Commercially available reagents and solvents were used without further purification, unless stated otherwise. Reactions were carried out in standard glassware under N₂ or Ar, and yields were not optimized. Demineralized H₂O: Millipore-Synergy-185 water purifier. Infrared (IR) spectra: Perkin-Elmer-1600-FTIR spectrometer; of weak (w), medium (m) or strong (s) bands in cm⁻¹. ¹H- and ¹³C-nuclear magnetic resonance (NMR) spectra: Bruker 400 MHz Avance III spectrometer; δ in ppm downfield from Me₄Si as internal standard. Standard pulse sequences and parameters were used for one-dimensional ¹H- and ¹³C-NMR experiments and for two-dimensional, gradient selected correlation spectroscopy (COSY), ¹H,¹³C-HSQC and ¹H,¹³C-heteronuclear multiple bond correlation (HMBC) experiments, respectively.

Size-exclusion chromatography (SEC) analyses were carried out at room temperature (ca. 22 °C) on a Viscotek GPC max VE 2001 GPC Solvent Sample Module connected to a Viscotek UV detector 2500, a Viscotek VE3580 RI detector and a Viscotek-270-Dual-Detector viscometer. Samples were eluted from a Macherey-Nagel VA 300/7.7 Nucleogel GPC 500-5 column at a flow rate of 1.0 mL min⁻¹ with tetrahydrofuran (THF, high performance liquid chromatography (HPLC)-grade). Universal calibrations were performed using commercial poly(styrene) standards. The polymer standard (ca. 40 mg) was accurately weighed and dissolved in THF (10 mL); then, these solutions (100 μL) were injected for the calibration. Molecular weight averages and polydispersity indices (PDIs) of the different polymers are listed in Table 2 below.

In a 100 mL round-bottomed flask, (E)-1-(2,6,6-trimethylcyclohex-3-enyl)but-2-en-1-one (δ-damascone, 20.0 g, 104.0 mmol), triethylamine (14.5 mL, 104.0 mmol) and 2-aminoethanethiol hydrochloride (11.8 g, 104.0 mmol) were dissolved in ethanol (50 mL) to give a colorless solution. 2,3,4,6,7,8,9,10-Octahydropyrimido[1,2-a]azepine (1,8-diazabicyclo[5.4.0]undec-7-ene (DBU), 1.6 g, 10.4 mmol) was added. The reaction mixture was then stirred at room temperature overnight. Cyclohexane (30 mL) was added to the reaction mixture to give a precipitate, which was removed by filtration. Ethanol was evaporated, and the reaction mixture was washed with NaCl (2 M, 2 × 50 mL), NaOH (10⁻³ M, 40 mL) and H₂O. The organic layer was dried with MgSO₄, filtered and concentrated to give a yellow oil of 4 and 5 in a ratio of 1.2:1 as a mixture of diastereoisomers (m = 27 g). ¹H-NMR (CDCl₃, 400 MHz) 4: δ 5.53 (m, 1H), 5.44 (m, 1H), 3.32 (m, 1H), 2.91 (m, 1H, diastereoisomer 1), 2.72 (m, 2H, diastereoisomer 2), 2.66 (br m, 1H), 2.53 (m, 1H, diastereoisomer 1), 2.51 (m, 1H), 2.50 (m, 1H), 2.22 (m, 1H), 1.96 (m, 1H), 1.69 (m, 1H), 1.31 (m, 3H), 1.00 or 0.98 (m, 3H), 0.95 (m, 3H) and 0.89 (m, 3H) ppm. ¹H-NMR (CDCl₃, 400 MHz) 5: δ 5.53 (m, 2H), 5.44 (m, 2H), 3.32 (m, 1H), 3.18 (m, 1H), 2.91 (m, 1H, diastereoisomer 1), 2.86 (m, 1H), 2.77 (m, 1H, diastereoisomer 1), 2.75 (m, 1H), 2.72 (m, 2H, diastereoisomer 2), 2.71 (m, 2H), 2.61 (m, 2H, diastereoisomer 2), 2.53 (m, 1H, diastereoisomer 1), 2.50 (m, 1H), 2.46 (m, 1H, diastereoisomer 1), 2.24 (m, 1H), 2.22 (m, 1H), 1.96 (m, 2H), 1.69 (m, 2H), 1.31 (m, 3H), 1.10 (m, 3H), 1.00 (m, 3H), 0.98 (m, 3H), 0.95 (m, 6H) and 0.89 (m, 6H) ppm. ¹³C-NMR (CDCl₃, 100.6 MHz) 4: δ 212.2 (s), 131.8 (d), 124.2 (d), 62.9 (d), 55.2 (t), 41.7 (t), 41.3 (br t), 34.8 (br t), 34.0 (d), 33.1 (s), 31.7 (d), 29.8 (q), 21.9 (q), 20.7 (q), 19.9 (q) ppm. ¹³C-NMR (CDCl₃, 100.6 MHz) 5: δ 214.1 (s), 212.2 (s), 131.8 (d), 124.2 (d), 62.9 (d), 55.2 (t), 54.9 (t), 48.5 (d), 46.1 (t), 41.7 (t), 34.0 (d), 33.1 (s), 31.7 (d), 31.3 (t), 29.8 (q), 21.9 (q), 20.7 (q), 20.4 (q) and 19.9 (q) ppm. IR (neat): 3387w, 3317w, 3018m, 2957s, 2928m, 2871m, 2830m, 1703s, 1652m, 1457m, 1386m, 1366s, 1280w, 1250w, 1212w, 1194w, 1153m, 1116m, 1080m, 1039w, 998m, 953w, 932w, 896w, 842w, 787w, 691s, 683s and 641m cm⁻¹.

In a 50 mL round-bottomed flask, poly(styrene-co-maleic anhydride) (6, 4.0 g, 18.78 mmol with respect to the anhydride group) and α-(2-aminoethyl)-ω-(2-methoxyethyl)-poly(propylene glycol-co-ethylene glycol) (7–9, Table 1, 1.88 mmol) were dissolved in acetone (20 mL) to give a yellow solution. The reaction mixture was stirred at 50 °C overnight. Copolymers 10–12 were obtained by precipitation from water, filtered and dried under vacuum at 50 °C for 24 h.

In a 100 mL round-bottomed flask, copolymers 10–12 (4.65 mmol with respect to the anhydride groups) and a mixture of 4 and 5 (2.35 g, at a ratio of 1.2:1, corresponding to 4.68 mmol of 4) were dissolved in acetone (40 mL) to give a yellow solution. The reaction mixture was stirred at 50°C overnight. Copolymers 13–15 were obtained by precipitation from n-heptane, filtered and dried under vacuum at 50 °C for 24 h.

A cationic triethanolamine (TEA)-esterquat surfactant formulation was prepared from the following ingredients: Stepantex^® VK90 (Stepan) 16.5%, an aqueous solution of calcium chloride (10%) 0.2% and water 83.3%. Depending on the δ-damascone content of the polymers, a total of 1.26 to 1.88% of poly(maleic acid)-based copolymers 13–16 was added to the softener formulation (5.00 g) in order to release a total amount of 8.7 mg of δ-damascone from each sample. All values are given in % by weight.

In a small vial, poly(maleic acid)-based copolymers 13–16, dissolved in ethyl acetate (0.5 mL, 13: 26.9 mg, 14, 15: 34.0 mg, 16: 22.7 mg, all releasing a total amount of 8.7 mg of δ-damascone) were added to the aqueous surfactant formulation (1.80 g), and the resulting mixtures were stirred overnight. The emulsions were then placed in a beaker (1 L) and diluted with demineralized cold tap water (600 g). One cotton sheet (Eidgenössische Materialprüfanstalt (EMPA, Switzerland), cotton test cloth Nr. 221, cut to ca. 12 × 12 cm sheets (average mass ca. 3.12 g and prewashed with an unperfumed detergent powder) was added to each beaker. The sheet was manually stirred for 3 min, left standing for 2 min and then wrung out by hand and weighed (average mass ca. 7.0 g) to ensure a constant amount of residual water. A solution with an equimolar amount of unmodified δ-damascone (8.7 mg in 0.5 mL of acetone) was added to another sample of the fabric softener formulation (1.80 g), which was used as the reference sample and treated as described above. All cotton sheets were line dried for 1 day.

Each of the line-dried cotton sheets was individually placed inside a headspace sampling cell (with a total internal volume of ca. 160 mL). The sampling device was thermostatted at 25 °C and exposed to a constant flow of air (ca. 200 mL min⁻¹). The air was filtered through active charcoal and aspirated through a saturated solution of NaCl to ensure a constant humidity of ca. 75%. For 2 h, the volatiles were adsorbed onto a waste Tenax^® cartridge and then for 15 min onto a clean Tenax^® cartridge (corresponding to a total sample volume of 3 L of air). The sampling was repeated twice every 2 h (105 min of trapping on the waste cartridge and 15 min of sampling onto a clean cartridge); the waste cartridges were discarded. The other cartridges were thermally desorbed on a Perkin Elmer TurboMatrix ATD 350 desorber coupled to an Agilent Technologies 7890A gas chromatograph equipped with a HP-1 capillary column (30 m, i.d. 0.32 mm, film thickness 0.25 μm) and a flame ionization detector. The volatiles were analyzed using a two-step temperature gradient starting from 60 °C to 130 °C at 15 °C min⁻¹ and then heating to 220°C at 40°C min⁻¹. The injection temperature was at 250 °C and the detector temperature at 250°C. External standard calibrations were carried out using five different concentrations of δ-damascone in acetone (varying between 7.09 × 10⁻⁵ and 2.63 × 10⁻² mol L⁻¹). Each calibration solution (0.2 μL) was injected onto three clean Tenax^® cartridges. All cartridges were then desorbed immediately under the same conditions as those resulting from the headspace sampling (see above). Since the quantities of released compounds were monitored down to very low concentrations, the calibration curve was forced through the origin of the coordinate system. Plotting the concentrations (in ng L⁻¹) against the peak areas gave a straight line (y = 1.452 x) with a correlation coefficient of r² = 1.0000. All experiments were carried out in duplicate.

δ-Damascone was conjugated to the poly(styrene-co-maleic anhydride) backbone by 1,4-Michael type addition of the thiol group from cysteamine hydrochloride onto the enone double bond to give the δ-damascone-releasing unit 4 (Scheme 1). DBU was used as a catalyst, and triethylamine was added to deprotonate the ammonium function of the linker to facilitate its solubility in ethanol. The addition of the cysteamine on the enone can theoretically occur by reaction of either the NH₂ or –SH group. We expected that the more stable product 4 was preferentially formed. To verify the selectivity between the two possible reactions, the crude reaction product was characterized by FTIR and NMR spectroscopy.

¹H-NMR spectroscopy showed that the reaction was not selective (see Figure 3a and Experimental Section).

Interpretation of the ¹³C-NMR spectrum revealed that a mixture of the desired compound 4 and diaddition product 5 was obtained (Figure 3b). Broad triplets at 34.8 and 41.3 ppm correspond to the two CH₂ groups of the thioether with a terminal NH₂ group. A signal, which would correspond to a CH₂ group next to the –SH function, was not detected. This assignment was confirmed by the absence of a peak between 2600 and 2650 cm⁻¹ in the IR spectrum, which would indicate the presence of a thiol group. In contrast, the presence of two peaks at 3,317 and 3,387 cm⁻¹ confirmed the presence of the NH₂ group in 4.

Subsequently, a comprehensive 2D-NMR analysis was carried out. The two triplets observed at 31.3 and 46.1 ppm in the ¹³C-NMR spectrum further confirmed the presence of the diaddition product 5. The carbonyl groups of the δ-damascone unit give rise to signals at 212.2 ppm (if linked via a thioether) and 214.1 ppm (if linked via a secondary amine, Figure 3c). Integration of these two signals resulted in a ratio of 2.2:1. After subtraction of the integral of the signal at 214.1 ppm (which corresponds exclusively to compound 5) from that at 212.2 ppm (corresponding to both compounds 4 and 5), a molar ratio of 1.2 to 1 was calculated for 4 and 5. Similar values were obtained from integration of signals in the ¹H-NMR spectrum. Because diaddition product 5 does not react with the maleic anhydride of the polymer backbone, its presence should not be a problem for the next reaction step. It can easily be removed during the purification of the modified copolymer.

To determine the styrene/maleic anhydride (S/MA) molar ratio, commercially available copolymer 6 was analyzed by ¹H-NMR spectroscopy. Integration of the signals centered at 7.20 ppm (corresponding to the aromatic ring of styrene (a, Figure 4)) and those at 2.20 and 3.30 ppm (resulting from the styrene and the maleic anhydride backbone (b and c, respectively, Figure 4)) showed that the S/MA molar ratio was around 1.1, which is close to the value of 1.3 given by the supplier. Nevertheless, the peaks are not well resolved, and quantitative analysis is difficult.

According to this ratio, the structure of copolymer 6 in Scheme 2 is represented as an alternating copolymer with a S/MA molar ratio of 1. With 1.1 equivalents of styrene and 1 equivalent of maleic anhydride, the molecular weight of the repeating unit corresponded to 213.51 g mol⁻¹. The number average molecular weight M_n of 6 was measured by analytical SEC to be 2100 g mol⁻¹, in contrast to a value of 1600 g mol⁻¹ given by the supplier. This molecular weight corresponds to a polymer consisting on average of 10 repeat units.

Amino functionalized ethylene oxide (EO)/propylene oxide (PO) copolymers (Jeffamine^®) (7–9 in Scheme 2) are commercially available, and their molecular composition was given by the supplier as illustrated in Table 1. Their EO/PO molar ratio was measured by ¹H-NMR spectroscopy. The signal centered at 1.10 ppm corresponds to the methyl group of the PO unit (d, Figure 4), whereas the signals between 3.10 and 3.60 ppm are from the protons of the methyne and methylene groups of the EO and PO units (e, Figure 4). The determination of the molar ratio was useful for the present study, because it indicated the polarity of the copolymer, with an increasing ratio of EO/PO, corresponding to an increase in polarity.

In copolymer 9, the EO/PO ratio corresponded to the values from the supplier, while the measured EO/PO ratios in copolymers 7 and 8 were lower than the values given by the supplier (0.11 and 0.14 as compared with 0.11 and 0.21, respectively, Table 1). The measured values were used as the basis for the discussions in the present work.

Copolymers 7–9 were grafted onto 6 by reaction in acetone at 50 °C. A constant grafting rate of 10 mol % was chosen to modify the polarity of copolymers 10–12 with the different Jeffamine^® 7–9 (Scheme 2). After purification, SEC showed one population, and an increase in molecular weight as compared with 6 was observed. The measured increase was in agreement with the expected structures (Table 2). The presence of peaks at 2972 and 2876 cm⁻¹ (corresponding to the C–H stretch of the CH and CH₂ groups) and at 1583 and 1540 cm⁻¹ (from the CO of the amide group) in the IR spectra of 7–9 confirmed the addition of the amino group to the copolymers on 6. The successful modification of 6 was also confirmed by ¹H-NMR spectroscopy. The amount of grafting was determined from integration of the signals centered at 7.20 ppm of 6 (a, 5 H, Figure 4) and the signals at 1.11 and 3.60 ppm from copolymers 7–9 (e and d, 3 H, Figure 4). The reaction was found to be complete under the present reaction conditions.

The grafting of 4 onto copolymers 6 and 10–12 was also carried out by reaction in acetone at 50°C. Taking the molar ratio of compounds 4 and 5 (1.2:1) into account, the amount of the mixture added to the reaction was chosen to correspond to a stoichiometric amount of 4. A complete conversion was obtained (Scheme 3). The presence of δ-damascone in copolymers 13–16 was confirmed by ¹H-NMR spectroscopy, where signals at 0.9 (f), 3.47 (g), 5.46 and 5.54 ppm (h, Figure 4) were observed. Integration of the signals at 5.46 and 5.54 ppm from the double bond of 4 (h, 2 H, Figure 4) and the signals at 7.20 ppm from the styrene units (a, 5 H, Figure 4) confirmed the complete conversion of 4. In addition, IR spectra showed not only the absence of peaks at 1773 and 1854 cm⁻¹ of the maleic anhydride in copolymers 6 and 10–12, but also the presence of new peaks at 3020 and 1702 cm⁻¹ (Figure 5), which were previously observed in 4. Finally, an increase in molecular weight was measured by SEC (Table 2).

The loading of δ-damascone in conjugates 13–16 was determined by ¹H-NMR spectroscopy (Table 3). The loading varied between 25.6 wt % in copolymers 14 and 15, 32.3 wt % in 13 and 38.3 wt % in 16.

The influence of the structure of the grafted poly(ethylene oxide)-poly(propylene oxide) (PEO-PPO) copolymers on the release of δ-damascone was investigated by dynamic headspace analysis [38,39] of the volatile molecule from a cotton surface after application of a fabric softener solution. Typical fabric softener formulations consist of about 15% of a cationic surfactant, such as triethanolamine (TEA)-esterquats of fatty acids [40,41], in water. For the measurements, conjugates 13–16 were dissolved in ethyl acetate at a concentration to potentially release a total of 8.7 mg (0.05 mmol) of δ-damascone. The solutions were then dispersed in 5.0 g of a TEA-esterquat fabric softening formulation consisting of an emulsion of cationic surfactant at 16.5% in water. The samples were constantly stirred to obtain homogeneous dispersions. The different fabric softener dispersions (1.8 g) were dissolved with water (600 mL), which corresponds to the water added to a fabric softener in a washing machine. To this diluted aqueous emulsion, a cotton sheet was added and stirred for 3 min to simulate the deposition of fabric softener and then left standing for another 2 min in the dispersion. The cotton sheet was then wrung out and weighed to leave 3.9 g of the dispersion on the sheet. The cotton sheet was finally line dried at room temperature for 1 day [27].

The cotton sheet was then placed inside a headspace sampling cell and exposed to a constant flow of air to constantly remove the volatiles evaporating from the cotton surface. The δ-damascone released from copolymers 13–16 was collected at constant time intervals on a Tenax^® cartridge and then analyzed by gas chromatography after thermal desorption [27]. Three measurements were taken, one each after 2, 4 and 6 h of sampling. The data illustrated in Figure 6 are average values of at least two measurements.

The water content of the polymers has been neglected in the current study because the polymers were formulated in an aqueous environment for the application. Furthermore, the headspace sampling was carried out with constant humidity in the air. We therefore assume that the water activity was constant for all measurements. Dynamic scanning calorimetry measurements carried out on the pure polymers between −25 and 110 °C (to represent realistic application conditions) did not show phase transitions. This indicated that residual water did not have an influence on the release kinetics.

After 2 h of sampling, headspace concentrations with relatively large standard deviations were measured, whereas relatively low standard deviations were obtained for the data acquired after 4 and 6 h of sampling. This suggests that the deviations observed at the beginning of the measurements are not the result of uncontrolled deposition of the conjugates on the cotton surface, but rather because an initial amount of time is necessary to equilibrate the headspace system before reaching a state in which headspace concentrations can be collected with good reproducibility. This phenomenon has been observed previously in a similar context [42,43]. Under the present reaction conditions, the highest headspace concentrations were measured after 4 h.

The headspace concentrations of δ-damascone released from conjugate 16 were found to be the lowest of the series, with 67 and 53 ng L⁻¹, measured after 4 and 6 h, respectively. These concentrations increased slightly for the release from conjugates 13 and 14, where 88 and 76 ng L⁻¹ were recorded after 4 h. The highest headspace concentrations of δ-damascone were measured above the sample of conjugate 15, with a value of 184 ng L⁻¹ after 4 h (Figure 6).

Figure 7 compares the performance of δ-damascone conjugates 13–16 by plotting the headspace concentrations measured after 4 h as a function of the EO/PO molar ratio in the Jeffamine^® side chain. Conjugate 13, with an EO/PO ratio of 0.11, gave rise to slightly higher headspace concentrations of δ-damascone (88 ng L⁻¹) than conjugate 16 (67 ng L⁻¹), which has no EO/PO side chains. Conjugate 14, with an EO/PO ratio of 0.14 in the side chain, gave rise to 76 ng L⁻¹ of δ-damascone above the dry cotton surface. As a consequence, the grafting of PPO or of copolymers with a low EO/PO molar ratio did not improve the performance of the profragrances in the targeted application.

However, grafting of a side chain with a high EO/PO ratio significantly improved the amount of δ-damascone released into the headspace. Moving from δ-damascone conjugate 16 (with no EO/PO side chains) to copolymer 15 (with a EO/PO ratio of 3.60 in the side chains) increased the amount of δ-damascone in the headspace from 67 to 184 ng L⁻¹, corresponding to an increase by a factor of more than three (Figure 7). This result was somehow unexpected. In our previous work, the grafting of a PEO side chain onto a δ-damascone-releasing poly(maleic anhydride) profragrance derived from 3 (Figure 2) resulted in a decrease in performance on dry cotton [27]. Although the presence of the polar PEG side chain increased the performance of the delivery system in water, it was less efficiently deposited onto the cotton surface, which resulted in lower headspace concentrations in the final application.

Our present data show that increasing the EO/PO molar ratio in the polymer side chain resulted in an increase of the headspace concentration of δ-damascone evaporated from the cotton surface. This suggests that the presence of PPO in the side chain has a positive effect in improving the performance of the delivery system. It seems that the side chain with an EO/PO ratio of 3.60 represents a good balance to be sufficiently hydrophilic to efficiently increase the release of δ-damascone from the conjugate, while the presence of PPO results in a certain hydrophobicity, which is useful for deposition on cotton. An EO/PO molar ratio is therefore suitable to achieve satisfactory deposition with sufficiently fast release kinetics.