Invited Paper VT-TuA7
Ultimate Limits in the Gas Composition Determination Within Small Sealed Volumes by Quadrupole Mass Spectrometry

Tuesday, November 11, 2014, 4:20 pm, Room 303

Miniaturization of modern sealed vacuum devices and higher demands for their stable operation on the long-term scale require accurate determination of the gas composition in the early stage of their operation, as well as after a long operating period. Since particular gases may have detrimental effect on the device performance even at low concentrations, accurate quantification of the gas mixture is an important as well as a challenging task. Among a few highly gas-sensitive methods capable to detect quantities below 10^-4 mbar L, the quadrupole mass spectrometry seems to be the most appropriate one for this task.

A two-step procedure, consisting of sample puncture inside an expanding chamber, followed by opening the leak valve to the quadrupole mass spectrometer, kept in the analytical chamber at ~3x10^-11 mbar, is proposed. A limited number of ion current readings are used for the reconstruction of the original total pressure and gas composition. Calibration of such instruments at particular partial pressure is regularly achieved at stable gas influx and constant pumping speed. Several discrete points have to be recorded to get the sensitivity of the instrument expressed in A/mbar.

In this presentation, a systematic approach for preparing the instrument for routine quantification of small gas amounts is described. In the first stage, the instrument was calibrated as the precise partial gas flow meter by an innovative in-situ calibration procedure by three different gases, hydrogen, argon and nitrogen. Each gas was admitted into the expanding chamber, having a precisely determined volume of 0.312 L and equipped by a capacitance manometer. By opening the leak valve, ion currents versus gas flux were recorded over three orders of magnitude, expressing the partial flux sensitivity in As/(mbar L). In the second stage, known gas quantities ~10^-4 mbar L of pure gas were admitted at different leak valve conductance to determine the instrument’s response. This data enabled minimizing the error by searching for a compromise between the number of the readings and the level of recorded ion currents. In the third stage, gas mixtures with various contents of three gases were prepared and analyzed. This evaluation enabled a much better prediction of the ultimate limits in reconstructing of the unknown gas mixture in a real device. Anyhow, uncertainty in evaluation increases by lowering the gas amounts as ion currents become indistinguishable from the background readings of the instrument.