Catherine M.
Aitchison
a,
Michael
Sachs
b,
Marc A.
Little
a,
Liam
Wilbraham
c,
Nick J.
Brownbill
ad,
Christopher M.
Kane
a,
Frédéric
Blanc
ad,
Martijn A.
Zwijnenburg
*c,
James R.
Durrant
b,
Reiner Sebastian
Sprick
*ae and
Andrew I.
Cooper
*a
aDepartment of Chemistry and Materials Innovation Factory, University of Liverpool, 51 Oxford St, Liverpool L7 3NY, UK. E-mail: aicooper@liverpool.ac.uk
bDepartment of Chemistry and Centre for Processable Electronics, Imperial College London, 80 Wood Lane, London W12 0BZ, UK
cDepartment of Chemistry, University College London, 20 Gordon Street, London WC1H 0AJ, UK. E-mail: m.zwijnenburg@ucl.ac.uk
dStephenson Institute for Renewable Energy, University of Liverpool, Crown Street, Liverpool L69 7ZD, UK
eDepartment of Pure and Applied Chemistry, University of Strathclyde, Thomas Graham Building, 295 Cathedral Street, Glasgow G1 1XL, UK. E-mail: sebastian.sprick@strath.ac.uk
First published on 28th July 2020
Most organic semiconductor photocatalysts for solar fuels production are linear polymers or polymeric networks with a broad distribution of molecular weights. Here, we study a series of molecular dibenzo[b,d]thiophene sulfone and fluorene oligomers as well-defined model systems to probe the relationship between photocatalytic activity and structural features such as chain length and planarity. The hydrogen evolution rate was found to vary significantly with bridge head atom, chain length, and backbone twisting. A trimer (S3) of only three repeat units has excellent activity for proton reduction with an EQE of 8.8% at 420 nm, approaching the activity of its polymer analogue and demonstrating that high molar masses are not a prerequisite for good activity. The dynamics of long-lived electrons generated under illumination in the S3 oligomer are very similar to the corresponding polymer, both under transient and quasi-continuous irradiation conditions.
Oligomers are far less studied as photocatalysts despite their extensive use in organic photovoltaics31,32 and in other heterogeneous photocatalysis applications.33 Oligomers of phenylene (n = 2–6)12,34 and pyridine35 show low but measurable photocatalytic activity, and it seems that it has been generally assumed, but not proven, that longer chain lengths are required for good photocatalytic activity. This could be based on the assumption that longer conjugation lengths and red-shifted light absorption onsets in higher molecular weight materials are needed. However, this has not been clearly demonstrated: indeed, we do not observe high molecular weights for polymer photocatalysts when studied via matrix-assisted laser desorption/ionisation mass-spectrometry,17,36 although this does not prove definitively that longer polymer chains do not exist in these materials. A study of low molecular weight phenyl–triazine oligomers also showed that these materials outperformed their polymeric analogues.37,38 Besides this, oligomers have the advantage of being well-defined chemical structures with known molecular weights and end-groups.
As molecular compounds, oligomers can also have a well-defined secondary structure. Unlike polymers, their arrangement in the solid-state structures and their molecular conformations can often be studied using X-ray diffraction. Long-range order has been suggested to enhance the photocatalytic activity in carbon nitrides,39 phenyl–triazine oligomers,37,38 and covalent organic frameworks compared to an amorphous analogue,26 although all these materials also show differences in driving-forces or unreacted functional groups from incomplete polymerisation. As a result, it is difficult to decouple the effects of composition and secondary structure on photocatalytic activity. We have shown recently that the packing motif of a photocatalytically active molecular crystal can be hugely important, with crystalline and amorphous versions of the same chemical displaying orders of magnitude differences in the photocatalytic hydrogen production rate.40 Intermolecular interactions were also shown to be a key structural feature in water soluble porphyrin oligomers, whereby aggregation was needed for high photocatalytic activity.41 While it is unclear whether crystallinity is beneficial for photocatalytic activity in all cases, the ability to correlate structural properties with changes in catalytic activity is useful for developing a better understanding of the factors affect activity, thus guiding the design of future photocatalysts. As well as their ability to form crystals, the solubility of oligomers in common organic solvents allows us to compare homogeneous and heterogeneous hydrogen production photocatalysts directly.
In this work, we synthesised three series of oligomers to investigate the effect of various chemical and structural properties, such as chain-length and backbone twisting, on physical properties such as light absorption, excited state lifetime, and hydrogen evolution activity. Dibenzo[b,d]thiophene sulfone has been shown repeatedly to be an active monomer unit in hydrogen producing organic photocatalysts;12,16,26 this is thought to be related to both its hydrophilicity and associated efficient generation of long-lived polarons. We therefore investigated oligomers of dibenzo[b,d]thiophene sulfone (S1–S3, Fig. 1) as well as the more easily processable (that is, more soluble) 9,9-dimethyl-9H-fluorene oligomers MeF1–MeF3 (Fig. 1). To test the impact of crystal packing more directly, we also designed a set of photocatalysts with either phenyl or mesitylene substituents on a 9H-fluorene or dibenzo[b,d]thiophene sulfone ‘core’ (PSP, PFP, MSM and MFM, Fig. 1).
Crystal structures of S1 (space group symmetry: C2/c) and MeF1 (I41/a) were determined by single crystal X-ray diffraction using the commercially sourced, as received, chemicals. Crystal structures MeF2 (P), MeF3 (P21/n), PSP (P21/n), MSM (Ia), PFP (P21/n) and MFM (P) were determined by single crystal X-ray diffraction, after growing suitable quality crystals by slowly evaporating the organic solvent mixtures used for purification (see Tables S1–S3† for full details). All of the oligomers were found to crystallise without solvent, and it was these solvent free crystalline phases that were used for the photocatalysis experiments. The poor solubility of S2 and S3 in organic solvents limited crystal growth under similar conditions. Consequently, suitable quality single crystals of S2 (P) and S3 (P) were grown by sublimation at temperatures >400 °C. The crystal structures of MeF1–MeF3 are dominated by edge-to-face and offset π–π interactions between neighboring oligomers (Fig. 2d–f). By comparison, the S1–S3 structures are dominated by face-to-face π–π stacking interactions between the sulfone dimers, which are rotated by 180° in each layer to prevent steric clashes between the sulfone oxygen atoms (Fig. 2j–i).
Powder X-ray diffraction (PXRD) analysis indicated that the bulk prepared crystalline materials of MeF1, MeF2, MeF3, PSP, MSM, and PFP matched the simulated PXRD patterns for the single crystal structures, and these phase pure crystalline materials were used for subsequent photocatalytic experiments. Analysis of the PXRD patterns for S2 and S3 indicated that these materials were semi-crystalline after purification (Fig. S17 and S18†), with SEM analysis showing that S2 was comprised of well-defined cuboidal crystals of circa 100 nm × 500 nm (Fig. S20†), while S3 was made up of globular particulates (Fig. S21†). Although the sublimed materials of S2 and S3 appeared to be more crystalline and of the same phase as the bulk purified semi-crystalline materials (Fig. S17 and S18†), we observed partial decomposition of the oligomers under the sublimation conditions. This prevented us from preparing large enough quantities of the sublimed S2 and S3 materials to allow photocatalytic experiments. Further analyses were therefore conducted on the as-purified semi crystalline materials. The PXRD pattern of MFM indicated that the bulk material was a mixed phase, and we could only determine the structure of the P structure during this study. MFM was therefore used as a mixed phase during the subsequent photocatalytic experiments.
The residual palladium content of the oligomers from their synthesis was measured via inductively coupled plasma mass spectrometry (ICP-MS) after digestion with nitric acid; commercial monomers MeF1 and S1, as well as PSP, showed palladium levels below the detection limit of the instrument (approximately 10 ppm, see ESI† for full details). MSM and PFP also had low Pd contents of 0.0015 and 0.0031 wt% (15 and 31 ppm) while MeF2, MeF3 and MFM had Pd levels between 0.01 and 0.02 wt% (100–200 ppm). The insoluble oligomers S2 and S3 showed the highest Pd levels (0.22 and 0.26 wt% or 2200 and 2600 ppm) probably because, unlike the other materials in this study, they could not be recrystallised from solution.
UV-Visible and photoluminescence spectroscopy was carried out on the oligomers in the solid-state (reflectance) and in chloroform solution (Fig. S23–S32†). The absorption onset of all the oligomers was also predicted using time-dependent density functional theory (TD-DFT) calculations (Table S9†). The families MeF1–MeF3 and S1–S3 both showed the expected redshift in absorption onset with increasing chain length while the absorption onset of the mesityl substituted oligomers MSM and MFM were blue-shifted relative to their phenyl substituted analogues PSP and PFP (ESI, Table S4†). This blue-shift for MSM and MFM results from their less planar structure and the reduced conjugation between the central ring and the two peripheral rings. The experimental and the TD-DFT predicted optical gap were found to be well correlated (Fig. 3b). The extinction coefficients of all soluble oligomers in chloroform were measured at the wavelength of their absorption maxima: the molar extinction coefficient increased from 16500 M−1 cm−1 for MeF1 to 42800 M−1 cm−1 for MeF2 and 71100 M−1 cm−1 for MeF3. The phenyl-substituted oligomers PSP and PFP had extinction coefficients of 35300 and 40300 M−1 cm−1 at their maximum of absorption, while the mesityl substituted analogues MSM and MFM were determined to be 18800 and 29800 M−1 cm−1.
Fluorescence quantum yields for all materials were measured in chloroform. We find an increase in photoluminescence quantum yield (PLQY, see ESI for details†) with increasing chain-length for both the dibenzo[b,d]thiophene sulfone oligomers (S1–S3) and 9H-fluorene oligomers (MeF1–MeF3): from 10% for MeF1 to 94% for MeF2 and to unity for MeF3. Likewise, we found an increase in the PLQY from 11% for S1 to 73% for S2 and 77% for S3. The mesityl substituted fluorene unit MFM had a very low PLQY of 9% while the phenyl analogue PFP was highly photoluminescent with a value of 94%. MSM and PSP had more similar PLQYs of 65% and 76%.
The photoluminescence lifetime of the oligomers was measured by time-correlated single photon counting (TCSPC) experiments. All oligomers showed multi-exponential photoluminescence decays and, in general, the oligomers containing dibenzo[b,d]thiophene sulfone showed longer lifetimes than the fluorene based materials. For example, MSM and PSP had weighted-average photoluminescence lifetimes of 2.68 and 2.47 ns whereas MFM and PFP were 0.58 and 1.02 ns (Fig. S36 and Table S5†). This trend was also observed for chloroform solutions; MSM, PSP, PFP and MFM had weighted average lifetimes of 3.86, 2.42, 0.57 and 0.05 ns, respectively (Fig. S38†). In general, longer chain lengths were found to have shorter photoluminescence lifetimes; in the solid state S3 had a weighted-average lifetime of 3.00 ns whilst this increased to 4.85 ns in S2 and 5.71 ns in S1. A sample of S1 was also tested with 0.36 wt% Pd added to give it similar Pd content to S2 and S3 but showed little change from pristine S1 (Fig. S33†) with a weighted average lifetime (5.64 ns), within error. Similarly, the addition of high levels of palladium to S2 and S3 had little effect on the PL lifetimes. S2 loaded with 3 wt% palladium had an identical lifetime to S2 without added palladium, while the photoluminescence lifetime of S3 loaded with additional 3 wt% Pd decreased slightly from 3.00 ns to 2.76 ns (Fig. S33†). The emission spectra of MeF oligomers in the solid-state were more complex with multiple maxima but again showed a decrease in lifetime with chain length; MeF3 had emissions at 428 nm and 446 nm with lifetimes of 0.78 and 1.07 ns, respectively, while MeF2 had lifetimes of 1.43 and 1.98 ns (Fig. S35†). MeF1 was significantly longer lived with weighted lifetimes of 2.87 and 7.15 ns at 326 and 350 nm. In this case, a sample of the MeF1 loaded with Pd to give equivalent levels to the dimer and trimer (0.016 wt%) did show a significant decrease in lifetime with a weighted average lifetime of 1.71 ns at 326 nm and 5.53 ns at 350 nm (Fig. S34†). In chloroform solution, all of the MeF oligomers had very short lifetimes but the general trend was preserved with average lifetimes of 1.15, 1.07 and 0.83 ns for MeF1–MeF3 (Fig. S37 and Table S6†).
Static light scattering experiments were conducted on suspensions of the oligomers in water to determine the particle size of the catalysts (Fig. S39–S42†). The particles of the dibenzo[b,d]thiophene sulfone oligomers were by far the smallest with both S2 and S3 having a significant fraction of material smaller than 1 μm. The Sauter mean diameters (D[3,2])42 were 3.96, 0.87 and 1.79 μm for S1, S2, and S3 respectively. The phenyl and mesityl oligomers were slightly larger with D[3,2] ranging from 5.25 μm for PSP up to 11.6 μm for PFP whilst the MeF oligomers had D[3,2] of 24.2, 13.9 and 33.4 μm for MeF1–MeF3. S1 dissolved in the TEA/MeOH/water mixture, but S2 and S3 suspensions were also measured under these conditions and had D[3,2] values of 0.68 and 0.57 μm.
Suspensions of the PSP, MSM, PFP and MFM oligomers in water were also tested for hydrogen evolution using the Na2S/Na2SO3 scavenger (Fig. S48 and S49†). The phenyl substituted dibenzo[b,d]thiophene sulfone material (PSP) had a hydrogen evolution rate of 24.1 μmol h−1 g−1 under broadband illumination (λ > 295 nm filter, 300 W Xe light source), higher than its 9H-fluorene analogue PFP when tested under the same conditions (13.8 μmol h−1 g−1). The mesityl substituted dibenzo[b,d]thiophene sulfone (MSM) and 9H-fluorene (MFM) photocatalysts were significantly less active under broadband illumination with rates of 5.5 and 4.8 μmol h−1 g−1. Under UV light irradiation (275–400 nm filter, 300 W Xe light source) PSP was again the most active material within the series with a HER of 29.6 μmol h−1 g−1, followed by its mesityl analogue MSM (20.4 μmol h−1 g−1). PFP on the other hand was the least active oligomer under these conditions with HER of 8.7 μmol h−1 g−1, slightly less than that of mesityl substituted MFM (10.0 μmol h−1 g−1).
Suspensions of the dibenzo[b,d]thiophene sulfone oligomers were the most active materials in this study: using Na2S/Na2SO3 mixtures under broadband illumination (λ > 295 nm, 300 W Xe light source), the HER increased with oligomer length from of 15 μmol h−1 g−1 for S1 to 81 μmol h−1 g−1 for S2 to 281 μmol h−1 g−1 for S3 (Fig. S50†). Using UV light irradiation (275–400 nm filter), the activity of the monomer S1 increased to 33 μmol h−1 g−1, similar to S2 (50 μmol h−1 g−1) but significantly lower than S3 (162 μmol h−1 g−1) (Fig. S51†). In these experiments, the monomer S1 was loaded with 0.36% Pd by photodeposition to give loadings that were comparable to the dimer and trimer. Without added Pd, S1 had a lower rate of 8 μmol h−1 g−1 at λ > 295 nm. A TEA/MeOH/water (1:1:1) mixture gave rise to significantly higher HERs of 26, 414 and 2073 μmol h−1 g−1 for the monomer (S1), dimer (S2), and trimer (S3), respectively (Fig. S54†). Increased activity with increasing chain length was also shown when only UV light was employed for photolysis, albeit to a lesser extent than when broadband illumination was used; using a 275–400 nm filter S1 had a HER of 20 μmol h−1 g−1, S2 101 μmol h−1 g−1 and S3 526 μmol h−1 g−1 (Fig. S55†). The visible light activity of these materials was also tested using a λ > 420 nm filter. The monomer (S1) produced no detectable amount of hydrogen over 5 hours under these conditions but the dimer (S2) retained some activity with a HER of 26 μmol h−1 g−1 whilst the trimer (S3) was reduced by less than 50% compared to broadband illumination with a HER of 1125 μmol h−1 g−1 (Fig. S55†). Again, it should be noted that the monomer (S1) is partially soluble in the TEA/MeOH/water (1:1:1) mixture, hence, the results cannot be compared directly with the results for S2 and S3, which are completely insoluble during the experiment. The external quantum efficiencies for S2 and S3 were determined using a 420 nm LED as the light source (ESI, General methods†). S2 has a relatively low EQE of 0.4% consistent with its limited light absorption at this wavelength. For S3, on the other hand, a high EQE of 8.8% was determined.
Dibenzo[b,d]thiophene sulfone (S1) was also soluble in the TEA/MeOH/water mixture but, unlike MeF1, a significant drop of in activity over 5 hours of broadband spectrum irradiation (λ > 295 nm, 300 W Xe light source) was observed. The 1H NMR spectrum of S1 after photocatalysis indicated breakdown of the catalyst. To investigate this further, a sample of S1 was irradiated for 72 hours in the photolysis mixture. Analysis of this material by solution NMR spectroscopy (1H, 13C ATP, 1H COSY and HSQC Fig. S62–S66†) and mass spectrometry indicated the breakdown product was the triethylamine salt of [1,1′-biphenyl]-2-sulfonic acid. This is possibly due to oxidation of S1 by either singlet oxygen or superoxide anions generated from residual oxygen in the system. To further support this interpretation, addition of a singlet oxygen and superoxide anion scavenger (nickel(II) dibutyldithiocarbamate)44,45 caused an increase of the HER under the same conditions to 53 μmol h−1 g−1 and an extended stability for at least 23 hours (Fig. S63†). The material collected after this photolysis experiment was found to be S1 by 1H NMR spectroscopy with no breakdown products present (Fig. S62†).
By contrast, we found that the PXRD patterns of dibenzo[b,d]thiophene sulfone oligomers S2 and S3 showed no change after 5 hours of irradiation (λ > 295 nm, 300 W Xe light source) in TEA/MeOH/water. Longer-term photocatalysis experiments of oligomer S2 irradiated for 50 hours under broadband illumination (λ > 295 nm, 300 W Xe light source) showed good longevity, with a slow drop off in activity and a reduction in the photocatalytic rate of less than 50% after 50 hours (Fig. S57†), which is comparable to many polymer photocatalysts.46 Oligomer S3 showed a more rapid drop off in photocatalytic hydrogen evolution rates from 2080 μmol h−1 g−1 over the first five hours to 576 μmol h−1 g−1 over 50 to 55 hours. However, when the catalyst was collected by filtration, washed and dispersed in new TEA/MeOH/water mixture we observed a recovery of the photocatalytic activity to 2066 μmol h−1 g−1 (Fig. S58†). It is possible that the reduction in activity is caused by accumulation of TEA oxidation products that inhibit photocatalysis. In addition, samples of S2 and S3 irradiated for 72 hours under broadband illumination (λ > 295 nm, 300 W light source) showed very similar PXRD patterns, UV and IR spectra to the as synthesised material (Fig. S67–S70†). It is possible that oxidation by reactive oxygen species does occur in S2 and S3 but at a slower rate than S1. The fact that S1 is stable in the Na2S/Na2SO3 photolysis mixture indicates that insolubility in the dispersant may be required for stability or it is possible that TEA plays a role in catalyst breakdown.
Fig. 5a shows transient absorption decay kinetics of S2 and S3 suspended in TEA/MeOH/water where the probe wavelength was set to 600 nm, corresponding to the transient absorption peak of S3 as shown in the inset. In spectral shape as well as spectral position, this transient absorption feature is in very good agreement with the absorption signature of electron polarons in P10 which we previously observed in the form of a peak at 630 nm,16 and we thus assign the 600 nm peak in the present study to electron polarons in S3. The shown kinetics demonstrate that these electron polarons have long lifetimes up to the millisecond timescale, which is sufficiently long to drive proton reduction.16 In contrast, S2 lacks a comparable polaron absorption peak in the visible range and generally exhibits much lower signal amplitudes over the probed spectral range. The main transient absorption feature above 1000 nm might be assigned to the onset of NIR polaron absorption in this material. In any case, assuming a comparable electron polaron extinction coefficient for S2 and S3, the electron polaron yield for S3 is considerably higher than that of S2, in line with its substantially higher hydrogen evolution activity.
To probe photogenerated charges under operando photocatalytic reaction conditions, we now turn to quasi-continuous illumination using 365 nm LED pulses with a duration of 2.7 s. As shown in Fig. 5b, we observe the buildup of a population of reaction intermediates when our S3 suspension is illuminated with the LED, and a saturation regime (steady state) is reached once an equilibrium between their generation, recombination, and reaction is established. As shown in the inset, the accumulating reaction intermediates exhibit an essentially identical absorption spectrum to the transiently observed electron polarons in the case of S3, suggesting that this signal is due to the buildup of electron polarons under illumination. This behavior of S3 is again very similar to that of its polymer analogue P10, where electron polarons build up during reaction as their transfer to catalytic Pd clusters is relatively slow.50 Like in the transient experiments (Fig. 5a), S2 exhibits lower signal amplitudes and a much broader spectrum with an absorption towards the NIR. Given the comparable Pd content of S2 and S3, this lower signal amplitude demonstrates that the lower yield of electron polarons in this material can also be observed under reaction conditions.
Overall, S3 behaves very similarly to P10 from a photophysical point of view, which is consistent with the observation that its activity approaches that of its polymer analogue (1125 vs. 2825 μmol h−1 g−1).50 Long-lived oligomer-centred electron polarons are generated upon photoexcitation and accumulate on the oligomer under quasi-continuous illumination, and the differences in polaron yield correlate with the measured hydrogen evolution activity.
We can use the analysis above to understand activities trends in Tables 1 and 2. We focus initially on the MeF1–MeF3 and S1–S3 oligomers, and in case of MeF1 and S1 on the activity measured for the samples where additional Pd was added, such that all materials compared have similar Pd loadings. When using triethylamine as sacrificial hole scavenger, the photocatalytic performance of both oligomer series appears to be more limited by light absorption than driving force, with the most active materials being MeF3 and S3, respectively. This does not only follow from the trend in activity with oligomer length but also the fact that when switching from UV to broadband illumination rates uniformly increase. When using the Na2S/Na2SO3 solution as sacrificial hole scavenger, S1–S3 behave as in the presence of triethylamine, though with 5–7× smaller hydrogen evolution rates. By contrast, MeF1–MeF3 appear limited by driving force with MeF1 being the most active material with hydrogen evolution rates, if anything, being higher under UV than broadband illumination. It is perhaps not surprising that MeF1–MeF3 rather than S1–S3 would be limited by driving force as MeF1–MeF3 have a smaller driving force for sacrificial hole scavenger oxidation than S1–S3. As far as we are aware, the first direct evidence of the link between oligomer length, oligomer properties, and oligomer photocatalytic activity, although we have previously observed related structure–property–activity relationships for co-polymers with varying composition17 and 2D networks with different linker lengths.62
Material | Pd amount by ICP-MS (wt%) | HER λ > 295 nm irradiationa,b (μmol h−1 g−1) | HERa 275 > λ > 400 nm irradiationb (μmol h−1 g−1) |
---|---|---|---|
a Photocatalyst (25 mg) suspended in Na2S/Na2SO3 (aq.) (0.2 M/0.35 M, 25 mL), rate calculated as linear regression fit over 5 hours. b See ESI for filter characteristics. c Pd below the detection limit of the instrument, see ESI for full details. | |||
PSP | > 0.001c | 24.1 ± 0.8 | 29.6 ± 0.4 |
MSM | 0.002 | 5.5 ± 0.2 | 20.4 ± 0.7 |
PFP | 0.003 | 13.8 ± 0.8 | 8.7 ± 0.3 |
MFM | 0.014 | 4.8 ± 0.4 | 10.0 ± 0.5 |
MeF1 | > 0.001c | 4.1 ± 0.3 | 6.5 ± 0.6 |
MeF1 + Pd | 0.016 | 14.2 ± 0.3 | 28 ± 1 |
MeF2 | 0.011 | 3.4 ± 0.2 | 4.4 ± 0.2 |
MeF3 | 0.014 | 10.1 ± 0.3 | 6.6 ± 0.1 |
S1 | > 0.001c | 8.1 ± 0.8 | 25.4 ± 0.2 |
S1 + Pd | 0.36 | 14.7 ± 0.5 | 33 ± 0.7 |
S2 | 0.22 | 81 ± 2 | 50 ± 1 |
S3 | 0.26 | 286 ± 4 | 162 ± 6 |
Material | Pd amounta (wt%) | HER λ > 295 nm irradiationb,c (μmol h−1 g−1) | HER 275 > λ > 400 nm irradiationb,c (μmol h−1 g−1) |
---|---|---|---|
a Determined via ICP-MS after digestion of the sample. b Photocatalyst (25 mg) suspended in TEA/MeOH/water (1:1:1, 25 mL), rate calculated as linear regression fit over 5 hours. c See ESI for filter characteristics. d Catalyst partially soluble in mixture used for photocatalysis experiment. e Rate determined as linear regression fit over 3 hours or less due to catalyst instability. f Measurement of the amount of photodeposited Pd not possible due to catalyst solubility. g Due to lack of material these experiments were conducted with photocatalyst (5 mg) in TEA/MeOH/water (1:1:1, 5 mL), control experiments were also run for S2 and S3 under these conditions and showed little change in rate from the large volume experiments (see Fig. 4). | |||
MeF1 | < 0.001d | 4.7 ± 0.2d | 2.6 ± 0.3d |
MeF2 | 0.011 | 13 ± 0.4 | 4.3 ± 0.3 |
MeF3 | 0.017 | 37 ± 1 | 28 ± 1 |
S1 | < 0.001d | 26 ± 3d,e | 20.1 ± 0.4d,e |
S2 | 0.22 | 414 ± 9 | 101 ± 1 |
S3 | 0.26 | 2073 ± 82 | 526 ± 9 |
S1 + 3 wt% Pd | —f | 43 ± 3d,e | — |
S2 + 3 wt% Pd | 2.1 | 1369 ± 24g | — |
S3+ 3 wt% Pd | 2.5 | 6550 ± 150g | — |
Comparing the activities of the MeF1–MeF3 and S1–S3 oligomers, the S1–S3 oligomers showed significantly higher photocatalytic activities compared to the MeF1–MeF3, analogues. This is consistent with our previous work comparing their polymer analogues.12 These oligomers all form close packed structures and are expected to have small internal surface areas. However, external surface area varies significantly between samples, as shown by the factor of 30 variation in D[3,2] values obtained from SLS, and could influence photocatalytic activity. Thus, in addition to the larger driving force and red-shifted light absorption onset discussed above, the much smaller particle sizes of the S1–S3 oligomer compared to the MeF1–MeF3 series (ESI, Fig. S39–S40†) likely to contributes to their much higher hydrogen evolution rates. Aside from primary particle size, the improved dispersibility of the sulfone-containing oligomers is thought to be related to their high hydrophilicity. We also note that the face-to-face stacking observed in the S1–S3 crystal structures, and the accompanying intermolecular orbital overlap, could facilitate improved transfer of excitons or charged species compared to the less-overlapped, edge-to-face packing that dominates in the MeF1–MeF3 structures.
Interestingly, the visible light activity of the trimer S3 (1125 μmol h−1 g−1) is of the same order of magnitude as its polymer analogue, P10, under equivalent conditions (2825 μmol h−1 g−1).50 An external quantum efficiency (EQE) of 8.8% at 420 nm was determined for S3, which is close to the 11.6% reported for P10 and higher than most conjugated polymer photocatalysts.20,63–65 The high performance of S3 is in line with our data showing that its potentials, optical gap, polaron signatures and polaron dynamics under transient and quasi-continuous illumination conditions are similar to P10. S3 challenges the general assumption that long chain lengths are required for significant hydrogen evolution activity, and, as a molecular material, demonstrates that these sulfone-containing photocatalysts can achieve high performance despite primarily relying on intermolecular exciton/charge transfer rather than exciton/charge transport along a polymer backbone.66 Our recent study on 1,3,6,8-tetrakis(p-benzoic acid)pyrene67 showed a highly crystalline, hydrogen-bonded form of the molecule, with idealised pyrene–pyrene stacking, had excellent activity for proton reduction, with an EQE of 4.1%. S3 is a semi-crystalline material, whose secondary structure contains π-stacking interactions over much smaller domain sizes, and yet it shows even higher activity. While it appears that molecular materials can be effective hydrogen production photocatalysts without extended backbones or the long-range stacking found in hydrogen bonded frameworks (HOFs), it seems likely that these materials might achieve even higher performances in the presence of more efficient charge transport, which could enable faster transfer to catalytic sites.50
While the polaron signatures for S3 are very similar to P10, optical signals from reaction intermediates under transient and quasi-continuous illumination are much lower for S2. No sharp polaron peak is observed in the visible range, which is in line with the significantly lower activity of S2 compared to S3 and suggests that S2 likely gives too low polaron yields to be determined reliably via spectroscopy. The performance of S1–S3 improves significantly upon addition of more Pd, which is in line with P10 where transfer of long-lived polarons to catalytic Pd sites limits the hydrogen evolution activity of the material.50 This interpretation is further corroborated by the polaron-like signature of S3, which suggests that long-lived electrons reside on the oligomer rather than on Pd clusters under both transient and charge accumulation conditions. On the other hand, the exciton lifetime in the S1–S3 series, as quantified via fluorescence lifetime measurements, is affected only little by added Pd, suggesting that the performance increase upon addition of more Pd is indeed due to improved catalysis rather than more efficient charge separation at early times.
Switching the focus to the MSM, MFM, PSP and PFP oligomers, it appears that aside from chain length, the effective conjugation length can have a marked effect on the activity of oligomers. Crystal structures show the backbone dihedral angles of MSM (64.0(5)° and 96.6(5)° for conformer 1 in the asymmetric unit; and 86.6(5)° and 115.4(5)° for conformer 2) and MFM (78.0(2)° and 109.4(1)° for conformer 1 in the asymmetric unit; and 91.8(1)° and 108.1(1)° for conformer 2) are significantly larger than the comparable dihedral angles for PSP (19.2(3)° and 33.5(3)°) and PFP (11.80(7)° and 14.09(7)°), which is due to the phenyl groups in PSP and PFP being substituted with the sterically hindered mesityl groups in MSM and MFM (Fig. S16†). This lack of planarity reduces the conjugation length, blue-shifting the absorption onsets in the solid-state by 40 nm when comparing PSP to MSM and by 96 nm for PFP compared to MFM, with similar shifts in solution measurements. Comparing PSP and PFP to MSM and MFM this twisting also appears to affect the intensity of absorption as reflected in the significantly higher extinction coefficient of the phenyl-bearing analogues which is consistent with the higher oscillator strength predicted by DFT. The loss in light absorption for the mesityl substituted oligomers as a result of their reduced conjugation length is reflected in their lower photocatalytic activity under broad band irradiation. However, when using UV light only, the activities of PSP and MSM are similar, as are those of PFP and MFM, where the former have at least twice the activity. The smaller effect of the reduced conjugation length in the UV could be because the UV-only range includes more of the middle-UV (275–300 nm), where the mesitylated oligomers absorb more strongly (Fig. S29†), than the broadband illumination spectrum, reducing the effect of loss of absorption at longer wavelengths. In this case, it is perhaps significant that the dibenzo[b,d]thiophene sulfone-containing PSP and MSM oligomers have a larger driving force for scavenger oxidation than their fluorene-based PFP and MFM analogues. Alternatively, the more hydrophilic nature of dibenzo[b,d]thiophene sulfone containing materials may improve interaction with water at the surface of the crystal. Differences in particle sizes, however, do not seem to play a role here; SLS measurements show that the particle sizes for PSP and MSM are broadly similar to PFP and MFM (Fig. S41†).
Footnote |
† Electronic supplementary information (ESI) available. CCDC 1999747–1999756. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/d0sc02675a |
This journal is © The Royal Society of Chemistry 2020 |