Photodegradation of Azathioprine in the Presence of Sodium Thiosulfate

The effect of sodium thiosulfate (ST) on the photodegradation of azathioprine (AZA) was analyzed by UV-VIS spectroscopy, photoluminescence (PL), FTIR spectroscopy, Raman scattering, X-ray photoelectron (XPS) spectroscopy, thermogravimetry (TG) and mass spectrometry (MS). The PL studies highlighted that as the ST concentration increased from 25 wt.% to 75 wt.% in the AZA:ST mixture, the emission band of AZA gradual downshifted to 553, 542 and 530 nm. The photodegradation process of AZA:ST induced: (i) the emergence of a new band in the 320–400 nm range in the UV-VIS spectra of AZA and (ii) a change in the intensity ratio of the photoluminescence excitation (PLE) bands in the 280–335 and 335–430 nm spectral ranges. These changes suggest the emergence of new compounds during the photo-oxidation reaction of AZA with ST. The invoked photodegradation compounds were confirmed by studies of the Raman scattering, the FTIR spectroscopy and XPS spectroscopy through: (i) the downshift of the IR band of AZA from 1336 cm−1 to 1331 cm−1, attributed to N-C-N deformation in the purine ring; (ii) the change in the intensity ratio of the Raman lines peaking at 1305 cm−1 and 1330 cm−1 from 3.45 to 4.57, as the weight of ST in the AZA:ST mixture mass increased; and (iii) the emergence of a new band in the XPS O1s spectrum peaking at 531 eV, which was associated with the C=O bond. Through correlated studies of TG-MS, the main key fragments of ST-reacted AZA are reported.


Introduction
Azathioprine (AZA) or 6-((1-methyl-4-nitro-1H-imidazol-5-yl)thio)-9H-purine, is an immunosuppressive drug generally used to prevent the rejection of new organ transplants (kidney or liver) [1] or in other medical treatment of some autoimmune diseases (e.g., rheumatoid arthritis [2] and pemphigus [3]), inflammatory bowel illnesses [4] (e.g., Crohn's diseases, ulcerative colitis), multiple sclerosis and so on [5]. Once in the body, AZA is converted to mercaptopurine, a compound belonging to the class of purines, which will eliminate the effect of some cells in the body's immune system [6], which may increase the probability of developing an infection [7]. However, many techniques for the determination of Figure 1 shows the UV-VIS spectra of the AZA:ST samples in dark conditions and after 6 h of UV irradiation.
The UV-VIS spectra of AZA:ST in dark conditions, namely before UV irradiation, show a band at 276 nm belonging to the π-π * electronic transition coming from type 1-methyl-4-nitro-1H-imidazol-5yl [20]. As the UV exposure time of AZA:ST increased, we observed a gradual diminution in the band absorbance at 276 nm and the emergence of a band in the 315-405 nm spectral range, which was associated with the n-π * electronic transition belonging to thiopurine and 6-mercapto-purine [21]. Regardless of the ST concentration in the AZA/ST mixture, exposure of the ST-interacted AZA to UV light highlighted the emergence of a band with a maximum around 346-350 nm. Recently, such a band was also reported in the case of the interaction of AZA with H 2 O 2 [22]. gence of new C=O functional groups. Some experimental evidence on the photo-oxidation reaction of ST-reacted AZA confirmed the photodegradation pathway proposed in this manuscript by IR spectroscopy, Raman scattering and X-ray photoelectron spectroscopy. The thermal stability of AZA and ST-reacted AZA and the key fragments were highlighted by TG-MS studies. The UV-VIS spectra of AZA:ST in dark conditions, namely before UV irradiation, show a band at 276 nm belonging to the π-π * electronic transition coming from type 1methyl-4-nitro-1H-imidazol-5yl [20]. As the UV exposure time of AZA:ST increased, we observed a gradual diminution in the band absorbance at 276 nm and the emergence of a band in the 315-405 nm spectral range, which was associated with the n-π * electronic transition belonging to thiopurine and 6-mercapto-purine [21]. Regardless of the ST concentration in the AZA/ST mixture, exposure of the ST-interacted AZA to UV light highlighted the emergence of a band with a maximum around 346-350 nm. Recently, such a band was also reported in the case of the interaction of AZA with H2O2 [22]. Other changes noted in the UV-VIS spectra of the ST-interacted AZA with an increase in the time of UV exposure were: (i) a downshift of the band from 276 nm (AZA unreacted with ST) to 274 nm (Figure 1a Depending on the ST concentration in the AZA:ST mixture, significant variations were observed ( Figure 2). According to our previous study, the AZA PLE spectrum is dominated by the two bands in the 280-325 nm and 330-450 nm ranges [14]. As the ST concentration in the ST-reacted AZA samples increased from 25 wt.% to 50 wt.% and to 75 wt.%, the intensity ratio of the two PLE bands above (I280-325/I330-450) became equal to 1.01, 1.23 and 1.27, respectively. As the UV exposure time of the ST-reacted AZA samples increased up to 6 h, we observed: (i) a gradual increase in the PLE band's intensity in the 280-325 nm spectral range and an upshift from 282 nm (the AZA sample) to 304 nm (the AZA:ST 25 wt.%: 75 wt.% sample, Figure 2a Depending on the ST concentration in the AZA:ST mixture, significant variations were observed ( Figure 2). According to our previous study, the AZA PLE spectrum is dominated by the two bands in the 280-325 nm and 330-450 nm ranges [14]. As the ST concentration in the ST-reacted AZA samples increased from 25 wt.% to 50 wt.% and to 75 wt.%, the intensity ratio of the two PLE bands above (I 280-325 /I 330-450 ) became equal to 1.01, 1.23 and 1.27, respectively. As the UV exposure time of the ST-reacted AZA samples increased up to 6 h, we observed: (i) a gradual increase in the PLE band's intensity in the 280-325 nm spectral range and an upshift from 282 nm (the AZA sample) to 304 nm (the AZA:ST 25 wt.%: 75 wt.% sample, Figure 2a These changes were accompanied by an increase in the intensity of the PL spectra of the AZA:ST samples, as observed in Figure 3. The chemical reaction of AZA with ST induced a downshift of the AZA PL band from 488 nm (2.54 eV) [14] to 492 nm (2.52 eV), 506 nm (2.45 eV) and 503 nm (2.46 eV) for the AZA:ST samples with the ST concentrations in the reaction mixture of 25 wt.%, 50 wt.% and 75 wt.%, respectively ( Figure 3). After exposure of the AZA:ST samples with ST concentrations in the reaction mixture equal to 25 wt.%, 50 wt.% and 75 wt.% to UV for time of 10 h, we observed that the maxima of the PL spectra of these samples were situated at 552 nm, 544 nm and 531 nm, respectively, and the values were higher than that of AZA alone (530 nm). Increases in the emission bands' intensity took place through UV exposure as follows:  [14].   An explanation for this behavior must take into account the chemical reaction of AZA with ST. To sustain this hypothesis, the IR spectra of ST-reacted AZA ( Figure 4) are shown.

Results and Discussion
The IR spectrum of ST shows three bands of high absorbance, peaking at 995 cm −1 , 1115 cm −1 and 1161 cm −1 , which are in the vicinity of those reported by K. Khairan et al. at 993 cm −1 , 1109 cm −1 and 1156 cm −1 [23]. The IR bands at 1652 cm −1 and 3400 cm −1 are associated with the bending vibration of H2O molecules in ST [24][25][26][27]. The IR bands of AZA and the three ST-reacted AZA samples are shown in Table 1.    An explanation for this behavior must take into account the chemical reaction of AZA with ST. To sustain this hypothesis, the IR spectra of ST-reacted AZA ( Figure 4) are shown. The IR spectrum of ST shows three bands of high absorbance, peaking at 995 cm −1 , 1115 cm −1 and 1161 cm −1 , which are in the vicinity of those reported by K. Khairan et al. at 993 cm −1 , 1109 cm −1 and 1156 cm −1 [23]. The IR bands at 1652 cm −1 and 3400 cm −1 are associated with the bending vibration of H 2 O molecules in ST [24][25][26][27]. The IR bands of AZA and the three ST-reacted AZA samples are shown in Table 1. An explanation for this behavior must take into account the chemical reaction of AZA with ST. To sustain this hypothesis, the IR spectra of ST-reacted AZA ( Figure 4) are shown.
The IR spectrum of ST shows three bands of high absorbance, peaking at 995 cm −1 , 1115 cm −1 and 1161 cm −1 , which are in the vicinity of those reported by K. Khairan et al. at 993 cm −1 , 1109 cm −1 and 1156 cm −1 [23]. The IR bands at 1652 cm −1 and 3400 cm −1 are associated with the bending vibration of H2O molecules in ST [24][25][26][27]. The IR bands of AZA and the three ST-reacted AZA samples are shown in Table 1.    The main changes observed in Figure 4,   Table 2 [14,29,30]. The reaction of AZA with ST induced the following changes in the Raman spectra of the two compounds: (i) an upshift of the Raman lines of AZA in the 1460-1550 cm −1 spectral range; (ii) a change in the intensity ratio of the Raman lines peaking at 1305 cm −1 and 1330 cm −1 (I 1305 /I 1330 ) from 3.45 (AZA) [12] Table 2 [14,29,30].

AZA [14]
The  (Figure 6b), we can calculate the Auger parameter, which was found to be equal to 2061.9 eV; the value was attributed to sulfites. Figure 6c highlights, in the case of the ST XPS S2p spectrum, two complex bands with maxima at 162.1 and 168.1 eV, and the values are in good agreement with the those reported by B.J. Lindberg et al. [31]. The S2p spectrum was fitted with doublet peaks (S2p 3/2 and S2p 1/2 ) due to spin orbit splitting. The doublet's separation was 1.2 eV and the peaks were constrained to a 2:1 area ratio during fitting. The bands at 162.1-163.3 eV and 168.1-169.3eV are attributed to Na 2 S 2 O 3 compounds, as stated in the literature [28,32]. The bands at 166.4-167.6 eV are attributed to the Na 2 SO 3 compound, while the bands at 169.4-170.6 eV are ascribed to Na 2 SO 4 , as also highlighted in the Na1s spectrum [28,33].    [34]; (ii) an AZA XPS O1s spectrum, which shows an intense band at 532 eV accompanied by a band of low intensity at 535 eV, associated with the C-O bonds and water traces [34]; (iii) an AZA XPS N1s spectrum that shows three bands at 398.6, 400.3 and 405.3 eV associated with the C=N-C bond [34], the C2NH secondary amine [34,35] and the NO2 group [36]; and (iv) an AZA XPS S2p spectrum, which highlights, by deconvolution, two bands at 163.4 and 164.6 eV coming from the thiol groups [35].    [34]; (ii) an AZA XPS O1s spectrum, which shows an intense band at 532 eV accompanied by a band of low intensity at 535 eV, associated with the C-O bonds and water traces [34]; (iii) an AZA XPS N1s spectrum that shows three bands at 398.6, 400.3 and 405.3 eV associated with the C=N-C bond [34], the C 2 NH secondary amine [34,35] and the NO 2 group [36]; and (iv) an AZA XPS S2p spectrum, which highlights, by deconvolution, two bands at 163.4 and 164.6 eV coming from the thiol groups [35].   [34]; (ii) an AZA XPS O1s spectrum, which shows an intense band at 532 eV accompanied by a band of low intensity at 535 eV, associated with the C-O bonds and water traces [34]; (iii) an AZA XPS N1s spectrum that shows three bands at 398.6, 400.3 and 405.3 eV associated with the C=N-C bond [34], the C2NH secondary amine [34,35] and the NO2 group [36]; and (iv) an AZA XPS S2p spectrum, which highlights, by deconvolution, two bands at 163.4 and 164.6 eV coming from the thiol groups [35].
(e,f) show the XPS Na1s spectrum and the Na KLL Auger spectrum, respectively, of ST-reacted AZA.
According to Figure 7a2,b2,c2d2, significant changes are induced as a consequence of the reaction of AZA with ST. We noted: (i) the emergence of a new band in (a) the XPS C1s spectrum peaking at 287.7 eV, which was associated with the C=O bond (Figure 7a2,b) in the XPS O1s spectrum peaking at 531 eV, which was correlated with the presence of the C=O bond ( Figure 7b2) The reaction products shown in Scheme 1 are sustained by: (i) the downshift of the IR band of AZA from 1336 cm −1 to 1331 cm −1 , attributed to N-C-N deformation in the purine ring, as a consequence of the transformation of the -N=CH-NH-bonds into the -NH-C(O)-NH-bonds; and (ii) the emergence of a new band in the XPS O1s spectrum peaking at 531 eV, which was associated with the C=O bond.
The TG-DSC curves corresponding to the AZA sample (Figure 8a) highlight its thermal stability up to 249 °C. The thermal decomposition occurring after this temperature takes place in two steps. In the first step (247-270 °C), a loss in mass of about 2.2% is observed on the TG curve. This is accompanied by a succession of two visible peaks on the DSC curve. The first is a small endothermic peak at 250 °C followed by an exothermic one with a higher intensity, centered at 269 °C. The second step of the thermal decomposition 270-500 °C has a mass loss of ~7.2% and is accompanied by a succession of endothermic peaks of low intensity (DSC curves) due to oxidation effects, considering that the sample was measured in synthetic air. The thermal decomposition behavior of Na2S2O3 is shown in Figure 8b. The first two endothermic peaks (DSC curves) with maxima at 58 °C and 95 °C, respectively, are most likely due to the process of evaporation of sulfur and of The reaction products shown in Scheme 1 are sustained by: (i) the downshift of the IR band of AZA from 1336 cm −1 to 1331 cm −1 , attributed to N-C-N deformation in the purine ring, as a consequence of the transformation of the -N=CH-NH-bonds into the -NH-C(O)-NH-bonds; and (ii) the emergence of a new band in the XPS O1s spectrum peaking at 531 eV, which was associated with the C=O bond.
The TG-DSC curves corresponding to the AZA sample (Figure 8a) highlight its thermal stability up to 249 • C. The thermal decomposition occurring after this temperature takes place in two steps. In the first step (247-270 • C), a loss in mass of about 2.2% is observed on the TG curve. This is accompanied by a succession of two visible peaks on the DSC curve. The first is a small endothermic peak at 250 • C followed by an exothermic one with a higher intensity, centered at 269 • C. The second step of the thermal decomposition 270-500 • C has a mass loss of~7.2% and is accompanied by a succession of endothermic peaks of low intensity (DSC curves) due to oxidation effects, considering that the sample was measured in synthetic air. The thermal decomposition behavior of Na 2 S 2 O 3 is shown in Figure 8b. The first two endothermic peaks (DSC curves) with maxima at 58 • C and 95 • C, respectively, are most likely due to the process of evaporation of sulfur and of some water molecules present on the surface. These are accompanied by two mass losses of 6.1% (the first step) and 4.43% (the second step) on the TG curve. Over the range from 123 • C to 140 • C, there is a large mass loss (~17.1%) accompanied by an endothermic peak centered at 137 • C. This effect can be assigned to thermal degradation of ST [37][38][39]. Due to the fact that the sample was analyzed in synthetic air, in the sequence of reactions specific to the thermal decomposition of Na 2 S 2 O 3 , an oxidizing effect of the released sulfur may also occur. The endothermic peak at 335 • C can be assigned to such a process [37][38][39]. The sequence of the next three exothermic peaks with maxima at 403 • C, 440 • C and 465 • C was associated with the vaporization of the gas mixture resulting from thermal decomposition, especially elemental sulfur. This hypothesis is supported by the boiling point of sulfur being 444.6 • C [37][38][39]. Figure 8c shows the thermal behavior of the AZA after UV irradiation. In this case, we observed an exothermic peak at 259 • C, succeeded by thermal decomposition in the range of 270-500 • C. Figure 8d shows the thermal behavior of the ST-reacted AZA (AZA:ST 50 wt.%: 50 wt.% sample). The thermal behavior of this sample sums up the effects identified in the two samples described above (AZA and Na 2 S 2 O 3 ). For the thermal decomposition carried out over the entire measured range (RT-500 • C), the TG curve registered a continuous decrease in a sequence of seven steps. The total loss of mass was~14%. One conclusion would be that the chemical reaction between AZA and ST generates better thermal stability than that of each separate analyzed compound. For the first time for these compounds, mass spectrometry was used to perform a qualitative analysis of the generated gas components. The data were acquired with MID (Multiple Ion Detection) measurements. In this method, the resulting gases were calculated by multiplying the intensities by previously determined factors [40]. Those calibration factors were determined by performing a gas-specific calibration measurement. In our work, the MS data were acquired through calibration of the ambient air species (argon, nitrogen and oxygen). The recorded MS curves are shown in Figure 9. The possible gases and/or molecular fragments resulting from the pyrolysis of the four samples are presented in Table 3. The key fragments were identified using the NIST database library [41]. some water molecules present on the surface. These are accompanied by two mass losses of 6.1% (the first step) and 4.43% (the second step) on the TG curve. Over the range from 123 °C to 140 °C, there is a large mass loss (~17.1%) accompanied by an endothermic peak centered at 137 °C. This effect can be assigned to thermal degradation of ST [37][38][39]. Due to the fact that the sample was analyzed in synthetic air, in the sequence of reactions specific to the thermal decomposition of Na2S2O3, an oxidizing effect of the released sulfur may also occur. The endothermic peak at 335 °C can be assigned to such a process [37][38][39]. The sequence of the next three exothermic peaks with maxima at 403 °C, 440 °C and 465 °C was associated with the vaporization of the gas mixture resulting from thermal decomposition, especially elemental sulfur. This hypothesis is supported by the boiling point of sulfur being 444.6 °C [37][38][39]. Figure 8c shows the thermal behavior of the AZA after UV irradiation. In this case, we observed an exothermic peak at 259 °C, succeeded by thermal decomposition in the range of 270-500 °C. Figure 8d shows the thermal behavior of the ST-reacted AZA (AZA:ST 50 wt.%: 50 wt.% sample). The thermal behavior of this sample sums up the effects identified in the two samples described above (AZA and Na2S2O3). For the thermal decomposition carried out over the entire measured range (RT-500 °C), the TG curve registered a continuous decrease in a sequence of seven steps. The total loss of mass was ~14%. One conclusion would be that the chemical reaction between AZA and ST generates better thermal stability than that of each separate analyzed compound. For the first time for these compounds, mass spectrometry was used to perform a qualitative analysis of the generated gas components. The data were acquired with MID (Multiple Ion Detection) measurements. In this method, the resulting gases were calculated by multiplying the intensities by previously determined factors [40]. Those calibration factors were determined by performing a gas-specific calibration measurement. In our work, the MS data were acquired through calibration of the ambient air species (argon, nitrogen and oxygen). The recorded MS curves are shown in Figure 9. The possible gases and/or molecular fragments resulting from the pyrolysis of the four samples are presented in Table 3. The key fragments were identified using the NIST database library [41].       The TG-DSC curves showed that the thermal decomposition of the samples is a complex process and takes place throughout the entire measured range (25-500 • C). The MS curves ( Figure 9) confirmed this fact by continuously recording the gases and/or molecular fragments resulting from these decompositions. The almost linear aspect of the MS curves is due to the rather low concentrations of the compounds resulting from the thermal degradation of the samples. A more concrete estimation of these is highlighted in Table 3 by the percentages of relative abundance. The large number of the resulting compounds is justified by the complex chemical reaction mechanisms specific to the analyzed materials. Most of the recorded gases are energy-stable and were identified with the abovementioned database. However, in the thiosulphate spectrum, there were two molecular fragments (m/e = 40 and m/e = 47, respectively) that could not be identified. They most likely correspond to metastable compounds obtained from the reaction of the sample with the working atmosphere (synthetic air). The two possible fragments are sodium hydroxide (m/e = 40) and nitrous acid (m/e = 47). To study the impact of ST on the photodegradation of AZA by UV-VIS spectroscopy, the following solutions of AZA and ST in C 2 H 5 OH were prepared: The IR spectra of the three samples of AZA:ST were registered with a Vertex 80 FTIR spectrophotometer from Bruker (Billerica, MA, USA).

AZA
The Raman spectra of the AZA:ST samples were recorded at λ exc = 1064 nm with a RFS100S FT Raman spectrophotometer from Bruker (Ettlingen, Germany).
The XPS spectra of ST, AZA and AZA:ST were achieved with a SPECS spectrometer (SPECS Gmbh, Berlin, Germany) endowed with a PHOBIOS 150 analyzer and an Al Kα source.
For the studies using UV-VIS and FTIR spectroscopy, Raman scattering and XPS spectroscopy, the photodegradation of AZA:ST was achieved by UV irradiation at 253 nm under a 350 W mercury-vapor lamp.
The PL and PLE spectra of the AZA:ST samples were measured with a Flurolog-3 spectrophotometer (FL3-221 model) from Horiba Jobin Yvon (Palaiseau, France), endowed with a Xe lamp of 500 W.
The thermal analysis experiments were undertaken with a SETARAM Setsys Evolution 18 Thermogravimeter (Caluire-et-Cuire, France) (Al 2 O 3 crucibles) in TG-DSC mode in the range of 20-500 • C. Samples with initial mass of ≈11 mg were measured in synthetic air (20%O 2 ; 80%N 2 ) with a flow gas rate of 16 mL/min. The heating rate was 5 • C/min. The accuracy of the heat flow measurements was ±0.001 mW and the temperature precision was ±0.01 • C. The analysis by mass spectrometry was performed by monitoring the gases with a Setaram QMS 301 Omnistar Pfeiffer mass spectrometer (Pfeiffer-Vacuum.com, Berlin, Germany), which was coupled to the Setaram Setsys Evolution 18 equipment.

Conclusions
In this article, new results concerning the reaction of AZA and ST assisted by UV light were reported. By using UV-VIS spectroscopy, photoluminescence, IR spectroscopy, Raman scattering and X-ray photoelectron spectroscopy, it was demonstrated that the photochemical reaction of AZA with ST induces: (i) the emergence of an isosbestic point at 340 nm, as a consequence of the emergence of new photodegradation products; (ii) a decrease in the I 280-320 /I 330-450 ratio in the PLE spectra from 0.98 to 0.77 as the ST weight in the AZA:ST reaction mixture increased; (iii) the PL band's intensity increased by~7-, 22-and~35-fold when the AZA:ST samples had an ST concentration equal to 25 wt.%, 50 wt.% and 75 wt.%, respectively (these values are smaller than those reported for AZA alone, as a result of the reaction of AZA with ST); (iv) the IR spectroscopy and Raman scattering analyses highlighted the changes in the purine ring's vibrational modes, as a consequence of the reaction of AZA with ST; (v) using XPS spectroscopy, the emergence of a band appearing to the C=O bond was reported; this result supported the idea that the proposed chemical reaction occurred when AZA reacts with ST; and (vi) the chemical reaction of AZA with ST generates a better thermal stability than that of each separate analyzed compound.