Several pertinent characteristics of the tubes are summarized in table 4. The large fluctuations on each near UV tube imply that several photoelectrons are produced for each main event, and the ratio of photoelectrons (pe) per event is a number characterizing the response of the tube to the cause of the dark counts. While a detailed statistical study is underway, we present in table 5 indicative values for the quantity pe/event. The count rates have been reduced by a correction factor to give the count rate due to cosmic rays; this factor was determined from the relative change with geomagnetic latitude, shown in table 4 (R), with the assumption that only cosmic rays contribute to the dark count in U1 and U2, which have no windows. The correction factor is Ci = (Ri-1)/(Ru-1) where Ri is the ratio Nmax/Nmin tabulated in table 4, i is the tube index, and Ru = 5, which is chosen as the comparison value for pure cosmic ray background. Evidently the corrections are only approximate and the values in table 5 are tentative.
Data for true orbit 4306, which corresponds to standard orbit 47 were reduced point by point and values of Sigma and Navg were computed for two separate 10 minute intervals, standard minutes 41-50 and 91-100. According to figure 4, the regions correspond with the minimum count rates observed on all six tubes, so the correction factors Ci tabulated in table 4 are valid. The footnotes to table 5 give the formalisms used to derive the number of primary particles/14 sec, I, and the yield in photoelectrons/primary event, Y, resulting from interaction of the primary particle with the window of tubes V1, V2, and V3. Y is apparently equal to unity for the windowless tubes, U1 and U2, and for the solar blind tube, U3. No recent data is available with V1 turned on for times of background minimum, so the relevant values are inferred from the text described in the footnote. The data all indicate that many photoelectrons per event are produced in the near UV tubes V1, V2, and V3, although there is a factor of 5-10 higher yield for the tubes V1 and V2, which have MgF2 windows, and the V3 tube, with a window of 7056 glass.
Based on the apparent single particle effects on U1 and U2 and the increase in pe/event for the three near UV tubes, we may assume as a working hypothesis that the main cause of background counts is associated with a flux of primary particles interacting with the windows. The count rate for U1 and U2 agrees qualitatively with the flux of primary cosmic rays striking these tubes; in particular, the variation with magnetic latitude is virtually identical (see Rogerson, J. B., Spitzer, L., Drake, J. F., Dressler, K., Jenkins, E. G., Morton, D. C., and York, D. G. 1973, Ap.J. Letters, 181, L97.) The detailed qualitative agreement of U3, V1, and V2 backgrounds with the background on U1 and U2 noted above, then implies that these tubes are also affected strongly by the cosmic-ray background. The observed gradual increase in the count rates for V1, V2, and V3 over the first 4000 orbits (table 2) can be interpreted as a gradual change in the effectiveness of the interaction mechanism of the cosmic rays and the window material. The alternative explanation that this initial increase is due to the build-up of radioactivity (and the subsequent Gamma ray production) in the tube environs is excluded by the variation with geomagnetic latitude shown after the build-up. The range in inferred primary flux for each tube is a factor of about 9, which can be accounted for by invoking differences in tube characteristics.
The main mechanism producing the observed photoelectrons in V1 and V2 is presumably fluorescence in the windows caused by the primary cosmic rays. Various experiments have shown that the secondary photoelectrons are produced very rapidly (within microseconds). (See Dressler, K. and Spitzer, L., Jr. 1967, Rev. Sci. Inst., 38, 436-438.) Hence, the count rates on V1 and V2 may be higher than is necessary due to the megahertz counting circuits (see table 4), whereas in slower circuits, the pulse separation would not be so fine, leading to a correspondingly lower number of detected photoelectrons per primary event. The net count rate on U3 remains low in spite of its LiF window, because the photocathode is insensitive to longer wavelengths where most of the fluorescence appears.
The qualitatively different behavior of V3, particularly following passage through the SAA, suggests that a second mechanism (denoted M2) is operative, which apparently makes a contribution to V1 and V2 as well, since the ratio of maximum to minimum counts for U1 and U2 (table 4) is higher than the same ratio for V1 and V2; this change of ratio has been used in computing the count rate due to cosmic rays, listed in table 5. The mechanism M2 apparently "fills in" what would otherwise be the minimum levels if only cosmic rays affected V1 and V2. The fact that M2 is secondary on V1 and V2, and that where M2 is most effective (V3), the initial time dependent build-up between orbits 1 and 4000 was relatively lowest, strengthens our working hypothesis that the initial build-up was due to changes in the fluorescence process as opposed to the phosphorescence attributable to passage through the SAA.
Following Lowrance ("Design Study of the Television System for the Large Space Telescope", Final report, Grant NGR 31-001-276), the different photoelectron production rates for the windowed tubes V1, V2, and V3 may be partly attributable to the different photocathode materials (since the fluorescence is apparently mainly evidenced by photons with wavelengths between 4000 Å and 6000 Å and even the similar bialkali cathodes of V1, V2, and V3 may differ significantly from one to another). The fact that V3 is most affected by the process M2 may be related to its markedly higher dark count on the ground, an effect apparently noticed in the response of the tubes on the Wisconsin OAO-2 ("Radiation-induced Dark Current in the OAO-2 WEP Photomultiplier Tubes", Charles F. Lillie, undated).
The phosphorescence of various window materials following irradiation by high energy electrons has been studied by Viehmann, Eubanks, and Bredekamp. (W. Viehmann, A. G. Eubanks, and J. H. Bredekamp, "Fluorescence and Phosphorescence of Photomultiplier Window Materials under Electron Irradiation", Goddard Space Flight Center preprint X755-74210). These results indicate that phosphorescence at levels of about 10-4 to 10-6 of the exposure level, with subsequent decay times of 10-100 minutes are typical for MgF2 windows and for UV glass windows, similar to the 7056 Glass window on V3. While the effect for MgF2 depends on the purity of the crystal, the phosphorescence described above is generally more marked for MgF2 than for UV glasses, according to the results of Vichmann et al., contrary to what one would infer from the Copernicus results. However, there may be other factors that affect the phosphorescence strength and decay; for example, the type of bialkali photocathode used in V3 is different from that in V1 and V2.
Evidently the Copernicus data indicates that delayed fluorescence produced by cosmic rays, with some 140 pulses following the passage of a primary proton within a few seconds (and probably within less than a millisecond) is responsible for the large background count observed in the near UV tubes. Phosphorescence following passage through the South Atlantic Anomaly is apparent in the V3 count, but must be a minor contribution to the V1 and V2 noise. There are still a number of unsolved problems in a detailed theory of the phenomena, and further research would be of interest.