Coumarin-4-ylmethyl- and p-Hydroxyphenacyl-Based Photoacid Generators with High Solubility in Aqueous Media: Synthesis, Stability and Photolysis

(Coumarin-4-yl)methyl (c4m) and p-hydroxyphenacyl (pHP)based compounds are well known for their highly efficient photoreactions, but often show limited solubility in aqueous media. To circumvent this, we synthesized and characterized the two new c4m and pHP-based photoacid generators (PAGs), 7-[bis(carboxymethyl)amino]-4-(acetoxymethyl)coumarin (c4mac) and p-hydroxyphenacyl-2,5,8,11-tetraoxatridecan-13-oate (pHP-t), and determined their solubilities, stabilities and photolysis in aqueous media. The two compounds showed high solubilities in water of 2.77 mmolL �0.07 mmolL 1 (c4m-ac) and 124.66 mmolL �2.1 mmolL 1 (pHP-t). In basic conditions at pH 9, solubility increased for c4m-ac to 646.46 mmolL � 0.63 mmolL , for pHP-t it decreased to 34.68 mmolL � 0.62 mmolL . Photochemical properties of the two PAGs, such as the absorption maxima, the maximum molar absorption coefficients and the quantum yields, were found to be strongly pH-dependent. Both PAGs showed high stabilities s24h �95% in water for 24 h, but decreasing stability with increasing pH value due to hydrolysis. The present study contributes to a clearer insight into the synthesis, solubilities, stabilities, and photolysis of c4m and pHP-based PAGs for further photochemical applications when high PAG concentrations are required, such as in polymeric foaming.


Introduction
(Coumarin-4-yl)methyl (c4m) and p-hydroxyphenacyl (pHP)based compounds are well known for their excellent photochemical properties such as their clean and highly efficient photo cleavage. [1] This was highlighted in an excellent review article by Givens et al., who pointed out that c4m and pHP derivatives are especially well suited for time-resolved biochemical and physiological applications. [1a] Furthermore, these two chromophores are easy to access synthetically, can cover a wide range of excitation wavelengths from 250 nm to 450 nm by adjusting their substituents and can be used without sensitizer. [1a-c,e,2] Therefore, c4m and pHP-based compounds have gained considerable attention in biochemical, [1g,2e,3] agricultural [4] and pharmaceutical applications. [5] They have been used for neurotransmitter release, [5b] enzyme catalysis, [3a] membrane acidification, [6] or for drug delivery of anticancer agents. [7] Barman et al. for instance used pHP-benzothiazolechlorambucil conjugates as a photoregulated drug delivery system due to its fast photocleavage and high biocompatibility. [5a] Moreover, c4m esters were employed to study proton migration in biological systems such as lipid bilayers. [6,8] They were also used as photoacid generators (PAGs) to release acidic compounds under UV irradiation in aqueous media. [2a,b] However, in many water-based applications where high PAG concentrations are required, like in the field of bioinspired hydrogels, [9] hydrogel modifications [10] or foaming of polymeric materials, [11] strong electrolyte PAGs are preferred compared to weak electrolyte PAGs like c4ms or pHPs. For such applications, diphenyliodonium compounds are often used as strong electrolyte PAGs, which were discovered by Crivello et al. in 1977. [12] However, many diphenyliodonium-based PAGs like diphenyliodonium nitrate or diphenyliodonium antimonate are toxic, which significantly limits their application. [13] In fact, such PAGs cannot be implemented into biological, physiologi-cal and medical applications. Nevertheless, Gargava et al. used diphenyliodonium nitrate as PAG to regulate the pH-dependent pore size of bioinspired hydrogel valves. [9a] Also Feng et al. applied diphenyliodonium nitrate as PAG to trigger light controlled shape memory hydrogels. [10] The process involved shape retention through coordination interaction between the imidazole groups of the poly(acrylamide-co-N-vinylimidazole) hydrogel and dissolved metal ions in the aqueous swelling agent. [10] Shape recovery of the hydrogel was achieved by switching off the complexation via PAG photolysis reaction due to the protonation of imidazole groups. [10] Diphenyliodonium nitrate was also used to phototrigger the self assembly of a 1,3 : 2,4-dibenzylidene-D-sorbitol hydrogel in a controlled way when the pH was lowered. [9b] There are approaches to circumvent the toxicity of diphenyliodonium nitrate by using other photoacid generators like diaryliodonium tetrakis (pentafluorophenyl) borate or diphenyliodonium hexafluorophosphate, but they frequently need additional photosensitizers. [11] Kovalenko et al. for instance applied commercially available diaryliodonium tetrakis (pentafluorophenyl) borate (Silcolease UV Cata211) with low toxicity, but needed 2-isopropylthioxanthone as photosensitizer to tailor the porous structure of polydimethylsiloxane foams. [11b] A further example where strong electrolyte PAGs were favoured was published by Schlögl et al. for the foaming of 3D printed polyacrylate films. [11a] In particular, they utilized diphenyliodonium hexafluorophosphate as PAG and a toxic anthracene photosensitizer in combination with carbonate particles to generate CO 2 as foaming agent. [11a] The implementation of PAGs enabled them to simultaneously foam and cure their 3D printed polymers. [11a] Such 3D printed porous materials are subject to current research. [14] Especially in the above described fields of hydrogel research and polymeric foaming, c4m-and pHP-based PAGs could be beneficial, since they do not exhibit a cationic core structure which often limits biocompatibility. However, c4m-and pHP-based PAGs are rarely used, presumably due to the restricted or undetermined solubility of c4m and pHP-based PAGs in aqueous media. In some publications, a water solubility of c4m and pHP derivatives was reported, [1g,2b-d,15] but a solubility in aqueous media was not yet quantified, which is a crucial parameter for the PAG selection process.
Thus, to develop PAGs that would be usable in aqueous media in high concentration, we designed a c4m and a pHP derivative where a bis(carboxymethyl)amino moiety or a tri (ethylene glycol) moiety, respectively, should ensure high water solubility. The synthesis of 7-[bis(carboxymethyl)amino]-4-(acetoxymethyl)coumarin (c4m-ac) and p-hydroxyphenacyl-2,5,8,11-tetraoxatridecan-13-oate (pHP-t) (Scheme 1) is described and the solubility in aqueous media is quantified. We aimed to synthesize c4m and pHP-based PAGs with solubilities well above 1 mM in aqueous media, which are referred to as good in the c4m and pHP literature. [1g,2b-d,15] We furthermore characterized these new c4m and pHP-based PAGs and determined their photochemical properties, their pH dependent stability and photolysis in aqueous media. We envision that our studies will contribute to an increased applicability of c4m and pHP-based PAG in aqueous media, where high PAG concentrations are needed, like in hydrogel research or the field of polymeric foaming.
The synthesis of c4m-ac is based on previous work by Hagen et al. and started with the alkylation of 7-amino-4methylcoumarin (1) with bromoacetic acid tert-butyl ester, [2c] followed by oxidation with SeO 2 to the corresponding aldehyde 1 b which was subsequently reduced with NaBH 4 to the primary alcohol 1 c. Deprotection of the carboxyl groups with trifluoroacetic acid yielded 1 d which was acetylated with acetic acid in the presence of 4-dimethylaminopyridine (4-DMAP) and 1ethyl-3-(3-dimethyl-aminopropyl) carbodiimide (EDC) to form c4m-ac. The 1 H NMR and 13 C NMR spectra of c4m-ac and its intermediates are given in Figures S1-5.
Generally, all characterization data indicate that the syntheses yielded c4m-ac, pHP-t and pHP-ac in sufficient purity for further characterization as described below.

Solubility Determination
As the solubility of many c4m-and pHP-based derivatives were only estimated roughly in previous studies, [1g,2b-d,15] we wanted to quantify the solubilities c max,w and c max,a of c4m-ac and pHP-t in water and alkaline solution, respectively. Solubilities were determined photometrically by diluting saturated solutions of the compounds to diluted concentrations c d . The value of c d was determined by applying appropriate UV-vis calibrations for c4m-ac and pHP-t ( Figure 1, Figure S10). [18] The solubilities c max,w and c max,a were then calculated with equation 2 (see Experimental Section) using the dilution factor. [19] Figure 1 demonstrates the absorbance of c4m-ac and pHP-t at different concentrations as well as at c d at the wavelength λ max of maximum absorbance in water and basic medium. In water, λ max of c4m-ac is at 366 nm and of pHP-t at 281 nm (Table 2). In alkaline solution, λ max of c4m-ac shifts to 377 nm and of pHP-t to 327 nm (Table 2). Resulting values for c d are listed Table S1 and c max,w and c max,a are shown in Table 1. c4mac shows a solubility c max,w in water of 2.77 mmol L À 1 � 0.07 mmol L À 1 and a solubility c max,a in alkaline solution of 646.46 mmol L À 1 � 0.63 mmol L À 1 . For pHP-t, a c max,w of 124.66 � 2.19 mmol L À 1 and a c max,a of 34.68 � 0.62 mmol L À 1 were found. Since PAG solubilities above 1 mmol L À 1 in aqueous solutions were referred to as good, [2b,20] c4m-ac and pHP-t exhibit excellent solubility in water and basic solution. For comparison purposes, the solubilities of the reference substance pHP-ac were determined to be 14.40 � 0.40 mmol L À 1 (c max,w ) and 21.46 � 8.58 mmol L À 1 (c max,a ) ( Figure S11).
In water, it becomes evident that the additional, hydrophilic tri(ethylene glycol) moiety present in pHP-t increased the solubility by a factor of 8.7 compared to pHP-ac, and thus the solubility enhancing effect of the tri(ethylene glycol) residue was clearly identified. In these conditions, the phenolic hydroxy   Table 2. Wavelength λ max at maximum absorption, molar absorption coefficient ɛ max at λ max , quantum yield Φ at the wavelength λ Φ given in parentheses, molar absorption coefficients ɛ Φ at λ Φ . Tested solvents were water (w) and alkaline solution (a) at pH 9. group present in both pHP-t-and pHP-ac can be expected to be in its neutral form. Similarly, in c4m-ac the carboxylic acid groups will be partly protonated, resulting in a relatively low solubility in water. In contrast, the solubility of c4m-ac in alkaline solution is boosted by a factor of 233 to the highest solubility observed in this study which can be explained by the more complete deprotonation of both carboxylic groups. A similar effect was observed for pHP-ac, however with only a moderate solubility increase by a factor of 1.5 due to the higher pK a value of the phenolic hydroxy group compared to carboxylic acid groups. Interestingly, for pHP-t the solubility decreased in alkaline conditions although one could expect that also its phenolic group is deprotonated to a similar extent like in pHP-ac. We tend to explain this observation with a disruption of the hydrogen bonds between water and the tri(ethylene glycol) residue of pHP-t in salt-containing alkaline solution. Similar salting-out effects were reported by Brunchi et al., who measured a decreasing solubility of poly(ethylene glycol) (PEG) in aqueous solution when adding electrolytes. [21] However, the solubility of pHP-t still was 1.6-fold higher in alkaline conditions than of pHP-ac.

Stability in Solution
As all PAGs investigated in this study contain at least one ester bond, hydrolysis may occur in aqueous solution. In order to quantify the influence of hydrolysis, the pH dependent stabilities s of c4m-ac, pHP-t and the reference compound pHPac in aqueous solution were investigated via HPLC. For this purpose, aqueous c4m-ac, pHP-t and pHP-ac solutions at pH 7, pH 8 and pH 9 were prepared and the PAG concentrations were measured after 1 h, 3 h, and 24 h storage time t s under light exclusion at room temperature. Additionally, the stabilities of c4m-ac, pHP-t, and pHP-ac in water without pH adjustment after dissolution leading to pH 3, pH 6, and pH 5, respectively, were investigated. PAG stabilities were calculated according to equation 7. The resulting pH dependent stabilities are shown in Figure 2 (c4m-ac, pHP-t) and Figure S12 (pHP-ac) and are summarized in Table S2.
Generally, the studied PAGs showed high stabilities (s 24h � 95 %) for 24 h in slightly acidic solution as obtained without pH adjustment. At pH 7, c4m-ac showed the highest stability after 24 h (s 24h = 99 %), whereas pHP-t showed only limited stability (s 24h = 85 %) under the same conditions, compared to an s 24h value of 94 % for pHP-ac. Upon increasing the pH value, stabilities generally decreased. At pH 8, c4m-ac still showed s 24h � 95 %, whereas for pHP-t s 24h dropped to 48 % compared to an unaltered value of 94 % for pHP-ac. At pH 9, all PAGs were significantly degraded with remaining concentrations of 11 % (c4m-ac), 0 % (pHP-t), and 53 % (pHP-ac). Summarizing, the PAGs showed decreasing stability with increasing pH and time.
Because in all PAGs in this study, an ester bond is present, these observations can be ascribed to the faster ester bond hydrolysis under more alkaline conditions. For example, increasing hydrolysis rates were published for pHP-based esters at higher pH (> 9), which is in line with our measurements (Table S2). [1a] The results demonstrate that at neutral to moderately alkaline conditions (pH � 8), c4m-ac is most resistant towards hydrolytic degradation. The stability of c4m-ac is in the range of other c4m-caged esters and amines from Hagen et al., which are described to be highly resistant to spontaneous hydrolysis at pH 7. [2b] In contrast, Hagen et al. also reported hydrolysis of c4m-caged aryl alcohols, thioaryl alcohol and carbamates up to 10 % at pH 7 during 24 h, which demonstrates the high stability of c4m-ac with less than 1 % hydrolysis under comparable conditions. [2b,15] As far as the stability of pHP derivatives is concerned, our results show that the leaving group present in the PAG influences its stability. At all pH values tested, pHP-t-showed  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57 faster hydrolysis than pHP-ac. Rather fast hydrolysis of esters neighboring an oligo(ethylene glycol) moiety were reported before by Claaßen et al. [22] and can presumably be explained by the negative inductive effect of the tri(ethylene glycol) residue, resulting in a better carboxylate leaving group. The influence of the leaving group on pHP-based compounds can also be found in the literature: On the one hand, quantitative stabilities were reported for pHP esters and other pHP derivatives like pHPadenosine triphosphate (ATP) in TRIS buffer at pH 7 after 24 h. [1g,23] On the other hand, pHP esters similar to pHP-t showed reduced stability. [2d,e] The di-alanine (Ala-Ala) pHP derivative pHP-Ala-Ala for instance hydrolyzed to 50 % in TRIS buffer at pH 7 in less than 4 h. [2e] In summary, hydrolysis is relevant for all PAGs studied, and has to be taken into account when considering to use these compounds in aqueous solution. Hydrolysis separates the acid from the chromophore, and therefore destroys the PAG functionality. Additionally, (unwanted) hydrolysis cannot be triggered and stopped as easily as (wanted) photolysis, and thus is a continuous process accompanying photolysis. Hence, PAG experiments, which are completed within a few minutes, can be performed at elevated pH values, but hydrolysis is certainly a disadvantage when solutions have to be stable for several hours. Therefore, photolysis conditions need to be chosen in such a way that hydrolysis plays only a minor role. For this reason, the photochemical properties of the PAGs are described in detail below.

Photochemical Properties
The reaction pathways for photolysis of c4m and pHP based compounds are well known as they were extensively investigated by Hagen et al. and Givens et al.,respectively. [1d,2b] Hagen et al. showed that c4m derivatives with the same coumarin group like in this study photolyse to 7-[bis (carboxymethyl) amino]-4-(hydroxymethyl)coumarin (c4m-OH) and the respective caged molecule. [2b] The photolysis of pHP derivatives in contrast is based on the Favorskii-rearrangement, which leads to p-hydroxy-phenylacetic acid and the corresponding caged compound, like acetic acid for pHP-ac. [1d] Both photoacid generation reactions of c4m and pHP based compounds are shown in Figure S17.
However, the efficiency of the photolysis reaction depends on the exact molecular structure and the solvent, which define the absorption coefficients and the quantum yields. Therefore, as a first step to assess the photochemical properties of c4m-ac and pHP-t as well as the reference compound pHP-ac, their UVvis absorption spectra both in water as well as alkaline solution were measured directly after dissolution (Figure 3 and Figure S13). All studied PAGs contain acidic groups, so it can be expected that their absorption spectrum is pH dependent. In order to assess if this was the case, the wavelength λ max at maximum absorption and the molar absorption coefficients ɛ max at λ max were extracted from the spectra (Table 2).
In water, the compound c4m-ac absorbed light up to 450 nm with a λ max at 366 nm and an ɛ max of 15 800 L mol À 1 cm À 1 (Table 2). This was in the range of other c4m-based compounds like c4m thioalcohol derivatives, which exhibited similar UV-vis absorption properties with λ max between 376 nm and 383 nm as well as ɛ max between 18 300 L mol À 1 cm À 1 and 20 000 L mol À 1 cm À 1 in hydroxyethyl piperazineethanesulfonic acid (HEPES) buffer at pH 7. [2b] In alkaline solution, the absorption band of c4m-ac shifted to longer wavelengths with a λ max of 377 nm and an ɛ max of 15 300 L mol À 1 cm À 1 . This bathochromic shift was based on solvatochromic effects in alkaline solution, since the polarity of the alkaline solution was higher compared to water. Similar observations were reported from Liu and coworkers, who described a red shift of the UV-vis spectra of 7aminocoumarins in more polar solvents. [24] Also Nad et al. published a λ max shift of 7-amino-4-trifluoromethylcoumarin from 347 nm to 378 nm by increasing the solvent polarity from hexane to methanol. [25] Both explained that the higher the polarity of the solvent, the more intermolecular interactions between the coumarin moiety and the solvent can evolve, which stabilizes the ground state and shifts the UV-vis absorption to lower excitation energies. [24][25] Compared to the two strong absorption bands of c4m-ac around 245 nm and 366 nm, pHP-t showed only one prominent π-π* absorption band from 240 nm to 320 nm with a λ max at 281 nm and an ɛ max of 13 500 L mol À 1 cm À 1 in water ( Table 2). The λ max of pHP-t shifted from 281 nm in water to 327 nm under basic conditions due to partly deprotonation of the phenol moiety. This phenoxide species contains an enlarged πelectron system, which leads to a bathochromic shift of the λ max in alkaline solution. This is in accordance with multiple pHP derivatives published by Givens et al. who also reported a pK a value of 8.0 for pHP-ac. [2d] In alkaline conditions, the absorption spectrum of pHP-t therefore is a superposition of deprotonated and protonated pHP-t. In fact, between 250 nm and 300 nm the absorption of the remaining protonated form is still visible. Similar to c4m-ac, ɛ max of pHP-t slightly decreased from 13 500 L mol À 1 cm À 1 in water to 12 900 L mol À 1 cm À 1 in alkaline solution (Table 2). For pHP derivatives this is quite unusual as most reported pHP phenoxide derivatives show significantly higher ɛ max than their protonated counterparts. [2d] The measurements on pHP-ac with an ɛ max of 11 600 L mol À 1 cm À 1 in water and an ɛ max of 20 400 L mol À 1 cm À 1 in basic solution confirmed the trend in the literature ( Figure S13, Table 2). [2d] Apart from the molar absorption coefficients, quantum yields Φ are equally important in defining the rate of a photolysis reaction, as the rate is determined by the product of Φ and ɛ. Therefore, the photolysis quantum yields were measured at wavelengths near λ max , i. e. 310 nm for pHP derivatives and 365 nm for c4m-ac (Table 2). For the photoacid generator photolysis, a high Φ value of 0.69 was found for pHP-t in water. In alkaline solution, Φ of pHP-t decreased to 0.07. Similarly, Φ of pHP-ac decreased from 0.46 in water to 0.02 in alkaline solution. The results for the two pHP-based compounds thus are in a similar range of other pHP-caged compounds with Φ values between 0.03 and 0.65 in water. [2d,26] Also the reduction of the quantum efficiency is in line with previous reports, where it was described that the conjugated phenoxide base has a much lower quantum yield than the protonated species. [2d,26a] This was attributed to a decreased intersystem crossing efficiency or competing nonproductive pathways. [2d,26a] The Φ of c4m-ac in contrast was not influenced by the pH and stayed at a relatively low value of 0.02 in water and basic solution, as the two carboxylic acid groups of c4m-ac were not part of the conjugated π-electron system. The Φ values of c4mac were in the range of other c4m-caged compounds reported by Hagen et al. with Φ between 0.01 and 0.30 in ACN/HEPESmixtures at pH 7.2. [2b,c] A comparison of the different photolysis efficiencies is now possible by comparing the product of Φ and the molar absorption coefficient ɛ Φ at the wavelength where the quantum yield was measured (Table 2). It becomes evident that the two pHP-based compounds showed higher photolysis rates than c4m-ac when irradiated during the quantum yield measurements. The fastest photoreaction was observed for pHP-t, which under such 'ideal' photolysis conditions is the most efficient PAG. The difference in photolysis rates could probably be further enhanced by irradiating closer to the respective λ max values, assuming that the quantum yields at these wavelengths are similar to the measured ones.

Photolysis with a Broadband Light Source
The data on molar absorption coefficients and quantum yields in the previous section give insight into photolysis rates when using monochromatic light sources or light sources with a narrow emission spectrum such as lasers and LEDs. However, broadband light sources are common in non-photochemical laboratories when rather short irradiation wavelengths are needed like for the pHP based compounds. This is especially true in the fields of PAG application described in the introduction such as polymer chemistry, hydrogel curing, and 3D printing. [27] To evaluate which of the studied PAGs are favorable under such circumstances, we investigated the photolysis of c4m-ac, pHP-t, and pHP-ac at three different pH values using a standard broadband light source by HPLC (see Figure S16 for an exemplary HPLC dataset). The emission spectrum of the light source ranged between wavelengths of 300 nm and 450 nm ( Figure SI15). The resulting PAG concentrations c against irradiation time t irr of c4m-ac and pHP-t are shown in Figure 4, the respective data of the reference substance pHP-ac in Figure S14.
Generally, all PAGs disappeared completely during irradiation ( Figure S16). The photolysis kinetics seem to follow a monoexponential decay of PAG concentrations. However, a correct physicochemical model describing the entire photolysis reaction will be more complex and the data were not fitted with a monoexponential function. Nevertheless, in order to compare the photolysis kinetics, the value of t irr , which corresponds to a decrease of the concentration to half of the initial concentration, was determined. For c4m-ac, this was the case after 1 min to 2 min, for pHP-t after about 6 min, and for pHP-ac after about 15 min. These values were independent of the pH value of the solution. Interestingly, these results seem to contradict the results found in the previous sections because 1) pHP-t was previously identified to show the most efficient photolysis reaction, and 2) absorption spectra and/or quantum yields were found to depend on pH.
These findings can be explained by an interplay between the spectral overlap of the emission spectrum of the light source and the corresponding quantum yields. For c4m-ac, the absorption spectrum overlaps to a great extent with the lamp spectrum both in water as well as in alkaline conditions. Therefore, although the quantum yields were measured to be quite low, the photolysis proceeded rapidly in both conditions. In contrast, both pHP-t and pHP-ac absorption spectra in water overlap only to a minor extent with the lamp spectrum. Therefore, although quantum yields were measured to be much larger than for c4m-ac, photolysis was slower than for c4m-ac. The red shift of pHP-t and pHP-ac absorption spectra in alkaline conditions improves the overlap with the lamp spectrum, but concomitantly the quantum yields decreased drastically. These opposing effects obviously cancel each other out, so that no overall change of photolysis rates were observed upon changing the pH value.
In fact, the stable photolysis rates with various pH values simplify the usage of c4m-ac and pHP-t. This way, is possible to tune the pH value according to other experimental pre-requisites without having to worry about the photolysis rate, given that the pH value is in a range where hydrolysis is not predominant. For prospective polymeric foaming experiments under these irradiation conditions, c4m-ac seems most advantageous compared to pHP-t due to the favorable absorption properties, the higher pH stability and the faster photolysis.

Conclusions
Our studies contribute to a comprehensive understanding of the synthesis, solubility, stability and photolysis of two highly water soluble c4m and pHP-based PAGs. This could help to satisfy the growing demand for water soluble PAGs especially in the field of hydrogel research and polymeric foaming. Yet, in such applications many strong electrolyte PAGs like diphenyliodonium salts are used, even if they are toxic or need additional sensitizer. This is mainly due to their good accessibility and high water solubility. With this work, we provided an alternative approach to such compounds by designing two easily accessible and highly water soluble c4m and pHP-based PAGs, as the substance classes of c4ms and pHPs are well suited for physiological applications and do not need additional sensitizers. [1a] The successful synthesis of c4m-ac and pHP-t showed their accessibility and the introduction of the hydrophilic groups did not interfere with the excellent photochemical properties of c4m and pHP-based PAGs. We also investigated other key parameters like the stability and the photolysis of these PAGs, which confirmed that c4m-ac and pHP-t are fairly stable and well photo cleavable in aqueous media under varying pH conditions.
These properties should enable the use of c4m-ac and pHPt for polymeric foaming, e. g. by using an alkaline carbonate solution and the in situ generated acid as foaming agent. We presume that c4m-ac and pHP-t are cytocompatible which would make them interesting candidates as PAGs e. g. for the production of 3D printed hydrogel foams as polymer scaffolds in tissue engineering. The question of cytocompatibility has to be addressed in future studies.

Instrumentation
The NMR spectra of the c4m-based compounds were recorded on a Bruker AVIII-HD 500 MHz instrument from Bruker (Billerica, USA) equipped with a N 2 cooled cryogenic probe head using DMSO-d 6 . The NMR spectra of the pHP-based compounds were recorded on a Bruker Avance 500 from Bruker (Billerica, USA) with CDCl 3 as solvent. Mass spectrometry was performed on an ESI-MS Bruker MicroTOFQ from Bruker (Billerica, USA) under nitrogen atmosphere.
For the measurement of the pH value an InLab 1 022 pH electrode from Mettler Toledo (Columbus, USA) in combination with a Lab 850 pH meter was used. Photolysis experiments were performed in an UV-H 255 UV chamber from Hartmann Feinwerkbau GmbH (Ober-Moerlen, Germany) with an irradiation intensity of approx. 40 mW cm À 2 between 300 nm and 450 nm ( Figure S15). The emission spectrum was determined with an Ocean Optics USB 2 000 + spectrometer from Ocean Optics Germany GmbH (Ostfieldern, Germany). The distance between the bottom of the sample and the UV source in the UV chamber was 8.5 cm. UV-vis spectra were recorded in a two beam UV-vis spectrometer UV-2450 from Shimadzu (Kyoto, Japan) with quartz cuvettes of 1 cm path length and a cross-sectional area of 1 cm 2 .
For quantum yield determinations c4m-ac was irradiated with a M365L2 LED from Thorlabs at 365 nm. For the irradiation of pHP-t and pHP-ac a M310L3 LED from Thorlabs (Newton, USA) at 310 nm was used. Both were operated by a DC4100 LED driver from Thorlabs (Newton, USA). The decay of c4m-ac was analyzed with a 1260 Infinity HPLC from Agilent Technologies Germany GmbH (Waldbronn, Germany), equipped with a diode array detector. Separation was done using MultoKrom® 100-5 C18 column (250 × 4.6 mm) from CS-Chromatographie Service GmbH (Langerwehe, Germany). Chromatograms were analyzed with ChemStation software from Agilent Technologies Germany GmbH (Waldbronn, Germany).
For stability and photolysis determinations, HPLC measurements were performed using an analytical CBM-20A/20Alite HPLC from Shimadzu (Kyoto, Japan) with a Superspher 100 RP-18 (125 mm × 4.0 mm) column from Merck Chemicals GmbH (Darmstadt, Germany) and a SPD-M20A photodiode detector from Shimadzu (Kyoto, Japan). For c4m-ac a mixture of ACN : H 2 O = 15 % : 85 %, for pHP-t a mixture of ACN : H 2 O = 17 % : 83 % and for pHP-ac a mixture of ACN : H 2 O = 10 % : 90 % was used for elution. Trifluoroacetic acid (0.1 %) was added to the HPLC solvent for pHP-t and pHP-ac. The flow rate was 1.0 mL min À 1 . The HPLC chromatograms were analyzed with LCsolution software from Shimadzu (Kyoto, Japan).

2,5,8,11-Tetraoxatridecan-13-oic acid (2 b)
The ester 2 a (0.98 g, 4 mmol, 1.0 eq) was dissolved in a 1 M methanolic solution of sodium hydroxide (22.00 mL, 20 mmol, 5.0 eq) and stirred for 72 hours at RT. The solution was adjusted to a pH value of 3 by adding aqueous HCl solution. The solvent was removed under reduced pressure and the residue was dissolved in diethylether. Unsoluble solid was separated through filtration and the organic phase was washed with water. By evaporating the solvents 2,5,8,11-tetraoxatridecan-13-oic acid (2 b) (0.75 g, 3 mmol, 86 %) could be obtained as colorless oil. 1

p-hydroxyphenacylacetate (pHP-ac)
2-Bromo-4-hydroxyacetophenone (1.48 g, 7 mmol, 1.0 eq) were dissolved in EtOH and a mixture of sodium acetate (1.12 g, 8 mmol, 1.2 eq) and AcOH (0.36 mL, 6 mmol, 0.9 eq) in water (3.6 mL) was added dropwise. The solution was heated for 3 h at 90°C under reflux conditions. The solvent was removed under reduced pressure and the remaining oily residue was dissolved in aqueous sodium carbonate solution. The aqueous phase was extracted with EtOAc and the combined organic phases were washed with brine and dried over sodium sulfate. After evaporation of the solvent, the raw product was recrystallized from EtOAc and purified by a column chromatography (n-hexane : EtOAc = 7 : 3). p-hydroxyphenacylacetate (pHP-ac) (0.97 g, 5 mmol, 73 %) was obtained as a white solid. 1

Solubility Determination
The solubilities of c4m-ac, pHP-t and pHP-ac were determined photometrically in water and alkaline solution (3 M KHCO 3 solution, pH 9). [18] For the determination of the solubility in water (c max,w ) and the solubility in alkaline solution (c max,a ) a calibration curve was prepared by measuring the UV-vis absorption spectra of the PAGs at four different concentrations in water and in basic solution ( Figure S10).
The concentrations were selected in such a way that the absorbance maxima were spread over the linear absorbance range (A λ = 0.1-0.9) of the Beer's law [Eq. (1)]. [28] A l ¼ e l c d (1) In this equation, A λ is the absorbance at a specific wavelength λ, ɛ λ is the molar absorption coefficient at the wavelength λ, c is the concentration and d is the path length. The calibration curve was used to determine the molar absorption coefficient ɛ λ of the PAG in the respective media. Three saturated PAG solutions were prepared and diluted until the absorbance maxima were in the linear absorbance range. To calculate the concentration of the diluted sample (c d ) the previously determined molar absorption coefficient ɛ λ was used. As shown in Equation (2), the solubility in water c max,w can be quantified by multiplying the concentration of the diluted PAG solution c d with the dilution factor d f .
The solubility in alkaline solution c max,a is determined respectively by using A λ , ɛ λ and d f of the alkaline solution. The solubility determination was performed in triplicates.

Quantum Yield Determination
All measurements were performed in 1 cm quartz fluorescence cuvette from Hellma GmbH (Müllheim, Germany). For irradiation of c4m-ac, a M365 L2 LED from Thorlabs at 365 nm was used and for pHP-t and pHP-ac, an M310L3 LED from Thorlabs (Newton, USA) at 310 nm was applied. Both were operated by a DC4100 LED driver from Thorlabs (Newton, USA). Light sources were calibrated using iron (III) ferrioxalate actinometry, following the literature procedure. [29] The photo reaction of c4m-ac was followed by HPLC and of pHP-t and pHP-ac by UV-vis spectroscopy. All quantum yields (Φ) were measured in triplicates and calculated as previously described. [30] For the Φ determination of c4m-ac, an aqueous sample of c4m-ac containing the internal standard phenylalanine was irradiated at 365 nm and the conversion was analyzed via HPLC. The conversion change (in %) of c4m-ac was plotted against irradiation time (t irr ). The plot was fitted using following exponential decay function [Eq.
(3)], where A 1 and t 1 are fit parameters and y 0 is a constant. [30] y ¼ A 1 e tirr t1 þ y 0 The initial rate of the concentration change (y') at the beginning of the irradiation can be calculated by deriving Equation (3) and inserting the corresponding fit parameters for t irr = 0 [Eq. (4)]: [30] y' ¼ À A 1 e À tirr t1 t 1 (4) For c4m-ac, Φ is then calculated by Equation (5), where c is the concentration of the irradiated solution, V is the volume of the irradiated sample, y' is the change in concentration, n p is the photon flux of the light source determined by actinometry and A λ is the absorbance of the PAG solution at the irradiation wavelength. [30] F ¼ c � V � y 0 n p � ð1 À 10 À A l Þ In contrast to c4m-ac, Φ of pHP-t and pHP-ac was determined by UV-vis spectroscopy. Therefore, pHP-t and pHP-ac samples with a high absorption (A λ > 3) at 310 nm were used to ensure complete absorption of radiant flux. The pHP-based PAGs were irradiated and the change of absorption was measured simultaneously using a photodiode array detector. A plot of the absorption change against irradiation time was prepared choosing a suitable wavelength with A λ < 1. The decrease of the absorption in the initial phase of the reaction (conversion < 10 %) was fitted by a linear regression. Φ of pHP-t and pHP-ac was calculated using Equation (6): Here, k is the slope from the linear regression, V is the volume of the irradiated sample, d is the pathlength of the cuvette, ɛ λ is the molar absorption coefficient of the wavelength used for the reaction control (here at 321 nm) and n p is the photon flux of the light source.

Stability Determination
C4m-ac, pHP-t and pHP-ac were dissolved in water at a concentration of 0.2 g L À 1 , which led to pH 3 for c4m-ac, pH 6 for pHP-t and pH 5 for pHP-ac. Aliquots of these PAG solutions were adjusted to pH 7 and pH 8 with 0.01 M NaOH. The samples were stored under light exclusion at RT for 1 h, 3 h and 24 h. After filtration, they were measured via HPLC. The elution time (t elu ) of c4m-ac was 6.3 min, of pHP-t 9.0 min and of pHP-ac 9.5 min. The PAG stability (s) was calculated according to Equation (7), where P t = 0 is the peak area of the PAG immediately after preparation (t = 0) and P t is the respective peak area after a storage time t s . s ¼ P t P t¼0 (7)

Photolysis
An aqueous solution of c4m-ac, pHP-t and pHP-ac was prepared with a concentration of 0.2 mg mL À 1 . This led to a solution of pH 3 for c4m-ac, of pH 6 for pHP-t and of pH 5 for pHP-ac, respectively. An aliquot of each PAG solution was adjusted to pH 7 and to pH 8 using 0.01 M NaOH. 1 mL samples of the PAG solutions were irradiated in a quartz glass cuvette using a hartmann.gs UV-H 255 UV chamber. The irradiation time t irr was adjusted to the photolysis speed of the compound. The photolysis of c4m-ac for instance was measured after 0.5 min, 1 min, 1.5 min, 2 min, 3 min, 4 min, 6 min and 8 min UV irradiation. For pHP-t, the photolysis was measured every two minutes for the first 10 minutes and then every 5 minutes. After 30 min, pHP-t was analyzed every 10 minutes until an overall t irr of 60 minutes. For pHP-ac, the concentration was measured every minute in the first 10 minutes and afterwards every five minutes up to 80 minutes .  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57