Recent results from sage

l~miliiBiimmilm~1

PROCEEDINGS SUPPLEMENTS

Nuclear Physics B (Proc. Suppl.) 35 (1994) 412-417 North-Holland

RECENT RESULTS FROM SAGE V.N. Gavrin, E.L. Faizov, A.V. Kalikhov, T.V. Knodel, I.I. Knyshenko, V.N. Kornoukhov, I.N. Mirmov, A.V. Ostrinsky, A.M. Pshukov, A.A. Shikhin, P.V. Timofeyev, E.P. Veretenkin, V.M. Vermul, G.T. Zatsepin Institute for Nuclear Research, Russian Academy of Sciences, Moscow 117312, RUSSIA

T.J. Bowles, S.R. EUiott*, J. S. Nico, W.A. Teasdale, D.L. Wark**, J.F. Wilkerson Los Alamos National Laboratory, Los Alamos, NM 87545 USA

B.T. Cleveland, T. Daily, R. Davis, Jr., K. Lande, C.K. Lee, P. Wildenhain University of Pennsylvania, Philadelphia, PA 19104 USA

M. L. Cherry Louisiana State University, Baton Rouge, LA 70803 USA R. T. Kouzes *** Princeton University, Princeton, NJ 08544 USA A radiochemical 71Ga-71Ge experiment to determine the primary flux of neutrinos from the Sun began measurements of the solar neutrino flux at the Baksan Neutrino Observatory in 1990. The number of 71Ge atoms extracted from initially 30 and later 57 tons of metallic gallium was measured in fifteen runs during the period of January 1990 to May 1992. The observed capture rate is 70 +19 (stat) + 10 (syst) SNU. This is to be compared with 132 SNU predicted by the Standard Solar Model.

1. INTRODUCTION A fundamental problem during the last two decades has been the large deficit of the solar n e u t r i n o flux o b s e r v e d in the r a d i o c h e m i c a l c h l o r i n e e x p e r i m e n t 1) compared with the Standard Solar Model (SSM) theoretical predictions2),3). Recent results of the Kamiokande II water Cherenkov e x p e r i m e n t 4) have confirmed this deficit. These results may be explained by deficiencies in the solar model in predicting the 8B neutrino flux or may indicate the possible existence of new properties of the neutrino.5)

The role new neutrino properties may play in the suppression of the solar neutrino flux can be determined by a radiochemical gallium experiment. An experiment using 71Ga provides the only feasible means at present to measure low energy solar neutrinos i?roduced in the proton-proton (p-p) reaction~°1. Exotic hypotheses aside, the rate of the p-p reaction is directly related to the solar luminosity and is insensitive to alterations in the solar models. An observation in a gallium experiment of a strong suppression of the tow energy solar neutrino flux requires the invocation of new neutrino properties.

0920-5632/94/$07.00 © 1994 - Elsevier Science B.V. All rights reserved. SSDI 0920-5632(94)00506-Q

EN. Gavrin et al./Recent results from SAGE

2. THE BAKSAN GALLIUM EXPERIMENT 2.1. Extraction Procedure The experimental layout as well as the chemical and counting procedures have been described previously and are only briefly outlined here. 7) Each measurement of the solar neutrino flux begins by adding approximately 700 ttg of natural Ge carrier equally divided among the reactors holding the Ga. After a typical 1 month exposure interval, the Ge carrier and any 71Ge atoms that have been produced by neutrino capture are chemically extracted from the Ga. The counting gas GeH 4 (germane) is then synthesized and purified by gas chromatography. The overall extraction efficiency is typically 80%. 2.2. Counting Procedure The GeH 4 is then mixed with a measured amount of Xe and inserted into a miniature low-background proportional counter. The counter is placed in the well of a NaI detector inside a l a r ~ passive shield and counted for 24 months. 1Ge decays by electron capture to 71Ga with an 11.4 day half life. The lowenergy K- and L-shell Auger electrons and X rays produced during electron shell relaxation in the 71Ga daughter atom are detected by the proportional counter. Pulse s h a p e d i s c r i m i n a t i o n b a s e d on rise t i m e measurements is used to separate the 71Ge decays from background. The counter is typically calibrated at one month intervals using an external 55Fe source. The K-peak acceptance w i n d o w is then determined by extrapolation from the 55Fe peak. The extrapolation procedure was verified by filling a counter with 71GeH4 together with the standard counter gas.

2.3. Extraction History The experiment began operation in May, 1988 w h e n purification of 30 tons of Ga commenced. Large quantities of long-lived 68Ge produced by cosmic rays while the Ga was on the surface were removed. New extraction procedures were implemented in

413

January 1990 which resulted in the elimination of radon contamination in the extractions. By January 1990, the backgrounds had been reduced to levels sufficiently low to begin measurements of the solar neutrino flux and monthly extractions were made. Useful solar neutrino data were not obtained after the July 1990 run due to a Cr engineering test run. Following this, 30 tons of n e w Ga were purified, the Ga tanks and extraction systems were extensively cleaned, and m o n t h l y extractions resumed. Data from all of the runs made through May 1992 are reported here, except for m o n t h s in which either the extraction sample was physically lost during the extraction process or runs in which no usable information was available d u e to counter or electronics failures. Additional runs made since May 1992 (most of which include 1 Ghz transient digitization of the pulse waveform) are still under analysis and a number of them are still counting. 2.4. Statistical Analysis The data analysis selects events that have no NaI activity in coincidence within the 71Ge K-peak acceptance window. The K-peak acceptance window in energy is a 2 FWHM wide energy cut centered on the K-peak and the inverse rise time cuts are 95% acceptance. Some runs in 1990 and 1991 showed gain shifts in the proportional counters of about 10%. As these shifts were monotonic, we accounted for them by linearly interpolating between sequential calibrations. The frequency of calibrating the counters was increased in order to m i n i m i z e the uncertainties in the corrections. A preliminary four hour cut has been made on the data after each shield o p e n i n g in order to eliminate possible contributions from short-lived radon d a u g h t e r s external to the counters. A maximum likelihood analysis 8) is then carried out on these events by fitting the time distribution to an 11.4-day half-life exponential decay plus a constant rate background. Table I shows the results of the maximum likelihood analysis. The probability that a measurement would statistically exceed the goodness-of-fit for each of the runs is also given in Table 1.

V.N. Gavrin et aL /Recent results from SAGE

414

Table 1. Statistical analysis of runs. Extraction Date Jan - 90 Feb - 90 Mar - 90 Apr - 90 Jul - 90 Jun - 91 Jul - 91 Aug - 91 Sep - 91 Nov - 91 Dec - 91 Feb - 92-a Feb - 92 -b Mar- 92 Apr - 92 May - 92

Ga Mass (Tons) 28.7 28.6 28.5 28.4 21.0 27.4 27.4 49.3 56.6 56.3 56.2 43.0 13.0 56.0 55.8 55.7

68% CL (SNU)

Best Fit (SNU)

0 94 109 0 0 13 50 401 68 43 144 0 91 183 51 0

500

I

I

Probability

0-66 19-158 0-227 0-103 0-210 0-109 0-105 237-561 19-118 1-89 73-176 0-39 0-175 110-264 12-101 0-67 I

4% 28% 68% 35% 32% 10% 26% 77% 72% 33% 70% 66% 94% 57% 17% 32%

I

I

SAGE I

400 • •

Individual Exa'actions Mean Value

300 Z

30 tons 200

100

I

I

10915.0 " 199~.5

I

1991.0

Itlt t! i

1991.5

1992.0"

D92.5

Extraction Date Figure 1. Best fit values and 1-o uncertainties for each of the runs, together with the best fit value and 1-o uncertainty for the combined 1990-92 data. With the increase in the mass of gallium and reduction of the backgrounds, the 1991 and 1992 data show clear evidence for the presence of 71Ge, as shown in figure 2, in which one can clearly observe the decay of the K-peak. The half-life for the decay is also consistent with the known 71Ge half-life.

2.5. S y s t e m a t i c E f f e c t s

A g o o d deal of w o r k was d o n e in d e t e r m i n i n g c o u n t e r efficiencies u s i n g counters filled with 37Ar and 71GeH4,Xe, A d d i t i o n a l m e a s u r e m e n t s of the Ge carrierconcentration were made using isotope dilution spectroscopy. This resulted in a

V.N.. Gavrin et al./Recent results from SAGE

I

0.08

0.08

].]

415

I

I

SAGE1992Data

SAGE1992Data

0.06

I

0.06

1st a~. 2rid

"~ 0.04

0.04

0.02

0.02

0.00

0.00

> 3 Meanlives

f] 0

4

8

12

16

20

, ,t~_.

0

4

8

12

16

20

Envy ~V) 0zV) Figure 2. The combined data for the 1992 runs as a function of time. The energy spectrum for a period equal to the first two mean lifetimes of 71Ge are shown on the left, while the (normalized) energy spectrum for the time greater than 3 mean lifetimes of 71Ge are shown on the right. The hatched areas shows the expected position of the K-peak. significant r e d u c t i o n in the systematic uncertainties in the chemical extraction and counting efficiencies, which now are typically 3.6% and 5.3%, respectively. The systematic uncertainty in extrapolating the inverse rise-time cuts is estimated using a cut that includes all events not in coincidence with the NaI counter which are within the energy cut of the K-peak acceptance window with no cut made on inverse rise time. This results in an uncertainties of 7 SNU (68% CL) for the combined 1990-92 data. The u n c e r t a i n t y in b a c k g r o u n d determination under the 71Ge decay curve due to possible time variations of the counter background was checked in a n u m b e r of w a y s 9 ) . All tests are consistent with the hypothesis that the apparent increase in background at late times is purely a statistical fluctuation. However, such a fluctuation could suppress the signal by causing an overestimation of the background at early times. In order to minimize any assumptions, we assigned an uncertainty to any possible time variation of the background for the 1990 data to be 30 SNU (68% CL)9). As there is no evidence for any time variation in the 1991, 1992, or the combined 1990-92 data, we assume the background is constant in time and do not assign any systematic uncertainty for a possible time variation in the background to the combined 1990-92 data sets.

Table 2. 1-a Systematic uncertainties for the combined data sets. Efficiencies Rise time Backgrounds Radon TOTAL

5 SNU 7 SNU 3 SNU 5 SNU 10 SNU

The final possible systematic effect is due to possible background reactions which could produce 71Ge and the possible presence of radon, which can mimic a 71Ge signal. The total background production rate in 30 tons of liquid gallium metal of all g e r m a n i u m activities from external neutrons, internal radioactivity, and cosmic ray muons has been calculated to be less than 2.5% of the SSM production rate7). This results in an uncertainty of 3 SNU (68% CL). Systematic effects from the possible presence immediately after shield openings of short-lived radon daughters external to the counters has been eliminated by making a four hour cut on the data after each shield opening. Further studies in this area are in progress. The data has also been examined to search for the possible presence of radon internal to the counters. Checks included looking at overflow events (due to alpha decays in the radon chain), looking outside of the K-peak acceptance

416

V.N. Gavrin et al./Recent results from SAGE

window, looking for delayed coincidences of events (due to subsequent decays in the radon chain), and fitting the data to allow for both 71Ge and radon. No evidence for internal radon was found and a systematic uncertainty of 5 SNU (68% CL) was assigned. Table 2 shows the systematic uncertainties assigned to the combined 1990- 92 data sets. 2.6. Results

The results 9) of the analysis of the five runs with rise time selection in the 1990 data indicated a flux of solar neutrinos of only 20 SNU and statistically one sigma above zero. However, the large systematic uncertainties in the 1990 data set led to an upper limit of the 71Ga capture rate of 79 SNU (90% CL). With the additional data from 1991-92, it appears that the increased count rates at late times observed in some of the runs were simply statistical fluctuations. Monte Carlo simulations of the data indicate that the 199092 data are statistically distributed as expected for a central value of 70 SNU. For the combined 1990-92 data, the capture rate was determined to be:

extraction efficiency of the natural Ge carrier and 71Ge track closely. However, we note that in inverse beta decay, the resultant 71Ge atom may be in an excited state or the 71Ge atom may be ionized. While very unlikely, these excitations may drive some chemical reaction which result in the 71Ge atom being tied up in a chemical form which we cannot efficiently extract. To test this, we have carried out a set of measurements in which we look at beta decay of Ga isotopes in the gallium. In the sudden impulse approximation, atomic excitations of an atom during beta decay should be the same as those in inverse beta decay. In this experiment, we have taken a few ,~rams of imllium and irradiated it to form ' UGa and 72Ga by (n,y) reactions. The 70Ga and 72Ga subsequently decay to stable 70Ge and 72Ge, which are then extracted from the irradiated gallium using the same procedure as in the full ~ale solar neutrino runs. The amounts of Ge and 72Ge are determined by mass spectroscopy. Table 3 shows the preliminary results from our first measurements, which indicates that 70Ge and 72Ge are observed in the amounts expected.

71Ga Rate = 70 + 19 (stat) + 10(sys0 SNU. Table 3. Results from Ga(n, ¥). This assumes that the extraction efficiency for 71Ge atoms produced by solar neutrinos is the same as that measured using natural Ge carrier. This corresponds to 36 counts assigned to 71Ge decay, compared to the SSM prediction of 68 counts. 2.7. Extraction Efficiencies While all available information leads one to expect that the extraction efficiency for 71Ge atoms produced by solar neutrinos should be the same as for the carrier, it is important to test this assumption. A test to search for possible losses in the extraction of 71Ge atoms was carried out by doping the Ge carrier with a known number (6555 + 359) of 71Ge atoms. The doped carrier was added to one of the reactors holding 7 tons of gallium and three successive extractions were carried out, The overall chemical extraction efficiency was determined to be 101 + 5%, while that for the 71Ge was 99 + 6/-8%, indicating that the

GeIsotope Expected Measured Efficiency 70Ge 9.3 ~tg 9.1 I~g 98 + 10 % 72Ge 17.8 ~g 16.4 ~tg 92 + 10 %

Finally, an experiment using a neutrino source is planned in order to test the overall extraction efficiency in situ. A suitable neutrino calibration source can be made using 51Cr, which decays with a 27.7 day half-life by electron capture, emitting monoenergetic neutrinos of 751 keV (90.2% BR) and 426 keV (9.8% BR). An engineering test run with a lower-intensity 5tCr source was carried out during the fall of 1990. A full-scale calibration run is scheduled for 1994 using a 0.5-MCi 51Cr source.

V.N. Gavrin et al./Recent results from SAGE

2.8. Current Status and Future Plans

With the combined 1990-92 data sets, SAGE is now observing a signal consistent with 71Ge produced by solar neutrinos. The first results from SAGE, and the present results appear consistent taking into account the systematic uncertainties. However, these results are still based on limited statistics. It is clearly necessary to accumulate more data with higher signal to noise and better efficiencies, as well as to test the extraction efficiency. Intensive work has been carried out to reduce noise pulsing and backgrounds in the L peak. Preliminary analysis indicates that beginning with the September 1992 run, we are able to count the L peak, almost doubling our counting efficiency. Preparations are also underway to fully calibrate the system using an intense artificial 51Cr source. We expect to be able to carry out this experiment in 1994. Finally, we are continuing to study possible systematic effects, including additional studies of possible background sources and Monte Carlo simulations. 2.9. Conclusions

Different SSMs predict that the total expected capture rate in 71Ga to be in the range2), 3)'of 125 to 132 SNU, with the dominant contribution (71 SNL0 coming from the p-p neutrinos. The minimum expected rate in a Ga experiment, assuming only that the Sun is presently generating nuclear energy at the rate at which it is radiating energy, is 79 SNU5). Observation of significantly less than 79 SNU in a gallium experiment is difficult to explain without invoking new neutrino properties. The first measurements from a gallium solar neutrino experiment have observed fewer 71Ge atoms than predicted by the SSM. From the 1990-92 data, we observe only 53% of the predicted flux. Assuming the extraction efficiency for 71Ge atoms produced by solar neutrinos is the same as for natural Ge carrier, the first measurements indicate that the flux may be less than that expected from p-p neutrinos alone.

417

2.10. Acknowledgments The SAGE collaboration wishes to thank A.E. Chudakov, G.T. Garvey, M.A. Markov, V.A. Matveev, J.M. Moss, J. Rhodes, S.P. Rosen, V.A. Rubakov, and A.N. Tavkhelidze. We are grateful to J.N. Bahcall, Yu. Smimov, and many members of GALLEX for useful discussions. We acknowledge the support of the Russian Academy of Sciences,, the Institute for Nuclear Research, the Russian Ministry of Sciences, the Division of Nuclear Physics of the US Dept. of Energy, the National Science Foundation, Los Alamos National Laboratory, and the University of Pennsylvania. *Present Address: L-296, Lawrence Livermore National Laboratory, Livermore, CA 94550 ** Present address: Dept. of Particle and Nuclear Physics, Oxford University, Keble Road, Oxford, OX1 3RH, England ***Present address: Batelle Pacific Northwest Laboratories, P.O. Box 999, Richland, WA 99352 REFERENCES

1) R. Davis et al., Proc. 25th Int. Conf. on High Energy Physics, edited by K.K. Phua and Y. Yamaguchi, (World Scientific, Singapore, 1991), p. 667 2) J.N. Bahcall and R. Ulrich, Rev. Mod. Phys. 60, 297 (1988). 3) S. Turck-Chieze, S. Cahen, M. Casse, and C. Doom, Ap. J. 335, 415 (1988). 4) K.S. Hirata et al., Phys. Rev. Lett. 65, 1297 (1990). 5) J.N. Bahcall, Neutrino Astrophysics, Cambridge University Press, 343 (1989). 6) V.A. Kuzmin, Soy. Phys. JETP 22, 1051 (1966). 7) V.N. Gavrin et al., Proc. of "Inside the Sun", Conference, Versailles, edited by G. Berthomieu and M. Cribier (Kluwer Academic, Dordrecht, 1989), p. 201. 8) B.T. Cleveland, Nucl. Instrum. Methods 214, 451 (1983). 9) A.I. Abazov et al., Phys. Rev. Lett. 67, 3332 (1991).