1 Introduction

Photochemical reactions are of particular interest in many different disciplines of current research. This applies to catalyst-based synthesis and mechanistic studies [1,2,3,4,5,6], medical application [7,8,9,10], development of photo-reactors [11], new production processes of polymers and gels [12, 13], photo-responsive materials [14,15,16,17,18], optical memories [19,20,21] and even chemical rotors [22,23,24]. For this purpose, it is necessary to identify all components involved and to be able to track these components during the course of the reaction. For simple systems, like A → B reactions, where the absorption bands of the two components are well separated and full conversion can be achieved, it is an easy task to obtain the UV/Vis spectra of both pure components. For more complex systems, however, for example with an unknown number of reaction components, and/or if the species have overlapping absorption bands or for more complex reaction schemes, this task becomes more difficult. This is especially true when a large number of components are involved and it is therefore particularly challenging to distinguish them. However, in order to fully understand the photochemical reaction behaviour, it is necessary to obtain pure component spectra for each individual component separately. With respect to photochemical reactions, absorption spectra and the resulting photochemical properties like absorption coefficients ε are of particular interest. With the help of these, the absorption characteristics and the mechanisms can be described in detail and further photochemical values like quantum yields Φ become accessible [25]. Today, there still seems to be a problem in obtaining the absorption spectra of pure components for complex photochemical systems with negative consequences for further applications [26, 27]. For systems with a small number of stable components, separation of the involved species for example via HPLC can help obtain spectra of the pure components. However, this is often time-consuming or impeded by limited lifetimes. Thus, if there are more than two or an unknown number of components with overlapping absorption bands involved, other methods have to be used [28, 29]. If the number of components can be estimated mathematically [30], indirect hard modelling (IHM) can be used [31, 32]. If the number of components is known, a system of equations can be set up from rate equations of each component with the aid of a reaction model [33,34,35]. Multivariate curve resolution (MRC) and optimised variants have proven to be methods that correctly represent the reaction process for each individual component and the corresponding spectra of all pure components, independent of the selected measurement technique. These methods have been shown to lead to even superior results if complementary measurement techniques are used [36,37,38,39]. However, it is not always necessary to use such methods to obtain pure spectra of components of a photochemical reaction. As we will demonstrate herein (Scheme 1), even very simple approaches based on the complementary use of spectroscopic techniques (NMR and UV/Vis) lead to correct results with little mathematical effort.

Scheme 1
scheme 1

Schematic depiction of the iterative procedure to determine photochemical properties using consecutive illumination, UV/Vis and NMR spectroscopic measurements

Using a combination of UV/Vis and NMR spectroscopy for this purpose was described for azobenzene as an example, however, other approaches are also feasible [27, 40, 41]. The approach relies on the quantitative nature of NMR together with its structural information for obtaining the identity and concentrations of individual species, thus providing the complementary information missing in the UV–Vis spectra of complex systems. We propose here, that it is sufficient for many simple but also more complex photochemical systems, for example for photochromic substances or chemical actinometers, and that it provides reliable values for photochemical properties.

This will be demonstrated on two compounds belonging to the class of dithienylethenes (DTEs, Scheme 2). Previous investigations on both DTE systems describe their high potential as electrically and photochemically switchable systems [42]. In particular, DTE1 (Scheme 2) shows an exceptional behaviour: its response to electrical stimuli is complete reversible without any signs of fatigue. From photochemical point of view, however, the cyclization reaction is not satisfactorily reversible and comes with irreversible formation of by-products [43]. In comparison, DTE2 shows high reversibility with little or no photobleaching, decomposition or any by-products. Although both systems were previously investigated photochemically, no pure component spectra or characteristic photochemical properties were determined. Hence, to obtain pure component UV/Vis spectra of all species involved in the two photoreactions shown, the combination of illumination, UV/Vis and NMR spectroscopy was executed in this work. With this powerful combination, the photo-cyclization can be monitored, the species identified and quantified. Thus, pure component spectra as well as photochemical properties for each individual species become accessible. It is to note, however, that for systems with limited life-time, this stepwise approach will naturally lead to erroneous results and thus a simultaneous and quantitative measurement of UV–Vis and NMR spectra would be very beneficial. Alternatively, deconvolution of the UV-spectra can be employed assuming the absorption bands to be of Gaussian shape, e.g. [44].

Scheme 2
scheme 2

Reaction scheme of 1 and 2 upon illumination with UV (365 nm) for photo-cyclization (reversible) and green light (525 nm) for photo-cyclo-reversion. The last reaction step only occurs for 1 upon continuing UV illumination of 1c and yields an irreversible species 1irr. This is not observed for 2 (no 2irr is observed) and a PSS can be reached

2 Experimental

2.1 Concept

As some knowledge of the reaction system is required to fully describe the total reactions process and absorption behaviour, we assume here that, the photoreactions of DTEs can be written as consecutive \(A\rightleftharpoons B(\to C)\) reactions and that the species do not possess limited life-times. As a first step, the species occurring in the sample have to be identified and quantified (requirement 1). Second, the absorbance of the sample at each time/composition point needs to be known to obtain the individual absorbances of the species (requirement 2). Both spectroscopic techniques, UV/Vis and NMR, reveal important and complementary information for the photochemical reactions. Using NMR spectroscopy, all occurring species can be identified and with the knowledge of the total concentration in the sample, concentrations of each individual species can be calculated from the ratio of the corresponding NMR signals (fulfilling requirement 1). From UV/Vis series, time-dependent absorbances of the total sample are accessible. If all possible species \(i\) in the system are known, the total absorbance at a specific wavelength \({A}_{\lambda ,\mathrm{tot}}\) can be written as the sum of the absorbances of each pure species \({A}_{\lambda , i}\). Further following Lambert–Beer's law, \({A}_{\lambda ,\mathrm{tot}}\) can be calculated with Eq. 1 as linear combination of the pathlength \(d\), the concentrations \({c}_{i}\) and the wavelength-dependent extinction coefficients \({\varepsilon }_{\lambda , i}\) of each species.

$${A}_{\lambda ,\mathrm{tot}}=d\bullet \sum{c}_{i}\bullet {\varepsilon }_{\lambda , i}$$

The total absorbance \({A}_{\lambda ,tot}\) in Eq. 1 is dependent on the illumination time \(t\) giving a system of linear equations (in which the concentration from NMR is used). With the measurements of the absorbance at different time points \({A}_{\lambda ,\mathrm{tot}}(t)\), the solution of the system of linear equations results in corresponding absorbance of each species (and thus fulfilling requirement 2). This allows the calculation of extinction coefficients at all wavelengths \({\varepsilon }_{\lambda , i}\) and thus represents pure component spectra for each species in the photo-cyclization of compounds 1 and 2.

2.2 Characterisation and measurement methods

UV/Vis spectra were recorded on a V-630 spectrophotometer (JASCO) with 1 mm quartz cuvettes and standard grade solvents. UV/Vis data were processed with JASCO Spectra Manager and exported as ASCII files. All 1H-NMR spectra were recorded on a 400 MHz proton resonance frequency (Bruker AVANCE III HD, equipped with a 5-mm Bruker PABBO broadband probe (BB/1H-19F/D) with z-gradient). 5 mm economy NMR tubes (Wilmad LabGlass, WG-5-E7) were used for the preparation of those samples. DMF-d7 (99.5 atom % D) was purchased from Sigma-Aldrich in separate 0.75 mL ampoules and used as received. All measurements were performed at 300 K without spinning of the sample. Detailed acquisition parameters can be found in the SI. Bruker TopSpin (Version 3.5 pl 2) was used for processing NMR data. All spectra were referenced to the DMF signal at 8.03 ppm. After Fourier transformation, phase and baseline corrections were executed automatically using TopSpin. To obtain spectra with different conversion ratios, ex situ illuminations with LEDs of specific wavelengths were performed for successive time intervals. For this purpose, a custom-made photo-reactor with UV (365 nm) and green (525 nm) LEDs featuring a socket for a 5 mm economy NMR tube was used [45], ensuring constant and reproducible illumination. The approach described consists of sequentially recording NMR and UV/Vis spectra alternating with periods of illumination. Prior to the start of the measurement series, green light (525 nm) was used to ensure that the highest possible proportion of open species o is present in the respective sample. At first, both measurements (NMR and UV/Vis) were performed on the solution of either 2 or 1 in DMF-d7 before illumination with UV light (365 nm) was conducted. The sample was prepared in an NMR tube and a 1H NMR was recorded. For UV/Vis spectroscopy, the solution was transferred into a quartz cuvette. After retransferring the solution into the NMR tube, the ring-opening photoreaction was initiated upon illumination with UV light (365 nm) in a custom-made photo-reactor [45] and the measurement procedure was repeated, resulting in 1H-NMR and UV/Vis spectra series.

3 Results and discussion

To demonstrate the beauty of this approach for the determination of spectra of pure components, two DTE systems are considered in which the number of components is known and follows a \(A\rightleftharpoons B\) or \(A\rightleftharpoons B\to C\) reaction scheme. These are the identical compounds we used in our previous study. They were synthesised by the Gallei group. Details on the synthetic procedure can be found in our previous publication [42]. The validity of our approach is first demonstrated with 2 due to its simple reaction system with separated absorption bands for the ring-opened and ring-closed form of 2 and is subsequently applied to the more complex system of 1, with overlapping absorption spectra for all isomers of 1 (Scheme 2). It is to note, that we did not investigate the conformational preferences of the open compounds in detail as we assume the interconversion of the respective conformers to be fast (as was previously observed for similar compounds) [46] so that no influence of the interconversion of conformers on the photo-cyclization is expected. The measurements start at pure 2o/1o (obtained via green light) and follow the iterative procedure NMR, UV/Vis and illumination (see Scheme 1) with UV light.

The photo-induced cyclization of both DTEs with UV light can be monitored in the respective spectra series either by NMR signals (B, E) in the aliphatic and aromatic region or by the absorbance maxima in UV/Vis spectra (A, D). The absorbance maxima for 2o are at 320 nm, for 2c at 610 nm, for 1o at 325 nm and for 1c at 537 nm. For both DTEs, a decrease of the open form and a simultaneous increase of the closed form are observed during illumination with UV light. Additionally, during illumination of 1 with UV light, a third species appears within the first minutes, which is structurally characterised and confirmed to be 1irr [42] (absorbance and concentration profiles are shown in the SI Figure S5). Due to this irreversible rearrangement reaction, no photo-stationary state can be reached and the illumination was stopped after 900 s (a continued illumination followed by UV/Vis spectroscopy is given in the SI Figure S6). In comparison, applying the same procedure to a sample of 2, the illumination with UV light leads to a photo-stationary state of 70% 2c within a few minutes of illumination confirming the results of our previous investigation [42]. Reaching a photo-stationary state indicates an equilibrium state without consecutive reactions, decomposition or photobleaching. This can be verified by NMR and UV/Vis spectroscopy in which no change of signal intensity in 1H or absorption spectra is observed. By integration of baseline separated 1H NMR signals (2: 7-Ho, thiophene proton for 2o and 7-Hc, thiophene proton for 2c as well as 13-Ho, methyl group at thiophene for 1o, 7-Hc, thiophene proton for 1c and 7-Hirr, thiophene proton for 1irr), concentrations of each species can be calculated. Following EQ1 using time-dependent absorbances and these concentrations, the system of linear equations can be solved with the method of least squares. This yields extinction coefficients for both DTEs and each respective species for the measured wavelength range (Fig. 1, Table 1).

Fig. 1
figure 1

UV/Vis spectra series (left, A and D) and 1H NMR (400 MHz) spectra series (middle, B and E) of 2 and 1 in DMF-d7 at 300 K upon UV illumination (365 nm) in a custom-made photo-reactor equipped with a NMR tube socket. Resulting pure component spectra of 2o and 2c (right, C) as well as 1o, 1c and 1irr (right, F)

Table 1 Calculated extinction coefficients at the maxima and excitation wavelengths of all species in both dithienylethenes in DMF

These represent the pure component spectra, which provide information about the compound’s photochemical ability to absorb light at specific wavelengths and are essential for understanding and applying these systems. As last step, in order to be able to assess whether the results obtained are of a correct order of magnitude, we compared our results with results from previous studies and used (back-)calculated absorption spectra to quantify the error. Since 2 was only investigated in our previous study [42], comparability to literature is limited and thus we can only compare our values to related compounds (for further information see SI Figure S3 and Table S4). 1 has already been synthesised, investigated and extinction coefficients were determined by Browne et al. [47] However, a different solvent (CH3CN (ACN)) was used resulting in slightly different and shifted values at the maxima (M), isosbestic points (I) and shoulders (S). In both cases, extinction coefficients agree very well and also wavelengths are reproduced reliably. The same is demonstrated by calculating the absorption spectra with the extinction coefficients obtained (Fig. 2, for whole UV/Vis series, see SI, Figure S4).

Fig. 2
figure 2

Calculated and experimental UV/Vis spectra series of the photo-cyclization reaction of 2 (top) and 1 (bottom) upon illumination with UV light (365 nm) in DMF-d7 at room temperature. The absolute difference \(\left|\Delta Abs\right|\) is given at the bottom part of each spectrum separately for every wavelength. Illumination times are set as specified in the legends

The comparison between the calculated and the experimental absorption spectra results in very small errors, which further supports the accuracy of the procedure used.

4 Conclusion

In this work, we present an approach to extract the pure component spectra of photochemical systems with more than one absorbing species and overlapping absorption bands using a combination of consecutive illumination, UV/Vis and NMR spectroscopy and simple mathematical tools. NMR spectroscopy allows the identification of all species together with their quantification. Thus, by obtaining the concentration from NMR and absorbance time profiles from UV/Vis spectroscopy, the wavelength-dependent extinction coefficients of all individual species can be determined. This iterative procedure is applicable not only to simple but also to more complex photochemical/chromic reactions. This allows obtaining pure component spectra. We further used (back-)calculated absorption spectra to confirm the accuracy of the extracted data and to quantify errors. The approach described here thus provides a method to reliably determine photochemical properties of pure components in complex mixtures.