Phase transition and proton ordering at 50 K in 3-(pyridin-4-yl)pentane-2,4-dione

At ambient temperature, the H atom in the short intramolecular hydrogen bond of 3-(pyridin-4-yl)pentane-2,4-dione is disordered; below 50 K, a structural phase transition and proton ordering are observed.

Abstract

Single-crystal neutron diffraction experiments at 100 and 2.5 K have been performed to determine the structure of 3-(pyridin-4-yl)pentane-2,4-dione (HacacPy) with respect to its protonation pattern and to monitor a low-temperature phase transition. Solid HacacPy exists as the enol tautomer with a short intramolecular hydrogen bond. At 100 K, its donor···acceptor distance is 2.450 (8) Å and the compound adopts space group C2/c, with the N and para-C atoms of the pyridyl ring and the central C of the acetyl­acetone substituent on the twofold crystallographic axis. As a consequence of the axial symmetry, the bridging hydrogen is disordered over two symmetrically equivalent positions, and the carbon–oxygen bond distances adopt intermediate values between single and double bonds. Upon cooling, a structural phase transition to the t ₂ subgroup P occurs; the resulting twins show an ordered acetyl­acetone moiety. The phase transition is fully reversible but associated with an appreciable hysteresis in the large single crystal under study: transition to the low-temperature phase requires several hours at 2.5 K and heating to 80 K is required to revert the transformation. No significant hysteresis is observed in a powder sample, in agreement with the second-order nature of the phase transition.

1. Introduction  

The substituted acetyl­acetone 3-(pyridin-4-yl)pentane-2,4-dione [HacacPy, (1), Fig. 1 ▸] represents a versatile reagent in coordination chemistry. The neutral molecule may act as a substituted pyridine; in its deprotonated form, acacPy⁻ has been used as an O,O′ chelating ligand towards divalent and trivalent cations (M) and as a ditopic ligand for the construction of oligo- or polynuclear binary coordination compounds (Mackay et al., 1995 ▸; Vreshch et al., 2003 ▸, 2004 ▸, 2005 ▸; Chen et al., 2004 ▸; Zhang et al., 2008 ▸; Knight et al., 2008 ▸; Knight & McCusker, 2010 ▸; Li et al., 2011 ▸; Merkens et al., 2014 ▸; Truong et al., 2017a ▸). In addition, the uncoordinated N atom in the pyridyl ring in [M(acacPy)₃] may provide electron density towards the σ hole in the strong halogen-bond partner 1,4-di­iodo­tetra­fluoro­benzene (Merkens et al., 2013 ▸). Despite these many structurally characterized derivatives, the structure of uncoordinated HacacPy itself has not yet been reported, to the best of our knowledge. In this contribution, we provide the missing information and communicate the crystal structure of (1) based on single-crystal neutron diffraction experiments conducted at 100 and at 2.5 K, and we report the reversible phase transition between the high-temperature phase (1α) and the low-temperature phase (1β). We refer to the related compound HacacPyen (2) in §4.

Figure 1.

Chemical diagram for HacacPy, (1), and for the closely related compound HacacPyen, (2). The twofold crystallographic axis in (1α) is shown as a dashed blue line.

Keto–enol equilibria in the uncoordinated species depend on the individual molecule, state of aggregation and solvent polarity (Truong et al., 2017b ▸). For consistency and in view of the fact that the deprotonated anions are commonly addressed as ‘acetyl­acetonates’, we adhere to the diketo nomenclature.

2. Experimental  

2.1. Synthesis and crystallization  

The ditopic ligand HacacPy was prepared as reported in the literature by Mackay et al. (1995 ▸). In the last purification step of the synthesis, the yellowish residue was sublimated and the product was obtained as colourless crystals of different sizes and quality (see Fig. S1, supporting information). The sublimation temperature allowed us to tune the crystal size; in order to obtain suitable single crystals for the neutron experiments, a sublimation temperature of approximately 328 K was chosen. For the purposes of purification and later use as a ligand in coordination chemistry, the temperature may be set to 343 K. The largest crystal had dimensions of 8.7 × 8.3 × 5.2 mm. For the experiments described here, a transparent single crystal of 4.4 × 4.3 × 3.3 mm was selected. The product was stored in a refrigerator (251 K) and under nitrogen for further use.

2.2. Single-crystal neutron diffraction  

Neutron diffraction was performed at the research neutron source Heinz Maier-Leibnitz (FRM II) of the Heinz Maier-Leibnitz Zentrum (MLZ) on the single-crystal diffractometer HEiDi. This instrument is a four-circle diffractometer with a closed Eulerian cradle and a single-point detector optimized for short wavelengths (Meven & Sazonov, 2015 ▸). Two intensity data sets at 100.0 (10) and 2.5 (5) K were collected, each at two different wavelengths (λ = 0.793 and 1.170 Å, Δλ/λ ≃ 5 × 10⁻³) without additional collimation to the natural collimation of about 1.0° of the beam path towards the sample. An Er filter (0.5 mm thickness) was used to suppress λ/2 and λ/3 contamination. Temperature stability during the experiment was ensured by a closed-cycle Sumitomo RDK101 cryostat which covers the range between 2.2 K and ambient temperature. A calibrated Cernox sensor at the very end of the cooling finger near the sample position was used for determining the sample temperature. The sample itself was connected to the finger with an aluminium pin and wrapped in aluminium foil to optimize thermal conduction to the cryostat. Contamination from external heat transfer was limited by a vacuum cap and a heat shield. The Bragg data were collected in rocking-scan mode. The program used for data reduction was PRON2013 (Meven, 2013 ▸).

2.3. Refinement strategy  

Refinements were conducted on F ² with SHELXL2014 (Sheldrick, 2015 ▸). Crystal data and convergence results for (1α) and (1β) have been compiled in Table 1 ▸.

Table 1. Crystal data and convergence results for (1α) and (1β).

Formula	C10H8NO2
M r	177.20
Crystal size (mm)	4.4 × 4.3 × 3.3
Phase	(1α)	(1β)
Temperature (K)	100.0 (10)	2.5 (5)
Space group, crystal system	C2/c, monoclinic	P , triclinic
a (Å)	11.0366 (4)	6.2755 (2)
b (Å)	13.2899 (5)	8.6368 (2)
c (Å)	6.3763 (2)	8.6570 (2)
α (°)		101.482 (3)
β (°)	99.548 (2)	96.2307 (13)
γ (°)		96.0785 (13)
V (Å3)	922.29 (6)	453.17 (2)
Z	4	2
(sinθ/λ)max (Å−1)	0.73	0.74
No. measured reflections	1688	2315
No. independent reflections	1055	1555
No. observed reflections [I > 2σ(I)]	718	1121
R int	0.0967	0.0733
R1 (obs)	0.0815	0.1069
R1 (all)	0.1260	0.1460
wR2	0.2208	0.2844
S	1.086	1.084
No. of restraints	6	67
No. of parameters	114	160

2.3.1. Intensity data collected at 100 K, i.e. those associated with (1α)  

The preliminary X-ray model was used for the refinement of the non-H atoms; space group C2/c with disorder for the enol moiety was confirmed. The position of the H atom in the intramolecular hydrogen bond deserves a comment. X-ray diffraction had allowed us to identify a local electron-density maximum in a general position close to but not on the twofold crystallographic axis as the electron density associated with the H site. The deepest local minimum (as expected for the negative scattering length of element type H) in a difference nuclear density after refinement of the neutron data amounted to −0.72 fm Å⁻³; the nucleus of H2, a H atom with chemically unambiguous occupancy attached to an aromatic C atom, was associated with a nuclear density of −0.93 fm Å⁻³. In contrast to the X-ray model, this local density minimum for the H atom in the intramolecular hydrogen bond was located in a special position on the twofold axis. The alternative models – a single H centred on the twofold axis and a disordered H in a general position close to the axis – could not be distinguished by tentative refinements, but the contour plot of the difference nuclear density (see Fig. S2, supporting information) suggested overlay of two neighbouring nuclear density distributions, in agreement with a split-position model as for the X-ray-derived electron density. The neutron diffraction experiment also revealed disorder for the acetyl­acetone methyl group. During refinement, this group was treated as disordered over two alternative orientations, with the sum of their occupancies constrained to unity. An even more complicated disorder model involving additional fractional site occupancies was discarded because the residual nuclear densities amounted to only −0.18 fm Å⁻³, to be compared with −0.93 fm Å⁻³ for a fully occupied H site as explained above. Tentative assignment of anisotropic displacement parameters for the H atoms in this disordered group required many additional restraints for convergence at physically meaningful values and did not result in a convincing improvement of agreement factors; the methyl H atoms were therefore treated as isotropic. All other atoms were assigned anisotropic displacement parameters (ADPs); the ADP of the disordered enol H atom was restrained to a more isotropic shape. Tentative introduction of an additional scaling factor between the data collected at two different wavelengths did not result in significant improvements of the agreement factors or changes of geometry or ADPs and was not pursued further. Nine reflections (F _obs ² >> F _calc ²) were considered outliers and excluded from the final refinement; without excluding these reflections, the target function minimized, wR2 amounted to 0.2439.

2.3.2. Intensity data collected at 2.5 K, i.e. those associated with (1β)  

A preliminary structure model was obtained by transformation of the monoclinic structure to space group P . As a result of the phase transition and the associated twinning, reflections were visibly split but could not be separately integrated due to profile overlaps. Convergence results are therefore not fully satisfactory, and no unrestrained refinement was possible. In order to maintain an acceptable ratio between the number of observed data and the number of variables and to ensure physically reasonable ADPs, only N and H were assigned ADPs in the final structure model. H-atom ADPs were restrained to isotropicity, and the crystallographically independent distances C3—C4 and C4—C5 were restrained to be similar. Three reflections (F _obs ² >> F _calc ²) were considered outliers and excluded from the final refinement; without excluding these reflections, the target function minimized, wR2 amounted to 0.2941. The relative domain sizes of the twin refined to 0.523 (9) and 0.477 (9).

2.4. Variable-temperature powder diffraction  

Temperature-dependent diffraction data were measured on flat samples with a Guinier powder diffractometer G645 (Fa. Huber, Rimsting) in asymmetric transmission geometry. A Cu anode (40 kV, 20 mA) and a focused germanium monochromator of the Johansson type generated Cu Kα₁ radiation (λ = 1.54059 Å). Intensities were measured with a scintillation counter at a step width of 0.01° in 2θ with a measurement time of 20 s per step in the range 8–30° in 2θ. The flat samples were placed in a closed-cycle refrigerator (Cryogenics) under high vacuum. The adjustment of the temperature for the measurement between 20 and 200 K was performed by heating against the cold head. For temperature measurement an Si diode was used. The experiment included (a) cooling from 200 to 20 K and heating to 200 K in 20 K steps; (b) cooling from 80 to 20 K and heating to 80 K in 5 K steps. Results for both temperature-dependent experiments have been compiled in the supporting information (see Fig. S3).

3. Results  

X-ray diffraction experiments on single crystals of 3-(pyridin-4-yl)acetyl­acetone [3-(pyridin-4-yl)­pentane-2,4-dione, Hacac­Py] had suggested that the compound adopts space group C2/c at 100 K, with the molecules disordered about a twofold crystallographic axis (see Fig. 1 ▸; the twofold axis is shown as a dashed blue line). Preliminary temperature-dependent powder diffraction experiments had shown splitting of powder lines upon further cooling to temperatures lower than 80 K. Both the necessarily very low data collection temperatures and the challenge to reliably refine hydrogen positions in the case of a phase transition associated with proton ordering induced us to propose single-crystal neutron diffraction experiments at two temperatures, well above and below the suggested transition temperature. We will first address the symmetry relationship between these high- (1α) and low- (1β) temperature phases.

3.1. Symmetry relationship  

Fig. 2 ▸ provides an overview of the unit cells for (1α) (at 100 K) and (1β) (at 2.5 K) and shows their symmetry relationship. Upon cooling, the monoclinic C-centred cell transforms to a triclinic cell of very similar dimensions (translationengleich). The index of the low-temperature subgroup C in the high-temperature supergroup C2/c is 2 and therefore the transition can be associated with a t ₂ type (Müller, 2013 ▸). In the (unconventional) C cell, the twofold symmetry axis lost upon cooling constitutes the twin law relating the domains in (1β). After transformation, refinement is conducted in a smaller conventional cell and space group P ; the required transformation matrix is given in Fig. 2 ▸.

Figure 2.

Phase transition and unit cells for (1α) (at 100 K) and (1β) (at 2.5 K).

3.2. High-temperature phase (1α)  

The outcome of the diffraction experiment at 100 K, i.e. above the expected phase transition temperature, confirms the previous results based on X-ray diffraction with averaged keto/enol geometries for the acetyl­acetone moiety. A displacement ellipsoid plot for a molecule in the high-temperature phase (1α) is depicted in Fig. 3 ▸.

Figure 3.

Displacement ellipsoid plot (PLATON; Spek, 2009 ▸) for a molecule of (1α); ellipsoids are drawn at 50% probability, H atoms of the disordered methyl substituent are shown as spheres of arbitrary radii. Selected interatomic distances (Å) and angles (°): O1—H1O 1.07 (3); O1ⁱ—H1O 1.45 (3); O1···O1ⁱ 2.450 (8); O1—H1O···O1ⁱ 152.3 (12); C8—O1 1.294 (5); C8—C9 1.404 (4). Symmetry operator (i): −x + 2, y, −z + 3/2.

The refinement also revealed rotational disorder for the methyl group associated with C7, with two alternative conformations of similar occupancy; their ratio refined to 0.53 (4): 0.47 (4). This detail had not been detected by inspection of the X-ray intensities collected at the same temperature. Packing calculations (PLATON; Spek, 2009 ▸) showed that this disordered methyl group requires a volume of 36 ų; we will come back to this number below.

3.3. Phase transition  

In preliminary X-ray diffraction experiments, we had collected temperature-dependent powder patterns for (1) and identified splitting of lines in a temperature range well below 80 K. This change in the powder pattern proved reversible. In our proposal for beam time, we suggested that this structural change might be accompanied by ordering of the proton H1O. The unfavourable atomic scattering factor of hydrogen for X-rays and the necessarily very low sample temperature during the diffraction experiment clearly favour neutron diffraction, despite the necessity to grow larger crystals and accept significantly longer data collection times. After completion of the diffraction experiment at 100 K with the result shown in §3.2, the sample was cooled to 2.5 K. Apparently, no qualitative change in the diffraction pattern could be perceived. To our surprise, splitting of reflections only occurred after several hours at the base temperature of 2.5 K; in contrast, upon heating the transition emerged without major delay in time at 80 K. Fig. 4 ▸ monitors the profile of the (0 2) reflection as a function of time and temperature; the right column shows that splitting occurs after ca 6 h.

Figure 4.

Reflection profiles as a function of temperature upon heating (left) and cooling (right) a sample of (1). The single reflection (0 2) in (1α), space group C2/c (e), (f), (g) splits into (2 3 ) and (2 3) in space group P (h); merging of these intensities (a), (b) to a single reflection (c), (d) is much faster when the sample is heated.

As shown in Fig. 2 ▸, the transformation matches the symmetry requirements for a second-order phase transition, but the observed hysteresis could also indicate a first-order transition for which such a symmetry relationship would be accidental but possible. In the case of a second-order transition, the unexpected hysteresis and the sluggishness of the phase transition could be explained by the size of the single crystal under study. In order to test this hypothesis, we repeated the variable-temperature diffraction experiments by X-rays on a powder sample, i.e. on much smaller crystallites. The main result is summarized in Fig. 5 ▸: (a) shows X-ray powder patterns of (1) after cooling to 100 K (black), (b) after cooling to 20 K (blue) and (c) after heating from 20 K to 100 K (red); splitting and re-merging of the reflection at 2θ ≃ 18.5° are clearly visible, the effect being slightly less pronounced for the reflection at 2θ ≃ 16°. Fig. 5 ▸(d) shows a magnification of the temperature-dependent intensity data collection in top view: both splitting of the reflections upon cooling and re-merging upon heating occur at approximately 50 K without significant hysteresis.

Figure 5.

Variable-temperature X-ray powder diffraction patterns (see §2.4). (a), (b), (c) Sections of diffractograms after cooling to 100 K (black) and 20 K (blue) and after re-heating to 100 K (red); (d) top view of the variable-temperature pattern, showing splitting and merging of reflections at ca 50 K.

The temperature-dependent X-ray powder diffraction experiments show that the phase transition is fully reversible and occurs without significant hysteresis in smaller crystallites, thus confirming our previous hypothesis that the transformation is second order and matches the symmetry requirements for a structural phase transition. As expected, only small atomic displacements are required for the transformation and packing is hardly affected (see Fig. S4, supporting information).

3.4. Low-temperature phase (1β)  

The structural phase transition to the low-temperature phase (1β) results in a twinned crystal with small twin obliquity of only fractions of a degree: only the sum of two (in general not equivalent) neighbouring intensities could be measured. Agreement factors are therefore not satisfactory and only a restrained structure model could be refined; details are given in §2.3. Fortunately, the data quality was sufficient to document proton ordering and concomitant formation of longer and shorter rather than averaged bonds in the acetyl­acetone moiety. A displacement ellipsoid plot and details of the intramolecular geometry are provided in Fig. 6 ▸. The methyl groups at C7 and C11 are associated with volume requirements of 34 and 31 ų, respectively, slightly less than in the disordered high-temperature phase.

Figure 6.

Displacement ellipsoid plot (PLATON; Spek, 2009 ▸) for a molecule of (1β); ellipsoids are drawn at 50% probability. Selected interatomic distances (Å) and angles (°) [values for the related compound 3,4-di­acetyl­hexa-2,4-diene-2,5-diol, see text]: O1—H1O 1.028 (12) [1.081 (2)]; O2—H1O 1.475 (14) [1.416 (2)]; O1···O2 2.432 (8) [2.434 (1)]; O1—H1O···O1ⁱ 152.3 (15) [153.90 (18)]; C8—O1 1.346 (11) [1.3087 (12)]; C8—C9 1.382 (11) [1.3915 (10)]; C9—C10 1.433 (11) [1.4399 (10)]; O2—C10 1.296 (13) [1.2672 (12)].

Disorder in the high-temperature phase (1α) precludes a meaningful comparison with diffraction results on related compounds. When putting the low-temperature phase (1β) into a wider context, we have to keep in mind the integration problems (see §2.3) and the resulting higher residuals and standard uncertainties. Many neutron diffraction studies have been devoted to short O—H···O contacts in general; based on neutron diffraction at room temperature, we had found an O—H distance of 1.044 (7) Å in a short intramolecular hydrogen bond of a coordination compound (Englert et al., 1999 ▸). A complete survey of the literature is beyond the scope of this article, but we have compiled the results of neutron diffraction experiments on hydrogen bonds with donor···acceptor distances less than 2.6 Å in the supporting information of a previous publication (Şerb et al., 2011 ▸). Here we only discuss the outcome of neutron diffraction experiments on related enol systems. All of them have been studied repeatedly; in order to facilitate comparison with the alternative references and the Cambridge Structural Database (CSD; Groom et al., 2016 ▸), each structure has been assigned its refcode in addition to the most relevant citation for our comparison. Benzoyl­acetone (BZOYAC; Herbstein et al., 1999 ▸) shows pairwise equal C—O and C—C bonds in the enol moiety over a wide temperature range, including results based on data collected at the lowest experimental temperature of 8 K. The authors find a single local nuclear density maximum for the bridging proton, but in contrast to the situation in (1α), their molecule is in a general position and the difference density contour plot does not suggest split positions for the proton. Herbstein et al. do not find any evidence for disorder even at very low temperatures and interpret their diffraction data as an example of a resonance-assisted hydrogen bond (Gilli et al., 1989 ▸).

In contrast, the bonding situation of the intramolecular hydrogen bond in 5-methyl-9-hy­droxy­phena­lenon (CIXKEA; Kiyanagi et al., 2005 ▸) is temperature dependent. The bridging H atom is disordered at room temperature, and the authors observed formation of a superlattice and two symmetrically independent molecules at T < 42 K. In one of these independent molecules, proton ordering is observed. In view of the rather complex system, the standard uncertainties of atomic coordinates and derived geometric parameters based on neutron data are comparatively large.

A series of neutron and X-ray diffraction experiments has shown that the short intramolecular hydrogen bond in di­benzoyl­methane (DBEZLM; Thomas et al., 2009 ▸) remains asymmetric over the whole temperature range between 100 and 280 K.

Although no phase transition has been documented for 3,4-di­acetyl­hexa-2,4-diene-2,5-diol (TACETA; Piccoli et al., 2008 ▸), this enol offers the closest analogy to (1). Based on time-of-flight Laue data, the authors find that the bridging proton is disordered at ambient temperature and ordered in an asymmetric intramolecular hydrogen bond at T < 110 K. The close relationship with (1β) is also reflected in a very similar O···O distance within the enol moiety; for comparison with our compound, geometry parameters for TACETA at 20 K have been included in brackets in the caption to Fig. 6 ▸.

4. Conclusions and future work  

We have successfully identified a reversible structural phase transition in 3-(pyridin-4-yl)pentane-2,4-dione (HacacPy); it converts the disordered high-temperature phase (1α) in space group C2/c to ordered but twinned crystals of (1β) in P , a translationengleich (Müller, 2013 ▸) subgroup of index 2 of (1α). From higher to lower temperature, the phase transition is remarkably sluggish. As no fully satisfactory individual reflection intensities could be extracted from the twinned (1β) crystals and the structure model for this low-temperature phase suffers from low precision, we plan to investigate a closely related substituted acetyl­acetone by low-temperature single-crystal neutron diffraction. Large crystals of compound (2) (Fig. 1 ▸) have already been obtained. Similar to (1), this ditopic molecule exists as an enol tautomer, but its acetyl­acetone and pyridyl moieties are separated by an additional ethyl­ene bridge. First tests have confirmed that these crystals of (2) are suitable for neutron diffraction and do not undergo any phase transition. We conclude this article by quoting the extensive study by Herbstein et al. (1999 ▸) on benzoyl­acetone; these authors suggest that the delicate distinction between truly delocalized structures and disordered superpositioning of two unsymmetrical structures should ideally be made by low-temperature neutron diffraction.

Supplementary Material

CCDC references: 1582128, 1582129

Funding Statement

This work was funded by RWTH Education Fund – Germany Scholarship grant . RWTH Graduiertenförderung grant . JARA cooperation at the Heinz Maier-Leibnitz Zentrum (MLZ) grant .