Four tops and the t tbar forward-backward asymmetry

0.16 MB·English

by  J. A. Aguilar-Saavedra

Four tops and the tt¯forward-backward asymmetry
J. A. Aguilar-Saavedra and J. Santiago
CAFPE and Departamento de Física Teórica y del Cosmos,
Universidad de Granada, E-18071 Granada, Spain
¯
New colour octet vectors below the TeV scale could explain the anomalous tt forward-backward
asymmetry observed at the Tevatron experiments, while being consistent with the current LHC
data. These models generally lead to four-top final states at the LHC at observable levels. We
compute the four-top production cross section at the LHC in a model with a massive colour octet
vector as a function its mass, its width and its coupling to the top quark. Octet masses in the
¯
vicinity of the tt threshold are generally excluded by present limits on the production of same-sign
dileptons and trileptons. Masses above 650 GeV are allowed, quiteindependently of the couplings,
but they can be probed with the luminosity of 5 fb−1 already collected at the LHC, up to around
2
800 GeV. The four-top production cross section is increased by a factor 2 with √s=8 TeV and
1 by up to almost two orders of magnitude with √s = 14 TeV, thus greatl∼y increasing the reach for
0
massive gluons after theLHC energy upgrade.
2

n
a Duetoitslargemass,thetopquarkisexpectedtoplay SM but are difficult to reconstruct completely (see for
J
arelevantroleinthediscoveryofnewphysicsbeyondthe instance section 12 of [26] and [27]). Here we show that

5 StandardModel(SM).Thefirsthintofsuchnewphysics simplersearches,basedonproductionofsame-signdilep-
2
couldbealreadyavailableintheformoftheanomalously tons and trileptons, are enough to probe and constrain

large tt¯forward-backward (FB) asymmetry observed at models of light gluons as an explanation of the tt¯asym-
]
h both Tevatron experiments [1, 2]. The fact that neither metry. Specifically,toestimatethepresentlimitsonfour-
p the Tevatronnor the LargeHadronCollider (LHC) have top production at the LHC we use (i) a supersymmetry-
-
observed any other anomaly in top or jet physics sets motivated search [28]; (ii) a search for fourth generation
p
′
e strong constraints on possible explanations in terms of b quarks [29], both performed by the CMS Collabora-
h new physics [3–5]. One of the few surviving explana- tion. The present analysis is of course relevant to any
[ tions, compatible with the present measurements of the model with color octect vector resonances that couple

2 tt¯invariant mass spectrum [6] and the charge asymme- strongly to the top quark and not only to the ones at-
v try at the LHC [7, 8], is a relatively light colour octet tempting to explain the top FB asymmetry.
8 vector boson (called here ‘gluon’ for brevity) with mass Let us consider a new massive gluon G. Its couplings
7 M . 1 TeV and with suppressed axial-vector couplings to SM gluons are fixed by gauge invariance whereas the
7 to the light quarks and sizeable axial-vector couplings ones to fermions gV,A are in principle free. The relevant
3 i
to top quarks [9–17]. The axial coupling ensures can- Lagrangianis
.
2 celation of the interference terms between the SM and
1 new physics contribution [18–20] in symmetric observ- G = 1D Ga DµGaν DνGaµ + 1M2GaGaµ
1 ables while preserving the contribution to the asymme- L −2 µ ν(cid:16) − (cid:17) 2 µ
v:1 ttroyp. qMuaasrskesanardouanldar1geTegVluorenquwiirdetlha,rguesucaolulypliwnigtshteoxtthrae +ψ¯iγµGaµλ2ahgiV +giAγ5iψi, (1)
Xi decay channels [11]. Masses close to the tt¯threshold can where i is a flavor index, λa are the Gell-Mann matrices
easilyhide inthe largeSM tt¯background,althoughthey
r and
a may also need extra decay channels to be invisible [12].
Masses lighter than the tt¯threshold can do with smaller D Ga ∂ Ga+g fabcgbGc
couplings and are essentially invisible in symmetric ob- µ ν ≡ µ ν s µ ν
servables [14]. is the SM covariant derivative, with a = 1,...,8, ga the
µ
In this paper we consider an alternative, yet unex- SMgluons,fabc theSU(3)structureconstantsandgs the
plored probe of these models. The massive gluon is a strongcouplingconstant. Inordertocontributetothett¯
colour octect vector resonance, thus its couplings to SM asymmetry, G must have non-vanishing couplings to the
gluons are fixed by gauge invariance. Due to the rela- topandfirstgenerationquarks,beingtheFBasymmetry
tively low masses relevant for the FB asymmetry, pair in qq¯→ tt¯proportional to the product gqAgtA, with q =
productionof suchobjects with subsequentdecayin two u,d. Searches for dijet resonancestypically constrain gA
q
top pairs can receive a fairly large cross section, which (aswellasthevectorcoupling)toberelativelysmall,the
is further increased by non-resonant contributions and precise bound depending on the gluon mass M (see for
by single gluon resonant production, especially if the instance[10,14]). Thisimpliesthattheaxialcouplingto
coupling of the new gluon to the top quark is sizeable. the topmustbe oforderunity orevenlarger,inorderto
(See [21–25] for preliminary studies of colour octet pair generate a sizeable asymmetry. For gluon masses above
production at hadron colliders followed by top decays.) the tt¯threshold, M 2m , pair production of massive
t
≥
Four-topfinalstateshaveaverysmallbackgroundinthe gluons followedby decays into top pairs is a largesource
2
of four-top final states, see Fig. 1 (left). Non-resonant H 400 GeV, where H is the scalar sum of the
T T
• ≥
diagrams such as the one depicted in the right panel, in p of all jets. Only those with p > 40 GeV and
T T
whichthe new gluonsarenot producedon-shell,are also η <2.5 are considered here.
| |
important for larger values of the gluon coupling to the
topquark,anddominatebothbelowthett¯thresholdand Missing energy(cid:0)ET 50 GeV.
• ≥
at large gluon masses.
The global efficiency of these cuts for our four-top sig-
nal, including the same-sign dilepton branching ratio, is
approximately of 2% for a wide range of heavy gluon
masses. (Requiring H 200 GeV and(cid:0)E 120 GeV
T T
≥ ≥
results in an efficiency only slightly smaller.) With this
selection,theCMSCollaborationmeasures7eventswith
−1
an integrated luminosity L = 0.98 fb , for a SM back-
ground prediction of 5.3 2.4 [28]. For the trilepton
±
channel we ask for
FIG.1: Samplediagramsforresonant(left)andnon-resonant
(right)contributiontofour-topproductioninthepresenceof three leptons ℓ=e,µ with pT >20 GeV and η <
• | |
newheavy gluons. Thethick linecorresponds to themassive 2.4;same-flavouropposite-chargepairsarerequired
gluon. to be outside a window M m <10 GeV (m
Z ll ll
| − |
is the invariant mass of the two leptons).
Two jets with p > 25 GeV and η < 2.4, at least
T
Given the fact that gA,V gA, the cross section for • | |
q ≪ t one b-tagged.
four-topfinalstatesisessentiallyindependentofthecou-
pling to light quarks. For definiteness, we take a purely The scalar sum H + pℓ +(cid:0)E 50 must be
axial coupling g gA = 0.2 to light quarks, which is • larger than 500 GeTV. Pℓ T T ≥
q ≡ q
around the upper limit for a wide range of heavy gluon
masses [14], and a right-handed one g /2 gA = gV to With such cuts the efficiency for our four-top signal is
t ≡ t t of 0.6%. With this selection, the CMS Collaboration
thetopquark,aspreferredbyB physicsconstraints[32].
−1
measures one event with a luminosity L=1.16 fb , for
(Setting g to zero the four-top cross section found is
q
an expected SM backgroundof 0.16 0.09 [29].
nearly identical in all mass range, except for a slightly
±
Upper bounds onfour-topproductioncanbe obtained
steeper rise atthe M 2m threshold.) The coupling to
t
∼ from either of these channels, as well as from their
the second generation is also constrained to be small by
combination, using the modified frequentist likelihood
dijet production and has even a smaller effect on our re-
method [30, 31]. These limits are evaluated using 106
sults. Forsimplicity,itissettozero. Ontheotherhand,
pseudo-experiments of the expected signal and back-
thefour-topcrosssectiondependsonthegluonmassand
ground samples. Statistical uncertainty effects are im-
itscouplingtothetopquark. Incasethatadditionalnew
plementedassumingGaussiandistributions[30]. Theob-
particles exist, the four-top cross section near and above
the tt¯ threshold also depends on the partial width for tained 95% CL bound on four-top production are
gluon decays into these new particles.
In order to test the sensitivity of existing analyses to σ4t ≤0.50 pb (2l),
four-top production, we have implemented our model in σ4t 0.70 pb (3l),
≤
MADGRAPH 5 [33] using FeynRules [34]. The matrix el- σ4t 0.36 pb (combined). (2)
≤
ement generated by MADGRAPH has been implemented in
Protos [35] for an efficient exploration of the model pa- As we have mentioned, the four-top cross section cru-
rameter space and computation of four-top production cially depends on whether the new gluon can decay to
cross sections. We have generated events for different additional non-SM particles. Thus, a detailed discussion
configurations of gluon masses and couplings for pp col- of the heavy gluon width is compulsory. Let us denote
lisions at a centre of mass (CM) energy √s = 7 TeV by Γ0 the partial width of the gluon to SM particles.
and passedthem through PYTHIA[36] andPGS4 [37]. All Below the M 2mt threshold, Γ0 receives the largest
our simulations are performed at leading order. For the contributionfro∼mdecaysG uu¯,dd¯,withasmallerone
same-sign dilepton final state we have applied the selec- fromfour-bodydecaysG →W+bW−¯b. (Atanyrate,for
→
tionandkinematicalcuts inRef.[28]andfoundthatthe masses M 320 GeV the four-top cross section is prac-
≤
analysis most sensitive to four-top production is the one tically independent of Γ, as we will explicitly see below.)
requiring Abovethisthreshold,Γ0 islargelydominatedbyon-shell
decays to tt¯. This is clearly seen in Fig. 2, in which we
two same-sign leptons ℓ±ℓ±, ℓ = e,µ with pseudo- plotΓ0 asafunctionofM,forfivevaluesofthecoupling
•
rapidity η < 2.4. Electrons must have transverse to the top quark g =1,2,3,4,5.
t
| |
momentum p > 10 GeV and for muons p > 5 We consider in first place models in which the new
T T
GeV is required. gluon only decays to SM particles, that is, Γ = Γ0.
3
103 fb−1 dataset, larger masses will be excluded. For exam-
ple, an upper limit σ4t < 0.1 pb (slightly better than a
102 gt = 5 naive 1/√L rescaling) seems likely, especially bearing in
mindthepossibilityofcombinationwiththesemileptonic
channel. Such limit will allow to probe gluon masses up
V) 10 to M 800 GeV.
e ∼
G Realistic models explaining the FB asymmetry with
(0 1 gt = 1 a new gluon above the tt¯ threshold often require new
G particles to enhance the gluon width, Γ > Γ0, so as to
maketheresonanceinvisibleinthett¯invariantmassspec-
10-1 trum [11, 14]. In this case, the four top cross section de-
creasesbyafactorR (Γ0/Γ)2,butnotexactlyequalto
∼
10-2 this ratioofwidths because ofthe contributionsfromdi-
100 200 300 400 500 600 700 800 900 1000 agramswithnon-resonantGexchange. WeplotinFig.4
M (GeV)
theratioofcrosssectionsfordifferentgluontotalwidths,
FIG. 2: Partial width of the heavy gluon to SM final states. σ4t Γ
The five lines, from bottom to top, correspond to gt = RΓ = | , (3)
1,2,3,4,5. σ4t|Γ0
for Γ=2Γ0,4Γ0. We only consider gt =4,5, since these
heavygluonmassesrequirealargetopcouplingtogener-
103 ate the FB asymmetry. We observe that this ratio devi-
G = G
102 0 0.5
10
gt = 5 0.4 gt= 5
b) 1 Estimated upper limit
p
(
10-1 0.3 G = 2G
s 0
RG
10-2
0.2
g = 5
10-3 g = 1 t
t 7 TeV
0.1 G = 4G
10-4100 200 300 400 500 600 700 800 900 1000 0 g = 4
M (GeV) t 7 TeV
0
400 500 600 700 800 900 1000
FIG. 3: Four-top cross section for Γ = Γ0 (decays to SM M
particles only) for the LHC with √s=7 TeV.The fivelines,
from bottom to top, correspond to gt =1,2,3,4,5. FIG. 4: Ratio RΓ in Eq. (3) between cross sections for dif-
ferent values of the gluon total width Γ, for M 2mt. The
≥
error bars represent the Monte Carlo uncertainty.
The four-top cross section is shown in Fig. 3 for g =
t
1,2,3,4,5. Clearly, for M 2mt the four-top cross ates from (Γ0/Γ)2 both at threshold andat high M, due
≥
section receives a boost from diagrams corresponding to precisely to the sizeable contributions from non-doubly-
on-shellproduction oftwo gluons with subsequentdecay resonantdiagrams. Note also that in our simulations we
to tt¯. Still, for these masses the contribution of non- haveconsideredthefixedwidthapproximation. Forvery
resonant diagrams (and diagrams with a single on-shell large widths its full energy dependence must be taken
gluon production) is important, as it can be found out intoaccount,whichreducesthe suppressionwithrespect
by the different cross sections for the several g values to the one depicted in Fig. 4 [11]. Thus, although the
t
considered. (Clearly, the cross section for on-shell pro- enlarged gluon width required in realistic models of the
duction of two gluons with subsequent decay to tt¯is in- FB asymmetry with gluon masses above threshold tend
dependent of g , as long as G tt¯is the dominant de- toreducetheconstraints,theydonotremovethemcom-
t
→
cay channel.) From this plot we can learn that models pletely. Furthermore, we will see below that once the
with gluon masses M = 350 650 GeV are quite gen- LHC energy is upgraded, the dramatic increase in the
−
erally excluded, unless there is an extra enhancement of production cross section will be enough to impose strin-
the width by decay to non-SM particles. As soon as the gent constraints even with enlarged widths.
limits on four-top production at LHC get more strin- Models with new gluons of masses M 300 GeV un-
gent, with dedicated analyses and the use of the full 5 der the tt¯threshold can generate sizeab∼le asymmetries
4
with g of order unity [14]. In this case, four-top pro- Moreover, the production cross section at the tt¯thresh-
t
duction is well below the present and foreseable limits. old is almost four orders of magnitude larger than the
Still, one may considera width enhancementfromdecay expected observation limit. Thus, even with a strong
to particles lighter than the top quark [16]. We show in suppresionduetoanenlargedwidth, modelswithalight
Fig. 5 the four-top cross section in this case, for width gluon below the TeV scale are expected to be probed at
enhancements Γ = Γ0+0.1M and Γ = Γ0+0.25M. In the LHC.
both cases the cross section for masses M 300 GeV
≤
is unchanged by the extra width, so models with very 105
light colour octets [15] may already be compromised by
G = G
limits on four-top production. On the other hand, four- 104 0
top production close to threshold is largely suppressed,
a fact which is expected since the extra width 0.1M, 103
0.25M to non-SM states is orders of magnitude larger
than Γ(G tt¯), see Fig. 2. Besides, achieving such a b) 102 gt = 5
→ p
width enhancement may not be natural and/or may re- (
10
quire too large couplings to the new particles. At any s
rate, an extra gluon width may hide the four-top signal
1
but gives rise to other new final states from the decay
of the heavy gluons, which have to be considered when 10-1
14 TeV
discussing the viability of any model. g = 1
t
10-2
100 200 300 400 500 600 700 800 900 1000
103 M (GeV)
g = 1
102 t FIG. 6: Four-top cross section for Γ = Γ0 (decays to SM
G = G
0 particlesonly)fortheLHCwith√s=14TeV.Thefivelines,
10 from bottom to top, correspond to gt =1,2,3,4,5.
b) 1 Estimated upper limit
p
(
10-1
s
G = G 0 + 0.1 M Insummary,inthispaperwehaveconsideredfour-top
10-2 production in models explaining the Tevatron tt¯asym-
metry with new ‘light’ gluons. Pair production of these
10-3 7 TeV G = G 0 + 0.25 M particles followed by decays into top pairs is a new, po-
tentiallylarge,sourceoffour-topfinalstates. Inorderto
10-4
100 150 200 250 300 350 400 cover all the relevant parameter space, we have studied
M (GeV)
thefour-topproductioncrosssectionasafunctionofthe
gluon mass and its coupling to the top. We have also
FIG. 5: Four-top cross section for new gluons below and
slightly above the tt¯threshold, for gt = 1, with and with- considered some examples of scenarios where the heavy
gluonwidth is increasedby decaysto additionalnon-SM
out an extra width enhancement.
particles. Our main results are summarized in Figs. 3, 4
and5forthe7TeVLHCandFig.6forthe14TeVLHC.
Let us now considerthe effect of the foreseenLHC en- The large four-topcrosssections found in a large partof
ergy upgrade. If the CM energy is increased to 8 TeV the parameter space, and their small SM backgrounds,
weobtainafactorof2 2.3increaseinthe four-toppro- make this channel a very promising probe of this class
duction cross section,−thus partially compensating the of models, capable to reach gluon masses up to 800 GeV
supression due to an enlarged gluon width (see Fig. 4). with the luminosity already collected at the LHC. An
Amuchmoredramaticincreaseofthesignalcrosssection LHCenergyupgradeto8(14)TeVimpliesanincreasein
is obtained for √s = 14 TeV, as we show in Fig. 6. The the four-top production cross section by a factor of 2
∼
four-top production cross section in enhanced by one to (10-500),thus improving dramatically the reachin these
almosttwo ordersofmagnitude, depending onthe gluon models.
mass, with respect to the one at √s = 7 TeV. Although Acknowledgements. We would like to thank M. Pérez-
the backgrounds also grow at this energy we can antici- VictoriaforusefuldiscussionsandF.Velosoforhelpwith
pate a very good sensitivity to four-top production. For statistical combination of limits. This work has been
example, the lowest point in Fig. 6 has a cross sectionof partiallysupportedbyMICINNprojectFPA2006-05294,
52 fb while an estimated 5 σ observation limit of 45 fb FPA2010-17915 and Junta de Andalucía projects FQM
is expected with 100 fb−1 of integrated luminosity [22]. 101, FQM 03048 and FQM 6552.
5
[1] T. Aaltonen et al. [CDF Collaboration], Phys. Rev. D [20] J. A. Aguilar-Saavedra and M. Pérez-Victoria, JHEP
83, 112003 (2011) [arXiv:1101.0034 [hep-ex]]. 1105, 034 (2011) [arXiv:1103.2765 [hep-ph]].
[2] V. M. Abazov et al. [D0 Collaboration], Phys. Rev. D [21] D. A. Dicus, B. Dutta and S. Nandi, Phys. Rev. D 51
84, 112005 (2011) [arXiv:1107.4995 [hep-ex]]. (1995) 6085 [hep-ph/9412370].
[3] J. L. Hewett, J. Shelton, M. Spannowsky, T. M. P. Tait [22] B. Lillie, J. Shu and T. M. P. Tait, JHEP 0804 (2008)
and M. Takeuchi, Phys. Rev. D 84, 054005 (2011) 087 [arXiv:0712.3057 [hep-ph]].
[arXiv:1103.4618 [hep-ph]]. [23] B.A.Dobrescu,K.KongandR.Mahbubani,Phys.Lett.
[4] J. A. Aguilar-Saavedra and M. Perez-Victoria, Phys. B 670, 119 (2008) [arXiv:0709.2378 [hep-ph]].
Rev.D 84, 115013 (2011) [arXiv:1105.4606 [hep-ph]]. [24] M. Perelstein and A. Spray, JHEP 1109 (2011) 008
[5] J. A. Aguilar-Saavedra and M. Pérez-Victoria, JHEP [arXiv:1106.2171 [hep-ph]].
1109, 097 (2011) [arXiv:1107.0841 [hep-ph]]. [25] G. Cacciapaglia, R. Chierici, A. Deandrea, L. Panizzi,
[6] CMS Collaboration, note CMS PAS EXO-11-055. S. Perries and S. Tosi, JHEP 1110, 042 (2011)
[7] CMS Collaboration, note CMS PAS TOP-11-014. [arXiv:1107.4616 [hep-ph]].
[8] ATLASCollaboration, noteATLAS-CONF-2011-106. [26] G. Brooijmans et al. [New Physics Working Group Col-
[9] R.Barcelo,A.Carmona,M.MasipandJ.Santiago,Phys. laboration], arXiv:1005.1229 [hep-ph].
Rev.D 84 (2011) 014024 [arXiv:1105.3333 [hep-ph]]. [27] G. Servant,DESY-PROC-2010-01.
[10] U. Haisch and S. Westhoff, JHEP 1108, 088 (2011) [28] M. Weinberg,arXiv:1110.2640 [hep-ex].
[arXiv:1106.0529 [hep-ph]]. [29] CMS Collaboration, note CMS PASEXO-11-036.
[11] R.Barcelo,A.Carmona,M.MasipandJ.Santiago,Phys. [30] T. Junk, Nucl. Instrum. Meth. A 434, 435 (1999) [hep-
Lett.B in press [arXiv:1106.4054 [hep-ph]]. ex/9902006].
[12] G.M.TavaresandM.Schmaltz,Phys.Rev.D84,054008 [31] A. L. Read,In *Geneva 2000, Confidencelimits* 81-101
(2011) [arXiv:1107.0978 [hep-ph]]. [32] Y. Bai, J. L. Hewett, J. Kaplan and T. G. Rizzo, JHEP
[13] E.Alvarez,L.DaRold,J.I.S.ViettoandA.Szynkman, 1103, 003 (2011) [arXiv:1101.5203 [hep-ph]].
JHEP 1109, 007 (2011) [arXiv:1107.1473 [hep-ph]]. [33] J. Alwall, M. Herquet, F. Maltoni, O. Mattelaer and
[14] J. A. Aguilar-Saavedra and M. Pérez-Victoria, Phys. T. Stelzer, JHEP 1106, 128 (2011) [arXiv:1106.0522
Lett.B 705, 228 (2011) [arXiv:1107.2120 [hep-ph]]. [hep-ph]].
[15] G. Z. Krnjaic, arXiv:1109.0648 [hep-ph]. [34] N. D. Christensen and C. Duhr, Comput. Phys. Com-
[16] A. Falkowski, G. Perez and M. Schmaltz, mun. 180, 1614 (2009) [arXiv:0806.4194 [hep-ph]].
arXiv:1110.3796 [hep-ph]. [35] J. A. Aguilar-Saavedra, Nucl. Phys. B 804, 160 (2008)
[17] R.Barcelo,A.Carmona,M.Chala,M.MasipandJ.San- [arXiv:0803.3810 [hep-ph]].
tiago, arXiv:1110.5914 [hep-ph]. [36] T. Sjostrand, S. Mrenna and P. Z. Skands, JHEP 0605,
[18] P. Ferrario and G. Rodrigo, Phys. Rev. D 78, 094018 026 (2006) [hep-ph/0603175].
(2008) [arXiv:0809.3354 [hep-ph]]. [37] PGS4 http://www.physics.ucdavis.edu/ conway/
∼
[19] C.Degrande,J.-M.Gerard,C.Grojean,F.Maltoniand research/software/pgs/pgs4-general.htm .
G. Servant, JHEP 1103, 125 (2011) [arXiv:1010.6304
[hep-ph]].