Structural differences between unannealed and expanded high-density amorphous ice based on isotope substitution neutron diffraction

Abstract

We here report isotope substitution neutron diffraction experiments on two variants of high-density amorphous ice (HDA): its unannealed form prepared via pressure-induced amorphization of hexagonal ice at 77 K, and its expanded form prepared via decompression of very-high density amorphous ice at 140 K. The latter is about 17 K more stable thermally, so that it can be heated beyond its glass-to-liquid transition to the ultraviscous liquid form at ambient pressure. The structural origin for this large thermal difference and the possibility to reach the deeply supercooled liquid state has not yet been understood. Here we reveal that the origin for this difference is found in the intermediate range structure, beyond about 3.6 Å. The hydration shell markedly differs at about 6 Å. The local order, by contrast, including the first as well as the interstitial space between first and second shell is very similar for both. 'eHDA' that is decompressed to 0.20 GPa instead of 0.07 GPa is here revealed to be rather far away from well-relaxed eHDA. Instead it turns out to be roughly halfway between VHDA and eHDA - stressing the importance for decompressing VHDA to at least 0.10 GPa to make an eHDA sample of good quality.

Keywords: Amorphous ice; neutron diffraction; water.

Graphical abstract

Figure 1.

Piston displacement curves for three distinct samples of about 1500 mg, namely (1) eHDA(0.20 GPa), (2) eHDA(0.07 GPa) and (3) LDA. An offset of 1 mm is applied at 1.1 GPa for clarity. Different colours correspond to different isotopologues, as indicated.

Figure 2.

Powder X-ray diffractograms recorded ex situ at ∼80 K in vacuum using a Siemens D5000 instrument equipped with a Göbel-mirror for parallel optics and an Anton Paar TTK450 chamber for horizontal sample geometry. Cu-K_α was used for (1) eHDA(0.20 GPa), (2) eHDA(0.07 GPa) and (3) LDA. The location of the first halo peak is indicated by a dashed vertical line. Top, middle and bottom (green, blue and black) diffractograms are for HDO, D₂O and H₂O, respectively.

Figure 3.

Mid-IR spectra recorded from the liquid samples at room temperature, after measurement at SANDALS and melting sample for (a) HDO and (b) D₂O. (1) eHDA(0.20 GPa), (2) eHDA(0.07 GPa) and (3) LDA.

Figure 4.

Fully corrected interference differential scattering cross section data, F(Q), for the samples as indicated.

Figure 5.

OO partial radial distribution function for the samples as indicated. Please note that some of these curves were published in advance in a review article, to which we contributed Ref. [38].

Figure 6.

OH partial radial distribution function for the samples as indicated. Please note that some of these curves were published in advance in a review article, to which we contributed Ref. [38].

Figure 7.

HH partial radial distribution function for the samples as indicated.

Figure 8.

(top) OO partial radial distribution function for uHDA and eHDA(0.07 GPa) to 8 Å, the error bars included represent the ensemble of structural configurations generated in the structure refinement process that are consistent with the supplied diffraction data. (bottom) difference function (g_oo(r)_uHDA – g_oo(r)_eHDA) (blue line).

Figure 9.

(top) Coordination number as obtained from integration of the OO partial radial distribution function to the given radius r for uHDA (black), VHDA (blue) and eHDA(0.07 GPa) (red). (bottom) differences in coordination number as indicated and calculated from the top curves.