Experimental and theoretical evidence for an ionic crystal of ammonia at high pressure

We report experimental and theoretical evidence that solid molecular ammonia becomes unstable at room temperature and high pressures and transforms into an ionic crystalline form. This material has been characterized in both hydrogenated (NH${}_3$) and deuterated (ND${}_3$) ammonia samples up to about 180 and 200 GPa, respectively, by infrared absorption, Raman spectroscopy, and x-ray diffraction. The presence of a new strong infrared absorption band centered at 2500 cm${}^{-1}$ in NH${}_3$ (1900 cm${}^{-1}$ in ND${}_3$) is in line with previous theoretical predictions regarding the ionization of ammonia molecules into ${{\mathrm{NH}}_2}^{-}$ and ${{\mathrm{NH}}_4}^{+}$ ions. The experimental data suggest the coexistence of two crystalline ionic forms, which our ab initio structure searches predict to be the most stable at the relevant pressures. The ionic crystalline form of ammonia appears stable at low temperatures, which contrasts with the behavior of water in which no equivalent crystalline ionic phase has been found.

Figure 1

Phase diagram of ammonia. The blue-, blue-gray-, and cyan-colored regions correspond to different molecular phases composed of NH${}_3$ units. The yellow region represents the tentative stability domain of the $\beta$ phase reported in the present paper, most likely composed of ${{\mathrm{NH}}_4}^{+}$ and ${{\mathrm{NH}}_2}^{-}$ species. The solid and dashed lines indicate the upstroke and downstroke transition pressures. At high pressures and temperatures, a dynamic ionization is observed in the superionic $\alpha$ phase [2], which is mainly composed of NH${}_3$, ${{\mathrm{NH}}_4}^{+}$, and ${{\mathrm{NH}}_2}^{-}$ species.

Figure 2

Example of an IR spectrum collected from the molecular phase V ($P{2}_1{2}_1{2}_1$) of the NH${}_3$ sample at 40 GPa. The upper graph shows the transmission (single beam) spectrum collected from NH${}_3$ at 40 GPa (blue solid curve) and from a ${\mathrm{N}}_2$ sample at 43 GPa (black dotted curve). The latter is used as the reference spectrum to calculate the absorbance of the NH${}_3$ sample, which is displayed in the lower graph (red solid curve). The internal modes of phase V (${\nu }_1–{\nu }_4$) are indicated. The strong absorbance of ${\nu }_2$ and ${\nu }_3$ saturates the absorption. The modes below 1000 cm${}^{-1}$ are lattice modes.

Figure 3

Evolution of the infrared absorption spectra of solid NH${}_3$ with pressure. The spectra have been offset for easier visualization. The blue curves are for the molecular phase V, and the red one is for the $\beta$ phase. The frequency window from $\sim 2000$ to $\sim 2300{\mathrm{cm}}^{-1}$ is obscured by the strong absorption band of the diamond anvils. Pressures are indicated on the right.

Figure 4

Experimental [(a) and (c) 300 K] and theoretical [(b) and (d) 0 K] Raman spectra of NH${}_3$ are shown across the phase transition between the molecular phase V ($P{2}_1{2}_1{2}_1$) and the $\beta$ phase. In (a) and (c), the blue and red curves, respectively, show the experimental spectra below and above the transition at the indicated pressures. The gray curves are spectra recorded with the laser spot focused onto the Re gasket, showing the contribution of the second-order Raman band of the diamond anvils. In (b) and (d), the blue, pink, and green bars, respectively, show the density-functional-perturbation-theory- (DFPT-) predicted Raman modes of the $P{2}_1{2}_1{2}_1$ phase V at 150 GPa and ionic $Pca{2}_1$ and $Pma2$ structures at 200 GPa. Global offsets of $-100$ and $-60{\mathrm{cm}}^{-1}$ have been applied to the theoretical spectra in (b) and (d), respectively. The star in (a) indicates the two peaks which appeared after exposure to hard x rays (33 keV) and identified as arising from the presence of ${\mathrm{H}}_2$ (see the text).

Figure 5

The experimental mid-IR absorption spectra of ammonia are shown at (a) 144 GPa and (b) 158 GPa, respectively, below and above the phase transition at 300 K (black solid curves). The blue, green, and pink bars show the theoretically predicted IR modes at 150 GPa and 0 K of the molecular $P{2}_1{2}_1{2}_1$, ionic $Pma2$, and $Pca{2}_1$ structures, respectively. In order to make the lower-frequency bands more visible, the theoretical IR intensities have been multiplied by 6 for mode frequencies below $2000{\mathrm{cm}}^{-1}$.

Figure 6

X-ray pattern and structures of ionic ammonia. (a) The experimental (background-subtracted) x-ray pattern from the NH${}_3$ sample at 194 GPa and 300 K is shown as black circles in the lower panel. The blue solid line is a Le Bail refinement obtained by considering a mixture of three phases: $Pma2,Pca{2}_1$, and rhenium from the gasket (peaks indicated by Re). In the upper part, the simulated x-ray powder patterns of the $Pca{2}_1$ (pink) and $Pma2$ (green) structures are depicted using the cell parameters determined from the Le Bail refinement: $Pma2\text{:}a=7.011\AA ,b=2.408\AA$, and $c=2.404\AA ;Pca{2}_1\text{:}a=8.540\AA ,b=2.415\AA$, and $c=3.883\AA$. (b) and (c) Representations of the ionic $Pca{2}_1$ and $Pma2$ structures, respectively. Red, blue, and pink balls, respectively, are H atoms and the N atoms of the ${{\mathrm{NH}}_2}^{-}$ and ${{\mathrm{NH}}_4}^{+}$ species. The pseudo-$\mathrm{hcp}(ABAB\cdots )$ or $\mathrm{fcc}(ABCABC\cdots )$ stacking of the nitrogen lattices are emphasized. The predicted cell parameters at 200 GPa for $Pma2$ are as follows: $a=6.9756\AA ,b=2.371\AA$, and $c=2.370\AA$; and for $Pca{2}_1$, they are as follows: $a=8.471\AA ,b=2.380\AA$, and $c=3.863\AA$. The differences between observed and predicted cell parameters are below 1.5%.

Figure 7

Difference in enthalpy with respect to the ionic $Pma2(Z=4)$ structure as a function of pressure. The $P{2}_1{2}_1{2}_1$ structure is that of the molecular phase V. The $Pma2$ ionic structure is identical to that reported in Ref. [9]. The other structures were obtained from new structural searches with a larger number of formula units per cell.