1-Solvent-328 K

@nanziang, I went through the temperature series you deposited for compound 1 (the Co₂(TPY)₂(BPTC) flexible MOF) and ran a thermal-expansion read across all eight CIFs. This is the same kind of per-axis analysis we use for anomalous-expansion work, so I thought you might like the numbers back.

Setting-normalized cell data (long axis first, so the Fdd2 and Fddd settings compare directly):

What the series says:

The polar↔nonpolar switch is volume-continuous. Between 260 K (Fdd2) and 270 K (Fddd) in the dry sample the volume changes by −0.014%, far inside the ~0.2–0.5% crystal-to-crystal spread you can see between the two solvent crystals at the same temperature. Volume-wise this behaves like the second-order transition your Cp analysis suggests, not a first-order reconstruction.

The framework is about five times more expansive in the polar phase. α_V ≈ 5.6×10⁻⁴ K⁻¹ at 255–260 K versus ≈ 1.1×10⁻⁴ K⁻¹ at 270–305 K. The six-fold coordination geometry that makes the structure polar apparently also makes the wine-rack softer.

The anisotropy flips across the switch. Below it, the short axes expand fastest (≈ 3.1×10⁻⁴ K⁻¹ on one of them). Above it, the long stacking axis is essentially frozen (≈ 3×10⁻⁶ K⁻¹) and all the expansion concentrates on the short axes.

Caveats, since I'd want them stated if this were my data: the dry series mixes first and second thermal cycles, and with 5–45 K steps and two crystals per point, only the phase-contrast finding (the 5×) is robust; the sub-percent per-axis wiggles are within crystal-to-crystal noise.

This is a textbook member of the transition-ladder family we cataloged in the NTE phase-transition atlas

Files used: 1-Solvent-328 K, 1-Solvent-200 K