Investigation of nucleation and growth phenomena during the thermal and light induced spin transition in the [Fe(1-bpp)₂][BF₄]₂ complex

Introduction

Molecular spin crossover (SCO) complexes are well known to display bistability between their high spin (HS) and low spin (LS) electronic configurations [1–3]. The SCO phenomenon can be triggered by a variety of external perturbations (temperature change, application of pressure or magnetic field, light irradiation, etc.) and entails a spectacular change of different material properties: color, density, magnetic susceptibility, dielectric permittivity, etc. The molecular spin state change in solids gives rise to elastic interactions between the molecules, leading to the emergence of cooperative phenomena, such as first-order phase transitions accompanied by heterogeneous phase separation. This latter aspect of the spin transition has been recently the focus of several studies using optical microscopy for imaging single crystals of various SCO compounds [4–20]. The main results are consistent with the general expectation in that the spin transition in cooperative systems occurs via heterogeneous nucleation and growth mechanism, while the spin crossover in weakly cooperative systems is a spatially rather homogeneous transformation. Remarkably, in high quality, cooperative crystals a very reproducible nucleation and growth of a single macroscopic domain with a well-defined “transition front” was reported [17, 18]. On the other hand, the mechanical effects accompanying the spin transition associated with the brittleness of the crystals have led in several cases to damage of the crystals (formation of cracks and other defects) over repeated cycling. The limited number of robust, high-quality single crystals represents thus a major obstacle for the deeper investigation of the spatio-temporal dynamics of the spin transition by microscopy techniques. In this paper, we report on the optical microscopy study of single crystals of the [Fe(1-bpp)₂][BF₄]₂ complex (1-bpp=2,6-di(pyrazol-1-yl)pyridine) (1, Scheme 1).

Scheme 1

Representation of the cationic [Fe(1-bpp)₂]²⁺.

This compound is known for its extremely abrupt cooperative spin transition around 260 K associated with a small (ca. 3 K wide) hysteresis loop [21–24] and for the remarkably high stability of the light-induced metastable HS form [25]. Indeed 1 is one of the rare SCO compounds in which a long-lived metastable HS phase (induced by light-irradiation) can be observed above liquid-nitrogen temperature.

Experimental section

Results and discussion

The variable-temperature magnetic susceptibility data of 1 is shown in Fig. 1. At room temperature sample 1 has a χ_MT value close to 3.1 cm³ K mol^–1 as expected for a d⁶ Fe(II) ion in the HS state (S=2). This value remains nearly constant until 260 K where the magnetic response decreases abruptly to achieve a diamagnetic value (LS state) – in agreement with the results reported in the literature [17, 19].

Fig. 1

Thermal variation of χ_MT (a) and of the optical reflectance (b) of 1 on heating and cooling through two successive thermal cycles. (The lines are guides to the eye.) Arrows indicate the temperatures for which snapshots (right panel) have been taken during the optical reflectance measurements.

The spin transition in 1 was also studied by means of optical reflectivity with the aim of monitoring the associated thermochromism, the HS/LS states having mustard yellow and dark brown colors, respectively. The variable temperature optical reflectivity in Fig. 1 reveals a very abrupt spin transition around 253 K with a small (2 K) hysteresis, which was perfectly reproducible over several thermal cycles. This hysteresis is not observable in the magnetic curve, which is also slightly less abrupt. The difference between the optical and magnetic response is simply linked to the fact that the optical measurement was done on a single crystal, while the magnetic measurement on an ensemble of crystals. Since each crystal has slightly different spin transition properties (due to the intrinsic defects) in the ensemble measurement the hysteresis smears out and the transition spreads over a larger temperature range. Intriguingly, the optical reflectivity decreases gradually well before the HS to LS spin transition. This colour change from 300 K to 252 K is clearly illustrated by the images recorded during the experiment (Fig. 1). This change is hardly visible in the magnetic data, which means that only a very small fraction of the molecules (ca. 2–3 %) are transformed to the LS state in this temperature region. Such premonitory phenomena are known to occur for various phase transitions [28–30] and are often associated with heterophase fluctuations, i.e., the formation and collapse of clusters of the low-temperature phase near the transition temperature. In the case of first-order spin transition phenomena such fluctuations have low probability due to the high elastic strain energy they involve, but we believe that in the vicinity of the critical point (where the two phases become miscible) this probability becomes higher as might be the case with compound 1. Nevertheless, this slight discrepancy between the two measurements (optical and magnetic) could also be explained by the limited penetration depth of light for the reflectivity measurement [31]. In order to confirm that this premonitory thermochromic effect is not associated with a structural phase transition and to try to understand its origin single crystal X-ray diffraction patterns, DSC curves as well as vibrational spectra (Raman and FTIR) were acquired between 200 and 300 K. Crystallographic unit cells parameters of 1 (see above) are in agreement for both spin states with those reported in the literature [21] and do not indicate a structural phase transition. The temperature dependence of the heat flow (DSC) in the cooling and heating modes is shown in Fig. 2a. The only anomaly in the heat flow for compound 1 appears in the cooling/warming modes at 251/255 K, which we can unambiguously assign to the first-order spin transition. The enthalpy (ΔH) and entropy (ΔS) changes associated with this thermal anomaly are 14.5 kJ mol^–1 and 71 J mol^–1 K^–1, respectively, which are typical values for iron(II) spin transition solids [1]. Figure 2b shows also selected IR absorption spectra associated with the pure HS state (303 K and 288 K) and the pure LS state (258 K and 213 K). A few marker bands can be chosen to follow the spin transition, for example, peaks at 1177 cm^–1 and 1224–1233 cm^–1 are characteristic of the LS state while the peaks at 1182 cm^–1 and 1222 cm^–1 only appear in the HS state. The Raman spectra of compound 1 at selected temperatures are shown in Fig. 2c. A characteristic marker peak appears in the HS spectrum at 1024 cm^–1, while two other peaks at 1047 and 1060 cm^–1 can be assigned to the LS state. It is interesting to notice that the premonitory phenomenon recorded above the transition temperature by the reflectance measurement is not revealed by these spectroscopic characterisations.

Fig. 2

DSC analysis (a), temperature dependent IR absorption spectra (b) and Raman spectra (spectra at 123 and 83 K correspond to the LIESST experiments) (c) of 1.

Figure 3a displays two selected images of 1 recorded by transmitted light microscopy in the HS (263 K) and LS (243 K) states. Through the HS → LS thermal transition the colour of the crystal changes from golden to dark-orange. This colour difference is mainly due to the higher absorbance of the LS molecules in the visible spectral range. Figure 3b presents a restrained thermal cycle imaging between 257.5 and 259.2 K. The HS → LS transition starts at the centre of the crystal and the LS phase grows through two different axes of propagation: first to the bottom left direction and then to the upper right direction. This behaviour is somewhat unusual since nucleation was usually observed at the edges of different SCO crystals [16, 18], but not completely unexpected since internal defects (cracks, etc.) can also allow for reduced nucleation energy barriers. During the opposite transition (LS → HS) nucleation and growth occurs with opposite sense, i.e., nucleation in the bottom left and upper right corners and propagation toward the center of the crystal. This reversible behaviour has already been observed in different SCO compounds and may be explained by the presence of small temperature gradients in the crystal [10].

Fig. 3

(a) Bright-field optical microscopy images in transmission mode of a crystal of 1 in the two spin states. (b) Selected snapshots of the spatio-temporal evolution of the spin transition on cooling (upper panel) and heating (lower panel). The corresponding movie is available in the Supplementary Information (SI). (c) Normalized variation of the optical transmittance of the crystal on cooling and heating. (The transmittance of the whole crystal was integrated for each optical microscopy image through the thermal cycle.)

A video showing the phase separation during the thermal spin transition in 1 is also shown in the SI. The corresponding normalized optical transmittance vs. temperature plot is presented in Fig. 3c. Due to the movement of the crystal during thermal cycles the measurement baseline is not very stable and the reversibility is also affected. Nevertheless this figure reveals the exceptionally abrupt nature of the spin transition in this sample, which is completed in a temperature interval of ca. 0.1 K around 257.4 K (on cooling) and 258.8 K (on heating).

In addition to the thermal spin transition, a study of the light-induced spin-state switching was also conducted in the same crystal. Indeed, Money et al. [25] have shown that upon light irradiation into the low-lying MLCT (metal-ligand charge transfer) absorption bands of 1 the ground ¹A state can be quantitatively transformed into the long-lived metastable ⁵T state at cryogenic temperatures. (At 70 K the complete decay back to the ground state takes more than 6 h in the dark [25].) This is the famous ‘light-induced excited spin state trapping’ or LIESST effect [32–34]. We explored the photo-switching of the spin states of our crystals first by means of Raman spectroscopy. As shown in Fig. 2c between ca. 300 and 250 K the Raman spectra are the same and correspond to the HS state. Upon further cooling to 230 K the Raman markers of the LS state appear and the transformation is quasi complete. The sample was then cooled to 123 and 83 K in order to carry out the LIESST experiments. The laser of the Raman spectrometer is used both for photoswitching the compound from the LS to the HS state and for the simultaneous detection of this phenomenon. At 123 K, only a partial transformation of the sample from the LS to the HS state is observed while a complete transformation is detected at liquid nitrogen temperature. In the next step we used the same laser for photoswitching the crystal, but instead of recording the Raman spectra we visualized the laser-induced changes in the crystal using the optical microscope of the Raman system. In this experiment the crystal was first cooled to 80 K in the dark to reach the LS state. Then, the laser beam (ca. 3 mW) was focused on the crystal during 10 s with an estimated spot size of 1.5 μm. The video recorded during the laser irradiation is available in the SI and selected snapshots are shown in Fig. 4a. Despite the fact that the laser irradiation was focused only into a small volume of the crystal we observed that the whole crystal transformed progressively into the HS state within <10 s. This transformation of the non-irradiated LS state to the HS state is a very surprising observation at first sight since the ground state of the system at this temperature is the LS state. There are two trivial explanations for this behaviour. First of all, one might suppose that the laser induced heating is so important that the whole crystal is heated above the thermal transition temperature. To test this hypothesis we have repeated the photo-switching experiment shown in Fig. 4a under the same irradiation conditions, but at a higher temperature (100 K). As shown in Fig. 4b at this temperature (in contrast to the 80 K experiment) only a rather partial photo-conversion was obtained after 10 s irradiation, which obviously contradicts the laser-induced heating hypothesis. Another plausible explanation for the switching of the whole crystal is that it is induced by the (white light) irradiation of the microscope lamp. This hypothesis was also discarded as we observed no photo-conversion at all upon prolonged exposure of 1 to the microscope lamp at 80 K. To explain our experimental observations we recall that Money et al. [25] have shown that under continuous irradiation there exists a light-induced instability phenomenon in 1, which leads to the so-called light-induced thermal hysteresis, LITH, loop between ca. 60 and 88 K. As it was discussed by Desaix et al. [35] this instability results from the competition between the photoexcitation and the relaxation, which is a non-linear process. The stability conditions (i.e., the width of the LITH loop) will depend on the laser beam intensity and it turns out that in our experimental conditions at 80 K the HS steady-state will be the stable state. The key novelty here is that the HS state is stabilised not only in the irradiated volume, but in the whole crystal, which is certainly a consequence of the collective behaviour of SCO molecules through long-range elastic interactions.

Fig. 4

(a) Reflected light optical microscopy snapshots of 1 recorded at 80 K while focusing a laser beam (633 nm, 3 mW) in the area indicated by a circle. (The image recorded at t=0 s corresponds to the onset of the laser irradiation. The laser light was filtered out using a narrow stop-band filter. The full video is available in the Supporting Information.) (b) Photo of the crystal recorded at two different temperatures after a 10 s laser irradiation. A full and a very incomplete conversion is observed at 80 K and 100 K, respectively.

After switching off the laser the HS → LS relaxation process was followed isothermally as a function of time using the microscope. A sequence of pictures showing this relaxation at 83 K is depicted in Fig. 5a. Although after multiple experiments a few surface defects started to emerge at the bottom of the crystal, the relaxation process was reproducible suggesting that the crystallinity of 1 remained unchanged. The relaxation was also plotted as the total variation of the optical transmittance (see Fig. 5b). The observed sigmoidal relaxation pattern is typical of a cooperative system and caused by a self-acceleration process as it was discussed by Hauser [32]. Surprisingly, the relaxation process occurs in an apparently homogeneous way in contrast to the temperature induced transition of the same crystal. While in the latter case a well-defined phase boundary develops and moves across the crystal, during the relaxation process we observed no phase separation (within the resolution of the optical microscope). This homogeneous character of the phase transformation seems counter-intuitive at first sight, but is in fact not inconsistent with the sigmoidal curve observed during the HS to LS relaxation process. Indeed, the cooperativity corresponds primarily to long-range intermolecular elastic interactions (due to the misfit of HS and LS molecular volumes). The energy cost for creating a HS/LS domain wall is highly unfavourable due to the lattice relaxation involved in such interface. Theoretical investigations have shown that in a perfect crystal, i.e., without defects, no clustering process can be observed [36], explaining why mean-field approaches are so popular in the SCO field. Nonetheless, the striking difference between the spatio-temporal development of the thermal spin transition and that of the HS → LS relaxation in 1 remains an open question and will require a deeper study. In particular, the crystal structures of the different thermo- or photo-induced phases will have to be compared carefully.

Fig. 5

(a) Bright field optical microscopy snapshots of the HS to LS relaxation of a crystal of 1 at 83 K following a laser irradiation induced switching of the crystal from the stable LS to the metastable HS phase. The image recorded at t=0 s corresponds to the laser switch off. The full video of the relaxation is available in the Supporting Information. (b) Normalized variation of the optical transmittance of the crystal. The transmittance of the crystal was integrated through the relaxation process for three different parts of the crystal (open symbols) and also for the whole crystal (closed symbols). The inset shows the three regions of the crystal selected for the analysis.

Conclusions

We succeeded to grow large single crystals of the [Fe(1-bpp)₂][BF₄]₂ SCO complex, which proved to be very robust over repeated cycling between the HS and LS phases allowing for a detailed optical microscopy investigation of the thermal and light-induced spin transitions. The thermal spin transition occurs in a reversible manner by a heterogeneous nucleation followed by the propagation of well-defined “transition fronts.” The HS → LS thermal transition is preceded by a premonitory spin conversion between ca. 300 and 260 K. Since we observed no sign of any other structural transformation in this temperature range we tentatively described this phenomenon as heterophase fluctuations, which may occur due to the proximity of the critical point. Shining laser light into a small volume fraction of the crystal at low temperatures (80 K, 100 K) leads to a photoinduced spin-state switching from the LS to the HS phase. Remarkably, at 80 K the photo-converted HS “nucleus” continues to grow until the whole crystal converts to the HS phase. We explained this spectacular phenomenon by the light-induced bistability of the system, which affects the whole crystal due to the long-range (elastic) intermolecular couplings. The relaxation curve of the light-induced HS phase has the well-known sigmoidal shape and the relaxation process appears spatially very homogenous. While this observation is not in conflict with theoretical predictions, the marked difference with the thermal transition (demixtion) is puzzling. Deep theoretical studies will be necessary to rationalise these experimental observations – this work is in progress in our group.