Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
Phenomenology of Higgsino NLSPs
Matthew Reece
Princeton Center for Theoretical Science
November 25, 2008
work in progress with Patrick Meade and David Shih
Matthew Reece Phenomenology of Higgsino NLSPs
Motivation
In this talk I’ll be discussing some signatures of Higgsino NLSP ingauge mediation. I’m choosing a few (preliminary, incomplete)elements of an ongoing project to discuss, but another goal of thetalk is to look at some of the capabilities of the LHC detectors(choosing ATLAS for concreteness) that we theorists don’t alwaysthink about.
I’ll also briefly review some facts about GMSB, and mention someof the Tevatron searches that place limits on minimal GMSB (andwhat they can tell us about non-minimal GMSB).
Matthew Reece Phenomenology of Higgsino NLSPs
SUSY Breaking Phenomenology
If the LHC discovers supersymmetry, possibly the most importantfollow-up question becomes: how is SUSY breaking mediated tothe Standard Model?
Broadly speaking, the main options are some form of gravitymediation with a special flavor-respecting structure (somemodification of anomaly mediation, Kahler-dominated modulimediation in IIB strings, ...) or gauge mediation.
In this talk I want to look at a few aspects of how gauge mediationcould show up at the LHC. The work is still preliminary.
Matthew Reece Phenomenology of Higgsino NLSPs
What is Gauge Mediation?
In its purest form, gauge mediation means that someSUSY-breaking hidden sector has the Standard Model gauge groupas a global symmetry, and the MSSM weakly gauges this globalsymmetry, with no other visible/hidden sector couplings.
In this case the MSSM soft supersymmetry breaking terms are allcalculable in terms of simple correlation functions of the hiddensector, even if it is strongly coupled (Meade, Seiberg, Shih).
More generally, one adds additional visible/hidden sectorinteractions, e.g. for µ/Bµ.
Matthew Reece Phenomenology of Higgsino NLSPs
Basic Gauge Mediation Phenomenology
Gauge mediation is inherently flavor-respecting.
One key feature for collider physics is the light gravitino LSP.Because the gravitino mass measures the size of gravity-mediated(Planck-suppressed) SUSY breaking, a GMSB solution to theflavor problem demands a light gravitino.
Minimal gauge mediation adds some messenger fields in the 5 and5 of SU(5). It predicts either a bino or stau NLSP.
Matthew Reece Phenomenology of Higgsino NLSPs
Minimal Gauge Mediation Spectrum
SPS Point 8: B NLSP GMSB (hep-ph/0202233)
Note squark/slepton mass ratio is large:√
3αs/α
Matthew Reece Phenomenology of Higgsino NLSPs
Why Revisit Gauge Mediation?
There has been a revival of interest in building models of gaugemediation, in part due to the work of ISS showing that metastableSUSY-breaking vacua are common and easy to construct.
Several recent papers have emphasized that gauge mediation can,in principle, look very different from minimal gauge mediation:Extra-Ordinary Gauge Mediation (Cheung, Fitzpatrick, Shih);General Gauge Mediation (Meade, Seiberg, Shih; also Carpenter,Dine, Festuccia, Mason); Dynamical µ/Bµ in NMSSM (Liu,Wagner); many others....
Apart from the interesting question of how we can know if whatthe LHC sees is GMSB, some of these models can also suggestconcrete experimental signatures that are amusing to think about.
Matthew Reece Phenomenology of Higgsino NLSPs
Higgsino NLSP
In this talk I’ll mostly be discussing the case of mixedbino–Higgsino NLSP. One can’t have small µ (with EWSB) inordinary gauge mediation, which always has squarks much heavierthan sleptons.
As Cheung &co. pointed out, one can construct GMSB modelsthat still unify but have split doublet and triplet messengers. Massrelations are then modified: one can have different “effectivemessenger numbers” Neff ,3 � Neff ,2.
Matthew Reece Phenomenology of Higgsino NLSPs
A Cancellation in the Running
As first pointed out by Agashe & Graesser, if the squarks andsleptons are relatively degenerate one can have a cancellation inthe running that accommodates a small µ. This can help to relievesome of the fine-tuning.
One balances a GMSB contribution to m2Hu
, proportional toα2(Mmess,2)
2/Neff ,2, against a stop loop contribution proportionalto α3(Mmess,3)
2/Neff ,3.
For this study we’re not interested in detailed questions of how µarises and how tuned the model is. We assume some scenariowhere µ is small, and see what the collider implications are.
Matthew Reece Phenomenology of Higgsino NLSPs
NLSP Branching Ratios
Neutralino decays to gravitino + (photon/Z/Higgs) are controlledby a quantity:
A =m5
χ01
32πF 2, (1)
times some factors that depend on masses and mixings.
Important qualitative feature: if F (and hence m3/2) is a little bitlarge, can have displaced vertices in a detector. If m3/2 issignificantly heavier, particles escape the detector entirely.
Bino to photon + gravitino
cτ =10−2cm
|N11 cos θW + N12 sin θW |2
(100 GeV
mχ01
)5( √F
100 TeV
)4
.
Matthew Reece Phenomenology of Higgsino NLSPs
Higgsino Branching Ratios
When Higgs and Z decays are on-shell, we can summarize thedecay widths as:
Γ(χ01 → G + γ) = 2 |N11cW + N12sW |2A
Γ(χ01 → G + Z ) =
(2 |N12cW − N11sW |2 + |N13cβ − N14sβ |2
)×
(1−
M2Z
m2χ0
1
)4
A
Γ(χ01 → G + h) = |N14cα − N13sα|2
(1−
M2h
m2χ0
1
)4
A
Matthew Reece Phenomenology of Higgsino NLSPs
NLSP BR to Photon + Gravitino
Branching ratio χ01 → Gγ, in the (M1, µ) plane with M2 = 2M1.
Matthew Reece Phenomenology of Higgsino NLSPs
Decays to Z and h
Tevatron signatures for Z and h decays were discussed in K.Matchev and S. Thomas, hep-ph/9908482
Let’s examine when we expect decays to h to be significant.Focusing on the decoupling limit mA,mH ,mH± � mh, whereα = β − π/2, we have:
Γ(χ01 → G + ZL)
Γ(χ01 → G + h)
=|N13cβ − N14sβ|2
|N13cβ + N14sβ|2
(1−m2
Z/m2χ0
1
1−m2h/m2
χ01
)4
. (2)
So, two important cases: large mχ01, or µ < 0 and tan β ≈ 1.
Matthew Reece Phenomenology of Higgsino NLSPs
CDF Search for γ + b + j + 6ET
]2
[GeV/cbjM0 100 200 300 400 500 600 700
2E
vent
s pe
r 24
GeV
/c
0
20
40
60
80
100
120
140
160
180
200
0 100 200 300 400 500 600 7000
20
40
60
80
100
120
140
160
180
200Data
, Real+fake bγFake
, Fake bγReal
cγ
bγ
Background Uncertainty
SearchTEbjγ -1
CDF Run II Preliminary 2.0 fb
www-cdf.fnal.gov/physics/exotic/r2a/20080410.gbjmet/public gbjmet/
CDF Public Note 9296Matthew Reece Phenomenology of Higgsino NLSPs
Clean photon events
In minimal gauge mediation, at a collider one can make wino pairsthat decay down through sleptons to the bino. (Bino-bino eventsare rare.) Even the sparsest events are likely to have some hardleptons.
With Higgsino NLSP, pair production of NLSPs is not so rare.Substantial numbers of events can be very clean, due to the smallsplitting: diphoton + nothing.
However, even if we make the splitting tiny, χ+1 χ0
1 events willfrequently have at least one hard jet, just from ISR.
Matthew Reece Phenomenology of Higgsino NLSPs
Constraints
A number of existing experiments constrain this scenario. There isa fairly general chargino mass bound from LEP of 103 GeV. Itbreaks down in the limit of extreme degeneracy of the charged andneutral Higgsino, but provided M1 and M2 are below a TeV, itshould be reliable. This is essentially a limit on µ.
If the chargino and neutralino were exactly degenerate, this wouldalso imply the NLSP has a bound of 103 GeV. However, mixingwith the bino splits the neutral Higgsinos and allows a lighterNLSP. In particular, mχ0
1≈ 90 GeV can be achieved without
violating the chargino mass bound. Such a light NLSP will decayalmost entirely to photons, so the usual minimal GMSB searchesinvolving photons come into play.
Matthew Reece Phenomenology of Higgsino NLSPs
D0 Diphoton Search
DØ Note 5427-CONF
Version: 3.4
Search for GMSB in diphoton final states by D0 at√
s = 1.96 TeV
The DØ CollaborationURL http://www-d0.fnal.gov
(Dated: July 26, 2007)
We report results of a search for Supersymmetry (SUSY) with gauge-mediated breaking in dipho-ton events using 1100 ± 70 pb−1 of data collected by the D0 experiment at the Fermilab TevatronCollider in 2002–2006. No excess of events above the standard model background is found. We setthe most stringent limits for a standard benchmark model on the lightest neutralino and charginomass of about 126 and 231 GeV, respectively, at the 95% C.L.
PACS numbers: 14.80.Ly, 12.60.Jv, 13.85.Rm
www-d0.fnal.gov/Run2Physics/WWW/results/prelim/NP/N54/N54.pdf
D0 Note 5427-CONF
Matthew Reece Phenomenology of Higgsino NLSPs
D0 Diphoton Search: 95% Exclusion Implications
The limits here are on Pdecay , the probability that the neutralinodecays inside the detector. This is a conservative preliminary plot,as we haven’t yet taken into account some aspects of the D0 studythat should strengthen the limit (work in progress).
Matthew Reece Phenomenology of Higgsino NLSPs
Long Lifetimes at the Tevatron?
CDF has an EM timing system added in Run II, motivated by the(in?)famous eeγγ 6ET event. Measures arrival time of electrons andphotons with a resolution of about 0.6 ns.Search for long-lived neutralinos decaying to photons(γ + j + 6ET ): 0804.1043. Limit for bino of 101 GeV for 5 nslifetime, from 570 pb−1.
D0 does not do timing, but it does pointing. Fits shower positionin the EM calorimeter and the central preshower detector to obtaina distance of closest approach to the beamline within 2 cm.Search for long-lived particles decaying to electron or photon pairs:0806.2223
Matthew Reece Phenomenology of Higgsino NLSPs
CDF Search for Long-Lived Particles Decaying to Photonsar
Xiv
:07
04
.07
60
v1
[h
ep-e
x]
5 A
pr
20
07
Search for Heavy, Long-Lived Particles that Decay to Photons at CDF II
A. Abulencia,24 J. Adelman,13 T. Affolder,10 T. Akimoto,55 M.G. Albrow,17 S. Amerio,43 D. Amidei,35
A. Anastassov,52 K. Anikeev,17 A. Annovi,19 J. Antos,14 M. Aoki,55 G. Apollinari,17 T. Arisawa,57 A. Artikov,15
W. Ashmanskas,17 A. Attal,3 A. Aurisano,42 F. Azfar,42 P. Azzi-Bacchetta,43 P. Azzurri,46 N. Bacchetta,43
W. Badgett,17 A. Barbaro-Galtieri,29 V.E. Barnes,48 B.A. Barnett,25 S. Baroiant,7 V. Bartsch,31 G. Bauer,33
P.-H. Beauchemin,34 F. Bedeschi,46 S. Behari,25 G. Bellettini,46 J. Bellinger,59 A. Belloni,33 D. Benjamin,16
A. Beretvas,17 J. Beringer,29 T. Berry,30 A. Bhatti,50 M. Binkley,17 D. Bisello,43 I. Bizjak,31 R.E. Blair,2
C. Blocker,6 B. Blumenfeld,25 A. Bocci,16 A. Bodek,49 V. Boisvert,49 G. Bolla,48 A. Bolshov,33 D. Bortoletto,48
J. Boudreau,47 A. Boveia,10 B. Brau,10 L. Brigliadori,5 C. Bromberg,36 E. Brubaker,13 J. Budagov,15 H.S. Budd,49
S. Budd,24 K. Burkett,17 G. Busetto,43 P. Bussey,21 A. Buzatu,34 K. L. Byrum,2 S. Cabreraq,16 M. Campanelli,20
M. Campbell,35 F. Canelli,17 A. Canepa,45 S. Carilloi,18 D. Carlsmith,59 R. Carosi,46 S. Carron,34 B. Casal,11
M. Casarsa,54 A. Castro,5 P. Catastini,46 D. Cauz,54 M. Cavalli-Sforza,3 A. Cerri,29 L. Cerritom,31 S.H. Chang,28
Y.C. Chen,1 M. Chertok,7 G. Chiarelli,46 G. Chlachidze,17 F. Chlebana,17 I. Cho,28 K. Cho,28 D. Chokheli,15
J.P. Chou,22 G. Choudalakis,33 S.H. Chuang,52 K. Chung,12 W.H. Chung,59 Y.S. Chung,49 M. Cilijak,46
C.I. Ciobanu,24 M.A. Ciocci,46 A. Clark,20 D. Clark,6 M. Coca,16 G. Compostella,43 M.E. Convery,50 J. Conway,7
B. Cooper,31 K. Copic,35 M. Cordelli,19 G. Cortiana,43 F. Crescioli,46 C. Cuenca Almenarq,7 J. Cuevasl,11
R. Culbertson,17 J.C. Cully,35 S. DaRonco,43 M. Datta,17 S. D’Auria,21 T. Davies,21 D. Dagenhart,17
P. de Barbaro,49 S. De Cecco,51 A. Deisher,29 G. De Lentdeckerc,49 G. De Lorenzo,3 M. Dell’Orso,46 F. Delli Paoli,43
L. Demortier,50 J. Deng,16 M. Deninno,5 D. De Pedis,51 P.F. Derwent,17 G.P. Di Giovanni,44 C. Dionisi,51
B. Di Ruzza,54 J.R. Dittmann,4 M. D’Onofrio,3 C. Dorr,26 S. Donati,46 P. Dong,8 J. Donini,43 T. Dorigo,43
S. Dube,52 J. Efron,39 R. Erbacher,7 D. Errede,24 S. Errede,24 R. Eusebi,17 H.C. Fang,29 S. Farrington,30
I. Fedorko,46 W.T. Fedorko,13 R.G. Feild,60 M. Feindt,26 J.P. Fernandez,32 R. Field,18 G. Flanagan,48 R. Forrest,7
S. Forrester,7 M. Franklin,22 J.C. Freeman,29 I. Furic,13 M. Gallinaro,50 J. Galyardt,12 J.E. Garcia,46
F. Garberson,10 A.F. Garfinkel,48 C. Gay,60 H. Gerberich,24 D. Gerdes,35 S. Giagu,51 P. Giannetti,46 K. Gibson,47
J.L. Gimmell,49 C. Ginsburg,17 N. Giokarisa,15 M. Giordani,54 P. Giromini,19 M. Giunta,46 G. Giurgiu,25
V. Glagolev,15 D. Glenzinski,17 M. Gold,37 N. Goldschmidt,18 J. Goldsteinb,42 A. Golossanov,17 G. Gomez,11
G. Gomez-Ceballos,33 M. Goncharov,53 O. Gonzalez,32 I. Gorelov,37 A.T. Goshaw,16 K. Goulianos,50 A. Gresele,43
S. Grinstein,22 C. Grosso-Pilcher,13 R.C. Group,17 U. Grundler,24 J. Guimaraes da Costa,22 Z. Gunay-Unalan,36
C. Haber,29 K. Hahn,33 S.R. Hahn,17 E. Halkiadakis,52 A. Hamilton,20 B.-Y. Han,49 J.Y. Han,49 R. Handler,59
F. Happacher,19 K. Hara,55 D. Hare,52 M. Hare,56 S. Harper,42 R.F. Harr,58 R.M. Harris,17 M. Hartz,47
K. Hatakeyama,50 J. Hauser,8 C. Hays,42 M. Heck,26 A. Heijboer,45 B. Heinemann,29 J. Heinrich,45 C. Henderson,33
M. Herndon,59 J. Heuser,26 D. Hidas,16 C.S. Hillb,10 D. Hirschbuehl,26 A. Hocker,17 A. Holloway,22 S. Hou,1
M. Houlden,30 S.-C. Hsu,9 B.T. Huffman,42 R.E. Hughes,39 U. Husemann,60 J. Huston,36 J. Incandela,10
G. Introzzi,46 M. Iori,51 A. Ivanov,7 B. Iyutin,33 E. James,17 D. Jang,52 B. Jayatilaka,16 D. Jeans,51 E.J. Jeon,28
S. Jindariani,18 W. Johnson,7 M. Jones,48 K.K. Joo,28 S.Y. Jun,12 J.E. Jung,28 T.R. Junk,24 T. Kamon,53
P.E. Karchin,58 Y. Kato,41 Y. Kemp,26 R. Kephart,17 U. Kerzel,26 V. Khotilovich,53 B. Kilminster,39 D.H. Kim,28
H.S. Kim,28 J.E. Kim,28 M.J. Kim,17 S.B. Kim,28 S.H. Kim,55 Y.K. Kim,13 N. Kimura,55 L. Kirsch,6 S. Klimenko,18
M. Klute,33 B. Knuteson,33 B.R. Ko,16 K. Kondo,57 D.J. Kong,28 J. Konigsberg,18 A. Korytov,18 A.V. Kotwal,16
A.C. Kraan,45 J. Kraus,24 M. Kreps,26 J. Kroll,45 N. Krumnack,4 M. Kruse,16 V. Krutelyov,10 T. Kubo,55
S. E. Kuhlmann,2 T. Kuhr,26 N.P. Kulkarni,58 Y. Kusakabe,57 S. Kwang,13 A.T. Laasanen,48 S. Lai,34 S. Lami,46
S. Lammel,17 M. Lancaster,31 R.L. Lander,7 K. Lannon,39 A. Lath,52 G. Latino,46 I. Lazzizzera,43 T. LeCompte,2
E. Lee,53 J. Lee,49 J. Lee,28 Y.J. Lee,28 S.W. Leeo,53 R. Lefevre,20 N. Leonardo,33 S. Leone,46 S. Levy,13
J.D. Lewis,17 C. Lin,60 C.S. Lin,17 M. Lindgren,17 E. Lipeles,9 A. Lister,7 D.O. Litvintsev,17 T. Liu,17
N.S. Lockyer,45 A. Loginov,60 M. Loreti,43 R.-S. Lu,1 D. Lucchesi,43 P. Lujan,29 P. Lukens,17 G. Lungu,18
L. Lyons,42 J. Lys,29 R. Lysak,14 E. Lytken,48 P. Mack,26 D. MacQueen,34 R. Madrak,17 K. Maeshima,17
K. Makhoul,33 T. Maki,23 P. Maksimovic,25 S. Malde,42 S. Malik,31 G. Manca,30 F. Margaroli,5 R. Marginean,17
C. Marino,26 C.P. Marino,24 A. Martin,60 M. Martin,25 V. Marting,21 M. Martınez,3 R. Martınez-Balların,32
T. Maruyama,55 P. Mastrandrea,51 T. Masubuchi,55 H. Matsunaga,55 M.E. Mattson,58 R. Mazini,34 P. Mazzanti,5
K.S. McFarland,49 P. McIntyre,53 R. McNultyf ,30 A. Mehta,30 P. Mehtala,23 S. Menzemerh,11 A. Menzione,46
P. Merkel,48 C. Mesropian,50 A. Messina,36 T. Miao,17 N. Miladinovic,6 J. Miles,33 R. Miller,36 C. Mills,10
M. Milnik,26 A. Mitra,1 G. Mitselmakher,18 A. Miyamoto,27 S. Moed,20 N. Moggi,5 B. Mohr,8 C.S. Moon,28
3
24University of Illinois, Urbana, Illinois 6180125The Johns Hopkins University, Baltimore, Maryland 21218
26Institut fur Experimentelle Kernphysik, Universitat Karlsruhe, 76128 Karlsruhe, Germany27High Energy Accelerator Research Organization (KEK), Tsukuba, Ibaraki 305, Japan
28Center for High Energy Physics: Kyungpook National University,Taegu 702-701, Korea; Seoul National University, Seoul 151-742,
Korea; SungKyunKwan University, Suwon 440-746, Korea29Ernest Orlando Lawrence Berkeley National Laboratory, Berkeley, California 94720
30University of Liverpool, Liverpool L69 7ZE, United Kingdom31University College London, London WC1E 6BT, United Kingdom
32Centro de Investigaciones Energeticas Medioambientales y Tecnologicas, E-28040 Madrid, Spain33Massachusetts Institute of Technology, Cambridge, Massachusetts 02139
34Institute of Particle Physics: McGill University, Montreal,Canada H3A 2T8; and University of Toronto, Toronto, Canada M5S 1A7
35University of Michigan, Ann Arbor, Michigan 4810936Michigan State University, East Lansing, Michigan 48824
37University of New Mexico, Albuquerque, New Mexico 8713138Northwestern University, Evanston, Illinois 6020839The Ohio State University, Columbus, Ohio 43210
40Okayama University, Okayama 700-8530, Japan41Osaka City University, Osaka 588, Japan
42University of Oxford, Oxford OX1 3RH, United Kingdom43University of Padova, Istituto Nazionale di Fisica Nucleare,
Sezione di Padova-Trento, I-35131 Padova, Italy44LPNHE, Universite Pierre et Marie Curie/IN2P3-CNRS, UMR7585, Paris, F-75252 France
45University of Pennsylvania, Philadelphia, Pennsylvania 1910446Istituto Nazionale di Fisica Nucleare Pisa, Universities of Pisa,
Siena and Scuola Normale Superiore, I-56127 Pisa, Italy47University of Pittsburgh, Pittsburgh, Pennsylvania 15260
48Purdue University, West Lafayette, Indiana 4790749University of Rochester, Rochester, New York 14627
50The Rockefeller University, New York, New York 1002151Istituto Nazionale di Fisica Nucleare, Sezione di Roma 1,
University of Rome “La Sapienza,” I-00185 Roma, Italy52Rutgers University, Piscataway, New Jersey 08855
53Texas A&M University, College Station, Texas 7784354Istituto Nazionale di Fisica Nucleare, University of Trieste/ Udine, Italy
55University of Tsukuba, Tsukuba, Ibaraki 305, Japan56Tufts University, Medford, Massachusetts 02155
57Waseda University, Tokyo 169, Japan58Wayne State University, Detroit, Michigan 48201
59University of Wisconsin, Madison, Wisconsin 5370660Yale University, New Haven, Connecticut 06520
(Dated: February 1, 2008; Version 5.1)
We present the first search for heavy, long-lived particles that decay to photons at a hadroncollider. We use a sample of γ+jet+missing transverse energy events in pp collisions at
√s =
1.96 TeV taken with the CDF II detector. Candidate events are selected based on the arrival timeof the photon at the detector. Using an integrated luminosity of 570 pb−1 of collision data, weobserve 2 events, consistent with the background estimate of 1.3±0.7 events. While our searchstrategy does not rely on model-specific dynamics, we set cross section limits in a supersymmetricmodel with eχ0
1 → γ eG and place the world-best 95% C.L. lower limit on the eχ01 mass of 101 GeV/c2
at τχ01
= 5 ns.
PACS numbers: 13.85.Rm, 12.60.Jv, 13.85.Qk, 14.80.Ly
∗With visitors from aUniversity of Athens, bUniversity ofBristol, cUniversity Libre de Bruxelles, dCornell University,eUniversity of Cyprus, f University of Dublin, gUniversity of Ed-inburgh, hUniversity of Heidelberg, iUniversidad Iberoamericana,
jUniversity of Manchester, kNagasaki Institute of Applied Science,lUniversity de Oviedo, mUniversity of London, Queen Mary Col-lege, nUniversity of California Santa Cruz, oTexas Tech University,
0704.0760
Matthew Reece Phenomenology of Higgsino NLSPs
Cuts in the CDF Search5
Preselection Requirements Cumulative (individual)Efficiency (%)
EγT > 30 GeV, E/T > 30 GeV 54 (54)
Photon ID and fiducial, |η| < 1.0 39 (74)*Good vertex,
Ptracks pT > 15 GeV/c 31 (79)
|ηjet| < 2.0, EjetT > 30 GeV 24 (77)
Cosmic ray rejection 23 (98)*Requirements after OptimizationE/T > 40 GeV, Ejet
T > 35 GeV 21 (92)∆φ(E/T , jet) > 1 rad 18 (86)2 ns < tγ
c < 10 ns 6 (33)
TABLE I: The data selection criteria and the cumulativeand individual requirement efficiencies for an example GMSBmodel point at mχ0
1= 100 GeV/c2 and τχ0
1= 5 ns. The ef-
ficiencies listed are, in general, model-dependent and have afractional uncertainty of 10%. Model-independent efficienciesare indicated with an asterisk. The collision fiducial require-ment of |zi| < 60 cm is part of the good vertex requirement(95%) and is estimated from data.
backgrounds come from cosmic rays and beam effectsthat can produce photon candidates, E/T , and sometimesthe reconstructed jet. We separate data events as a func-tion of tγc into several control regions that allow us toestimate the number of background events in the finalsignal region by fitting to the data using collision andnon-collision shape templates as shown in Fig. 1.
Collision photons are subdivided in two subclasses:correct and incorrect vertex selection [13]. An incorrectvertex can be selected when two or more collisions occurin one beam bunch crossing, making it possible that thehighest reconstructed
∑tracks pT vertex does not produce
the photon. While the fraction of events with incorrectvertices depends on the final event selection criteria, thetγc distribution for each subclass is estimated separatelyusing W → eν data where the electron track is droppedfrom the vertexing. For events with a correctly associ-ated vertex, the tγc distribution is Gaussian and centeredat zero with a standard deviation of 0.64 ns [13]. Forthose with an incorrectly selected vertex the tγc distribu-tion is also Gaussian with a standard deviation of 2.05 ns.
The tγc distributions for both non-collision backgroundsare estimated separately from data using events withno reconstructed tracks. Photon candidates from cos-mic rays are not correlated in time with collisions, andtherefore their tγc distribution is roughly flat. Beam halophoton candidates are produced by muons that origi-nate upstream of the detector (from the p direction) andtravel through the calorimeter, typically depositing smallamounts of energy. When the muon deposits significantenergy in the EM calorimeter, it can be misidentified as aphoton and cause E/T . These photons populate predomi-nantly the negative tγc region, but can contribute to thesignal region. Since beam halo muons travel parallel tothe beam line, these events can be separated from cosmic
ray events by identifying the small energy deposited inthe calorimeter towers along the beam halo muon trajec-tory.
The background prediction uses control regions out-side the signal time window but well within the 132 nstime window that the calorimeter uses to measure theenergy. The non-collision background templates are nor-malized to match the number of events in two time win-dows: a beam halo-dominated window at {−20, −6} ns,selected to be 3σ away from the wrong vertex collisionbackground, and a cosmic rays-dominated window at{25, 90} ns, well away from the standard model andbeam halo contributions. The collision background isestimated by fitting events in the {−10, 1.2} ns windowwith the non-collision contribution subtracted and withthe fraction of correct to incorrect vertex events allowedto vary. In this way the background for the signal regionis entirely estimated from data samples. The systematicuncertainty on the background estimate is dominated byour ability to calibrate the mean of the tγc distributionfor prompt photons. We find a variation of 200 ps onthe mean and 20 ps on the standard deviation of the dis-tribution by considering various possible event selectioncriteria. These contribute to the systematic uncertaintyof the collision background estimate in the signal regionand are added in quadrature with the statistical uncer-tainties of the final fit procedure.
We estimate the sensitivity to heavy, long-lived parti-cles that decay to photons using GMSB models for dif-ferent χ0
1 masses and lifetimes. Events from all SUSYprocesses are simulated with the pythia Monte Carloprogram [16] along with the detector simulation [17]. Theacceptance is the ratio of simulated events that pass allthe requirements to all events produced. It is used inthe optimization procedure and in the final limit settingand depends on a number of effects. The fraction of χ0
1
decays in the detector volume is the dominant effect onthe acceptance. For a given lifetime this depends on theboost of the χ0
1. A highly boosted χ01 that decays in
the detector typically does not contribute to the accep-tance because it tends to produce a photon traveling inthe same direction as the χ0
1. Thus, the photon’s arrivaltime is indistinguishable from promptly produced pho-tons. At small boosts the decay is more likely to happeninside the detector, and the decay angle is more likelyto be large, which translates into a larger delay for thephoton. The fraction of events with a delayed photon ar-rival time initially rises as a function of χ0
1 lifetime, butfalls as the fraction of χ0
1’s decaying outside the detectorbegins to dominates. In the χ0
1 mass region considered(65 ≤ mχ0
1≤ 150 GeV/c2), the acceptance peaks at a
lifetime of around 5 ns. The acceptance also depends onthe mass as the boost effects are mitigated by the abilityto produce high energy photons or E/T in the collision, asdiscussed in Ref. [8].
The total systematic uncertainty of 10% on the ac-
0704.0760Matthew Reece Phenomenology of Higgsino NLSPs
Clean γ + 6ET?
With light, largely Higgsino NLSP, and long lifetime, can havesome events where the signal is one hard photon and no other hardobjects. Very clean – is it a good search channel at the Tevatron?
SM physics backgrounds include the irreducible Z (→ νν) + γ,W → eν with e → γ fake, γγ with one lost photon.
Instrumental backgrounds: beam halo, cosmic rays.
Matthew Reece Phenomenology of Higgsino NLSPs
Non-collision backgrounds
Peter Wagner’s CDF thesis (August 2007, Texas A&M) is a goodreference for learning about these issues.
Beam halo comes from interactions of beam bunches with materialnear the beam pipe, before reaching the detector. This canproduce muons moving roughly parallel to the proton beam, whichpass through several calorimeter cells. One cell may get a largedeposit and become a photon candidate. (Typically at early times.)
Cosmic rays, on the other hand, will sometimes interact in the EMcalorimeter and produce a fake photon, without leaving significantamounts of energy elsewhere in the detector. These will have flattiming distributions.
Matthew Reece Phenomenology of Higgsino NLSPs
Early/late asymmetry as indication of new physics?
Suppose the early-time beam halo events can be filtered outentirely from the way they deposit energy in the calorimeter.
Physics backgrounds will have a Gaussian timing distribution. (By“timing”, we mean time the hit is registered in EM Timing versustime it should take a photon to propagate from the vertex.)Cosmics will have a flat timing distribution.
New physics could occur at positive timing, because the pathlength for a neutralino to its decay position plus the photon fromthe decay position to the calorimeter is longer than the direct pathfrom the vertex to the calorimeter. Is (late - early)/(late + early) anew physics analyzer?
Matthew Reece Phenomenology of Higgsino NLSPs
Wrong vertices
These are relatively clean events, and a good photon candidate hasno associated track. So the vertex must be found from tracks fromthe underlying event. If a min-bias event happening elsewhere inthe detector has a higher
∑|pT |, the event might be
reconstructed as if that is the true vertex.
How often does this happen? Underlying event is usuallysomewhat harder than min bias, but at high instantaneousluminosity we can have a few min bias events happening at once,each with some probability to have higher
∑|pT |. Assuming
Pythia (Rick Field’s Tune A) is modeling this correctly for theTevatron, can happen 25-30% of the time.
(This is of course an experimental question; we can only give arough guide to whether it’s important!)
Matthew Reece Phenomenology of Higgsino NLSPs
Vertex distribution at the Tevatron
The z vertex distribution at the Tevatron is broad (easily givesnanosecond delays). Larger time delays correlate with largermeasured ET , so a wrong vertex allow more events to pass cuts!(z vtx distribution at LHC is much narrower.)
10
Vertex Time (ns)-6 -4 -2 0 2 4 6
Eve
nts/
0.15
ns
0
1000
2000
3000
4000
5000
6000
(a)
Vertex Z Position (cm)-100 -50 0 50 100
Eve
nts/
2.5
cm
0
1000
2000
3000
4000
5000
(b)
Vertex Z Position (cm)-40 -20 0 20 40
Mea
n V
erte
x Ti
me
(ns)
-0.2
-0.1
0
0.1
0.2
(c)
FIG. 8: Plots (a), (b), and (c) show the t0, z0, and their cor-relation respectively for the reconstructed highest ΣpT vertexin W → eν events. The fits in (a) and (b) are both a singleGaussian. The falloff in the (b) at |z| " 60 cm is due to therequirement that all tracks have |z| < 70 cm. In the searchthe vertex is required to have |z| < 60 cm.
T (ns)∆-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
Eve
nts/
0.07
ns
1
10
210
310
410
(a)
Z (cm)∆-1.5 -1 -0.5 0 0.5 1 1.5
Eve
nts/
0.04
cm
1
10
210
310
410
(b)
FIG. 9: The difference in t and z between two arbitrarilyselected sets of tracks from the same reconstructed vertex ina W → eν dataset with the electron track removed from thevertexing. This is a measure of the vertex resolution. (a) isfit with one Gaussian while (b) is fit with two. Note that thefactor of
√2 is already taken out.
the electron does not originate from the highest ΣpT ver-tex in the event.
The efficiency of the vertex reconstruction algorithm isinvestigated using two separate methods. The efficiencyas a function of the number of tracks is determined byselecting events that contain a cluster with a high trackmultiplicity. Next, various random subsets of the tracksare taken that belong to this cluster to see if they alonecould produce a cluster. Figure 11 shows the ratio ofsubset samples in which a cluster is reconstructed to allcases tried for a given set of tracks as a function of thenumber of tracks in the various subsets. The algorithm isover 90% efficient if 4 tracks are present, where the ineffi-
True Time
Q True
False Time
Q False
CDF vertex distribution from 0804.1043.
Matthew Reece Phenomenology of Higgsino NLSPs
Challenging? Timing in W → eν, with e → γ
The upshot is that even the background distributions will be veryasymmetric between early and late times. The amount ofasymmetry depends on cuts; e.g. for W → eν, the electron pT hasan edge at 40 GeV. So if a cut is placed on photon ET above 40GeV, it will be selecting wrong-vertex events and enhancing theasymmetry in the background.
One outcome of all this is that re-interpreting a CDF analysis ofγ + 6ET with timing as a constraint on the Higgsino NLSP scenariois an almost hopeless task for a theorist. We are trying to convincesome of the relevant experimentalists to set a limit on (at leastsome slice of the parameter space of) this model. Our clean searchchannel is not so clean!
Matthew Reece Phenomenology of Higgsino NLSPs
Delayed Z Boson Events
While the details differ and work remains to be done to understandthe exclusion contours, in the case of delayed (non-pointing)photons there is a substantial amount of literature to draw on.
On the other hand, with Higgsino NLSP one can have delayed,non-pointing Z or Higgs bosons, which have been studied less.Understanding what these events look like in the detector and whatwe can learn from them is interesting and potentially challenging.
I’ll discuss one of the most accessible cases: a delayed Z thatdecays to e+e−. Understanding what to do with these eventsrequires some detailed discussion of the ATLAS detector.
Matthew Reece Phenomenology of Higgsino NLSPs
The ATLAS DetectorAT
LAS
Technical Design R
eportC
alorimeter P
erformance
13 January 1997
2 P
erformance for electrons and photons
91
Figure2-i
Longitu
din
al v
iew
of a
quadra
nt o
f the E
M c
alo
rimete
r.
BARREL ENDCAP
WHEELOUTER
INNERI N N E R D E T E C T O R
B = 2 T
warm wallAl cryostat
Al cryostatwallswarm
(isogrid)cold wallAl cryostat
coldPresampler
superconductingsolenoid coil
scintillatorID services+cables
feed
thro
ugh
1 m
2 m 4 m
Pb(1.5mm) Pb(1.1mm)2.10cm/X0 2.65cm/X0
Pb(1.7mm)
Pb(2.2mm)
=0.8
=1.375=1.475
=1.68 =1.8
=2.5
=3.2
!
! ! ! !
!
!
Matthew Reece Phenomenology of Higgsino NLSPs
ATLAS Electromagnetic Calorimeter
The ATLAS electromagnetic calorimeter uses LAr and lead. In thebarrel it extends to |η| < 1.475, while the endcap covers1.375 < |η| < 3.2. The fast response time of LAr allows precisiontiming, used to reject pile-up and to detect long-lived particles.
Matthew Reece Phenomenology of Higgsino NLSPs
ATLAS ECAL Granularity
ATLAS Technical Design ReportCalorimeter Performance 13 January 1997
92 2 Performance for electrons and photons
Figure 2-ii Readout granularity of the EM calorimeter.
!" = 0.0245
!# = 0.02537.5mm/8 = 4.69 mm�!# = 0.0031
!"=0.0245x4 36.8mmx4�=147.3mm
Trigger Tower
TriggerTower!" = 0.0982
!# = 0.1
16X0
4.3X0
2X0
1500
mm
470
mm
#
"
# = 0
Strip towers in Sampling 1
Square towers in �Sampling 2
1.7X0
Towers in Sampling 3�!"$ !# = 0.0245$ 0.05
Matthew Reece Phenomenology of Higgsino NLSPs
ATLAS ECAL: Key Numbers
The basic measurements made by the ECAL are:
Energy: resolution δE/E ∼ 10%/√
E/GeV ⊕ 0.7%
Position in η, ϕ: resolution ση = 0.002, σϕ = 0.004
Direction in η: σθ = 0.060/√
E/GeVArrival time: σt = 100 ps
The use of these quantities for precision mass determination inordinary gauge mediation, using events with leptons andnonpointing photons, has been discussed by Kawagoe, Kobayashi,Nojiri, and Ochi (hep-ph/0309031).
Matthew Reece Phenomenology of Higgsino NLSPs
ATLAS Vertexing
The beam spot is essentially Gaussian with σz = 5.6 cm(σx ,y = 15µm). We would like to know the vertex position muchmore precisely for this study.
The ATLAS TDR contains a range of estimates for the precision ofthe primary vertex, which depends on physics process and onluminosity. Pile-up, obviously, makes the issue more difficult.
For now we’ll go to the pessimistic end of the TDR range andsmear the vertex with a Gaussian of width 100 µm. Pile-up couldmake this too optimistic, but this is just a first estimate....
Matthew Reece Phenomenology of Higgsino NLSPs
ATLAS Tracking
TRT: straws parallel to the beamline give accurate informationabout direction in the (r , ϕ) plane.
Software can find photons that convert. Can this be adapted tolook for displaced Z vertices? Need to be sure not to restrict tothings that point back to the beamline.
I won’t use this information in my reconstruction, but it should beused: it’s redundant information, to some extent, but doing a fit toall the information we have should help overcome limitations fromexperimental resolutions.
Matthew Reece Phenomenology of Higgsino NLSPs
A Sample Point
I’m going to run through an example of some events. The pointchosen is M1 = 320 GeV, M2 = 640 GeV, µ = 140 GeV,tanβ = 20, mG = 25 eV, and for simplicity all squarks, sleptons,and the gluino are decoupled so that we just focus on productionof charginos and neutralinos for now.
Chargino/Neutralino Masses
mχ01
= 134.0
mχ02
= −150.5
mχ03
= 324.0
mχ04
= 702.0
mχ±1= 142.6
mχ±1= 702.0
Matthew Reece Phenomenology of Higgsino NLSPs
Reconstructing the decay vertex
We would like to solve for the decay vertex position (xd , yd , zd)and time td . We assume the two particles that gave us the signalin the ECAL are massless, so we have two equations
c(ti − td) = |xi − xd |2 (3)
The pointing measurement tells us zi−zd√(xi−xd )2+(yi−yd )2
. These four
equations allow us to solve for (xd , yd , zd , td).
Discrete ambiguities are reduced by demanding that td < ti . Afurther reduction comes from noting that we can compute thevelocity of the neutralino:
(vx , vy , vz) =
(xd
ctd,
yd
ctd,zd − zvtx
ctd
), (4)
which must square to a number less than one.Matthew Reece Phenomenology of Higgsino NLSPs
Reconstructing the Higgsino mass
Reconstructing the decay vertex position and time is alreadyinteresting, as we can try to infer from it the neutralino lifetime andhence the parameter F characterizing the scale of SUSY breaking.
In fact there is more that we can do; as we already noted we knowthe neutralino velocity (vx , vy , vz)χ, so the only unknown quantityin its 4-momentum is the energy Eχ. If we assume a masslessgravitino, we have:
m2G
= (Eχ − E1 − E2)2 − (Eχvχ − p1 − p2)
2 = 0, (5)
and we can solve for Eχ and use it to compute mχ, up to quadraticambiguity.
Matthew Reece Phenomenology of Higgsino NLSPs
Higgsino mass results: before smearing
Even before smearing, have some spread from radiation in Pythia.Also note the unphysical low-mass solutions for mχ (it’s below theZ mass, so clearly nonsensical!)
Matthew Reece Phenomenology of Higgsino NLSPs
Higgsino mass results: after smearing
With all observables smeared by the appropriate Gaussians, theresult is not sharp, but there is a cluster of results near the correctanswer 134 GeV.
We still haven’t used the ϕ direction information from tracking, soI’m optimistic that this can be cleaned up somewhat.
Matthew Reece Phenomenology of Higgsino NLSPs
Error in Reconstructed Decay Time
We would like to be able to use the reconstructed vertex and timeto understand the χ0
1 lifetime and hence the gravitino mass. Here’sthe fractional error trecon−tsim
tsim:
Percentage errors can be large, but many are within 20%. Again,need to try to clean this up, doing a full fit with tracking.
Matthew Reece Phenomenology of Higgsino NLSPs
Conclusions
The phenomenology of Higgsino NLSP is not as extensivelyexplored as bino and stau, and much work remains to be done.
The case with long-lived NLSP gives interestingphenomenology, with delayed photon, Z, or Higgs. Thedelayed photon case can be distinct from ordinary GMSBbecause the sleptons may not be light, and Higgsinos aredirectly produced.
The ATLAS electromagnetic calorimeter gives precisiondirection and timing information that could play a key role inunderstanding the mass spectrum and the SUSY breakingscale.
Much more fun to be had.
Matthew Reece Phenomenology of Higgsino NLSPs