head groups has also been investigated210,211. When lipids were exposed to a more polar environment the C=O stretching shifts towards lower wavenumbers210,211. This occurs due to the involvement of the lone pair in the non-covalent interaction, which withdraws electron density from the C=O bond, weakens it, and consequently lowers the characteristic resonance frequency.
104 | 212
Figure 22: CH2 in-plane bending mode. A.I) ATR-FTIR spectra for CH2 in-plane bending mode at different UV areal irradiation doses. A.II) The area underneath the peaks versus UV areal irradiation doses and A.III) Δ peak position versus areal irradiation doses. AI) ATR-FTIR spectra, II) the analysis of the area underneath the peak versus UV irradiation energies, where arrows indicate irradiation energies at which the deconvolution was studied, and III) the peak shift versus UV irradiation energies are shown. The region within the two vertical dashed lines in A.I indicates the wavenumber areas that were interpreted as one peak. B) Overlaying plot of the original spectra and the sum of the deconvoluted spectra, where the good overlap ensures the successful deconvolution. C) Deconvoluted spectra of CH2 in-plane bending ATR-FTIR signal at 0 J/cm2 (in black), blue polymer (40.6 J/cm2 in blue), and red polymer (162.5 J/cm2 in red).
CH2 in-plane bending and CH2 symmetric stretching indicate an anisotropic packing
The information on the CH2 side chains can be obtained by analysing the in-plane bending of CH2 (Figure 22A) since it is sensitive to interdigitation changes, hence the packing, of alkyl chains. Van der Waals forces are the main responsible for influencing this type of bond movement. The deconvolution of this peak revealed two components (Figure 22A.I). The peak
105 | 212
with a lower wavenumber (1463 cm-1 in yellow) slightly decreases its intensity during the UV exposure, while the one with a higher wavenumber (1465 cm-1 in violet) initially increases, followed by a slight decrease. This cannot be explained by uniform packing or expansion, because if the sample were compacting/expanding in all directions we would expect a monotonous decrease/increase in both components. Previously, the deconvoluted double peaks from the CH2 in-plane bending signal has been interpreted as a consequence of the interactions of alkyl chains in different directions212. In the their work, high and the low wavelength peaks have been associated to the short and the long lattice of the fundamental cells respectively212, while the difference between these wavenumbers has been linked to the lattice angle of the two conformations212,213. Our result implies that there is a direction in the diacetylene lattice that is rather slightly expanding (un-packing) during the initial state of the polymerisation among other directions that are contracting (packing). This finding is interesting because such a lattice deformation of PDA during polymerisation has been monitored previously by gradient incidence X-ray electron diffraction (GIXD) experiments, where a lattice constant in one direction increases and the other direction decreases during the monomer-to-blue state polymerisation58. This detection of an additional expansion of the TRCDA assemblies at the initial state of the polymerisation is a benefit from the deconvolution analysis. Typically X-ray crystallography or theoretical works show that the area per molecule declines by around 1% during the polymerisation (monomer-to-blue polymerisation)58 and it does by 2 - 21%58,101 during the blue-to-red transition. The shrinkage of PDAs upon polymerisation has been observed also by atomic force microscopy214,215 (AFM) and scanning tunnelling microscopy61 (STM), thus the compression is a well-recognised phenomena but the expansion has been reported much less. For the peak position, a slight shift towards a higher wavenumber was observed for both peaks, indicating packing216.
106 | 212
Figure 23: CH2 symmetric stretching mode. A.I) ATR-FTIR spectra for CH2 symmetric stretching mode at different UV areal irradiation doses. A.II) The area underneath the peaks versus UV areal irradiation doses and A.III) Δ peak position versus areal irradiation doses.
To further study the effects of the polymerisation on the alkyl chain packing, the signals arising from the CH2 symmetric stretching vibrations217,218 were also analysed (Figure 23A.II). These wavenumber regions are known to be correlated with the conformational order of the alkyl tails216,218,219,220, thus have been studied extensively to clarify the chain conformation and the lattice structure216. Note that the corresponding asymmetric CH2
stretching vibration was also observed but discarded from the analysis, because they are known to be difficult to interpret due to the influences from Fermi resonances220,221,222.
107 | 212
The deconvolution revealed two main peaks (violet and yellow spectra in Figure 23A.I) apart from the broad background peaks (grey spectra) that overlap with CH2 asymmetric stretching mode. The area under the absorption curve in CH2 symmetric stretching reflects the inter-alkyl chain interactions, thus also packing220. When the alkyl tails are more compact (packing), the restricted movement of these vibrational modes results in the decrease of the
The deconvolution revealed two main peaks (violet and yellow spectra in Figure 23A.I) apart from the broad background peaks (grey spectra) that overlap with CH2 asymmetric stretching mode. The area under the absorption curve in CH2 symmetric stretching reflects the inter-alkyl chain interactions, thus also packing220. When the alkyl tails are more compact (packing), the restricted movement of these vibrational modes results in the decrease of the