Nonclassical Rotational Inertia in Helium Crystals

It has been proposed that the observed nonclassical rotational inertia (NCRI) in solid helium results from the superflow of thin liquid films along interconnected grain boundaries within the sample. We have observed NCRI in large ${}^4\mathrm{He}$ crystals grown at constant temperature and pressure, demonstrating that the superfluid grain boundary model cannot explain the phenomenon.

Figure 1

(a) Schematic drawing of TOs. The epoxy coating is 0.025 cm thick on the bottom of each cell, and $0.05/0.04\mathrm{cm}$ thick on the $\mathrm{BeCu}/\mathrm{AgCu}$ walls. The 0.25 cm diameter BeCu cold spot is recessed (to the dotted line) from the epoxy layer. In the AgCu TO the cold surface is flush with the epoxy layer. Some relevant parameters for the $\mathrm{BeCu}/\mathrm{AgCu}$ TO are $h=0.483/0.597\mathrm{cm}$, $d=0.914/1.016\mathrm{cm}$, $\tau =0.933/1.27\mathrm{ms}$ (resonant period at 20 mK). (b) Mass loading of BeCu TO during growth of two samples. The minimum loading (solid at 25 bar) of the $\mathrm{BeCu}/\mathrm{AgCu}$ TO is $740/2000\mathrm{ns}$.

Figure 2

(a) Comparison of several BC, CP, and CT samples grown in the BeCu TO with two different impurity levels. $T_F$ is the temperature where solidification is complete. For eight 1 ppb crystals grown at CT/CP, $0.26\%\le \mathrm{NCRIF}\le 0.38\%$ ($\pm 0.01\%$) and $T_O=79\pm 5\mathrm{mK}$. Among 12 BC samples, $0.46\%\le \mathrm{NCRIF}\le 2.0\%$ and $80\mathrm{mK}\le T_O\le 275\mathrm{mK}$. The two 300 ppb samples grown in the BeCu TO are also shown. The maximum speed at the rim of the cell is $<5\mu \mathrm{m}{\mathrm{s}}^{-1}$ in all cases. (b) The four 300 ppb samples grown in the AgCu TO. Rim speeds are $<8\mu \mathrm{m}{\mathrm{s}}^{-1}$.

Figure 3

(a) NCRIF after each sequential anneal for a BC sample quenched from the bcc phase. A CT crystal grown from the superfluid is potted for comparison. (b) Asymmetric reduction in $Q^{-1}$ following each anneal. All data were obtained with rim speeds $<3\mu \mathrm{m}{\mathrm{s}}^{-1}$.

Figure 4

Normalized NCRIF in various samples. There is a wide spread in the data from the original KC experiment [1]. BC samples presented in Fig. 2 possess a high temperature tail of NCRI. CT/CP samples have a considerably sharper onset. A two-thirds power law is plotted for comparison.