Experiments for the measurement of the Townsend coefficients for ionization by collision were performed on pure ${\mathrm{N}}_2$ and ${\mathrm{N}}_2$ contaminated with Na vapor and Hg vapor. Cathodes of Pt, Na, and Hg were used. Values of $\frac{\alpha }{p}$ in ${\mathrm{N}}_2$ contaminated with Hg vapor agree with the values found by previous investigators using ${\mathrm{N}}_2$ of similar purity. With pure ${\mathrm{N}}_2$ at the lower pressures and higher $\frac{X}{p}$, the values were seventeen percent lower than the corresponding values in the Hg contaminated ${\mathrm{N}}_2$. At high pressure and low $\frac{X}{p}$, the curves differed by relatively little. With Na contamination which may have been accompanied by ${\mathrm{H}}_2$ contamination, a larger increase in $\frac{\alpha }{p}$ was observed, which amounted at low pressure and high $\frac{X}{p}$ to as much as twenty-four percent higher than the value for the pure ${\mathrm{N}}_2$. The high value of $\frac{\alpha }{p}$ for Hg contaminated ${\mathrm{N}}_2$ is explained as being partly due to the direct ionization of Hg vapor atoms by electrons and partly being due to the action of metastable ${\mathrm{N}}_2$ molecules on Hg. Imprisoned radiation is probably not an important factor. Because of the low vapor pressure of the Na, the direct electron ionization and the ionization by metastable molecules can hardly be invoked to explain the increase in $\frac{\alpha }{p}$. One must conclude either that the vapor pressure of the Na at room temperature is far above the extrapolated values taken from tables, which is unlikely, or that ${\mathrm{H}}_2$ introduced in the distillation of Na or ${\mathrm{N}}_2$ make a volatile compound with Na under excitation or action of ultraviolet light. The shapes of the $\frac{\beta }{p}$ curves plotted as $f\left(\frac{X}{p}\right)$ show abrupt rises at certain values of $\frac{X}{p}$, definite plateaus, and subsequent gradual increases. These effects have not been observed in previous investigations where Hg contamination predominates, those curves rising asymptotically. Greater significance attaches to the $\frac{\beta }{\alpha }=\gamma =\frac{\eta \theta g}{\alpha }$ curves, which in the case of Pt and Na rise abruptly from zero at a given $\frac{X}{p}$ to a high peak, then fall abruptly and rise again gradually at higher $\frac{X}{p}$. The rise for the Pt electrode at $\frac{X}{p}=60\mathrm{}$ is sharp and occurs at a much lower $\frac{X}{p}$ than any previously observed when low electron current densities are used. The curve with the Na cathode rises at a higher $\frac{X}{p}$ than that for Pt to a lower maximum and then, after a fall, rises again. This is surprising, in view of the fact that the Na surface has a far lower work function and far greater photoelectric efficiency than Pt. The sharp peak in pure ${\mathrm{N}}_2$ is ascribed to the action of radiation either via metastable molecules or by some other mechanism in liberating electrons from the cathode. In the presence of either Hg or Na vapor (possibly contaminated with ${\mathrm{H}}_2$), the metastable or radiative action is completely destroyed, the energy going to the ionization of the Hg or of the Na and ${\mathrm{H}}_2$ atoms, if present. In the case of Na, it is possible that only the lower energy states are destroyed so that some of the higher energy states are able to reach the Na cathode. Hence, the small peak at higher $\frac{X}{p}$ with Na. Deducting the peaks leads to curves for Na and Pt and to a lesser extent Hg which intercept the axis at an $\frac{X}{p}$ between 200 and 300. In this region, positive ions have an energy of the order of between two and five volts. Such ions can liberate secondary electrons from the cathode, and we have the effect of a true $\gamma$.

• Received 1 November 1937

DOI:https://doi.org/10.1103/PhysRev.53.293