Caracterización de la isoforma II de DGKz

one strip, so the gains can be compared directly.

5.3.5 Comparison of Light Injection Gain to single-photoelectrons Gain

For the far detector three sets of data, on three different dates, were taken for the single-photoelectron fit (22/06/07, 21/02/08, 28/04/09), which cover 670 days.

Figure 5.8: An example of a single-photoelectron charge distribution with the fit given by function 5.11. The pedestal is fitted separately. The x axis is in units of ADC counts. The fit consists of Gaussian distributions weighted by Poisson statistics. Also shown are the first and second p.e. peaks and a term to describe when a photon passes through the photocathode and produces an electron.

5.3. GAIN CALIBRATION WITH SINGLE-PHOTOELECTRONS 87 Date taken DB DB error Fit Fit error Ratio Ratio error

25 Jun 07 79.81 0.89 79.09 1.06 1.009 0.030 21 Feb 08 81.05 0.94 80.17 0.40 1.011 0.020 24 Apr 09 82.41 0.97 81.40 0.65 1.012 0.023

Table 5.3: The gains for the FD, in the database from the LI system and the gains found by the fit to the s.p.e. spectrum in the far detector. It can be seen that they agree to within errors.

Date taken DB DB error Fit Fit error Ratio Ratio error 6 Jun 08 121.94 1.63 103.80 4.30 1.175 0.087 11 Sep 09 126.55 1.03 112.15 1.72 1.128 0.029

Table 5.4: The gains for the ND, in the database from the LI system and the gains found by the fit to the s.p.e. in the near detector. It can be seen that they agree to within errors.

For the near detector only two such datasets were taken and fitted (06/06/08, 10/09/09) covering 461 days. Figure 5.9 shows the gains from the s.p.e. fit method vs. the gains in the database, for each of the two detectors. This shows that the two methods agree well in the FD: there is only a narrow spread of values and the projection of the means in the 2D histogram lies along the best fit line.

However, in the ND there is a large spread between the fit gain value and the LI gain value. A projection of the mean value of all the fit values corresponding to a given database gain, shows some structure as these values do not follow that best fit line (figure 5.9(b)). The low gain discrepancy could be coursed by trig-ger or sparsification efficiencies from crudeness in the electronics. The structure above 110 ADC is likely to be caused from nonlinearity effects that are handled differently between the two methods.

In order investigate whether the gain changes over time by the same amount between the datasets, the mean of the gain values for all channels that were included, in both the database and s.p.e. fit method was calculated for each dataset. Table 5.3 shows that in the FD the two method are in agreement with

each other, while table5.4shows that in the ND the database gains found by the LI are consistently higher. This is not unexpected as in the FD there is dead time after every time a PMT records an event, in the ND there is continuous readout so a trigger is set on the number of ADCs. Due to time restraints in taking the data this was not optimised for this investigation. Figure 5.10 gives a graphical representation of these tables.

5.4 Summary

To find ∆m²₃₂and sin²2θ₂₃the energy of the interacting particles must be known to high precision. The MINOS detectors have been calibrated with cosmic muons, and the built-in light injection system. These systems ensure that the detectors respond in the same way no matter where the interaction happens within them, and also that the response is the same between detectors throughout the running period. The relationship between the response of the detectors and different par-ticles was investigated with the calibration detector. When the first MINOS beam ν_µ-CC result was released [82] the total response remained the same through time. In order to check that the formula that uses high-intensity light from the LI system to find the gains of the PMTs was correct, the single-photoelectron peak was determined from using data from natural s.p.e. scintillation from the detector.

This s.p.e. method agreed with the LI method for the far detector. Due to inef-ficiencies at low light levels for the M64 the methods do not agree for the ND. It needs further investigation to determine whether the change by gain is the same in the two methods. This would be achieved by taking a further high statistics data run in the future.

5.4. SUMMARY 89

differnce between spe and db gains

-2 -1 0 1 2

differnce between spe and db gains

-2 -1 0 1 2

Figure 5.9: A comparison of the gains found by the s.p.e. and the LI methods. a) The FD entries have a narrow spread around a straight line. The profile points of the 2D histogram are aligned on the best fit line. b) The ND has a larger spread between the gains found by the two methods. There is also some structure to the distribution of the majority of the points which depends on the gain. c) Shows the distribution of difference in gains between the two methods for the FD. The histogram is centred around zero and has a narrow width, which shows good agreement between the two methods. The red curve is a Gaussian fit made to the data. d) The histogram difference in gains for the ND is offset from zero which shows that the two methods are not in agreement. The broadness of the histogram shows that there is more than just an offset between the two methods.

This broadness maybe caused by the inefficiencies in the electronics at low light level.

Number of days

Figure 5.10: Gains from the database and the s.p.e. fit as a function of time.

a) It can be seen that the far detector values agree to within errors. b) The near detectors points are quite different for each method. c) It can be seen in the ratio of gains that all points for a given detector are consistent with each other. The ND point on 362 days has large error bars due to lack of statistics in the fit data set.

Chapter 6

MINOS Analyses

“All in all, a 100 % successful trip!”

“But, sir, we lost Mr. Rimmer, sir.”

“All in all, a 100 % successful trip!” (Cat, Kryten - series 6 Rimmer-world)

The MINOS detectors were designed to make the most precise measurement of

∆m²₃₂ and the experiment now has the world-leading result along with the best measurement of sin²2θ₂₃ of a man-made neutrino beam (∆m²₃₂ = 2.43^+0.13_−0.13 × 10⁻³eV at 68 % C.L., sin²2θ₂₃ > 0.90 at 90 % C.L.; see section 6.1). MINOS has also made world-leading, or very competitive, measurements on a range of other topics. It has been able to set a limit on the fraction of νµ changing to a sterile neutrino, and it has found no evidence for more than three active neutrino flavours (section 6.2). It has also released results on the measurement of the as yet unmeasured third mixing angle (section 6.3) which suggest a non-zero value for θ13. MINOS has also been able to check CP T invariance by checking that ∆m²₃₂, sin²2θ₂₃ are the same as ∆m²₃₂, sin²2θ₂₃ (section 6.4). This chap-ter gives an overview of beam oscillation results; however, the MINOS detectors have also been used to make a range of other measurements: MINOS has mea-sured the velocity of the neutrinos and thus put limits on its mass [108]. It has also taken data with the far detector that have shed new light on the charge ra-tio of K and π producra-tion at TeV energies by cosmic rays [109]. Furthermore, it has also been used to measure atmospheric temperature change over northern

Minnesota [110]. In addition, MINOS has been able to make a measurement of

∆m²₃₂ from atmospheric neutrinos in the far detector and to measure the differ-ence between νµ and νµ disappearance and thus able to place a limit on CP T violation in the lepton sector. Data from the near detector has been used to set a limit on Lorentz violation [111] and to make various neutrino cross-section mea-surements.

6.1 The MINOS Charged Current ν

Analysis

The νµ-CC analysis is that for which the MINOS detectors were built, namely the measurements of ∆m²₃₂and sin²2θ₂₃. In order to measure these parameters, the energy spectrum of data events in the far detector is compared to a predicted spectrum. This is obtained by extrapolating to the far detector the spectrum mea-sured in the near detector. This relative measurement utilises the two detectors, thus reducing the error from unknown ν cross-sections. A dip in the ratio between the data spectrum and the predicted spectrum provide the values for the oscilla-tion parameters: the depth of the dip gives sin²2θ23; the energy of the dip gives

∆m²₃₂by

E[ GeV]

1.27 × 735[km] ' ∆m²₃₂. (6.1) The latest MINOS result [1] constrains the oscillation parameters to

∆m²₃₂= 2.43^+0.13_−0.13 × 10⁻³eV²at 68 % C.L. and sin²2θ₂₃>0.90 at 90 % C.L.. It also disfavoured decay and decoherence to 3.7 and 5.7 σ respectively to oscillation.