1. Systematic Errors Studies in the RHIC/AGS Proton-Carbon CNI Polarimeters Andrei Poblaguev Brookhaven National Laboratory The RHIC/AGS Polarimetry Group: I. Alekseev, E. Aschenauer, G. Atoian, A. Bazilevsky, A. Dion, H. Huang, Y. Makdisi, A.Poblaguev, W. Schmidke, D. Smirnov, D. Svirida, K. Yip, A. Zelenski 1PSTP 2011, St. Petersburg6/5/2016

2. Layout of the RHIC facility H jet (pp) polarimeter provides absolute polarization measurements at RHIC RHIC pC polarimeters provide polarization monitoring including polarization profile measurements AGS pC polarimeter provides polarization monitoring (mainly used for technical control and special beam studies) 2PSTP 2011, St. Petersburg6/5/2016

3. Proton-Carbon Polarimeter kinematics Plan view Event selection in RHIC/BNL pC polarimeters: 3PSTP 2011, St. Petersburg6/5/2016

4. Polarization Measurement Spin dependent amplitude: Rate in the detector: 1. Spin Flip (one detector): 2. Left-right asymmetry (two detectors) Square-root formula: Combining “spin flip” and “left/right asymmetry” methods allows us to strongly suppress systematic errors A theoretical model for A N (t) (a fit to the BNL E950 data) 4PSTP 2011, St. Petersburg6/5/2016

5. AGS CNI Polarimeter 2011 PSTP 2011, St. Petersburg5 1,8 - Hamamatsu, slow preamplifiers 2,3,6,7 - BNL, fast preamplifiers 4,5 - Hamamatsu, fast preamplifiers 3 different detector types: Larger length (50 cm ) Regular length (30 cm ) Silicon Strip Detectors: 90 degree detectors (2,3,6,7 ) 45 degree detectors (1,4,5,8) Strip orientation Dead Layer 6/5/2016

6. Schema of Mesurements PSTP 2011, St. Petersburg6 WFD α-source measurements ( 241 Am, 5.486 MeV) “Banana fit” t-t 0 = t A (x DL,αA) 6/5/2016

7. An example of data selection PSTP 2011, St. Petersburg7 Wrong determination of mean time It must be a vertical line if detector is properly calibrated If t 0 is known, a model independent calibration can be done 6/5/2016

8. The AGS pC polarimeter is succesfully used for the relative measurements Beam Intensity, I Polarization profile measuremens (jump quads study) Study of Polarization dependence on beam intensity 8PSTP 2011, St. Petersburg6/5/2016

9. Is absolute polarization measurement possible with a proton-Carbon polarimeter ? A systematic errors study is necessary to answer this question. Are results dependent on detector configuration ? Do we know the Analyzing Power A N (t) ? Could we properly calibrate detectors ? Do we understand energy losses in the target ? Can we control rate dependence of polarization measurements ? … 9PSTP 2011, St. Petersburg6/5/2016

10. 10PSTP 2011, St. Petersburg Polarization measured by all 3 types of detectors is consistent within 1-2% accuracy ! Can we explain slope difference for 90 and 45 degree detectors by rate effect ? All 2011 data was included in the fit. Results of the fit should be used for comparison only Polarization, P(1.2), is given for intensity 1.2×10 11 Polarization vs Beam Intensity (Late CBM), Vertical Target3, all 2011 runs Polarization dependence on detector type 6/5/2016

11. Hamamatsu (45 degree) vs. BNl (90 degree) detectors PSTP 2011, St. Petersburg11 Polarization dependence on detector type No visible variations of the polarization ratio during 4-month Run 2011! 6/5/2016

12. A N measurement for assumed 65% polarization PSTP 2011, St. Petersburg12 Analyzing Power A N (t) Poor consistency between theory and measurements Wrong energy calibration and energy losses in the target may contribute to the discrepancy Results depend on the target (rate ?, energy losses ?) Potentially, analyzing power may be measured by the pC polarimeter (up to a normalization constant) 6/5/2016

13. Dead-Layer corrections Stopping range parametrization: “standard parametrization”, p=1/d constant energy loss, p=E loss polinomial Carbon Energy from measured amplitude: 13PSTP 2011, St. Petersburg Enrgy Calibration 6/5/2016 L 0 is stopping range derived from MSTAR dE/dx (used in “standard” calibration)

14. Inverse task: If E(αA) is known then we can determine L(E) and dE/dx If t 0 is know then we can measure Carbon energy as a function of the amplitude αA and thus we can measure dE/dx (in deadlayer length units) WARNING: In such a way we measure effective dE/dx which may be different from ionization losses dE/dx. If t 0 is unknown we can make a fit, that is to try all possible t 0 and select one which provides best data consistency. It might provide us with value of t 0 and calibration of the measured amplitude E Carbon = E(αA). WARNING: the fit may work incorrectly if parameterization of stopping range L(p, αA) can not approach well true effective dE/dx. 14PSTP 2011, St. Petersburg A model independent calibration of the amplitude Enrgy Calibration 6/5/2016

15. New calibration method vs standard one The function L(E) = p 0 L 0 (E) + p 1 L 0 2 (E) fits data much better then “standard” calibration function p 0 L 0 (E) Significant difference in the value of t 0 Significant difference (up to 15% ) in the energy scale Better fit of data does not guarantee better calibration ! 15PSTP 2011, St. Petersburg6/5/2016

16. Comments about t 0 determination in the fit Including t 0 to the fit: (τ is time of flight for 1 MeV carbon ) If then (good calibration) However, if may be approximated by variations of the then result of the calibration is unpredictable may be masked by faked correction 16PSTP 2011, St. Petersburg6/5/2016

17. More realistic example is rate of good events is total DAQ rate Simplified example Only one carbon signal may be taken by the DAQ Detection efficiency: where r is average rate per bunch. An estimate of the rate effect Rate effect - is a strip pair number - is average rate per strip (millions events per spill) - is rate in strip i (events per bunch), n = 0.0528 - is relative rate in the strip I assume factor k is the same for all strips Rate contribution Machine contribution 17PSTP 2011, St. Petersburg6/5/2016

18. Polarization dependence on beam intensity (averaged over all 2011 runs) : The measured value of the rate effect factor agrees well with a pileup based estimate Vertical Target3, all 2011 runs: Strip Pairs 18PSTP 2011, St. Petersburg Rate effect 6/5/2016

19. Target dependence of the Polarization measurements PSTP 2011, St. Petersburg19 AGS pol., during H-jet meas. at injection Intensity -1.5 Polarization vs intensity, Horiz. target #1, JQ-on Polarization vs intensity, Vertical target#3, JQ-on Slope difference is consistent with our estimates We can explain 4±1 % of polarization difference by rate effect. Where the rest 4.6±1.7% come from? Enrgy losses in the target 6/5/2016

20. Energy losses in the target 20PSTP 2011, St. Petersburg φ Target Beam Energy range 400-900 keV Calculation Angle Target Thickness (μg/cm 2 ) 4816 00.9910.9820.965 450.9870.9750.951 800.9500.9030.825 850.9030.8020.610 0 - 3600.9700.9480.911 Measured/True Polarization Results are independent on target width ! 125 μm target Effect of energy losses in the target may be significant may be unpredictable Enrgy losses in the target 6/5/2016 dE/dx A N (t) (d ~ 30 nm)

21. Summary Different types of detectors were tested in the Run 2011 Results of polarization measurements were consistent within 1-2% accuracy No significant variation of the results of measurements were observed during the whole 4 month run. The polarimeter has a capability to measure analyzing power up to the arbitrary normalization factor, but accurate study of the systematic errors is needed for that. Standard energy calibration method was found to be unreliable, new method of calibration are suggested but more development is still needed. Experimental evaluation of the rate effect is consistent with estimation of pileup contribution. More accurate control of energy losses in the target is needed. 21PSTP 2011, St. Petersburg6/5/2016