Sacha KoppUniversity of Texas at Austin

on behalf of the MINERvA Collaboration

Flux Measurements in the NuMI Beam

What are the challenges?

• Experiments limited by– statistics

– knowledge of flux (E)

You Thought You Wanted…(E)

LE BEAM pME BEAM

What You Really Wanted…(xF,pT)

Pz (GeV/c)

BUT

pHE BEAM

Pz (GeV/c)Pz (GeV/c)

Why Do I say (xF,pT)?! • It is the dominant

uncertainty in predicting the neutrino flux.

• Calculating the (E) requires of

– particle production off the target

– ray tracing through beam optics

• As demo, ab initio flux for NuMI compared to MINOS data. Flux error bars dominated by particle production off the target (calculated prior to tuning to nu data).

“Medium” EnergyBeam Setting

“High” EnergyBeam Setting

“Low” EnergyBeam Setting

MINOS Data

Calculated flux

Compare Hadron Production Models

Model pT

(GeV/c)GFLUKA 0.37

Sanf.-Wang 0.42

CKP 0.44

Malensek 0.50

MARS – v.14 0.38

MARS – v.15 0.39

Fluka 2001 0.43

Fluka 2005 0.36

Fluka2001

Fluka2005

MARS–v.14

MARS–v.15

A perpexing situation for a poor experimentalist…

A Two-Part Proposal to Determine Beam Flux

Hardware (extra targets)

Dedicated RunningFor Special Configurations

The “Two Buddha’s” Near Sensoji Temple, Asakusa, Tokyo

Data Sets for Flux Determination• Special runs to constrain beam energy spectrum

– Vary target position and horn current – 3-5 ton detector fiducial mass– Total ~ 1020 POT before and after “ME Config” for these studies

Hardware Requirements (I)

pre-target beamline target hall hot work cell

target pile re-circulating air cooling system

Move horn 2 to ME setup– Extend the stripline

Switch to ME target– Fixed location upstream – Target doesn’t fit into horn

We want to delay the 2nd part

Target

Drop-shaft from surface

power supply for horns

Hardware Requirements (II)

• After NOvA upgrades target motion will no longer be possible– Required special module -- requires lots of moving parts (deemed risky)– Module experiences too much thermal motion under high heat load.

Allows for 2.5 m of target motion to vary the beam

energy

Baffle

Target

Target/Baffle Carrier• The loss of this module

removes ability to flexibly change E

• MINERvA requests that Lab provides one of these + spare for our studies.

• Could be useful to NOvA too! They need a spare (backup) target!

Hardware Requirements (III)• Target will be re-designed for for NOvA

– Still 6.4mm wide, but not as tall– No longer fits inside horn for LE config

• We therefore request– a fresh LE target to be used for our first set of test runs– a spare LE target to ensure success of test runs after ME switch (could

be used as spare for NOvA, too).

Past Techniques

Used to Wrestle with Beam Flux

High School Sumo ChampionshipsSumo Hall, Ryōgoku, Tokyo, 5 Aug. 2007

• Neutrino data• Muon Monitors• Hadron Production

Experiments

12/25

FNAL NBB

Fluxes came from these

In situ Muon Monitor Flux

• CERN PS• CERN

WANF• IHEP• FNAL

E616• FNAL

NuMI• Typical

~20%

NuMI Mon Flux

• Measurement of of hadron flux using Mon event rates (no error from x-sec)

• L. Loiacono, PhD thesis

• 20% errors

External Production Exp’t

A. Aguilar et al., arXiv:0806.1449

MiniBooNE

• Flux prediction based on HARP.

• As a check: compare the HARP flux to QEL events.

• Scale flux by 1.21!

• What about K2K?! Never see their plot!

… and yet …

• reconciling the MiniBooNE event rate with HARP flux• Consistent within errors• Possible shift?

In situ Flux Using Neutrinos

P. Astier et al., Nucl. Instr. Meth. A 515 (2003) 800.

NOMAD

Also see papers by

• L. Ahrens et al, Phys. Rev. D 34, 75 - 84 (1986)

• K. McFarland, et al., arXiv:hep-ex/9806013

19

NuMI Flux Tuning

• Fit all 7 beam runs.

• Fit νμ and νμ

spectra• But uses

inclusive events!

• To be replicated by MINERvA using QELs

Phys. Rev. D77, 072002 (2008).

The Call of the Mermaid

“I’ll just let [Harp/NA69/NA49/MIPP/SPY] solve my problems”

-- Hans Christian Anderson

Does any one recall the fate of the person that answers the mermaid’s call?

Atherton400 GeV/c p-Be

Barton100 GeV/c p-C

SPY450 GeV/c p-Be

Data Upon Which Models are Based

NuMI HE BeamNuMI HE Beam

NuMI LE BeamNuMI LE Beam

p (GeV/c)

p T (

GeV

/c)

Modern Data Sets are $%#&! Good!

• Modern data sets better than original ‘beam surveys’– single particle detection– particle ID– large acceptance

• So can’t we just use this to map (xF,pT)??

pT (GeV/c)

d/d

p T (

mb/

GeV

/c)

eg: C. Alt et al, Eur.Phys.J.C49:897-917,2007

No! (1) Thick Target Effects

• Most ptcle production exp’ts on thin targets

• Nu production target ~ 2int

• Reinteractions!

• 20-30% effect

Min

iBoo

NE

NuMI

CNGS

J-PARC

figure courtesy Z. Pavlovic

No! (2) In-beam variations

24/25

• Temperature in NuMI target hall varies by 8°C as beam power cycles.

• Causes change in horn current ~1 kA

• Observe direct variation in beam flux (Mons)

• Thermal variations in your beam MC?

NuMI-only

figure courtesy L. Loiacono

NuMI-Collider Combined mode

No! (3) Beam Degradations?

Each data point is one

month’s data

Neutrino Energy (GeV)

Eve

nts

/ 10

16P

OT

/ G

eV • Started after installation of new target.

• Have ruled out horns (swapped)

• Have ruled out He leak in decay volume

• Problem mitigated when swapped in new target

figure courtesy M. Dorman

CNGS: Earth B Field?!

NeutrinoFocus +

Anti-neutrino

Focus

They See shift of 6.4 cm (consistent with 0.3 Gauss)

figure courtesy E. Geschwendtner

A Cautionary Tale• CERN PS team did particle prod

@ IHEPJ.V. Allaby, et al., Phys. Lett. 29B 48 (1969)

• In-situ flux using Mons suggested X2 off?!D. Bloess, et al, CERN-69-28 (1969),

Nucl. Inst. Meth. 91 (1971) 605.

• Particle production round two – ok to 15%J.V. Allaby, et al., CERN-70-12.

MIPP Data Not Currently Competitive

• MIPP uncertainties were bigger than the 5% uncertainty from the MINOS nu data fit (backgrounds, statistics, etc)

• Must improve by factor 9 in statistics and extend (xF,pT)

J. Paley, NuFact07

Could MIPP Provide a Stronger Measurement?• We would not accept a MIPP-like measurement alone.• MIPP provides a MC input, not a measurement of the nu flux for the experiment!

– MIPP data feeds into our MC, just like the muon lifetime, K BR’s, etc.– It provides no handles that our MC is right (real life variations and uncertainties).

• A cross section experiment like MINERvA must have in situ measurement– Other experiments have all made statements like “Using HARP/MIPP/SHINE, we will

measure to 5%.” These are strong promises!– Past cross section experiments measured fluxes (NBB experiments like CCFR,

CHARM/CDHS). Others are normalized to these.– Biggest uncertainties in cross section measurements from WBB’s are always the flux.

• MIPP and E938 are complex experiments!– Backgrounds to the desired particle species (eg K/pi) are substantial– Kinematic coverage relevant for a neutrino beam has never been achieved.– We demonstrated better coverage of (xF,pT) range in MINOS in situ data

At the very least, we must pursue both in situ and external (MIPP-like) experiments!

The Five Foundations of

Beams

5-story pagoda ofSensoji Temple, Asakusa, Tokyo

Neutrino Beams 101

• Feynman scaling in xF ~ pL/p0

• No scaling for

• ‘Cocconi divergence’

• Tell me what p you want, I’ll tell you what angle to focus.

proton

p0

pTpz

p + A → + X

Neutrino Beams 102

• ‘Cocconi divergence’

• Neutrino divergence

• Reduce divergence ~3, flux goes up by ~ 25

We’ll be sensitive to “edges” where focusing fails.

L. Ahrens et al, Phys. Rev. D 34, 75 - 84 (1986)

Neutrino Beams 103Beam MC

i

i

X i

B

B

Neutrino Beams 104

Focu

sing

pe

ak

figure courtesy Ž. Pavlović

• Focusing errors are fairly small.

• Mostly pile up at edges of focusing system.

figure courtesy Ž. Pavlović

“High”Energy

targetHorn 1

Horn 2

“Low”Energy

protonHorn 1

Horn 2

target

Pions with pT=300 MeV/c and

p=5 GeV/cp=10 GeV/cp=20 GeV/c

Vary beam energy by sliding the target in/out of the 1st horn

Neutrino Beams 105

Opportunity: Flexible Beam Energy

figure courtesy Ž. Pavlović

M. Kostin et al, “Proposal for Continuously- Variable Neutrino Beam Energy,” Fermilab-TM-2353-AD (2002)

in situ Particle Production Off the

Target Measurements using Flexible Beam

Configurations

Sensoji Temple, Asakusa, Tokyo

NuMI Beam Configurations• Can vary

– Horn current (pT kick supplied to ’s)

– Target Position (xF of focused particles)

• Plots show (xF ,pT) of contributing to neutrino flux.

• Similar plots exist for kaons

• Acquired data from 8 beam configurations (here are shown 4)

LE010/185kALE010/0kA

LE100/200kA LE250/200kA

Parameterizing Hadron Production (I)• We tried to

parameterize the Fluka’05 (xF, pT) distributions with an empirical formula.

• In this fit,– A = A(xF)– B = B(xF)– C = C(xF)

• This form is quite similar to BMPT (which has a D(pT)2 term – small??)

• I cannot motivate the (pT)3/2 other than the fact that it fits the Fluka spectrum rather well.

2/3

)( TCpT

T

eBpAdp

dN

Fluka ‘05

Parameterizing Hadron Production (II)

FFF xbxx 11

• Both A(xF) and B(xF) fit reasonably well to following shape

• All this says, of course, is that Fluka’05 is roughly consistent with BMPT• The values of the exponents are not in agreement with BMPT’s paper, but this

is a thick target parameterization, and they quoted invariant cross section.

Parameterizing Hadron Production (III)• Used empirical form

similar to BMPT to parameterize Fluka2005:

• Fit was to a MC of our thick-target yield estimated by Fluka2005.

• Tune parameters of the fit to match ND data.

ND Spectra After Reweighting (I)

ND Spectra After Reweighting (II)

ND Spectra After Reweighting (III)

ND Spectra After Reweighting (IV)

ND Spectra After Reweighting (V)

ND Spectra After Reweighting (VI)

ND Spectra After Reweighting (VII)

ND Spectra After Reweighting (VIII)

(xF,pT) weights

• Result of the fit is set of weights in (xF,pT) plane that should be applied to /K yields

• Data prefers more low pT pi’s

weightsweights

Region of LE beam focusingRegion of

insignificant focusing

Constraint of fit on ratio

• Anti-nu’s tune the flux off the target

• Out put of fit agrees well with recent NA49 data

Constraint of fit on K/ ratio• Recent

data from FNAL E907 to which we can compare

Improvement in Flux Uncert.Before

use this line

use this line

After

• We obtain greater precision than MIPP or E938 proposal.

What To Take Away From MINOS• Proof of principle that can

disentagle underlying (xF,pT) from focusing effects.

• Precision is ~5% in neutrino flux.

• If we use a “known” or “standard candle” channel, this could be turned into a flux.

• Not possible with MINOS data, but possible with MINERvA QELs!

• Gives a flux shape, can be normalized at E>25 GeV.

(Fluka2005)

Question: Why Repeat this in MINERvA?

• MINOS data is shown here as a proof of principle that neutrino data in a flexible beam is sensitive to underlying hadron production.

• Since MINOS used inclusive events, it should not be interpreted as a measurement, however. We never cared if there was an overall scale factor in flux or cross section (just wanted to predict far detector!)

• MINERvA will use QELs, for which the shape of the cross section is reasonably flat and known to ~10%.

• If we use the inclusive events >30GeV to normalize the flux, then this analysis in MINERvA, repeated with QELs, can give the flux shape.

But Time is Of the Essence!

(could we make due with less beam?)

• Our revised request is for o 4e20 POT in the LE beamo 0.9e20 POT for special runs

in the LE configurationo 0.9e20 POT for special runs

after the ME conversion

• This is a bit less than our original request submitted to Program Planning

the eleventh hour

We revised to 70% of POT• I took #POT from

MINOS and x10

• For MINERvA we want to do this study with only QELs

• In LE configuration, much of x-sec is QEL, so we’re OK.

• At high energy configurations, more events, but these are Res+DIS, so don’t want to take credit for this.

Revise 4 3 target positions?

• Must make sure to map out full (xF,pT) space

• Error reported here is MINUIT float. Not reflected is how well model agrees with data

• Some of these go down, reflecting that those configurations were in tension with the others.

• Could probably combine LE100+150---> LE125?

Must we repeat in ME configuration?

• Yes.

• The flux derived in the LE configuration of horns will be valid only in LE configuration

• New horn configuration leads to new focusing uncertainties

• New passive material will be different in beam line; this component will need to be measured.

Performing Measurements in ME Configuration

• Spectra have lots of rich structure – a good thing!

Performing Measurements in ME Configration

• We can vary (xF,pT) just as well as in the LE configuration

LE010/200kA LE100/200kA

ME010/185kA

ME010/0kA ME010/185kA

ME250/200kA

ME150/200kA

ME100/200kA

Summary of Our Philosophy

• Accuracy of measurement is 5-10%– 5% from the flux fit– 10% from normalization mode– Better than any other technique.

• We are requesting resources from the Lab to do this– extra moveable LE target + carrier– Special running of 0.9e20POT in LE and ME configurations

• Analysis emphasizes direct measurements to measure flux in situ.– There are always uncertainties in using

external production experiments– Must ensure what’s happening during

MINERva experiment!