Infrared Spectrum and UV-Induced Photochemistry of Matrix-Isolated Phenyl 1-Hydroxy-2-Naphthoate

: The conformational stability, infrared spectrum, and photochemistry of phenyl 1-hydroxy-2-naphthoate (PHN) were studied by matrix isolation infrared spectroscopy and theoretical computations performed at the DFT(B3LYP)/6-311++G(d,p) level of theory. The main intramolecular interactions determining the relative stability of seven conformers of the molecule were evaluated. Ac-cording to the calculations, the twofold degenerated O–H ··· O=C intramolecularly hydrogen-bonded conformer with the phenyl ring ester group ± 68.8 ◦ out of the plane of the substituted naphtyl moiety is the most stable conformer of the molecule. This conformer is considerably more stable than the second most stable form (by ~15 kJ mol − 1 ), in which a weaker O–H ··· O–C intramolecular hydrogen bond exists. The compound was isolated in cryogenic argon and N 2 matrices, and the conformational composition in the matrices was investigated by infrared spectroscopy. In agreement with the predicted relative energies of the conformers, the analysis of the spectra indicated that only the most stable conformer of PHN was present in the as-deposited matrices. The matrices were then irradiated at various wavelengths by narrowband tunable UV light within the 331.7–235.0 nm wavelength range. This resulted in the photodecarbonylation of PHN, yielding 2-phenoxynaphthalen-1-ol, together with CO. The extension of the decarbonylation was found to depend on the excitation wavelength.

Woolfe and Thistlethwaite [9,10] have reported on the photophysics of methyl 3hydroxy-2-naphthoate and phenyl 1-hydroxy-2-naphthoate. They observed that the first compound gives rise to a dual fluorescence emission, with one component exhibiting a large Stokes shift, in contrast to the second compound, which exhibits normal fluorescence [9,10]. Law and Shoham have also investigated the photophysics of methyl 3hydroxy-2-naphthoate in different solvent media and at various temperatures, confirming the previous observations regarding its unusual fluorescence pattern, which they interpreted as being the result of a photoinduced excited-state intramolecular proton transfer (ESIPT), leading to the formation of the enol tautomer of the compound that is responsible

Experimental Details: Matrix Isolation Infrared and Photochemical Experiments
A commercial sample of PHN was provided by Sigma-Aldrich, St. Louis, MO, USA (powder; 99% purity), and was used without any further purification. In order to prepare the cryogenic matrices, the sample was placed in a mini glass oven attached to the vacuum chamber of the cryostat, and was sublimated using a thermoelectrical heating system, whosemain component is a DC Power Supply VITEECOM, model 75-HY5003. The vapor of the compound was co-deposited with a large excess of the host matrix gas (Ar or N 2 , obtained from Air Liquide) onto a cold CsI window assembled at the tip of the cryostat, and cooled to 10 K by an APD Cryogenics DE202 closed-cycle refrigerator system. The solute to matrix ratios were kept at~1:1000 in order to guarantee the adequate isolation of the compound. The temperature of the matrix sample was measured at the sample holder by using a LakeShore Model 331S Temperature Controller, with 0.1 K accuracy.
The infrared spectra were collected in the 4000-400 cm −1 range, with 0.5 cm −1 resolution, using a Nicolet 6700 FTIR spectrometer (Waltham, MA, USA) equipped with a mercury cadmium telluride (MCT) detector and a KBr beam splitter. The instrument was purged by a stream of dry/CO 2 -filtered air in order to avoid interference from atmospheric H 2 O and CO 2 vapors.
The photoexcitation of the matrix-isolated PHN was performed through the outer KBr window of the cryostat using tunable narrowband UV light (full-width at half-maximum of 0.2 cm −1 ) provided by a Spectra Physics MOPO-SL optical parametric oscillator (OPO) pumped by a pulsed (pulse duration 10 ns, repetition rate 10 Hz) Quanta Ray Pro-Series Nd-YAG laser. The excitations were undertaken within the 331.7-235.0 nm wavelength range, and were performed with different pulse energies and exposure times (see Table S1 in the Supplementary Materials for details).

Computational Details
All quantum chemical computations were performed with Gaussian 09 (revision A.02) [18]. The geometries were fully optimized at the DFT level of theory using the B3LYP functional [19][20][21] and the 6-311++G(d,p) basis set [22]. The harmonic vibrational frequencies and infrared intensities were calculated at the same level of theory, the frequencies being subsequently scaled by a factor of 0.978 in order to approximately correct them for the neglected anharmonicity and method/basis set limitations. The assignment of the vibrational modes was performed with the help of the animation module of GaussView (version 5.0) [23]. In the simulated spectra, the infrared bands were built using Lorentzian functions centered at the calculated (scaled) frequencies, and with a full-width at half-maximum (fwhm) of 2 cm −1 .

Geometries and Relative Energies of the PHN Conformers
The relative energies of the conformers of PHN, calculated at the B3LYP/6-311++G(d,p) level, are presented in Table 1. The optimized structures of the various conformers are represented in Figure 1, while their Cartesian coordinates are provided in Table S2 (Supporting Information). Table 2 gives the optimized values for the conformationallyrelevant dihedral angles of the different conformers of PHN. The photoexcitation of the matrix-isolated PHN was performed through the outer KBr window of the cryostat using tunable narrowband UV light (full-width at half-maximum of 0.2 cm −1 ) provided by a Spectra Physics MOPO-SL optical parametric oscillator (OPO) pumped by a pulsed (pulse duration 10 ns, repetition rate 10 Hz) Quanta Ray Pro-Series Nd-YAG laser. The excitations were undertaken within the 331.7-235.0 nm wavelength range, and were performed with different pulse energies and exposure times (see Table S1 in the Supplementary Materials for details).

Computational Details
All quantum chemical computations were performed with Gaussian 09 (revision A.02) [18]. The geometries were fully optimized at the DFT level of theory using the B3LYP functional [19][20][21] and the 6-311++G(d,p) basis set [22]. The harmonic vibrational frequencies and infrared intensities were calculated at the same level of theory, the frequencies being subsequently scaled by a factor of 0.978 in order to approximately correct them for the neglected anharmonicity and method/basis set limitations. The assignment of the vibrational modes was performed with the help of the animation module of GaussView (version 5.0) [23]. In the simulated spectra, the infrared bands were built using Lorentzian functions centered at the calculated (scaled) frequencies, and with a full-width at half-maximum (fwhm) of 2 cm −1 .

Geometries and Relative Energies of the PHN Conformers
The relative energies of the conformers of PHN, calculated at the B3LYP/6-311++G(d,p) level, are presented in Table 1. The optimized structures of the various conformers are represented in Figure 1, while their Cartesian coordinates are provided in Table S2 (Supporting Information). Table 2 gives the optimized values for the conformationally-relevant dihedral angles of the different conformers of PHN.   Table S2. a Angles in degrees. The indicated values correspond to the form shown in Figure 1; for all of the conformers, there is a symmetry-related minimum wherethe dihedral angles assume the symmetric values. b The carbon atoms in bold correspond to the two ring carbon atoms where the substituents are bound; the C-C-O-C dihedral corresponds to the smaller dihedral made by the phenyl ring of the ester group relative to the plane of the carboxylate moiety.
The molecule of PHN has four a priori conformationally-relevant internal rotation axes, which can be defined as a function of the C-C-O-H, O=C-C-C, C-O-C=O and C-C-O-C dihedrals provided in Table 2. The first and second dihedrals determine the relative orientation of the OH and phenylcarboxy substituents, while the third and fourth dihedrals are related to the conformation assumed by the phenyl ester group. One can group the seven different conformers of PHN into two groups, the first comprising conformers 1-3, which bear an O-H···O intramolecular hydrogen bond and are of lower energy, and the second including the remaining four conformers (4-7), in which no intramolecular hydrogen bond exists, and that are of higher energy. All conformers are doubly-degenerated by symmetry, so that all structures shown in Figure 1-the dihedrals of which are specified in Table 2-have symmetry-related equivalent forms in which all of the dihedral angles have symmetric values. For simplicity, in the discussion below, we will only refer to the structures depicted in the Figure. In the first group of conformers, the two lower energy forms (1 and 2) have the ester group in the intrinsically more stable cis conformation (C-O-C=O dihedral equal to~0 • ) [24,25], whereas conformer 3 has this group in a distorted trans geometry where the C-O-C=O dihedral is −118.2 • (the trans arrangement corresponds to a C-O-C=O dihedral equal to~180 • ). The reasons for the higher stability of the cis ester arrangement have been pointed out in detail elsewhere [24]. Essentially, the main factor responsible for the stabilization of a cis carboxylic ester arrangement in relation to the trans one is electrostatic in nature, and results from the different relative orientation of the dipoles associated with the C=O and O-C (ester) bonds in the two types of geometries. In the cis geometry, this alignment is nearly anti-parallel (attractive, stabilizing), while in the trans it is approximately parallel (repulsive, destabilizing). In conformer 3, the distortion relative to the trans conformation results from the strong steric repulsion in that geometry between the phenyl and naphtol moieties (see Figure 1). Conformer 1, in which a strong intramolecular hydrogen bond is established with the carbonyl oxygen atom, is the most stable conformer of PHN, with conformer 2, where a weaker intramolecular hydrogen bond is established with the ester oxygen atom, being higher in energy than form 1 by 14.75 kJ mol −1 . In all of the conformers 1-3, the phenyl ester group is considerably out of the carboxylic plane in order to minimize repulsions within the carboxylic ester group (for 1 and 2) and with the naphtol moiety (in the case of 3). Conformer 3 is higher in energy than the most stable PHN conformer by 33.13 kJ mol −1 .
The four higher-energy non-hydrogen bonded conformers (4-7) have relative energies from ca. 55 to about 81 kJ mol −1 above conformer 1. In all these conformers, the dominant intramolecular interactions are of repulsive type, and are established between the lone electron pairs of the hydroxylic and the carbonylic or carboxylic ester oxygen atoms. In the case of the highest energy conformers 6 and 7, to these repulsions one must also add the extra energy resulting from the unfavorable trans arrangement of the ester group and the phenyl-naphtol repulsions (in the case of 6). As for the lower energy conformers, the high-energy conformers of PHN also have the phenyl ester group considerably out of the carboxylic plane in order to minimize repulsions.
It is interesting to examine in more detail the intramolecular hydrogen bonds present in conformers 1-3. Table 3 presents some structural parameters (bond distances and angles) that are useful to estimate the different strengths of the hydrogen bonds in these conformers. Some of these structural parameters can also be correlated with the different stability of the cis and trans ester arrangements. As seen in the Table, the strongest hydrogen bond occurs in conformer 3, as measured by the values of the H···O and O···O distances and, to a less extent, by the O-H bond lengths found in the three conformers under consideration. The hydrogen bond in conformer 1 is slightly less strong than that present in conformer 3, while that present in conformer 2 (which is established with the ester oxygen atom as acceptor instead of the carbonylic oxygen atom as in 1 and 3) is much weaker, in particular if one takes into account the smaller value for the O-H···O angle and the much longer H···O distance (see Table 3). If we assume that the energy difference between 1 and 2 is mostly due to the different strengths of the intramolecular hydrogen bonds present in these two forms, we can estimate that the one that is present in the most stable conformer is about 10-15 kJ mol −1 stronger than that which is present in conformer 2. On the other hand, assuming that the hydrogen bonds in 1 and 3 do not differ by more than 3-5 kJ mol −1 in energy, the (trans-ester)-(cis-ester) intrinsic energy difference can be roughly estimated as being of ca. 30 kJ mol −1 in PHN, a value that is similar to those found in other carboxylic esters [24,25]. The relative values of the C-O (naphtol) and O-H bond lengths and the H···O distances correlate with the relative strength of the intramolecular hydrogen bonds in the three conformers, the first and the last becoming shorter and the second becoming longer when the hydrogen bond becomes stronger, as expected. On the other hand, the relative values of the C=O bond lengths require a somewhat more elaborated explanation, because two factors are relevant to their determination. On the one side, there is the well-known, more important π-electron delocalization in the cis-ester arrangement compared to the trans-ester geometry, which favors a longer C=O bond in 1 and 2 compared to 3 [24,25], and on the other side, there are the relative strengths of the hydrogen bonds in 1 and 3, in which the carbonyl oxygen atom acts as the acceptor, and the non-participation of this atom in the hydrogen bond in 2, which favors a longer C=O bond in the order 3 > 1 > 2. On the whole, the two factors lead to the observed order for the C=O bond lengths: 1 > 3 >> 2.
The higher degree of π-electron delocalization in the cis-ester conformation (as in 1 and 2), compared to the trans-ester geometry (as in 3) is also in accordance with the shorter C-O (carboxylic) bond length in 1 compared to 3 (form 2 has the longest C-O bond among the three conformers because of its involvement in the intramolecular hydrogen bonding).
In the context of the present work, the relative energies of the different conformers are of particular importance. It is clear from Table 1 that, in the room temperature (298.15 K) gas phase equilibrium, conformer 1 accounts for 99.5 % of the total population. Because the gaseous beam of the compound (and matrix gas: Ar; N 2 ) was at room temperature in the matrix isolation experiments described in the next sections, one can expect that only conformer 1 shall be present in the matrices in a detectable amount. Conformer 2 is predicted by the calculations to account for only 0.5% of the population, which is too low to allow its experimental detection, while the remaining conformers have a negligible population.

Infrared Spectra of Matrix-Isolated PHN
In the present study, PHN was isolated in low temperature (10 K) Ar and N 2 matrices, as described in Section 2. The recorded experimental mid-IR spectra of the as-deposited matrices in the 1800-1100 and 1100-450 cm −1 spectral regions are shown in Figures 2 and 3, in which they are compared with the B3LYP/6-311++G(d,p)-calculated infrared spectrum of conformer 1 (the spectral range 3500-2900 cm −1 is shown in Figure S1). The proposed band assignments are provided in Table 4, together with the calculated frequencies and infrared intensities.  in the argon and N2 matrices, respectively, in good agreement with the predicted value (777 cm −1 ). Besides this, the bands observed at about 800, 755, 735 and 630 cm −1 are also observed in this spectral region in both the argon and N2 matrices, and have significant contributions from the δO-C=O, γC=O and γC-O(H) vibrations. Both the frequencies and relative intensities of these bands agree fairly well with the calculated data, as can be seen in Figure 3 and Table 4.  As seen from Figures 2 and 3 and Table 4, the calculated spectrum for conformer 1 reproduces the experimental data very well, demonstrating the sole existence of this conformer in the matrices and making the assignment of the experimental spectra straightforward. The excellent reproduction of the experimental data by the calculations shall be highlighted, because frequently the DFT method does not describe hydrogen bonding precisely, which is an essential intramolecular interaction in the case of the experimentallyrelevant conformer of the studied molecule. Nevertheless, the functional/basis set used in the present study has been shown to be reliable for infrared spectra predictions for many other intramolecularly-bonded molecules formed exclusively by atoms of the first and second row of the periodic table [26][27][28]. In here, only the assignments for the most prominent characteristic vibrations of PHN will be briefly addressed.  The stretching mode of the H-bonded O-H group can be expected to give rise to a very broad band, spreading across a wide range of frequencies. Accordingly, in the experimental spectra, a broad feature is observed within the 3460-3350 cm −1 range, with maxima at 3435 and 3375 cm −1 in the case of the argon matrix spectrum, and at 3429 and 3375 cm −1 for the N 2 matrix, which are ascribed to the νO-H stretching mode. It is interesting to note that the calculations predict this vibration at 3314 cm −1 , i.e., somewhat shifted to lower frequencies when compared to the experimental data. Such a result may indicate that the intramolecular hydrogen bond strength is overestimated to some extent by the DFT calculations, a result that has been found for other molecules bearing strong intramolecular hydrogen bonds, e.g., malonic acid [29]. The fact that the experimentally-observed bands have two major maxima indicates that, in both matrices, PHN molecules have been trapped in two main matrix sites, a conclusion that is reinforced by the observation of other bands exhibiting a doublet-type structure in other spectral regions, as pointed out below.
The bands originating in the aromatic C-H stretching modes give rise to the structured features spreading from 3261 and 2970 cm −1 , and from 3229 and 2895 cm −1 , in the argon and N 2 matrices, respectively. These spectral features result from extensive overlapping bands due to the different νC-H stretching modes (which can also be expected to be site split) and also comprise contributions resulting from Fermi resonance interactions involving the overtones and combination modes of the δC-H bending and νC-C stretching ring modes whose fundamentals are observed in the 1550-1450 cm −1 range. This is in accordance with the findings of earlier theoretical studies that found a strong contribution of non-fundamental modes to the C-H stretching region [30].  Table 4).
With the exception of the feature observed at 1255 cm −1 , the remaining three modes give rise to four overlapping bands resulting from the simultaneous occurrence of matrix site splitting (two major sites, as seen also for νO-H; see above) and Fermi resonance interactions (tentative interacting modes are respectively the overtone of the mode whose fundamental is observed at 845 cm −1 , the combination tone 687 + 660 cm −1 , and the overtone of the fundamental at 605 cm −1 , in the argon matrix, which in N 2 appear at 846, 691+661 and 604 cm −1 , respectively). It is also interesting to note that the calculations predict the νC=O stretching at a slightly lower frequency value compared to the observed ones, which supports the above conclusion that the strength of the intramolecular hydrogen bond is somewhat overestimated by the calculations.
A remarkable result that can be extracted from the calculations is the fact that the δO-H bending vibration is mixed extensively with other vibrations, contributing significantly to eight normal modes, whose frequencies span from ca. 1270 up to about 1630 cm −1 (see Table 4). As expected, all of these modes thus give rise to bands with a significantly high infrared intensity, which is confirmed by the experimental data (see Table 4 and Figure 2). To a lesser extent, the same can be also stated for both the C-O and C-O(H) stretching vibrations, which are predicted to contribute significantly to four and five normal modes, respectively (in some cases the same to which the δO-H vibration contributes), all of them giving rise to considerably intense experimental infrared bands.
The remaining bands observed in this spectral region in the matrix isolation spectra of PHN are mostly ascribable to vibrations of the aromatic rings and could be assigned straightforwardly in view of the very good agreement with the calculated data (see Table 4).

1100-450 cm −1 Region
In the higher frequency range of this spectral region, from 1100 to 800 cm −1 , most of the bands are of low intensity and originate from vibrations of the aromatic moieties, and the experimental spectra are globally very well reproduced by the calculations. Only two vibrations with significant contributions from other molecular fragments are observed in this part of the spectrum. The first corresponds to the intense bands observed in argon and N 2 at 1088 and 1089 cm −1 , respectively, which are ascribed to a vibration with a relevant contribution from the νC-O(H) stretching (predicted at 1076 cm −1 ). The second appears at 845/840 and 846 cm −1 in the spectra obtained in the argon and N 2 matrices, respectively, which-according to the calculations-will also have a non-negligible contribution from the δC-O-C bending (calculated at 830 cm −1 ; see Table 4). On the other hand, below 800 cm −1 , the number of bands ascribable other than the aromatic groups is larger. The most prominent band is due to the τO-H torsion, which is observed at 774 and 777 cm −1 in the argon and N 2 matrices, respectively, in good agreement with the predicted value (777 cm −1 ). Besides this, the bands observed at about 800, 755, 735 and 630 cm −1 are also observed in this spectral region in both the argon and N 2 matrices, and have significant contributions from the δO-C=O, γC=O and γC-O(H) vibrations. Both the frequencies and relative intensities of these bands agree fairly well with the calculated data, as can be seen in Figure 3 and Table 4.

Narrowband UV-Induced Decarbonylation of PHN
The photochemistry of matrix-isolated PHN (in both argon and N 2 matrices) was here investigated for the first time. The matrices were irradiated using narrowband UV-light (see Section 2 for details); the excitation wavelength range used was chosen whilst taking into account the TD-DFT-B3LYP/6-31G(d) calculated UV spectrum of the compound [17]. After each irradiation, an infrared spectrum was collected and analyzed in order to probe the changes taking place in the sample. The irradiation experiments started using UV light with λ = 331.7 nm, which was subsequently decreased stepwise down to λ = 235.0 nm (the sequences of irradiations performed are described in Table S1, which gives the irradiation wavelengths, laser pulse energies and exposure times used). Noticeable spectral changes indicating the transformation of PHN into other molecules were observed for all wavelengths used, the efficiency of the process increasing with the energy of the excitation UV light used.
The spectra of the irradiated matrices reveal the photogeneration of carbon monoxide (CO), which gives rise to the bands at 2138 cm −1 (in the Ar matrix) and at 2139 cm −1 (in the N 2 matrix), which were assigned to the well-isolated monomer, and at 2126/2122/2118/2114 cm −1 (Ar) and 2126/2117 cm −1 (N 2 ) due to CO molecules interacting with other species in the matrices (Figure 4) [31][32][33][34][35]. These results doubtlessly indicate the occurrence of photodecarbonylation. Together with CO, 2-phenoxynapthalen-1-ol (PNO) shall be produced, which is confirmed by the comparison of the experimental spectra of the irradiated matrices with the calculated spectra of this molecule. This comparison is presented in Figure 5, which shows the 1300-1050 cm −1 spectral region of the difference infrared spectra (spectra of the irradiated matrices at the end of the full sequence of irradiations, minus the spectra of the as-deposited matrices), with the simulated difference spectrum built based on the B3LYP/6-311++G(d,p) calculated infrared spectra of cis PNO (with positive intensities) and of PHN (with negative intensities). In this figure, the calculated spectrum of the cis conformer of PNO is presented, because it was calculated to be more stable than the trans form (by 22.9 kJ mol −1 ) (see Figure 6). The full calculated infrared spectrum of PNO is provided in Figure S2.
The calculations predict that, globally, the infrared spectrum of PNO is less intense than that of PHN. Together with the low efficiency of the decarbonylation reaction (the estimated amount of reactant consumed at the end of the full sequences of irradiation, based on the decrease of intensity of its bands, is less than 5%), this fact makes it possible to detect only the most intense bands of the photoproduct. However, in the 1300-1050 cm −1 spectral region (see Figure 5), in which almost all of the most intense bands of PNO are predicted to occur, the emerging bands fit fairly well those calculated for this compound.  Table S4).
(with positive intensities) and of PHN (with negative intensities). In this lated spectrum of the cis conformer of PNO is presented, because it wa more stable than the trans form (by 22.9 kJ mol −1 ) (see Figure 6). The full ca spectrum of PNO is provided in Figure S2.   Table S1) of PHN isolated in Argon (middle) and N2 (bottom) matrices. The simulated difference spectrum built based on the B3LYP/6-311++G(d,p) calculated IR spectra of PNO (cis form) and PHN (conformer 1) is presented in the (top) of the figure. In the calculated spectra, the wavenumbers were scaled by 0.978. The calculations predict that, globally, the infrared spectrum of PNO is less inten than that of PHN. Together with the low efficiency of the decarbonylation reaction (t estimated amount of reactant consumed at the end of the full sequences of irradiatio based on the decrease of intensity of its bands, is less than 5%), this fact makes it possib to detect only the most intense bands of the photoproduct. However, in the 1300-10 cm −1 spectral region (see Figure 5), in which almost all of the most intense bands of PN   Table S1) of PHN isolated in Argon (middle) and N 2 (bottom) matrices. The simulated difference spectrum built based on the B3LYP/6-311++G(d,p) calculated IR spectra of PNO (cis form) and PHN (conformer 1) is presented in the (top) of the figure. In the calculated spectra, the wavenumbers were scaled by 0.978.
, FOR PEER REVIEW 1 Infrared difference spectra (1300-1050 cm −1 region, in which the most intense bands of PNO occur; baseline ) showing the results of the full sequence of the performed irradiations (see Table S1) of PHN isolated in Argon and N2 (bottom) matrices. The simulated difference spectrum built based on the B3LYP/6-311++G(d,p) calculated of PNO (cis form) and PHN (conformer 1) is presented in the (top) of the figure. In the calculated spectra, the bers were scaled by 0.978. The calculations predict that, globally, the infrared spectrum of PNO is less intens than that of PHN. Together with the low efficiency of the decarbonylation reaction (th estimated amount of reactant consumed at the end of the full sequences of irradiation based on the decrease of intensity of its bands, is less than 5%), this fact makes it possibl  Table S3.

Conclusions
Phenyl 1-hydroxy-2-naphthoate (PHN) was investigated by matrix isolation infrared spectroscopy, complemented by DFT(B3LYP)/6-311++G(d,p) calculations. The investigations were focused on the evaluation of the conformational space of the molecule and reasons for the different stability of its conformers, and on the UV-induced phototransformation of matrix-isolated PHN in argon and N 2 matrices (at 10 K).
The twofold degenerated O-H···O=C intramolecularly hydrogen bonded conformer with the phenyl ring ester group ±68.8 • out of the plane of the substituted naphtyl moiety is the most stable conformer of the molecule, being considerably more stable than the second most stable form (by over 14.75 kJ mol −1 ), in which a weaker O-H···O-C intramolecular hydrogen bond exists. The non-hydrogen bonded conformers were characterized by dominant repulsive intramolecular interactions involving the lone electron pairs of the oxygen atoms as well as the aromatic rings, and were predicted to have much higher energies than the H-bonded forms (over 55 kJ mol −1 ).
The most stable conformer of PHN was successfully trapped both in argon and nitrogen matrices, and was shown to photodecarbonylate upon UV irradiation within the wavelength range 331.7-235.0 nm, with the decarbonylation process showing a low efficiency which increases for shorter irradiation wavelengths. Together with CO, 2-phenoxynaphthalen-1-ol (PNO) was produced and successfully detected in the photolysed matrices. No evidence of the occurrence of any other photo-induced process was gathered. On the whole, the present study contributes to the better understanding of the structure, intramolecular interactions and photochemistry of PHN, a relevant member of the hydroxynaphtoate family of compounds.