Measurement and simulation of the neutron response and detection efficiency

Anna Ferrari VCI 2007, Wien, February 22nd, 2007 1

Measurement and simulationMeasurement and simulationof the neutron response and detection efficiencyof the neutron response and detection efficiency

of a Pb – scintillating fiber calorimeterof a Pb – scintillating fiber calorimeter

A. FerrariA. Ferrari Fondazione CNAO (Milano)Fondazione CNAO (Milano)


The KLOE Pb-scintillating fiber calorimeterThe KLOE Pb-scintillating fiber calorimeter

1.2 mm1.35 mm

1.0 mm

Active material: •1.0 mm diameter scintillating fiber (Kuraray SCSF-81, Pol.Hi.Tech 0046), emitting in the blue-green region: Peak ~ 460 nm.• Core: polystyrene, =1.050 g/cm3, n=1.6

High sampling structure:• 200 layers of 0.5 mm grooved lead foils (95% Pb and 5% Bi).• Glue: Bicron BC-600ML, 72% epoxy resin, 28% hardener.• Lead:Fiber:Glue volume ratio = 42:48:10

Designed and put in operation as e.m. calorimeter

Good performance in time and energy response:

(E)/E = 5.7 %/√E(GeV) (t)= 54 ps/√E(GeV)

and high photon efficiency see NIMA 482 (2002) 364-386


Why looking for neutron detection efficiency ?Why looking for neutron detection efficiency ? Detection of neutrons of few to few hundreds of MeV is traditionally performed with organic

scintillators (principle of operation: elastic neutron scattering on H atoms, with production of protons detected by the scintillator itself) efficiency scales with thickness ~1%/cm

see C. Birattari, A.Ferrari, M.Pelliccioni et al., NIMA 297 (1990) 250-257, NIM A 338 (1994) 534-543

On the other hand, the extended range rem countersextended range rem counters used in radiation protection are based on a structure scintillator/medium-high Z material, which enhances the neutron efficiency

an intense Monte Carlo study has been performed with the FLUKAFLUKA code, which is well validated for the hadronic physics, till the low energy region

an experimental test has been carried out with the neutron beam of the The Svedberg Laboratory of Uppsala (October 2006) [with TARI program support]

An estimate with KLOE data ( n are produced by K- interactions in the apparatus walls) gave: for low energy neutrons (Ekin ≤ 20 MeV) , confirmed by KLOE MC (expected: 10% )

- Measurement of the neutron e.m. form factors in the time-like region (DANTE) - Search for deeply bounded kaonic nuclei (AMADEUS)

n are importantfor the DANE-2 program @ LNF


The neutron beam line at TSLThe neutron beam line at TSL

A quasi-monoenergetic neutron beam is produced in the reaction 7Li(p,n)7Be. 42% of neutrons at the max energy The absolute neutron flux in the peak is measured after the collimator by 2 monitors of the beam intensity. Accuracy: ~ 10%

5.31 m

KLOE calorimeter moduleKLOE calorimeter module

EKIN (MeV)


(1)

(3)

(2)

3 large data sets collected with different beam intensities: 1.5 kHz/cm2, 3.0 kHz/cm2 and 6.0 kHz/cm2

The experimental setup and the data setThe experimental setup and the data set

( 2 ) Beam position monitor: array of 7 scintillating counters, 1 cm thick.

( 1 ) Old prototype of the KLOE calorimeter: 60 cm long, 3 x 5 cells (4.2 x 4.2 cm2), read out at both ends by Hamamatsu/Burle PMTs

( 3 ) Reference counter: NE110, 10×20 cm2, 5 cm thick

A rotating frame allows for: - vertical positions (data taking with n beam) - horizontal positions (calibration with cosmic rays)

n

Y XZ

For each configuration, several scans with different trigger thresholds Typical run: 0.5-1.5 Mevents, 1.7 kHz DAQ rate Cosmic rays run (beam off) for calibrations with MIPs.

last plane not integrated in the acquisition system


The measurement of the global efficiencyThe measurement of the global efficiency

RNEUTRON: from beam monitor via neutron flux intensity measured by TSL.

RTRIGGER: use coincidence between sides.

•Scintillator: T1 trig = Side 1×Side 2•Calorimeter: use the analog sum of 12 PMs/side (first four planes) T1 trig = A×B

Global efficiency measurementGlobal efficiency measurementintegrated on the full spectrum

I. The methodI. The method

fLIVE: live time fraction : for preliminary measurement, assume full acceptance and no background

= RTRIGGER

RNEUTRON × fLIVE ×


II. The scintillator efficiencyII. The scintillator efficiency The measurement of the scintillator efficiency gives a cross calibration of the measurement method and of the beam monitor accuracy, with small corrections due to the live time fraction

The energy scale was calibrated with a 90Sr source. 10% accuracy for horizontal scale (threshold) and the vertical one ()

Threshold (MeV e equiv. energy)

(%

)

Results agree with “thumb rule” (1%/cm): 5% for 5 cm thick scintillator (with a threshold of 2.5 MeV electron equivalent energy)

Agreement within errors with previous published measurements in the same energy range, after a rescaling of them to our thickness


III. The calorimeter efficiencyIII. The calorimeter efficiency

Energy scale setting done by MIP calibration of all channels, and using the MIP/MeV scale factor used in KLOE

10% uncertainty on both horizontal and vertical scales

Stability wrt very different run conditions: a factor 4 variations of both live time fraction (e.g. fLIVE=0.2 0.8) and beam intensity (1.5 6.0 kHz/cm2).

Very high efficiencyVery high efficiency, about 4 times larger than the expected if only the amount of scintillator is taken into account: ~ 8% for 8 cm of scintillating fibers.

(%

)

Compare with our scintillator efficiency measurement, scaled by the scintillator ratio factor 8/5

Thr (MeV e equiv. energy)


The neutron spectrum from ToFThe neutron spectrum from ToF

n

Correct raw spectra for T0 and convert into ns

Since the trigger is phase locked with the RF ( time structure: 45 ns), rephasing is needed for neutrons with Ekin < 50 MeV (5.3 m far from the target)

ToF (ns)

From ToF spectrum obtain of the neutron Assuming neutron mass, obtain Ekin


Energy vs ToFEnergy vs ToF

Energy released vs ToF Energy released vs ToF

ToF (ns)

Ener

gy (M

eV e

q. e

l.)

Energy (MeV eq. el.)threshold: 15 mV

Charge response Charge response

The collected charge is here expressed as the energy of an electron that gives the same charge response


LEAD

GLUE FIBERS

base module

replicas

200 layers

Using the FLUKA tool LATTICEthe fiber structure of the whole calorimetermodule has been designed.

In the base module the calorimeter is simulated in detail, both under the geometrical point of view and with respect to the used materials

The calorimeter simulation with FLUKAThe calorimeter simulation with FLUKA

All the compounds have been carefully simulated. - for the fibers, an average density between cladding and core has been used : ρ = 1.044 g/cm3 - glue: 72% epoxy resin C2H4O, =1.14 g/cm3, + 28% hardener, =0.95 g/cm3

Polyoxypropylediamine C7H20NO3 90%

Triethanolamine C6H15NO3 7%

Aminoethylpiperazine C6H20N3 1.5%

Diethylenediamine C4H10N2 1.5%

hardener composition


The readout simulation The readout simulation

The simulation of the Birks effectThe simulation of the Birks effect

Fluka gives energy deposits in the fiber.

The light is propagated by hand at the end of the fiber taking into account the attenuation.

The energy read-out has been simulated by including: the generation of photoelectronsthe generation of photoelectrons the constant fraction distribution the constant fraction distribution the discriminator threshold.the discriminator threshold. No trigger simulation is included at the moment.

dL/dx = k dE/dx / [ 1 + c1 dE/dx + c2 (dE/dx)2] c1 = 0.013c2 = 9.6×10-6

The energy deposits are computed in Fluka taking into account the Birks effect, that isthe saturation of the light output of a scintillating material when the energy release is high, due to the quenching interactions between the excited molecules along the path ofincident particles:

In literature and in GEANT:


Proton beam

Li targetn 5.5°

The simulation of the beam lineThe simulation of the beam line

Z(cm)

Y(c

m)

Shielding(concrete and steel)

Calorimeter

7Li Target

Gaussian angular distribution(Journal of Nuclear Scienceand Technology, supplement 2(2002), 112-115)

At the Li-target

At the calorimeter

Ekin(MeV)

The beam line has been simulated starting from the neutrons out of the Litium target

At the entrance of the

beam monitor


Neutron interactions in the calorimeterNeutron interactions in the calorimeter Each primary neutron has a high probability to have elastic/inelastic scattering in Pb

target Pel(%)

Pinel(%)

Pb 32.6 31.4

fibers 10.4 7.0

glue 2.3 2.2In average, secondaries generated in inelastic interactions are 5.4 per primary neutron,counting only neutrons above 19.6 MeV.

neutrons

above 19.6 MeV

62.2%

photons 26.9%

protons 6.8%

He-4 3.2%

deuteron 0.4%

triton 0.2%

He-3 0.2%

Typical reactions:

n Pb n1 γ + n2 n n Pb n1 γ + n2 n + p + res.nuc. n Pb n1 γ + n2 n + 2p +res.nuc.

In addition, secondaries created in interactions of low energy neutrons (below 19.6 MeV) are - in average - 97.7 particles per primary neutron.

neutrons 94.2%

protons 4.7%

photons 1.1%

Simulated neutron beam: Ekin = 180 MeV


A typical inelastic processA typical inelastic process

n

Z(cm)

p

n1

n2

n3

n4

X(cm

)primary vertex

En = 175.7 MeV En (p) = 126 MeV

The enhancement of the efficiency appears to be due to the huge inelastic production of neutrons on the lead planes. These secondary neutrons: - are produced isotropically; - are produced with a non negligible fraction of e.m. energy and of protons, which can be detected in the nearby fibers; - have a lower energy and then a larger probability to do new interactions in the calorimeter with neutron/proton/γ production.


Neutron yield inside the calorimeterNeutron yield inside the calorimeter

X(c

m)

Z(cm)

beam

Neutron fluence

Ekin (MeV)

(E

)

cos(θ)

dN/d

Ω (n

sr-1 p

er p

rim

) 1° plane4° plane

Isotropic angular distributions

from inelastic scattering

Energy distributionEnergy distribution


The proton yield inside the calorimeterThe proton yield inside the calorimeter X

(cm

)

Z(cm)

beam

Proton fluence

(E

)

Ekin (MeV) cos(θ)

dN/d

Ω (p

rot s

r-1 p

er p

rim

)

Protons are mainlyconcentrated alongthe direction of the primary beam

Energy distribution Energy distribution Angular distribution Angular distribution


A key point: the high sampling frequencyA key point: the high sampling frequency

The energy deposits of the ionizing particles (protons and excited nuclei) are distributed mainly in the nearby fibers: the high sampling frequency is crucial in optimizing the calorimeter

Z depo

sit –

Z prim

.ver

t(c

m)

Xdeposit -Xprim.vert (cm)

neutron lateral profile neutron lateral profile proton lateral profile proton lateral profile

Interaction vertex Interaction vertex in leadin lead (protons and res nuclei) (protons and res nuclei)


Ekin (GeV)

Protons

Neutrons

E.m. energyOthers

E(r

il)/E

(tot)(r

il)

Particle contribution to the energy responseParticle contribution to the energy response

EE(tot)(tot) (ril) (ril) = = ΣΣ E Epp(ril)(ril) + + ΣΣ E Enn

(ril)(ril) + + ΣΣEEemem(ril)(ril) + + ΣΣEEresnucresnuc

(ril)(ril)

Particle contribution to the energy released in the fibers:Particle contribution to the energy released in the fibers:

Evaluating the particle contribution to the energyresponse, we have to take into account:

- the contribution of the highly ionizing particles: protons and excited nuclei; - the contribution of the e.m. energy

The neutron contribution is not to take into account in general, because the neutrons transfer energy to the nuclei of the fibers basically as invisible energy. We don’t know if at least a part of this energy can produce somehow scintillating photons.

For this reasons, we evaluate first the efficiency without taking into account the neutron energy deposits, then we do the exercise to re-calculate the efficiency also including the neutron contribution, as a superior limit to the true value.


Integrated efficiency: 50%

The simulated efficiency vs energyThe simulated efficiency vs energy ε

(%)

Ekin (MeV)

Preliminary

Preliminary

No cut in released energy!

No trigger simulation Simulation of the discriminator threshold applied only at the cell level not at the cluster level

By taking into account neutron

energy deposits: ε ≈ 56%

To be read as a superior limit !!


Data vs Monte CarloData vs Monte Carlo The whole cluster algorithm procedure is under study A first comparison at the cell level has

been made (in this example: threshold = 15 mV)

The agreement Data/MC is good, except for

the lower energy region

n

To be included in the simulation:

local shielding, metallic supports, …

to simulate the background due to the

neutrons scattered on the materials in the

experimental hall


Tcell(ns)

MC● Exp

Ekin(MeV)

ε(%

) A study on the detection efficiency vs energy A study on the detection efficiency vs energy

A fast Monte Carlo procedure has been used to test the sensitivity of the time distribution to

the shape of the efficiency curve

n

• two efficiency functions (Fermi-Dirac) are used in Monte carlo generation time distributions for the central cell in the first plane are calculated and

compared with the experimental data (threshold = 15 mV)

In this example:

Tcell(ns)

MC● Exp

Ekin(MeV)

ε(%

)


ConclusionsConclusions

We think that this work is the starting point for the study and the development of a new, compact, chip, fast and efficient neutron detector

A first comparison Data/Monte Carlo has been done and it is satisfactory. Work is in progress to tune the Monte Carlo.

The first measurement of the detection efficiency of a high sampling lead-scintillating fiber calorimeter to neutrons, in the kinetic energy range [20, 178] MeVhas been performed at The Svedberg Laboratory, Uppsala .The efficiency integrated over the whole neutron energy spectrum ranges between 40% and 50%, at the lowest trigger threshold used.

A detailed Monte Carlo study, carried out with FLUKA, showed that that the origin of a such enhancement is related both to an effect shower-like, due to the inelastic processes in the Pb-scintillating fiber structure of the calorimeter, AND to the high sampling fraction used of this detector.

New tests on neutron beam in different energy regions are in program



SPARES


Details on DAQDetails on DAQ• Scintillator trigger: Side 1 – Side 2 coincidence (T1=S1×S2)• Calorimeter trigger: based on analog sum of the signals of the first 4 plan out of 5 (T1=A×B).• Trigger signal is phase locked with RF signal (T1 free).• Vetoed from retriggering by a 5-35 ms busy signal and by the DAQ busy.• The final trigger signal is: T2 = T1free.AND.NOT(BUSY).

• T1free, T2, and the n monitor signals are acquired by a scaler asynchronous from DAQ.• Fraction of live time: T2/T1free; essential for the efficiency evaluation.

T2/T

1 FREE

Thr (mV)• Neutron rate proportional to neutron monitor via neutron flux intensity (I0) and peak fraction (fP)• An absolute rate calibration should be provided by scintillator counter.• Calorimeter scintillator efficiency rate is almost independent from beam monitor.


Time structureTime structure

4.2 ms

2.4 ms

40 ns41 ns

5 ns FWHM

RF Macro structure

Calorimeter Trigger signal


Test of phase lockingTest of phase locking

Beam RF

T1(Free)

Test done with a random trigger at 60KHz



S1(ADC counts)

Thr (mV)

1.15 count/mV

Trigger threshold calibration: mV to ADC counts

Scintillator calibrationScintillator calibration

source to set the energy scale in MeV: 90Sr endpoint 0.564 MeV; 90Y endpoint 2.238 MeV.Fit of the b spectrum, with ADC counts to MeV factor as a free parameter.

0.021 MeV/countS1

ADC counts


Calorimeter calibrationCalorimeter calibration

A 1.6 count/mVB 2.0 count/mV

AD

C c

ount

smV

• Cell response equalized: MIP peak at 600 ADC counts.• Trigger threshold calibration: - HP attenuators used for A and B not to exceed the dynamic range of the ADC; different attenuation factors: fA=2.0, fB=1.7. - Threshold in counts studied with different methods.• Energy scale set with MIPs using the conversion factor from KLOE: a MIP in a calorimeter cell corresponds to an electron of 35 MeV.


100 mV

40 mV

15 mV


Simulation of the energy read-out

fiber (active material)

energy deposit given by FLUKAThe light is propagated by hand at the end of the fiber using the parametrization:

Kuraray

Politech

The number of photoelectrons generated by the light collected by each fiber is evaluated:

Attenuation

na , bpe− fib generated

according to a Poisson distribution

the constant fraction distribution is simulated (15% fr., 10 ns t.w.) to obtain the time

Ea,b(fib) = E(dep) ·[0.35 e-x(a,b)/50 + (1- 0.35) e–x(a,b)/430 ]

Ea,b(fib) = E(dep) ·[0.35 e-x(a,b)/50 + (1- 0.35) e–x(a,b)/330 ]

ta,b(fib) = t(dep) + X(a,b) /17.09

na,b (pe-fib) =E(fib)(MeV)(a,b) · 25t(a,b)

(p.e.) = t(a,b)(fib)

+ tscin+ 1ns (smearing)

na,b (pe-cell) = ∑ t(pe)<300ns na , bpe− fib


Passive material

Lead foils: 95% Pb 5% Bi homogeneous compound

Material simulation and compounds

Polyoxypropylediamine C7H20 NO3

Aminoethylpiperazine C6H15 N3

Diethylenediamine C4H10 N2

1.5%1.5%

Triethanolamine C6H15 NO3

90%7%

Active material (fibers+cladding)

Polystyrene C2H3 homogeneous materialaverage density between cladding and core = 1.044 g/cm3

72% Epoxy resin C2H4O ( = 1.14 g/cm3)

Glue: 28% Hardener ( = 0.95 g/cm3)


COMPARISON WITH DATA - Photon response

linearity response well reproduced curve: detector resolution dots: FLUKA simulation

Energy response

In the cluster algorithm the cluster energy is multiplied by an empirical factor 1/0.145*1/0.768 (11.1% sampling fraction)

σ(E)/E = 5.7% /√E (GeV)


CLUSTER POSITION – longitudinal resolution

Resolution of the cluster centroid position of the along the module.

A B

curve: detector response

dots: FLUKA simulation

Energy deposit

σ(E)= 1.2 cm /√E (GeV)

Z = (ta –tb) vlight / 2


A MC study is in program, based on the Cesare Bini work on KLOE data:A MC study is in program, based on the Cesare Bini work on KLOE data:

selected proton sample selected proton sample from KLOE datafrom KLOE data

The hadronic response: (2) protons


Ek~30 MeV p ~ 240 MeV Range ~ 1 mm (Pb)

The proton energy response

Angular and kinetic energy distribution of the proton data

A MC proton production A MC proton production according with these distributions according with these distributions must be donemust be done

Documents

Measurement and simulation of the neutron response and detection efficiency