Photo-Fries rearrangements of 1-naphthyl esters in the glassy and melted states of poly(vinyl acetate). Comparisons with reactions in less polar polymers and low-viscosity solvents

Abstract

The photo-Fries and associated photoreactions of four 1-naphthyl acylates have been examined in two types of poly(vinyl acetate) (PVAc) films above and below their glass transition temperatures, T_g. Because of the ‘templating’ effect of the esters on their reaction cavities, especially below T_g, the distributions of photo-Fries products, as mandated by the intermediate acyl/1-naphthyloxyl singlet radical pairs, are determined largely by the initial conformations of the guest molecules. Even above T_g, at 50 °C, where segmental chain motions of PVAc are relatively rapid, the influence of the cages in directing product formation is apparent. The radical-pair recombination rates for formation of the keto precursor of 2-(2-phenylpropanoyl)-1-naphthol upon irradiation of 1-naphthyl 2-phenylpropanoate in PVAc are reduced drastically as the temperature is lowered from above to below T_g. Comparison of results in PVAc with those in low-viscosity solvents (ethyl acetate and hexane) and low-polarity polymer films (polyethylene and polypropylene) indicate that interactions between the radicals produced from irradiation of 1 and the acetate pendant groups of PVAc, as well as the nature of its chain motions above and below T_g, influence enormously the course of the photo-Fries rearrangements.

---

Introduction

Photochemical reactions of 1-naphthyl acylates are sensitive to the local environments in which they occur. For that reason, their selectivities are convenient and informative probes of the characteristics of anisotropic reaction cavities.¹ Transformations of these esters proceed almost exclusively from excited singlet states, via homolytic scission of the acyl–oxy bond to form geminal acyl/1-naphthyloxyl radical pairs.^2,3 These radicals lead to formation of the three major isolated photoproducts, 1-naphthol (NOL) and 2- and 4-acyl-1-naphthols (2-AN and 4-AN, classical Fries products from cage-recombinations and tautomerizations of the initial keto intermediates³). NOL is usually a cage-escape product, but it can be formed in-cage as well if the cavity walls possess easily abstractable H-atoms. No attempt has been made by us to analyze the aldehyde products that are also formed in many cases as a result of H-atom abstraction by acyl radicals.

Some acyl radicals undergo in-cage decarbonylation at rates competitive with acyl/1-naphthyloxyl recombinations. The resultant photoproducts are 2- and 4-alkyl-1-naphthols (2-BN and 4-BN) (via initially formed keto tautomers) and 1-alkylnaphthalenes (BzON) (Scheme 1).² In such cases, rate constants for in-cage acyl/1-naphthyloxyl recombinations can be calculated.² Decarboxylation, leading to 1-alkylnaphthalenes (BzN), is usually a minor concerted reaction of excited singlet states, but can be important in some media.⁴

Scheme 1 Mechanistic steps involved in formation of photoproducts from 1. Processes in the rectangle are observed only for 1d.

Previously, we examined the photochemical rearrangements of several aryl esters in morphologically complex, low-polarity, polymeric media, especially polyethylenes (PE) of differing crystallinity and polypropylenes (PP) of differing tacticities, above their glass transition temperatures (T_g).^2,5 Our principal objectives in those studies were: (1) to elucidate how reaction cavities of constraining, anisotropic media affect the ground-state conformations and excited-state properties of reactants, as well as their key intermediates; and (2) to identify the factors that influence significantly the reaction pathways in locally anisotropic environments.

In the polyolefinic media, in-cage rates of recombination of the acyl/1-naphthyloxyl radical pairs were found to be faster than relaxation of the cavity walls, and product formation was mediated by the size and shape of the starting ester; there is a ‘templating’ effect.^5c In addition, the size and shape of the radicals constituting the in-cage pair, as well as interactions between a guest (or its intermediates) and the walls of the reaction cavity in which it resides, were found to be capable of influencing photoproduct selectivity.^5c,d,6

Here, the reaction media are two types of a completely amorphous polymer: poly(vinyl acetate) [PVAc-83 (M_wca. 83000) and PVAc-140 (M_wca. 140000)], consisting of short, polar chain branches attached to a low-polarity polymer backbone. Experiments are also conducted in low-viscosity solvents: ethyl acetate (whose polarity is like that of PVAc) and hexane. The reacting molecules are four 1-naphthyl acylates: 1-naphthyl acetate (1a), 1-naphthyl myristate (1b), 1-naphthyl benzoate (1c) [eqn. (1)], and 1-naphthyl 2-phenylpropanoate (1d) [eqn. (2)]. In addition to the objectives cited above, the current work seeks to elucidate especially the influences of performing the photoreactions above and below T_g of PVAc (i.e., over a small temperature range that results in very large changes in the flexibility of the reaction cavity walls). It also explores variables such as polymer composition, polymer morphology, temperature (above and below T_g), and free volume (as it relates to the size, shape, and wall stiffness of reaction cavities¹). We demonstrate that subtle changes in temperature can influence enormously the course of the photo-Fries reactions in these media.

Results and discussion

Irradiations of 1 in isotropic solutions

Irradiations of 1a–c in ethyl acetate or hexane produced 2-AN, 4-AN, and NOL as the only detected products in relative yields that were unchanged (within experimental error) at ≤25% conversion (Table 1). Averages from experiments at different conversions are presented. Since the amount of NOL at 20 °C is ca. 20% from 1a–c in ethyl acetate (as well as from 1b in hexane, where it is predominantly a cage-escape product), the different singlet radical pairs must have similar rate ratios for in-cage recombination and cage escape. However, the rearrangement product ratios, 2-AN/4-AN, depend somewhat on the nature of the acyl fragment (and the solvent). In both ethyl acetate and hexane, the 2-AN/4-AN ratios and relative yields of NOL increase with temperature, indicating that the rate of cage escape is more dependent on temperature than are the rates of acyl addition to the 2- and 4-positions of 1-naphthyloxyl, and that acyl radicals leaving their initial location (near the 1-position of 1-naphthyloxyl) are less likely to migrate to the 4-position within their cage than to separate from their 1-naphthyloxyl partners (i.e., cage escape) at higher temperatures.

Table 1 Relative photoproduct yields (%) from irradiations (λ > 300 nm) of 0.5 mM 1a–c or 2 mM 1d in ethyl acetate (E) and hexane (H) under nitrogen atmospheres

1	Solvent	T/°C	Conversion (%)	2-AN	4-AN	NOL	2-AN/4-AN
a 2-BN (3.5 ± 0.3), 4-BN (9.9 ± 0.1), BzON (2.4 ± 0.4), BzN (0.2 ± 0.1), and (Bz)2 (8.8 ± 0.1) formed also; see text.
a	E	20	2–8	54.5 ± 4.2	22.2 ± 2.0	23.3 ± 6.2	2.5
b	E	0	3–6	62.4 ± 1.0	25.6 ± 3.5	12.0 ± 4.5	2.4
	E	20	4–8	58.4 ± 1.1	19.1 ± 2.4	22.5 ± 1.3	3.1
	E	50	7–11	46.1 ± 1.6	11.6 ± 2.2	42.3 ± 3.7	4.0
	H	0	2–3	54.3 ± 5.4	32.0 ± 1.5	13.7 ± 5.8	1.7
	H	20	3–5	54.6 ± 3.7	25.2 ± 2.9	20.2 ± 1.1	2.2
	H	50	4–6	37.5 ± 1.0	14.5 ± 2.7	48.0 ± 1.9	2.6
c	E	20	5–16	40.4 ± 2.0	41.6 ± 4.6	18.0 ± 5.9	1.0
d ^a	E	22	12–23	56.6 ± 1.4	11.4 ± 0.2	7.2 ± 2.1	5.0

---

Four additional photoproducts {BzON, 2-BN, 4-BN and two diastereomers of 2,3-diphenylbutane [(Bz)₂] from cross-coupling} were formed from 1d, besides the normal Fries and cage-escape products. In hexane, the 2-AN/4-AN ratio from 1d was not changed appreciably when sufficient concentrations of thiophenol to trap most of the decarbonylated photoproducts were added.² On this basis, we conclude that the vast majority of Fries products from 1d in low-viscosity liquids is formed in-cage but decarbonylation products are formed primarily out-of-cage.

Temperature dependence of photoproduct distributions from 1a–c in PVAc films

Irradiations of 1a in both PVAc films resulted in 2-AN/4-AN ratios that were equal to or much higher than those in ethyl acetate (Table 2). The precision of large ratios (i.e., when the yields of 4-AN are small) is low, especially at low conversions and when a significant fraction of the photoproduct mixture is NOL. In this case and with the other 3 esters, there is no indication from UV spectra that 1-naphthyloxyl radicals add to PVAc chains. Although problems associated with the reflection and diffraction of radiation in the films precluded quantum yield measurements, temperatures above T_g led to significantly higher conversions of ester than those below T_g at the same photon fluence. A much smaller dependence of conversion on temperature was observed in hexane and ethyl acetate solutions.⁷ As such, a larger fraction of the radical pairs is able to proceed to products rather than being forced by cavity constraints to return to 1.

Table 2 Relative photoproduct yields (%) from irradiations of ca. 1 mmol kg⁻¹1a, 1b, or 1c in PVAc at different temperatures under nitrogen atmospheres

1	Film	T/°C	Conversions (%)	2-AN	4-AN	NOL	2-AN/4-AN
a	PVAc-83	0	1–3	65.5 ± 3.7	2.3 ± 0.5	32.2 ± 3.8	28
		20	3–4	81.7 ± 3.4	2.0 ± 0.1	16.3 ± 3.4	41
		50	3–5	80.2 ± 3.2	4.0 ± 0.4	15.8 ± 3.3	20
	PVAc-140	0	1–3	90.9 ± 0.1	1.6 ± 0.1	7.5 ± 0.2	57
		20	3–4	85.9 ± 1.0	3.0 ± 0.5	11.1 ± 1.3	29
		50	3–6	61.3 ± 5.7	2.5 ± 0.6	36.1 ± 5.1	24
b	PVAc-83	0	2–3	66.3 ± 2.7	32.0 ± 2.5	1.7 ± 0.2	2.1
		20	3–5	74.1 ± 3.0	23.1 ± 3.2	2.8 ± 0.2	3.2
		50	5–7	83.3 ± 2.4	13.2 ± 2.8	3.5 ± 0.4	6.3
	PVAc-140	0	2–4	81.2 ± 1.2	16.9 ± 1.7	1.9 ± 0.6	4.8
		20	3–6	85.2 ± 0.8	11.7 ± 1.0	3.1 ± 0.3	7.3
		50	5–8	87.9 ± 0.3	8.3 ± 0.7	3.8 ± 0.3	12
c	PVAc-83	0	3–5	93.0 ± 1.2	7.0 ± 1.2	<0.2	13
		20	4–7	92.2 ± 1.0	7.8 ± 0.9	<0.2	12
		50	5–9	86.6 ± 1.4	13.4 ± 1.4	<0.2	6.5
	PVAc-140	0	2–5	89.4 ± 0.7	10.6 ± 0.7	<0.2	8.4
		20	4–7	90.9 ± 0.5	9.1 ± 0.5	<0.2	10
		50	5–8	87.9 ± 3.3	12.1 ± 3.3	<0.2	7.3

---

High yields of NOL, comparable to those found in ethyl acetate, are obtained in PVAc only from 1a, the ester with the smallest acyl group. This suggests that cage escape by acetyl radicals is impeded only slightly by the polymer cages, but is attenuated greatly when larger acyl groups like benzoyl and myristoyl from 1b and 1c, respectively, are involved. As a corollary, the results in Table 2 indicate that 1-naphthyloxyl escapes more slowly from PVAc cages than the singlet radical pairs combine.

Due to the low relative yields of NOL from 1b and 1c and of 4-AN from 1a in the PVAc films (as well as the low sensitivity of NOL to our analytical techniques), no clear temperature dependence on their formation was apparent in these systems. With increasing temperature, the relative yields of NOL from 1aincreased and those of 2-ANdecreased by comparable amounts in PVAc-140; the trends of the two products are reversed in PVAc-83, but the losses and gains are, again, compensated. These results suggest that radical pairs from 1a follow different courses in the less and more branched PVAc films. Higher temperatures facilitate cage escape as well as radical abstraction from the cage walls by one of the radicals in the pair. Thus, differences in the frequency of chain branching in the two PVAc films can influence both cage escape by the small acetyl radical and the rate that it abstracts hydrogen atoms; the probability that branching points (with weaker tertiary C–H bonds) are near a radical pair is greater in PVAc-140. The results are intriguing, but we are reluctant to interpret them further with the limited data set available.

In solution, the relative yields of photo-Fries rearrangement products depend on the π-electron densities at various atoms of the 1-naphthyloxyl radical as well as steric factors related to each site. From EPR data, the 1-naphthyloxyl free electron densities are 0.350 and 0.456 at the 2- and 4-positions, respectively.⁸ On that basis alone, the 2-AN/4-AN product ratios should be <1. In fact, they are somewhat >1 when irradiations are conducted in hexanes or ethyl acetate and are ≫1 in PVAc (vide infra for an explanation based on reaction cavities). At one temperature, 2-AN/4-AN ratios in PVAc-83 and PVAc-140 differ, but not in a systematic fashion. We suspect that the subtle differences between the influences imposed by the two films on the dynamics of the radical pairs from 1 are masked by the large uncertainties in our measurements, especially when the ratios are high; the branching differences between the two films suggested by T_g measurements do not appear to play a significant role in the courses of the photoreactions.

Temperature dependence of photoproduct distributions from 1d in PVAc films

Under our irradiation conditions, only 1d yields measurable quantities of decarbonylated rearrangement products in addition to the normal photo-Fries products (Table 3). Surprisingly, no 4-AN was detected in PVAc films even though there were significant amounts of 4-BN, especially at the lower temperatures. In that regard, the benzoyl radical from 1c and 2-phenylpropanoyl radical from 1d react very differently despite their similar shapes.

Table 3 Relative photoproduct yields (%) from irradiations of 3–7 mmol kg⁻¹1d in PVAc films under nitrogen atmospheres at different temperatures and the calculated photo-Fries rate rearrangement constants, k_2A, from eqn. (3)

Film	T/°C	Conversion range (%)	NOL	(Bz)2	BzN	BzON	2-BN	4-BN	2-AN	4-AN	10^−6k2A/s^−1
PVAc-140	5	10–11	10.3 ± 0.7	<0.1	2.9 ± 0.1	35.4 ± 0.5	1.1 ± 0.2	38.4 ± 3.4	11.9 ± 2.2	<0.1	3.4 ± 0.6
	22	15	22.0 ± 0.7	<0.1	1.5 ± 0.1	28.1 ± 0.4	2.5 ± 0.2	28.9 ± 0.7	17.0 ± 0.3	<0.1	12 ± 0.3
	50	12	9.9 ± 0.6	<0.1	0.4 ± 0.2	4.8 ± 0.4	1.2 ± 0.2	3.4 ± 0.8	80.3 ± 0.5	<0.1	860 ± 73
PVAc-83	5	10–12	9.0 ± 1.9	<0.1	4.2 ± 0.3	31.0 ± 2.4	1.8 ± 0.8	27.4 ± 1.6	26.6 ± 1.1	<0.1	9.4 ± 0.6
	50	13–17	5.2 ± 0.4	<0.1	0.9 ± 0.1	6.5 ± 0.6	1.6 ± 0.1	5.0 ± 0.3	80.8 ± 0.4	<0.1	620 ± 33

---

The lack of (Bz)₂ upon irradiation of 1d in PVAc indicates that, at least in this medium, the Fries and decarbonylated rearrangement photoproducts are derived only from intramolecular (in-cage) recombination processes; they are not from 1-naphthyloxyl and either 2-phenylpropanoyl or 1-phenylethyl radicals whose origin is different ester molecules.² As a result, eqn. (3) can be used to calculate k_2A, the rate constant for formation of the keto precursor of 2-AN.²

k _2A ≅ k_−CO[2-AN]/{[2-BN] + [4-BN] + [BzON]} (3)

The more complex photoproduct mixtures and k_2A values from 1d offer more detailed information about reaction selectivity than can be gleaned from 1a–c: photoproduct distributions in PVAc, especially when compared to those from reactions in isotropic media of similar polarity and functionality, provide information about the influence of the polymer on the total reaction course, while rate constants probe the environmental factors that modulate the dynamics of in-cage radical-pair motions leading to individual products.

The rate constant k_−CO for decarbonylation of the 2-phenylpropanoyl radical (generated here from 1d) has been measured as a function of temperature in the low-viscosity and low-polarity medium, isooctane,⁹ but not in more polar or viscous environments.^10,11 Although the high microviscosity and anisotropy of PVAc cages are not expected to affect the magnitude of k_−CO,¹¹ polarity should somewhat because the dipole of 2-phenylpropanoyl is larger than its constituents, CO and 1-phenylethyl. For example, the decarbonylation rate constant for phenylacetyl at room temperature in hexane (relative permittivity 1.89) is ca. 3 times that in acetonitrile (relative permittivity 38.8)¹¹ and twice that in methanol (relative permittivity 33.1). The relative permittivity of PVAc, reported to be 3.5 at 50 °C,¹² is actually lower than that of ethyl acetate, 6.4. On that basis, k_−CO in PVAc should be very similar to the values in isooctane (relative permittivity 1.96), and they are employed in our calculations.

Because our experiments do not distinguish between NOL formed from different pathways (Scheme 1), the k_2A calculated from eqn. (3) at 5 and 50 °C may be over-estimated slightly. Regardless, the potential errors from NOL are small and the rate constants in Table 3 should be precise relative to each other. The data demonstrate that the state of PVAc, as modulated by temperature, has a significant effect on the courses and rates of product formation; the relative yields of 2-BN, 4-BN, and BzON are much smaller above T_g than below it, and the k_2A values above T_g are >60 times larger than those below it.

Almost no NOL was detected upon irradiation of 1b–d in PE films, but much larger quantities were produced from 1a.^5c Furthermore, very low yields of NOL and no detectable (Bz)₂ were produced upon irradiation of 1d in either isotactic or syndiotacticPP at 5 °C (i.e., above their T_g);² 1-naphthyloxyl recombines more rapidly with myristoyl, benzoyl, and 2-phenylpropanoyl than they can escape from their initial PE cages. On these bases, and from the absence of detectable amounts of (Bz)₂ from 1d, we believe that a significant fraction of the NOL from irradiations of 1a or 1d in PVAc films arises from hydrogen abstraction by 1-naphthyloxyl radicals within their initial cages, competing with singlet radical-pair recombinations. In-cage formation of NOL is competitive with radical-pair recombination and cage escape here because PVAc has much more easily abstractable H-atoms (along chains at carbon atoms bearing acetate pendant groups) than does PE or PP.

Photodecarboxylation, a process attributed to excitation of naphthyl esters held in specific conformations,^4c is not important in PVAc films. No BzN was detected from 1a–c in PVAc or in isotropic solutions.

Properties of PVAc films and their influence on rearrangements of 1

Basic properties of PVAc. Atactic PVAc is typically a completely amorphous polymer¹² as a result of its many branches induced by chain transfer reactions that accompany the free-radical polymerizations of vinyl acetate.¹³ In addition, PVAc is more polar and, especially at temperatures above T_g, more homogeneous microscopically than partially crystalline PE and PP, the polyvinylic polymers we have employed previously to investigate photo-Fries rearrangements. The T_g values of PVAc polymers should be constant in the high molecular weight ranges of PVAc-83 and PVAc-140, regardless of molecular weight distributions (e.g., the polydispersity of PVAc-83 is 3–3.5).¹⁴ However, the higher T_g value of PVAc-140 (33 °C) indicates that it is more highly branched than the shorter PVAc-83 (27 °C),¹⁵ and, as noted, the walls of its reaction cavities may be more reactive. However, the physical properties of the walls of the reaction cavities afforded by both PVAc polymers (NB, the motions of the chains constituting the walls) should influence the guest molecules in a similar fashion. Macroscopically, decreasing the temperature increases the viscosity of PVAc, especially below T_g; microscopically, motions of radicals from 1 are slowed because segmental motions of polymer chains are attenuated at lower temperatures above T_g and they nearly stop below T_g. Based on the reported dipole moment per monomer unit of PVAc below (2.3 D at 20 °C) and above T_g (1.77 D at 150 °C),¹² the acyloxy groups may also rearrange somewhat near T_g.

Influence of PVAc polarity on radical-pair recombinations. Generally, the greater polarizability of radicals (R˙) makes their diffusion markedly slower than that of their parent molecules (R–H).¹⁶ The magnitude of the decrease is dependent on the nature of the radical and the solvent molecules (i.e., how they are able to interact). For instance, H-bonding can play an important role in mediating radical-pair diffusion and reorientation.¹⁷ The large dipoles of acetate pendant groups of PVAc should allow the radical centers of 1-naphthyloxyl and acyl fragments from 1 to interact more strongly with PVAc than with either PE or PP. As a result, radical-pair recombination processes can be slower in PVAc than in nonpolar polymer films of comparable viscosity. These interactions should also enhance the probability of in-cage formation of NOL in PVAc (vide supra) under conditions where the rates of chain relaxations are the same for PVAc, PE, and PP. Since the relative permittivity, a more quantitative measure of PVAc medium polarity, should be nearly the same above and below T_g, the major reason for the observed temperature dependences of the photoproduct selectivities and rate constants is attributed to changes in the reaction cavities.

Reaction cavities in PVAc. Inhomogeneities in reaction environments become more apparent below T_g. Long chain segments become locked into specific conformations and only very short chain or pendant group segmental motions remain possible on the short time scales needed for nuclear motions that transform molecules of 1 to their products.^18,19 The time scale for segmental motion of polymer chains below T_g is on the order of 100 s,²⁰ but the product-determining radical-pair recombination processes studied here occur in less than one microsecond (as estimated from the radical-pair recombination rates of 1d). Based on correlation frequencies characteristic of molecular motions in PVAc from mechanical, dielectric, and NMR measurements, attenuated acetate side chain motions persist far below T_g;²¹ frequencies of ca. 10⁵ s⁻¹ can be estimated for rotations about (chain)C–OAc bonds just below T_g. Some reorientation of radical pairs from molecules of 1 in the glassy state is possible, but only over short distances and in directions dictated by the side chains of the polymer.²² The influence of PVAc cavities, with limited free volume and hard or soft walls depending on the temperature, on the mobility of a radical pair is presented in cartoon form in Fig. 1. Here, fast and slow motions of the polymer chains (indicative of ‘soft’ and ‘hard’ walls, respectively) are defined according to the time scale for photo-Fries processes by 1 and its intermediates.¹

Fig. 1 Cartoon representations of PVAc reaction cavities with (a) hard walls (i.e., below T_g) and (b) soft walls (i.e., above T_g). Significant radical-pair reorientation is impeded when walls are hard and the free volume is small.

A similar relationship between relaxation of cavity walls and radical-pair reaction times applies to our previous studies of 1a–c photoreactions in PE.^5c Typically, a single, time-averaged reaction cavity (i.e., a ‘single effective environment’) need be invoked to explain the influence of a low-viscosity, isotropic solvent because the cavities are able to adapt rapidly to changes mandated by a reacting guest molecule.

The above data suggest strongly that more than one type of reaction cavity for 1 may be present in PVAc films, especially below T_g. However, the principal piece of experimental evidence that bears on this issue, the independence of photoproduct distributions from conversion (up to our self-imposed limit, ca. 25%) at temperatures above and below T_g, indicates that only one operationally distinguishable cavity type need be invoked, and we abide by Ockham's razor, especially in the calculation of k_2Afrom 1d.²

Mean free volumes in reaction cavities, the molecular volumes of 1, and their influence on photo-Fries rearrangements. The amount and location of free volume available in reaction cavities are key contributors to the fates of radical pairs from 1.^23,24 The motion of PVAc polymer chains and, consequently, the mean free volumes of cavities increase much more rapidly with increasing temperature above T_g than below it.^15,25,26 The rates of reaction and diffusion of guest molecules within PVAc are affected by both the size of its reaction cavities^22c,27 and the translational and rotational motions of its chain segments along the walls.²¹

From positron annihilation studies, the mean free volumes of holes in a different PVAc to that employed here are ca. 85 ų at 5 °C (below T_g) and ca. 105 ų at 50 °C (above T_g).²⁶ The mean free volumes of the holes in PVAc-83 and PVAc-140 films above or below their T_g are probably comparable. The van der Waals volumes in Table 4²⁸ indicate that even the smallest ester, 1a, does not fit comfortably in the vast majority of holes in a native polymer (i.e., one undisturbed by the presence of guest molecules). However, our doping process for incorporating molecules of 1 into PVAc films involves swelling of the polymer chains by cyclohexane or hexane (see Experimental section). This process allows the guest molecules to diffuse rapidly into the films by increasing the spaces between chains. Once the solvent is removed, the chains contract around the dopant molecules, but not necessarily in the same arrangements as in the native films. Nevertheless, as in PE and PP, a significant ‘templating’ effect by the ‘stiff’ walls of the PVAc reaction cages is expected, and a distribution of cage types should be present below T_g: in addition to decreasing somewhat the hole sizes, lowering temperature increases wall stiffness to a much greater degree (vide supra). In turn, shape anisotropies and, potentially, the distribution of site types experienced by molecules of 1 are increased.

Table 4 Calculated van der Waals volumes (V_M) of 1

	1a	1b	1c	1d
V M/Å^3	166	370	220	254

---

Since the cited hole free volumes are mean values without information about their distribution, and the molecular volumes of 1 and the hole free volumes have been calculated by different methods, the degree to which the reaction cages constrain motions of the reacting 1 can be discussed only qualitatively. Nevertheless, an acyl radical is much nearer the 2-position than the 4-position of 1-naphthyloxyl at the moment of formation of the radical pair from an electronically excited 1 (Fig. 2). The longer the distance an acyl radical center must migrate to add to the 4-position of 1-naphthyloxyl, restrictions to shape changes imposed by slowly relaxing PVAc cavity walls (especially below T_g), and the intrinsically slower translational and rotational motions expected of the radicals within the more polar PVAc (than PE or PP) cages must contribute to the remarkable regioselectivity of photoproduct formation observed.

Fig. 2 Cartoon representation of the formation of 2-AN, 4-AN, and NOL from 1 in the restricted environments of PVAc cages.

A radical with a long alkyl chain, like myristoyl from 1b, can experience much stronger dispersive interactions with PE chains than with those of PVAc. The potentially greater importance of dispersive interactions is underscored by the absence of discernible 4-AN from irradiations of 1b in PE films (i.e., only 2-AN is detected),^5c while both AN isomers of 1b are found in PVAc. The long alkyl chain of myristoyl can pack more tightly and interact more strongly in more extended conformations with segments of PE polymer chains than with those of PVAc. Kinked chains of 1b in PVAc may help to template the formation of both 2-AN and 4-AN.

Both 2-AN and 4-AN were formed from irradiations of 1d in PE and PP films.^2,5b The absence of 4-AN from irradiations of 1d in PVAc films at temperatures both above and below T_g is remarkable given the formation of 4-BN (as well as the other expected decarbonylation products) and the presence of 4-AN from irradiations of 1d in ethyl acetate. For reasons similar to those mentioned above, the PVAc reaction cavities should restrict severely the motions of a bulky 2-phenylpropanoyl radical, making difficult its approach to the 4-position of 1-naphthyloxyl in a trajectory compatible with bond formation.² The approaches necessary for bond formation between a decarbonylated radical and either the oxygen atom or 2- or 4-positions of 1-naphthyloxyl are less demanding than those of the corresponding acyl (Fig. 3). As a result, we believe that the fraction of 2-phenylpropanoyl that would otherwise lead to the formation of the keto precursor of 4-AN, were its migration within the cage more rapid than the rate of decarbonylation, loses CO in the restricted environment afforded by PVAc before approaching (along a bond-making trajectory) the 4-position of 1-naphthyloxyl.

Fig. 3 Approximate orientations for radical additions to 1-naphthyloxyl by (a) 2-phenylpropanoyl and (b) 1-phenylethyl radicals. Addition to C-4 is shown. The z-axis is orthogonal to the 1-naphthyloxyl plane.

Influence of the glassy versus the melted state of PVAc. The enormous changes in the photoproduct selectivities and k_2A from 1d in PVAc films between 5 and 50 °C (Table 3) are a consequence of the large changes in the reaction microenvironments that occur on passing from below T_g to above it. Over a somewhat broader temperature range (5–60 °C) in PE (where the amorphous parts and interfacial parts of the films, in which 1d resides, remain in their melted state), the photoproduct distributions and k_2A rate constants vary much less and the rate constants conform to a single Arrhenius behavior.² As expected, logarithms of rate constants from 1d in PVAc-140versus inverse temperature do not fall on a common straight line.

The larger reaction cavities and more flexible walls of PVAc above T_g allow more and faster motions of the guest molecules and their intermediates. Conversely, the cessation of most segmental chain motions below T_g^19,21 freezes molecules of 1 in specific conformations and leads to severe resistance to motions of the radical pairs that deviate from the overall shapes of their ester precursors.

As exemplified by the results from 1b in ethyl acetate and hexane (Table 1), the magnitude of the 2-AN/4-AN photoproduct ratios is small and they are only slightly dependent on temperature between 0 and 50 °C. In PVAc, the larger magnitudes of these ratios and their small temperature dependence (Table 2) from 1b, as well as from 1a and 1c (i.e., the esters whose acyl radicals have the least conformational freedom), suggest that the ‘templating’ effect of their reaction cavities is retained for the most part above T_g. Even well above T_g (at 50 °C), the mean free volumes of holes in PVAc²⁶ are much smaller than the size of even the smallest naphthyl ester investigated, 1a (Table 4). The ‘softer’ walls of the reaction cavities above T_g allow additional motions, but the guests are still constrained.

As mentioned, no 4-AN was detected upon irradiations of 1d in PVAc at any of the temperatures explored, and the relative percentage of 2-AN decreased dramatically (at the expense of increasing decarbonylation products) as temperature was decreased from above to below T_g. The BzON/2-BN/4-BN ratio of decarbonylation products in ethyl acetate at room temperature is 0.69 ∶ 1.0 ∶ 2.8. At 50 °C (i.e., above T_g), these ratios are very similar in the two PVAc films (4.0 ∶ 1.0 ∶ 2.8 in PVAc-83 and 4.1 ∶ 1.0 ∶ 3.1 in PVAc-140), but BzON (the decarbonylated product requiring the least radical-pair motion and orientational specificity to be formed) is obtained in the largest yields. The total relative yield of the decarbonylation products dramatically increases in PVAc below T_g, suggesting that the rates of decarbonylation of 2-phenylpropanoyl are affected less by the state of the polymer film (i.e., temperature) than are the rates of radical-pair motions: the ratios of the sum of the decarbonylated product yields divided by the yield of (the sole) normal Fries product, {[BzON] + [2-BN] + [4-BN]}/[2-AN], increase by 14 (PVAc-83) and 54 (PVAc-140) between 50 and 5 °C (Table 3). Consistent with this hypothesis, between 50 and 5 °C, k_2A decreases by >250 (PVAc-140) and >60 (PVAc-83) while k_−CO (in isooctane) is calculated to decrease by only ∼5 fold.⁹

The disparity in the changes between k_−CO and either k_2A or the {[BzON] + [2-BN] + [4-BN]}/[2-AN] product ratios indicate, again, that modifications to the structure and flexibility of the PVAc reaction cavities induced by the state of the polymer play a significant role in determining the fates of the radical pairs generated from the esters; segmental chain motion is a key factor in facilitating (and directing) radical-pair recombination. In evidence of this, k_2A from 1d increased only 4–12 fold between 5 and 60 °C in unstretched PE films (where the state of the films does not change), and is 10–100 fold larger in PE and PP at 5 °C (i.e., well above their T_g) than in PVAc.² Furthermore, although the magnitudes of the decreases in mean hole free volume of PVAc upon decreasing the temperature through T_g are similar to those found when PE films are cold-stretched at one temperature,^5cPE film stretching results in only small decreases in k_2A.² Film stretching restricts radical-pair recombination processes due principally to decreases in free volume, but decreasing the temperature to below T_g restricts the free volume and stiffens significantly the cavity walls.

Conclusions

Changes induced in PVAc films by lowering the temperature from above to below the glass transition temperature are sensed by acyl/1-naphthyloxyl singlet radical pairs generated upon electronic excitation of 1-naphthyl acylate guest molecules. The changes are detected by both the reactivity of the radical pairs and the selectivity of their recombination modes. From comparisons of the results in the PVAc films with those in low-viscosity solvents (ethyl acetate and hexane) and in low-polarity polyethylene and polypropylene films, we conclude that the major contributor to the large temperature dependence observed in PVAc is changes in the stiffness of reaction cavity walls (i.e., rates of segmental chain motions) rather than in the volumes of the reaction cavities. In addition, the influence of the media on the photochemistry is a sensitive function of the nature of the acyl groups of 1 and how the acyl radicals of each ester are able to interact with their local environment.

These explanations and the model that follows from them do not explain all of the results. For instance, throughout the temperature range explored, BzON/4-BN ratios from irradiations of 1d in PVAc are ca. 1 ∶ 1, and the relative yields of 2-BN remain small and change less than those of the other decarbonylated products. If decarbonylation is less likely from radical pairs whose initial disposition in PVAc reaction cavities favors formation of the keto precursor of 2-AN or reformation of the starting ester than from radical pairs inclined to yield the keto precursor of 4-AN, the BzON/4-BN ratios should not have remained nearly constant below and above T_g. The results suggest that radical pairs that migrate within their radical cages and lose the ‘history’ of that motion are more prone to form BzON and 4-BN (as decarbonylated products) than 2-BN. Fathoming the exact cause of this curious observation and several others will require additional experimentation. However, it is clear that PVAc films offer interesting opportunities to tune dramatically the courses of other photochemical reactions by small changes in temperature. The lack of chromophores in PVAc that absorb strongly above ∼300 nm adds to the attractiveness of this polymer as a matrix for photochemical studies.

Experimental

Most methods of sample preparation, irradiation procedures, and analytical protocols, as well as syntheses and characterizations of reactants and photoproducts, have been reported elsewhere.^2,5b,c,29

Instrumentation

Melting points (corrected) were determined on a Leitz SM-LUX-POL microscope equipped with cross polarizers, a Leitz 585 thermostatting stage, and an Omega HH21 Microprocessor thermometer connected to a K thermocouple. Differential scanning calorimetry (DSC) to determine glass transition temperatures (see Fig. S1, ESI†) was conducted on a TA 2910 instrument interfaced to a TA Thermal Analyst 3100 controller. Aluminium pans were used and experiments were performed under a slow stream of nitrogen. The heating rate was 10 °C min⁻¹ and the cooling rates were proportional to the difference between the temperature of the sample compartment of DSC and room temperature. IR spectra (Nujol mulls or KBr pellets) were recorded on a MIDAC FT-IR with Spectra-Calc software. ¹H NMR spectra were obtained on Varian Mercury 300 MHz NMR spectrometer interfaced to a Sun SparcStation 5. UV–VIS spectra were recorded on a Perkin-Elmer Lambda 6 UV–VIS spectrophotometer. Gas chromatography (GC) was performed on a Hewlett-Packard 5890 gas chromatograph equipped with a flame ionization detector, a 0.25 μm Alltech DB-5 (15 m × 0.25 mm) column, and a Hewlett-Packard 3393A integrator. GC-MS was performed on a FISONS MD-800 instrument using the same type of DB-5 column. High performance liquid chromatography was conducted on a Waters chromatograph with a UV detector (254 nm), an electronic integrator, and either an Alltech 250 × 4.4 mm 5 μm silica gel column or a Waters Symmetry C18 3.9 mm × 150 mm column.

Reagents

Solvents (ACS or HPLC grade, Aldrich) and reagents were used without further purification unless indicated otherwise. 2-Myristoyl-1-naphthol, 2-acetyl-1-naphthol and 4-acetyl-1-naphthol were obtained from Mr Xiaochun Wang. 1-Naphthyl acetate (1a) (ACROS, 99%) was sublimed under vacuum to yield white crystals, mp 45–46 °C (lit. 45–46 °C³⁰).

Film preparations and doping

PVAc films were prepared by two methods: (a) PVAc-83 (M_wca. 83000; polydispersity 3–3.5) and PVAc-140 (M_wca. 140000) pellets from Aldrich were dissolved in methylene chloride and cast into films on clean glass slides. Air-dried films (>2 weeks) were immersed in 3 cyclohexane aliquots during more than one week to remove additives and were dried under vacuum (<0.1 mmHg for ≥3 hours) before use. (b) PVAc pellets were dissolved in warm CH₂Cl₂, precipitated by slow addition of petroleum ether at 0 °C, filtered, and dried under vacuum at room temperature overnight. The polymer was then cast into films on clean glass slide from methylene chloride solutions and dried under vacuum before use.

Films were doped by immersing them in hexanes or cyclohexane solutions of 1 for 12–30 h at room temperature (for 1a–c) or 40 °C (for 1d). The surfaces of the films were rubbed with a tissue soaked in hexane to remove occluded ester molecules, and then dried under vacuum (<0.1 Torr for ≥3 h). The dopant concentrations, ca. 1 mmol kg⁻¹ (1a–c) or 3–7 mmol kg⁻¹ (1d), were calculated from UV–VIS absorption spectra (assuming absorption coefficients for the esters in ethyl acetate), and film thicknesses [measured by a micrometer to be ca. 1 mm (1a–c) or 100 μm (1d)].

Irradiations and analyses

Films and isotropic solutions were placed inside septum-sealed Pyrex vials and either purged or bubbled with nitrogen for >15 min before being irradiated (>300 nm) in a thermostatted water bath with a 450 W medium pressure Hg Hanovia arc. Irradiated films containing 1a–c were placed in 20 ∶ 1 (v/v) hexanes–ethyl acetate for 48 h at room temperature and those containing 1d were placed in cyclohexane baths for >5 h in the dark. Thereafter, the UV–VIS absorption spectra of the irradiated/extracted and native, undoped films were indistinguishable, indicating >95% removal of unreacted 1 and its photoproducts. At least 2 irradiations were performed for each PVAc type or solvent to different conversions (<25%). Irradiated isotropic solutions were analyzed directly by chromatography, 1a–c by HPLC and 1d by GC. The liquid extracts from PVAc films were either analyzed directly or concentrated, if necessary, by slowly bubbling dry nitrogen through them before analysis. Photoproducts were identified by comparison of retention times of peaks with those of authentic samples and by co-injection of reaction solutions with authentic samples. Relative concentrations were determined from peak areas and the ratios of their molar absorption coefficients in the elution solvent at the wavelength of detection (HPLC) or relative detector response factors (GC). The distributions of products showed no systematic deviations within the range of conversions explored. Relative photoproduct yields are the averages from at least 6 separate GC or HPLC analyses (≥2 irradiated samples and ≥3 chromatograms per sample); precision errors are one standard deviation. Total yields of photoproducts are normalized to 100%.

Acknowledgments

We thank the National Science Foundation for its support of this research and Professors Teresa Atvars and V. Ramamurthy for useful discussions.

References

1. R. G. Weiss, V. Ramamurthy and G. S. Hammond, Photochemistry in organized and confining media, Acc. Chem. Res., 1993, 26, 530–536.
2. W. Gu and R. G. Weiss, Mediation of photochemical reactions of 1-naphthyl phenylacylates by polyolefin films. A ‘radical clock’ to measure rates of radical-pair cage recombinations in ‘viscous space’, Tetrahedron, 2000, 56, 6913–6925.
3. (a) N. P. Gritsan, Y. P. Tsentalovich, A. V. Yurkovskaya and R. Z. Sagdeev, Laser flash photolysis and CIDNP studies of 1-naphthyl acetate photo-Fries rearrangement, J. Phys. Chem., 1996, 100, 4448–4458; (b) V. I. Sternberg, Organic Photochemistry, ed. O. L. Chapman, Arnold, New York, 1967, vol. 1, ch. 3; (c) M. A. Miranda, in Organic Photochemistry and Photobiology, eds. W. M. Horspool and P.-S. Song, CRC Press, Boca Raton, 1995, ch. 47; (d) J. W. Meyer and G. S. Hammond, Mechanism of photochemical reactions in solution: LXX. Photolysis of aryl esters, J. Am. Chem. Soc., 1972, 94, 2219–2228.
4. (a) R. A. Finnegan and D. Knutson, Photodecarboxylation process: preparation of some hindered aryl ketones, Chem. Ind., 1965, 1837–1838; (b) R. A. Finnegan and D. Knutson, Stereochemistry of photodecarboxylation and photodecarboxylation reactions of aryl esters. The photolysis of (S)-(+)-3,5-di-tert-butylphenyl 2-methylbutanoate, J. Am. Chem. Soc., 1967, 89, 1970–1972; (c) W. Gu, D. J. Abdallah and R. G. Weiss, Conformational control of photo-induced decarboxylation of simple aryl esters. Enhancement by templating effects in polyethylene films, J. Photochem. Photobiol., A, 2001, 139, 79–87.
5. (a) C. Cui and R. G. Weiss, Photo-Fries rearrangements of 2-naphthyl acylates as probes of the size and shape of guest sites afforded by unstretched and stretched low density polyethylene films. A case of remarkable selectivity, J. Am. Chem. Soc., 1993, 115, 9820–9821; (b) W. Gu, M. Warrier, V. Ramamurthy and R. G. Weiss, Photo-Fries reactions of 1-naphthyl esters in cation-exchanged zeolite Y and polyethylene media, J. Am. Chem. Soc., 1999, 121, 9467–9468; (c) W. Gu, A. J. Hill, X. Wang, C. Cui and R. G. Weiss, Photo-rearrangements of five 1- and 2-naphthyl acylates in three unstretched and stretched polyethylene films. Does reaction selectivity correlate with free volumes measured by positron annihilation lifetime spectroscopy, Macromolecules, 2000, 33, 7801–7811; (d) W. Gu and R. G. Weiss, Photo-Fries rearrangements of phenyl phenylacylates in polyethylene films. Comparison of reactivity and selectivity with 1-naphthyl phenylacylates, J. Org. Chem., 2001, 66, 1775–1780.
6. W. Gu, M. Warrier, B. Schoon, V. Ramamurthy and R. G. Weiss, Understanding the influence of active (zeolite) and passive (polyethylene) reaction cages on photo-Claisen rearrangements of aryl benzyl ethers, Langmuir, 2000, 16, 6977–6981.
7. S. Bi, Photo-Fries reaction of 1-naphthyl myristate, 1-naphthyl benzoate, and 1-naphthyl acetate in poly(vinyl acetate) films, MS Thesis, Georgetown University, Washington DC, 2001.
8. W. T. Dixon, W. E. J. Foster and D. Murphy, The electron spin resonance spectra of naphthoxyl radicals, J. Chem. Soc., Perkin Trans. 2, 1973, 15, 2124–2127.
9. N. J. Turro, I. R. Gould and B. H. Baretz, Absolute rate constants for decarbonylation of phenylacetyl and related radicals, J. Phys. Chem., 1983, 87, 531–532.
10. L. Lunazzi, K. U. Ingold and J. C. Scaiano, Absolute rate constants for the decarbonylation of the phenylacetyl radical, J. Phys. Chem., 1983, 87, 529–530.
11. Y. P. Tsentalovich and H. Fisher, Solvent effect on the decarbonylation of acyl radicals studied by laser flash photolysis, J. Chem. Soc., Perkin Trans. 2, 1994, 729–733.
12. M. K. Lindemann, in Polymer Handbook, 4th edn., eds. J. Brandrup, E. H. Immergut and E. A. Grulke, Wiley, New York, 1999, pp. V77–V85.
13. O. Nuyken and J. Crivello, in Handbook of Polymer Synthesis, ed. H. R. Kricheldorf, Marcel Dekker, New York, 1991, pp. 164–172.
14. J. M. G. Cowie, Polymers: Chemistry and Physics of Modern Materials, 2nd edn., Chapman and Hall, New York, 1991, ch. 12 and 15.
15. See for instance: S. Matsuoka, Relaxation Phenomena in Polymers, Carl Hanser Verlag, Munich, 1992.
16. M. Terazima, Is the translational diffusion of organic radicals different from that of closed-shell molecules, Acc. Chem. Res., 2000, 33, 687–694.
17. See for instance: P. Franchi, M. Lucarini, G. F. Pedulli, L. Valgimigli and B. Lunelli, Reactivity of substituted phenol toward alkyl radicals, J. Am. Chem. Soc., 1999, 121, 507–514.
18. K. Yoshii, S. Machida and K. Horie, Local free volume and structural relaxation studied with photoisomerization of azobenzene and persistent spectral hole burning in poly(alkyl methacrylate)s at low temperatures, J. Polym. Sci,. Part B: Polym. Phys., 2000, 38, 3098–3105.
19. R. B. Seymour and C. E. Carraher, Jr., Polymer Chemistry, 3rd edn., Marcel Dekker, New York, 1992, pp. 27–29.
20. C. A. Angell, Formation of glasses from liquids and biopolymers, Science, 1995, 267, 1924–1935.
21. M. J. R. Hoch, F. A. Bovey, D. D. Davis, D. C. Douglass, D. R. Falcone, D. W. McCall and W. P. Slichter, Nuclear magnetic resonance in poly(vinyl acetate), Macromolecules, 1971, 4, 712–715.
22. Recently, some details of the nature of segmental motions of polymers related to PVAc near T_g have been revealed using spectral changes in or reaction rates of small probe molecules.^22a–d As expected, more heterogeneous environments are found as the temperature is lowered to slightly above T_g. (a) L. A. Deschenes and D. A. V. Bout, Single molecule studies of heterogeneous dynamics in polymer melts near the glass transition, Science, 2001, 292, 255–258; (b) Z. Wang, D. A. Holden and F. R. W. McCourt, Generation of nonrandom chromophore distributions by the photo-Fries reaction of 2-naphthyl acetate in poly(methyl methacrylate), Macromolecules, 1990, 23, 3773–3779; (c) J.-W. Park, M. D. Ediger and M. M. Green, Chiral studies in amorphous solids: the effect of the polymeric glassy state on the racemization kinetics of bridged paddled binaphthyls, J. Am. Chem. Soc., 2001, 123, 49–56; (d) H. Schäfer, U. Albrecht and R. Richert, Dispersive first-order reactions. II. Photochromic model system, Chem. Phys., 1994, 182, 61–68.
23. M. H. Cohen and D. J. Turnbull, Molecular transport in liquids and glasses, J. Chem. Phys., 1959, 31, 1164–1169.
24. J. M. Zielinski and J. L. Duda, Predicting polymer/solvent diffusion coefficients using free-volume theory, AIChE J., 1992, 38, 405–410.
25. (a) J. Y. Ye, T. Hattori, H. Nakatsuka, Y. Maruyama and M. Ishikawa, Microscopic dynamics of the glass transition investigated by time-resolved fluorescence measurements of doped chromophores, Phys. Rev. B, 1997, 56, 5286–5296; (b) M. Faetti, M. Giordano, D. Leporini and L. Pardi, Scaling analysis and distribution of the rotational correlation times of a tracer in rubbery and glassy poly(vinyl acetate): An electron spin resonance investigation, Macromolecules, 1999, 32, 1876–1882.
26. Y. Kobayashi, W. Zheng, E. F. Meyer, J. D. McGervey, A. M. Jamieson and R. Simha, Free volume and physical aging of poly(vinyl acetate) studied by positron annihilation, Macromolecules, 1989, 22, 2302–2306.
27. B. T. Muellers, J.-W. Park, M. S. Brookhart and M. M. Green, Glassy state and secondary structures of chiral macromolecules: polyisocyanates and polyketones, Macromolecules, 2001, 34, 572–581.
28. A. Bondi, van der Waals volumes and radii, J. Phys. Chem., 1964, 68, 441–451.
29. C. Cui, X. Wang and R. G. Weiss, Investigation of the photo-Fries rearrangements of two 2-naphthyl alkanoates by experiment and theory. Comparison with the acid-catalyzed reactions, J. Org. Chem., 1996, 61, 1962–1974.
30. D. J. Crouse, S. L. Hurlbut and D. M. S. Wheeler, Photo-Fries rearrangements of 1-naphthyl esters in the synthesis of 2-acylnaphthoquinones, J. Org. Chem., 1981, 46, 374–378.

---

Footnotes

† Electronic supplementary information (ESI) available: Fig. S1. See http://www.rsc.org/suppdata/pp/b1/b107011h/

‡ Current address: Department of Chemistry and Biochemistry, University of Colorado, Boulder, CO 80309, USA.

---