As part of the NASA CRRES chemical release program in September 1990, two Ba and also one each Sr and Ca canisters of a boron‐titanium thermite mixture, which vaporizes the element on ignition, were released near perigee after dusk in the South Pacific to study the critical velocity effect proposed by Alfvén. The critical velocities of these three elements are 2.7, 3.5, and 5.4 km/s respectively, all well below the orbital velocity of 9.4 km/s. On September 10, 1990, a Sr and Ba pair (G‐13, or critical ionization velocity (CIV) I) was released near Rarotonga at ∼515 km altitude in a background electron density of 3.4 × 106 cm−3. On September 14, 1990, G‐14 or CIV II released a Ca and Ba pair west of New Caledonia near 595 km at an electron density of 1.5 × 106 cm−3. Ions of all three elements were observed with low‐light level imagers from two aircraft after they had transited up the magnetic field lines into the sunlight. Emissions from the spherically expanding neutral gas shells below the solar terminator, observed with cameras filtered for the Ba+ ion line at 4554 Å and also in unfiltered imagers for ∼15 s after release, are probably due to excitation by hot electrons created in the CIV process. The ions created clearly lost much of their energy, which we now show can be explained by elastic collisions: Ba+ + O. Inventories of the observed ions indicate yields of 0.15% and 1.84% for Ba in the first and second experiments, 0.02% for Sr and 0.27% for Ca. Ionization from all the releases continued along the satellite trajectory much longer (>45 s) than expected for a CIV process. The ion production along the satellite track versus time typically shows a rapid rise to a peak in a few seconds followed by an exponential decrease to a level essentially constant rate. The characteristic distances for CIV I and II are 47 and 62 km, respectively. We interpret the early time rise and exponential fall to be due to CIV ionization, of 0.014% (CIV I) and 0.40% (CIV II) for the Ba releases. The later ions produced at a constant rate probably have origins from other such processes as stripping and associative ionization collisions with atmospheric constituents primarily O, and charge exchange with O+, He+, and H+. We suggest that the much larger Ba ionization rate in CIV II than CIV I is due to the fact that the release occurred in the peak Ca density where hot electrons were already present.