Transcript Document 7796324

TAUP 2001 - LNGS 8-12 September 2001
Updated results from GNO
at LNGS
C.Cattadori – INFN Milano
on behalf of the GNO collaboration
Summary:
• Introduction
• Major changes GNO vs GALLEX
• Update of GNO meas. of solar ne interaction rate
16 new solar runs + 6 blanks
and combination with GALLEX
• Neural Network analyis
• Future plans
• Conclusions
Motivations: Measure the low energy solar neutrinos interaction rate,
whose flux is strictly related to solar luminosity (i.e. model independent),
with an accuracy of 5 SNU, examine its constancy over 1 solar cycle with
a sensitivity of ~ 15 %.
71Ga(n ,e)71Ge
e
Basic interaction
EC
n Signal Composition:
(BP00 SSM)
Expected Signal
(SSM)
(Ethr = 233 keV)
t = 16.5 d
71Ga
PP
73 SNU (56%)
7Be
35 SNU (27%)
CNO 9 SNU ( 7%)
8B
12 SNU (10%)
Tot 129 SNU +8–6 1s
1.2 n int. per day, but due to decay during
exposure + ineff., 9 71Ge decay detected per
extraction (28 days exposure)
Technique
Radiochemical - Target: 30 t of natGa
(12 t of 71Ga) in 102 t of Ga3Cl acid sol.
Detector description
and operation
See f.i. PL B490(2000)16
PL B314(1993)445
What is a run?
Extract
GeCl4
12 h
Add 1 mg
of carrier
t0
In tank
Wait
21-28 d for SR
1 d for blanks
Counter in
shielding
6 months
In synthesis lab
GeCl4
10 h
GeH4 + Xe
in counter V =1cc
Stop counting
Remove counter
What did change between GALLEX and GNO ?
• Collaboration (a restricted part of the GALLEX coll. + l’Aquila Univ.)
•Analog electronics (LV,HV, preamp and amp,analog BW from 100 MHz  300 MHz)
•Digital electronics (no multiplexing, 1 digitizer per line @ 5Gs/s and DAQ
The resulting noise figure is < 2 mV r.m.s. @ 300 MHz on digitizer
•Counter calibration with an X ray tube
•More severe criteria to select counter to be used in solar runs
• Revaluation of residual bckgr due to internal Rn
• Counter type and passive shielding
What did not change
• Extraction and sinthesys plants
• Target and tank.
This presentation
GNO II results relative to
16 exposures 27-28 days long called Solar Runs (SR)
6 exposures
1 day long called Blank Runs (BL)
From 13-jan-2000 untill 3-may-2001
• 17 extraction performed, only 1 run lost due to HV instability
• counting live time 97% ; DAQ stops only for calibrations, monthly counter
connection and improvements and or maintenance .
Analysis
The results presented here are obtained from our Standard Analysis based on
two parameter (E-RT) Selection criteria
If the neural network analysis will pass all the validation steps, all the GNO
runs will be reanalized and both the single run and the global run values could
eventually change.
GNO results updated at 3-may-01
Selection
SR
SNU
N
71Ge
N.
Bckgr
19
71.1 +16.4 –15.1
45
128
GNO-I K
59.5 +12.9 –11.9
43
76
GNO-I L+K
64.2 +10.1 –9.5
88
204
81.5 +18.2 –17.0
43
82
GNO-II K
68.9 +14.3 –13.0
41
59
GNO-II L+K
74.1 +11.2 –10.7
84
141
GNO-I L
GNO-II L
GNO L
16
35
64.0 +
GNO K
GALLEX K
GALLEX L+K
9.4
– 8.9
68.9 ± 7.3 (stat) ± 3.2 (sys)
GNO L+K
GALLEX L
75.9 +12.1 –11.5
65
74.4 ± 10
79.5 ± 8.2
77.5 ± 6.2 (stat) ± 4.5 (sys)
Refinement of the GNO I results
•Algoritm for determination of E window (starting from Ce calibration)
•Re-evaluation of Rn inefficiencies in the removal of Rn bckgr events
eineff = 7% ± 5% (eineff = 9% ± 5% until PLB490 (2000)16)
•Rn cut (3h dead time after ovfl only in first 30 days)  increase live time of meas of
1%
•Introduced in side reaction subtraction the atm neutrino contribution (0.3 ± 0.3)
SNU. Side reaction subtraction new value 4.55 SNU (old value 4.25 SNU)
•2 counter filled with 69Ge for absolute determination of volume efficiency. The
newly determined values are used for the 6 runs (2 in GNO I + 4 in GNO II).
PLB490 (2000)16
This update
GNO I L ev.
80 +17.5 – 16.2
71.1 +16.4 –15.1
GNO I K ev.
57.2 +12.4 – 11.4
59.5 +12.9 –11.9
GNO I L+K
65.8 +10.2 – 9.6
64.2 +10.1 –9.5
GNO II single Solar Runs
Run
lable
Start exp.
End exp.
Exp
time
Chem
Yield
Countin
g
A024
13-01 / 9-02-00
27
91.2
165
77
+51
A025
9-02 / 8-03-00
28
94.8
168
29
+51
A026
8-03 / 5-04-00
28
97.7
167
3
A028
6-04 / 3-05-00
27
94.7
168
143 +57 –47
A029
3-05/31-05/00
28
93.0
166
81
+51
A030
31-05/28-06-00
28
98.1
Lost
A032
29-06/26-07-00
27
93.9
168
69
+58
A033
26-07/23-08-00
28
94.1
168
89
+48
For GNO I single runs see PLB 490(2000)16
SNU
L+K
+36
–39
–33
–24
–39
–46
–38
GNO II single Solar Runs (follows)
Run
lable
Start exp.
End exp.
Exp
time
Chem
Yield
Counting
A034
23-08/20-09-00
28
97.2
167
47
+52
A036
21-09/18-10-00
27
95.4
167
60
+41
A037
18-10-/15-11-00
28
93.9
168
-12 +36 –24
A038
15-11/12-12-00
27
98.1
167
119 +54 –44
A040
13-12/10-01-01
28
96.2
164
103 +60 –49
A041
10-01/07-02-01
28
95
167
88 +44 –34
A042
7-02-01/7-03-01
28
97.8
140+
51 +47 –36
A044
8-03/4-04-01
27
92.3
115 +
118 +63 –53
A045
4-04-01/3-05-01
29
95.1
100 +
62 +40 –32
For GNO I single runs see PLB 490(2000)16
SNU
L+K
–40
–29
Starting point: pulses from counters
800 mV
1100 mV
A typical 0.5 keV event
TDF 2
TDF 1
400 ns
400 ns
Bi Po TDS
800 ms
8 ms
counts/day/run
Time dist. of selected events
GNO
Superimposed the N(t)=a + b exp(-t/t)
which gives the best Log(L)
35 runs
counts/day/run
Time [d]
Energy dist. of fast ev.
Red – fast ev. t < 50 days
Black – fast ev. t > 50 days
Energy [keV]
Time dist. of selected events.
GALLEX
65 runs
from 0.1 c/d/run to 0.07 c/day/run
Time [d]
GNO II Blank Runs
Why Blank Runs?
To check each 3 month the proper functionality of the whole setup
(no tailig effect,background from isotopes other than 71Ge, ecc)
Start exp.
End exp.
Exp
Time
[d]
Chem
Yield
[%]
Counting
[d]
Excs
Cnts
Bckg
Cnts
160 d
A023
12-01/13-01-00
27
91.2
165
1.4
8.6
A027
5-04 / 6-04-00
28
94.8
166
0.0
15.0
A031
28-06 / 29-06-00
28
97.7
164
0.0
11.0
A035
20-09 / 21-09-00
27
94.7
166
1.7
5.3
A039
12-12/13-12-00
28
93.0
165
1.3
8.7
A043
7-03/8-03-01
28
98.1
166
0.7
9.3
3.7
59.3
0.85 ±
0.73
9.65 ±
3.2
Run
lable
Average
per run
counts/day/run
Time dist. of selected ev.
Superimposed the N(t)=a + b exp(-t/t)
which gives the best Log(L)
t=16.48 days
35 GNO solar runs
Time [d]
6 GNO blanks
Time [d]
GALLEX + GNO 73.9  4.7 (stat) 4.0 (sys)
Single runs SNU values distributions
GALLEX
Test of the hypothesis
of constant signal over the entire
period of GNO
Ratio of L test
2(L/L) is distributed as a c2
2[L(single runs)/L(global dist)] = 31.2
D.O.F. = 34
C.L. = 60%
GNO
2L[(GNO1 + GNO2)/L(global dist)]
= 0.4816
D.O.F. = 1
C.L. = 50%
Single runs SNU values distributions
GALLEX
+
GNO
2L[(GNO1 + GALLEX single runs)/L(global dist)]
=93.677
D.O.F. = 83
C.L. = 19%
Single run excess counts distribution
Red histogram of single run GALLEX + GNO
Blue simulation of 1000 runs where N71 and Bckgr
fluctuate following a Poisson distribution
around actual values
N71 found in run
Seasonal variations expected in GNO
in the 2n MSW scenario
Fogli et al. hep/ph/9910387
A powerful alternative analysis:
Neural Network
71Ge
b/g
The selection of events from solar runs can
be carried out using a 3-levels neural
network instead of Rise Time cuts (standard
analysis)
K main flag
In the input neurons of the neural
network there are five parameters coming
from the fit of the pulse and
characterizing its shape.
(L ev: RT, c2, spread of the charge cloud)
(K ev : the same as for L+ A1/A2 and dt
for double peaks)
Distribution functions of the output flag for a sample
of data containing both 71Ge and b/g events
71Ge
b/g
The output is a flag value in 0÷1; if the
flag is greater than 0.5 the event is
accepted as 71 Ge, else it is rejected as
background
L main flag
Fit of the pulses
(Fit-data) amplified
Fit function
Charge dist. function
Training of the Neural Network
The NN is trained with genuine 71Ge pulses (from calibrations) as true
examples and b/g from 137Cs or 222Rn as false examples.
The number and the type of the training examples must be carefully
chosen, as the overall efficiency of the network is very sensitive to these
parameters
Event selection efficiencies
Noise rejection efficiencies
NN
RT
NN
RT
L evnts
95.3%
96.6%
L evnts
87.8%
65.7%
K evnts
94.0%
96.3%
K evnts
77.1%
73.5%
Conclusions
Although the Ge-efficincies are slighly smaller than the corrisponding values for the
rise-time analysis, the NN is able to reject backgrond from b/g more efficiently. When the
Rn meas. is analyzed with the fit+NN and after the Rn cuts (3h after alpha ev, 15 min
before BiPO ev.) the inefficiency is 0. (inefficiency is 7% ± 5% for RT selection)
Time distribution of selected events
35 GNO runs
Single runs SNU values distributions
GNO
NN selection
Future plans
•Last validation steps of the new NN analysis  increase S/N.
• Next publication probably before end 2001
•Direct and absolute determination of volume efficiencies of all counters
by 69Ge measurements  reduce systematic < 3% (spring 2002 ?)
•Production of a new 2.5 MCi
51Cr
source to perform a 3rd irradiation
Intensity
exp/theo
I (may 1995)
63.4 +1.1 -1.6 1.01 ± 0.11
II (sept 1996)
69.1+3.3 -2.1
Average
Later corrections
0.85 ± 0.11
0.93 ± 0.08 (PLB420(1998)114)
0.89 ± 0.07
Now take the echem from 71As exp. (1.00 ±0.01)%  the source exp. gives
results on cross section 71Ga(ne,e)71Ge
Proposal for a new source experiment
Possibility to irradiate 11.5 Kg of Cr at SM3 reactor of RIAR
(Dimitrovgrad) for 50 days to obtain a source 2.6 Mci on site (LNGS)
The expected accuracy will be better than 8%
When taken togheter with others exposure results this will lead to an
Accuracy on s(750 keV) 5%
New detector for 71Ge counting
R&D work in an advanced phase but the technique is very different
And an eventaual implementation on site not so near.
Conclusions
•GNO experiment is running smoothly and with a very high duty cycle since
may 1998 and GALLEX + GNO since 1991
•The updated GNO results (35 solar runs) is 68.9 ± 7.3 (stat) ± 3.2 (sys)
and when combined with GALLEX 73.9  4.7 (stat) 4.0 (sys)
Exp/SSM = 0.57 + 0.047 %
• no seasonal variation is observed in the GALLEX+ GNO data
Winter-Summer = -9  10 SNU
• the systematic error will be reduced before middle 2002 at 3% level
• a new pulse analysis has been developped and if definetely validated will
be applied at the whole GNO data set.
•if refined at a 5% level, Gallium measurements give important constraints
in the oscillation scenario f.i. (non)observation of seasonal effects, and when
the 7Be meas will be available the very important PP flux will be derived.
• high values of the GALLIUM exp. favour SMA,
• Low values of
“
“
“LOW and LMA
To reach the 5% accuracy level we have to improve
•Statistic ( increase the target mass and continuously measure)
Join GNO and SAGE targets
Why not?
•Systematics (work is going on)
•Knowledge of the cross section better then 5% (new source exp)