Pressure-Induced Magnetic Crossover Driven by Hydrogen Bonding in CuF2(H2O)2(3-chloropyridine)

Hydrogen bonding plays a foundational role in the life, earth, and chemical sciences, with its richness and strength depending on the situation. In molecular materials, these interactions determine assembly mechanisms, control superconductivity, and even permit magnetic exchange. In spite of its long-standing importance, exquisite control of hydrogen bonding in molecule-based magnets has only been realized in limited form and remains as one of the major challenges. Here, we report the discovery that pressure can tune the dimensionality of hydrogen bonding networks in CuF2(H2O)2(3-chloropyridine) to induce magnetic switching. Specifically, we reveal how the development of exchange pathways under compression combined with an enhanced ab-plane hydrogen bonding network yields a three dimensional superexchange web between copper centers that triggers a reversible magnetic crossover. Similar pressure- and strain-driven crossover mechanisms involving coordinated motion of hydrogen bond networks may play out in other quantum magnets.


I. Room Temperature Vibrational Properties
In order to investigate the 0.8 GPa antiferromagnetic to ferromagnetic crossover in CuF 2 (H 2 O) 2 (3-chloropyridine), we carried out high pressure spectroscopic measurements.
The infrared and Raman spectra at ambient conditions (Fig. S1) display a large number of vibrational modes. Our lattice dynamics calculations allow us to assign the peaks below 500 cm −1 as displacements involving the bipyrimidal copper environment, for example the F-Cu-F symmetric and asymmetric stretches. These features are highly collective. Modes above this threshold originate from the 3-chloropyridine ring, or are well-known organic functional group vibrations such as the O-H (3200 cm −1 ) stretch). Assignment of these modes (Tables S1 and S2) enables us to understand the role of the lattice in the pressure-driven magnetic crossover.

II. Variable Temperature Vibrational Properties
Although CuF 2 (H 2 O) 2 (3-chloropyridine) displays an orthorhombic (Pnma) to monoclinic (P2 1 /c) structural transition between 200 and 100 K S1 , our variable temperature infrared measurements reveal no signature of symmetry changes down to 4.2 K (Fig. S2). In other words, although more vibrational modes are expected in the lower symmetry monoclinic phase as compared to the orthorhombic phase, the spectral patterns are equivalent. We therefore assume that pressure-induced changes in the room temperature orthorhombic phase are mirrored in the low temperature monoclinic phase. Short-range antiferromagnetic ordering is observed below 10 K, and long range ordering below 2.2 K in CuF 2 (H 2 O) 2 (3chloropyridine) S1 . Our low temperature infrared spectra reveal no changes with the onset of short range magnetic ordering, indicating that antiferromagnetic ordering does not change the local structure. This allows us to use our room temperature high pressure measurements to understand the low temperature 0.8 GPa magnetic crossover.

III. Infrared Signatures of the Structural Transition Between 4 and 5.5 GPa
Beyond revealing the mechanism of the magnetic transition, we discovered an additional rather sluggish structural transition between 4 and 5.5 GPa. While there are many signatures of this higher pressure transition, the Raman spectra shown in the main text are the most revealing. The infrared active modes support this discovery as well (Fig. S3). As with the Raman active modes, the infrared features sensitive to the transition are mostly related to the pentacoordinate copper environment. The disappearance of five modes, the appearance of a new mode at 490 cm −1 , and the 190 cm −1 mode splitting all signal the structural transition.
That they appear at low frequency indicates that there is a real lattice component to this

IV. Reversibility Considerations
We have shown in the main text that the 0.8 GPa includes a significant contribution from the lattice, making it a magnetoelastic transition. Moreover, a new structural crossover was discovered between 4 and 5.5 GPa. No other transitions were observed up to 11.5 GPa, although many modes broaden significantly hinting at an onset of amorphization (Fig. S4).
Upon the release of pressure, the system springs back into the low-pressure orthorhombic phase. Judging by the character of the spectrum, no damage was done. This reversibility allows for simple magnetostructural switching of CuF 2 (H 2 O) 2 (3-chloropyridine) through the application and release of pressure, making it a candidate material for piezomagnetic applications.

V. Calculation Details
To understand the spectral results as well as the magnetic properties of the molecular crystal, we employed a multi-scale approach where both the molecular unit as a building block was modeled using molecular orbital theory and the magnetic properties under pressure were calculated using super cell techniques and band structure methods. We realize that many of the vibrational modes in the crystal are dominated by the inter-atomic vibrations of the molecular building block as these are linked in the crystal through hydrogen bonds.
In order to learn about the molecular vibrational modes we calculated the frequencies of a single unit of CuF 2 (H 2 O) 2 (3-chloropyridine) taken from the crystal geometry. The calculations were based on density functional theory with generalized gradient approximation for exchange and correlation potential. Since CuF 2 (H 2 O) 2 (3-chloropyridine) is bonded through a hydrogen bonding network, we have used Grimme's functional S2 including dispersion correction with the B97D S2 functional and SDD S3 basis as implemented in Gaussian09 S4 . The infrared and Raman frequencies were calculated without optimizing the geometry in the gas phase. Understandably, this led to a number of imaginary frequencies most of which come from 3-chloropyridine and H 2 O. To circumvent this problem we have calculated the infrared and Raman spectra for individual 3-chloropyridine and H 2 O units at the same level of theory. Figure S5 summarizes our results, and mode assignments are given in Tables S1   and S2. within the generalized gradient approximation is used except as otherwise stated. Full ionic and supercell relaxation are performed at a given external pressure implementing the conjugate gradient algorithm. Convergence criteria for total energy and ionic force components were set at 10 −4 eV/cell and 10 −3 eV/Å, respectively. Since the crystal is bound by a network of hydrogen bonds, we used the modified density functional theory method of Grimme S2 to properly treat the long range dispersive interactions. The Brillouin Zone was represented by means of 3x3x1 Monkhorst and Pack S8 k points grid. Calculations using a denser grid 6 showed no qualitative improvement in the calculated results.
Because CuF 2 (H 2 O) 2 (3-chloropyridine) magnetically orders below 2.1 K, we chose the low-temperature monoclinic phase to simulate the phase transition. We used enthalpy (H=U+pV) rather than merely total energy to assess the relative stability of different magnetic configurations as a function of pressure. Based on the experimental observations, both the ferromagnetic and antiferromagnetic configurations were relaxed at 0, 0.6, 0.8, 1, 2, 2.5, and 3 GPa pressure. The changes in relative enthalpy and volume with increasing external pressure were calculated. The relative enthalpy was found to be positive around 0.8 GPa implying an antiferromagnetic to ferromagnetic crossover.