rb and new physics: a comprehensive analysiscliff/papers/rb.pdf · 2002-11-12 · rb and new...

26
R b and new physics: A comprehensive analysis P. Bamert, 1 C. P. Burgess, 1 J. M. Cline, 1 D. London, 1,2 and E. Nardi 3 1 Physics Department, McGill University, 3600 University Street, Montre ´al, Que ´bec, Canada H3A 2T8 2 Laboratoire de Physique Nucle ´aire, Universite ´ de Montre ´al, C. P. 6128, succ. centre-ville, Montre ´al, Que ´bec, Canada H3C 3J7 3 Department of Particle Physics, Weizmann Institute of Science, Rehovot 76100, Israel Received 1 March 1996 We survey the implications for new physics of the discrepancy between the measurement of R b at the CERN e e collider LEP and its standard model prediction. Two broad classes of models are considered: i those in which new Zb ¯ b couplings arise at the tree level, through Z or b -quark mixing with new particles, and ii those in which new scalars and fermions alter the Zb ¯ b vertex at one loop. We keep our analysis as general as possible in order to systematically determine what kinds of features can produce corrections to R b of the right sign and magnitude. We are able to identify several successful mechanisms, which include most of those which have recently been proposed in the literature, as well as some earlier proposals e.g., supersymmetric models . By seeing how such models appear as special cases of our general treatment we are able to shed light on the reason for, and the robustness of, their ability to explain R b . S0556-2821 96 05617-2 PACS number s : 13.38.Dg, 13.65. i, 14.40.Nd I. INTRODUCTION The standard model SM of electroweak interactions has been tested and confirmed with unprecedented precision over the past few years using measurements of e e scattering at the Z resonance at the CERN e e collider LEP 1 and the SLAC linear collider SLC 2 . A particularly striking ex- ample of the impressive SM synthesis of the data came with the discovery, at the collider detector at fermilab CDF and D0 collaborations 3 , of the top quark with a mass which is in excellent agreement with the value implied by the mea- surements at LEP. The biggest, and only statistically important, fly to be found so far in the proverbial SM ointment is the experimen- tal surplus of bottom quarks produced in Z decays, relative to the SM prediction. With the analysis of the 1994 data as described at last summer’s conferences 1,2 , this discrep- ancy has become almost a 4 deviation between experiment and SM theory. The numbers are R b b / had 0.2219 0.0017, while R b SM 0.2156. 1 The SM prediction assumes a top quark mass of m t 180 GeV and the strong coupling constant s ( M Z ) 0.123, as is obtained by optimizing the fit to the data. There are other measurements which differ from their SM predictions at the 2 level: R c 2.5 , A FB 0 2.0 , and the inconsistency 2.4 between A e 0 as measured at LEP with that obtained from A LR 0 as determined at SLC 2 . In fact, since the R c and R b measurements are correlated, and because they were announced together, some authors refer to this as the ‘‘R b R c crisis.’’ One of the points we wish to make in this paper is that there is no R c crisis. If the R b discrepancy can be resolved by the addition of new physics, one then obtains an acceptable fit to the data. In other words, R c , as well as A FB 0 and A LR 0 , can reasonably be viewed simply as statistical fluctuations. On the other hand, it is difficult to treat the measured value of R b as a statistical fluctuation. Indeed, largely be- cause of R b , the data at face value now exclude the SM at the 98.8% confidence level. If we suppose that this disagree- ment is not an experimental artifact, then the burning ques- tion is the following: What does it mean? Our main intention in this paper is to survey a broad class of models to determine what kinds of new physics can bring theory back into agreement with experiment. Since R b is the main culprit we focus on explaining both its sign and mag- nitude. This is nontrivial, but not impossible to do, given that the discrepancy is roughly the same size as, though in the opposite direction to, the large m t -dependent SM radiative correction. The result is therefore just within the reach of one-loop perturbation theory. Our purpose is to survey the theoretical possibilities within a reasonably broad framework, and we therefore keep our analysis quite general, rather than focusing on individual models. This approach has the virtue of exhibiting features that are generic to sundry explanations of the Z bb ¯ width, and many of the proposals of the literature emerge as special cases of the alternatives which we consider. In the end we find a number of possible explanations of the effect, each of which would have its own potential sig- nature in future experiments. These divide roughly into two categories: those which introduce new physics into R b at tree level, and those which do so starting at the one-loop level. The possibilities are explored in detail in the remainder of the article, which has the following organization. The next section discusses why R b is the only statistically significant discrepancy between theory and experiment, and summarizes the kinds of interactions to which the data points. This is followed by several sections, each of which examines a dif- ferent class of models. Section III studies the tree-level pos- sibilities, consisting of models in which the Z boson or the b quark mixes with a hitherto undiscovered particle. We find several viable models, some of which imply comparatively large modifications to the right-handed b -quark neutral- PHYSICAL REVIEW D 1 OCTOBER 1996 VOLUME 54, NUMBER 7 54 0556-2821/96/54 7 /4275 26 /$10.00 4275 © 1996 The American Physical Society

Upload: others

Post on 06-Aug-2020

0 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Rb and new physics: A comprehensive analysiscliff/papers/Rb.pdf · 2002-11-12 · Rb and new physics: A comprehensive analysis P. Bamert,1 C. P. Burgess,1 J. M. Cline,1 D. London,1,2

Rb� and new physics: A comprehensive analysis

P.�

Bamert,1 C.�

P. Burgess,1 J.�

M. Cline,1 D.�

London,1,2 and� E. Nardi3�

1Physics Department, McGill University, 3600 University Street, Montreal, Quebec, Canada H3A 2T82Laboratoire�

de Physique Nucleaire,� Universite de

Montreal,� C. P. 6128, succ. centre-ville, Montreal, Quebec,

Canada H3C 3J73�Department�

of Particle Physics, Weizmann Institute of Science, Rehovot 76100, Israel Received1 March 1996�

We�

surveytheimplicationsfor newphysicsof thediscrepancybetweenthemeasurementof Rb� at� theCERN

e� � e � collider� LEP andits standardmodelprediction.Two broadclassesof modelsareconsidered: � i� � thosein�

which new Zb� ¯b couplings� ariseat the treelevel, throughZ

�or� b-quarkmixing with new particles,and � ii �

thosein which newscalarsandfermionsalter the Zbb

vertexat oneloop. We keepour analysisasgeneralaspossible� in orderto systematicallydeterminewhatkindsof featurescanproducecorrectionsto Rb

� of� the rightsign� andmagnitude.We areableto identify severalsuccessfulmechanisms,which includemostof thosewhichhave�

recentlybeenproposedin the literature,aswell assomeearlierproposals e.g.,! supersymmetricmodels" .By seeinghow suchmodelsappearasspecialcasesof our generaltreatmentwe areableto shedlight on thereasonfor, andthe robustnessof, their ability to explainRb

� . # S0556-2821$ 96% 05617-2& '

PACSnumber( s� ) :* 13.38.Dg,13.65.+ i, 14.40.Nd

I.,

INTRODUCTION

The standardmodel - SM. /

of0 electroweakinteractionshasbeen1

testedandconfirmedwith unprecedentedprecisionoverthe2

pastfew yearsusingmeasurementsof e3 4 e3 5 scattering6 atthe2

Z7

resonance8 at theCERNe3 9 e3 : collider; LEP < 1= and� theSLAC.

linear collider > SLC. ?A@

2B . A particularly striking ex-ample� of the impressiveSM synthesisof thedatacamewiththe2

discovery,at the collider detectorat fermilab C CDF� D

and�D0�

collaborationsE 3F G ,H of the top quarkwith a masswhich isin excellentagreementwith the value implied by the mea-surements6 at LEP.

TheI

biggest, and only statistically important, fly to befoundJ

sofar in theproverbialSM ointmentis theexperimen-tal2

surplusof bottomquarksproducedin Z decays,K

relativetothe2

SM prediction. With the analysisof the 1994 data asdescribedK

at last summer’sconferencesL 1,2M ,H this discrep-ancy� hasbecomealmosta 4N deviation

Kbetweenexperiment

and� SM theory.The numbersare

RbO PRQ

bO /S T

hadU 0.2219V W

0.0017,V

while RbO X SM. Y[Z

0.2156.V \

1]The SM prediction assumesa top quark massof m^ t_ ` 180GeVa

andthe strongcouplingconstantb sc (d M Ze )f g 0.123,

Vasis

obtained0 by optimizing the fit to the data.ThereI

areothermeasurementswhich differ from their SMpredictionsh at the i 2j level: Rck l 2.5mon ,H A FB

p0q rtstuvr2.0wox ,H and

the2

inconsistencyy 2.4zo{ between1

A e|0q as� measuredat LEPwith} that obtainedfrom A

~LR0q

as� determinedat SLC � 2� � . Infact,J

sincethe R�

ck and� R�

bO measurements� are correlated,and

because1

theywereannouncedtogether,someauthorsrefer tothis2

as the ‘‘ RbO � Rck crisis.’’; One of the points we wish to

make� in this paper is that there is no R�

ck crisis.; If the R�

bO

discrepancyK

canbe resolvedby the additionof new physics,one0 thenobtainsanacceptablefit to thedata.In otherwords,Rck ,H as well as A FB

p0q �����and� A LR

�0q ,H can reasonablybe viewedsimply6 asstatisticalfluctuations.

On�

the other hand, it is difficult to treat the measuredvalue� of R

�bO as� a statisticalfluctuation. Indeed,largely be-

cause; of RbO ,H the dataat face value now exclude3 the

2SM at

the2

98.8%confidencelevel. If we supposethat this disagree-ment� is not an experimentalartifact, then the burningques-tion2

is the following: What doesit mean?Our�

main intentionin this paperis to surveya broadclassof0 modelsto determinewhatkindsof newphysicscanbringtheory2

backinto agreementwith experiment.SinceR�

bO is�

themain� culprit we focuson explainingboth its sign andmag-nitude.This is nontrivial,but not impossibleto do,giventhatthe2

discrepancyis roughly the samesize as, though in theopposite0 direction to, the large m^ t_ -dependentSM radiativecorrection.; The result is thereforejust within the reachofone-loop0 perturbationtheory.

Our�

purpose is to survey the theoretical possibilitieswithin} a reasonablybroadframework,andwe thereforekeepour0 analysisquitegeneral,ratherthanfocusingon individualmodels.This approachhasthe virtue of exhibiting featuresthat2

aregenericto sundryexplanationsof the Z7 �

bb� ¯ width,}

and� manyof theproposalsof the literatureemergeasspecialcases; of the alternativeswhich we consider.

In the end we find a numberof possibleexplanationsofthe2

effect, eachof which would haveits own potentialsig-nature� in future experiments.Thesedivide roughly into twocategories:; thosewhich introducenew physicsinto Rb

O at�tree2

level, and thosewhich do so starting at the one-looplevel.�

TheI

possibilitiesareexploredin detail in theremainderofthe2

article, which has the following organization.The nextsection6 discusseswhy Rb

O is the only statisticallysignificantdiscrepancyK

betweentheoryandexperiment,andsummarizesthe2

kinds of interactionsto which the data points. This isfollowed by severalsections,eachof which examinesa dif-ferentclassof models.SectionIII studiesthe tree-levelpos-sibilities,6 consistingof modelsin which theZ

7boson1

or theb�

quark� mixes with a hitherto undiscoveredparticle.We findseveral6 viable models,someof which imply comparativelylarge modifications to the right-handedb

�-quark neutral-

PHYSICAL REVIEW D 1 OCTOBER1996VOLUME�

54, NUMBER 7

54�

0556-2821/96/54� 7� � /4275� �

26� �

/$10.00�

4275 © 1996The AmericanPhysicalSociety

Page 2: Rb and new physics: A comprehensive analysiscliff/papers/Rb.pdf · 2002-11-12 · Rb and new physics: A comprehensive analysis P. Bamert,1 C. P. Burgess,1 J. M. Cline,1 D. London,1,2

current; couplings.SectionsIV andV thenconsiderloop con-tributions2

to RbO . SectionIV concernsmodificationsto the

t� -quark sector of the SM. Although we find that we canreduce8 the discrepancyin R

�bO to2 �

2� �

,H we do not regardthisas� sufficientto claim successfor modelsof this type.SectionV�

then considersthe generalform for loop-level modifica-tions2

of the Zb7 ¯b

�vertex� which arisefrom modelswith new

scalars6 andfermions.The generalresultsarethenappliedtoa� numberof illustrative examples.We are able to seewhysimple6 models,like multi-Higgs-doubletandZee-typemod-els  fail to reproducethe data, as well as to examinetherobustness8 of the difficulties of a supersymmetricexplana-tion2

of RbO . Finally, our generalexpressionsguideusto some

examples  which do¡

makeexperimentallysuccessfulpredic-tions.2

SectionVI discussessomefutureexperimentaltestsofvarious� explanationsof theRb

O problem.h Our conclusionsaresummarized6 in Sec.VII.

II.,

THE DATA SPEAKS

Takenat face value, the currentLEP/SLC dataexcludesthe2

SM at the 98.8% confidencelevel. It is natural to askwhat} new physicswould be requiredto reconciletheoryandexperiment  in the eventthat this disagreementsurvivesfur-ther2

experimentalscrutiny. Before digging through one’stheoretical2

repertoirefor candidatemodels,it behoovesthetheorist2

first to askwhich featuresarepreferredin a success-ful explanationof the data.

An efficientway to dosois to specializeto thecasewhereall� new particlesare heavyenoughto influenceZ

7-pole ob-

servables6 primarily through their lowest-dimensioninterac-tions2

in an effectiveLagrangian.Then the variouseffectivecouplings; may be fit to the data,allowing a quantitativesta-tistical2

comparisonof which onesgive thebestfit. Althoughnot� all of thescenarioswhich we shalldescribeinvolve onlyheavy particles,many of them do and the conclusionswedrawK

usinganeffectiveLagrangianoftenhavea muchwiderapplicability� thanonemight at first assume.Applicationsofthis2

type of analysisto earlierdata ¢ 4,5£ ¤

have¥

beenrecentlyupdated¦ to includelastsummer’sdata § 6¨ © ,H andthepurposeofthis2

sectionis to summarizethe resultsthat werefound.ThereI

are two main typesof effective interactionswhichplayh an importantrole in the analysisof Z

7-resonancephys-

ics, and we pausefirst to enumeratebriefly what theseare.ªFor moredetailsseeRef. « 4¬®­ . The first kind of interaction

consists; of the lowest-dimensiondeviations to the elec-troweak2

bosonself-energies,andcanbeparameterizedusingthe2

well-known Peskin-Takeuchiparameters1 S¯

and� T ° 7± ² .The secondclass of interactionsconsistsof nonstandarddimension-fourK

effective neutral-currentfermion couplings,which} may be definedasfollows:2

³eff´NCµ ¶ e3

s· w¸ c¹ w¸ Z º f» ¼o½¿¾ÁÀ g L

� fà ÄÆÅg L� fà ÇÉÈ

L� ÊÌË g R

Í fà ÎÆÏg RÍ fà ÐÉÑ

RÍ Ò f»

. Ó2� Ô

In this expressiong L� fà and� g R

Í fà denoteK

the SM couplings,which} are normalized so that g L

� fÃ Õ I 3� fÃ Ö Q

× fÃs· w¸2Ø and�

g Rfà ÙÛÚ

Q× fÃs· w¸2Ø ,H where I

Ü3� fà and� Q

× fÃ

are� the third componentofweak} isospin and the electric chargeof the correspondingfermion, f

». s· w¸ Ý sin6 Þ

w¸ denotesK

thesineof theweakmixingangle,� and ß L

�(àRÍ

)á âRã 1äæå 5

ç è /2.S

Fittingé

theseeffective couplingsto the data leadsto thefollowingJ

conclusions.(1)ê

What must be explained. Although the measuredval-ues¦ for severalobservablesdepartfrom SM predictionsat the2� ë

level�

and more,at the presentlevel of experimentalac-curacy; it is only the R

�bO measurement� which really must be

theoretically2

explained.After all, some2ì fluctuationsarenot surprisingin any sampleof twenty or moreindependentmeasurements.� í Indeed,

îit would be disturbing,statistically

speaking,6 if all measurementsagreedwith theory to within1ï .ð This observationis reflectedquantitativelyin the fits ofRef. ñ 6¨ ò ,H for which the minimal modification which is re-quired� to accommodatetheR

�bO measurement,� namelythead-

ditionK

of only new effective Zb7 ¯b

�couplings,; alreadyraises

the2

confidence level of the fit to acceptable levelsóõôminö2Ø /SN÷

DFø ù 15.5/11ascomparedto 27.2/13for a SM fitú . We

therefore2

regardtheevidencefor otherdiscrepancieswith theSM,.

suchasthevalueof R�

ck ,H as beinginconclusiveat presentand� focusinsteadon modelswhich predictlargeenoughval-ues¦ for Rb

O .(2)ê

The significance of Rck . Sincethe 1995summercon-ferencesJ

havehighlightedthe nonstandardmeasuredvaluesfor theZ branching

1ratio into both

�c and� b

�quarks,� it is worth

makingtheabovepoint morequantitativelyfor theparticularcase; of thediscrepancyin R

�ck . This wasaddressedin Ref. û 6¨ ü

by1

introducingeffectivecouplingsof the Z7

to2

both b�

and� c¹quarks,� andtestinghow muchbettertheresultingpredictionsfit the observations.Although the goodnessof fit to Z-poleobservables0 does

¡improve�

somewhatý with} þmin2 /SN÷

DF ÿ 9.8/9� �

,H

1Thethird parameterU also� appearsbut doesnot play a role in theZ-pole observables.

2�Herewe introducea slight notationchangerelativeto Ref. � 4� in

that our couplings � g L�

,� R�f

correspond� to g L�

,� R�f

of Ref. � 4� .�

FIG. 1. A fit of the Zb� ¯b couplings� � g L

�,� R�b�

to�

Z�

-pole datafromthe 1995 SummerConferences.The four solid lines respectivelydenotethe 1� , 2� , 3� � , and4� error! ellipsoids.The SM predictionlies at the origin � 0,0

& �.� This fit yields � s� (M Z)

� �0.101& �

0.007.&

4276 54BAMERT,�

BURGESS,CLINE, LONDON, AND NARDI

Page 3: Rb and new physics: A comprehensive analysiscliff/papers/Rb.pdf · 2002-11-12 · Rb and new physics: A comprehensive analysis P. Bamert,1 C. P. Burgess,1 J. M. Cline,1 D. London,1,2

it doessoat theexpenseof driving thepreferredvaluefor thestrong6 coupling constantup to � sc (d M� Z)

f 0.180V !

0.035,V

indisagreementK

at the level of 2" with} low-energydetermina-tions,2

which lie in therange0.112# 0.003V $

8% &

. This changeinthe2

fit value for ' sc (d M z( )f is given by the experimentalcon-straint6 that the total Z

7width} not changewith the additionof

the2

new Zc7 ¯c¹ couplings.; 3

�Once�

the low-energydetermina-tions2

of ) sc (d M z( )f arealso included, * minö2Ø /SN÷

DFø not only drops

back1

to the levelstakenin the fit only to effectiveZbb�

cou-;plings,h but the best-fitpredictionfor R

�ck again� movesinto a

roughly8 2+ discrepancyK

with experiment.It is neverthelesstheoreticallypossibleto introducenew

physicsh to accountfor Rck in a way which doesnot drive upthe2

value of the strong coupling constant.As arguedonmodel-independent� groundsin Ref. , 6¨ - ,H and more recentlywithin} the contextof specificmodels . 9,10

� /,H an alterationof

the2

c¹ -quark neutral-currentcouplings can be compensatedforJ

in the total Z7

width} by also altering the neutral-currentcouplings; of light quarks,suchasthe s· . We put thesetypesof0 modelsasidein the presentpaper,consideringthemto beinsufficiently motivatedby the experimentaldata.

(3)ê

LH vs RH couplings. TheI

data do not yet permit adeterminationK

of whetherit is preferableto modify the left-handed0 LH 1 or0 right-handed2 RH3 Zbb

�coupling.; Themini-

mum valuesfor 4 2Ø

found in Ref. 5 6¨ 6 for a fit involving LH,RH,7

or both couplings are, respectively, 8 min2 /SN÷

DF 9 LH: ;

< 17.0/12, = min2 /SN÷

DF > RH7 ?A@

16.1/12, or B min2 /SN÷

DF C both1 D

E 15.5/11.(4)ê

The size required to explain RbO . The analysisof Ref.F

6¨ G

also� indicatesthesizeof thechangein theneutral-currentb�

-quark couplingsthat is requiredif theseare to properlydescribeK

the data.The bestfit valueswhich arerequiredaredisplayedK

in Fig. 1, and are listed in Table I. Table I alsoincludes�

for comparisonthe correspondingtree-level SMcouplings,; aswell asthe largestSM one-loopvertexcorrec-tions2 H

those2

which dependquadraticallyor logarithmicallyon0 the t� -quarkmass4

Im^ t_ J ,H evaluatedat s· w¸2Ø K 0.23.

VFor making

comparisons; we takem^ t_ L 180 GeV.AsM

we now describe,the implicationsof the numbersap-pearingh in TableI dependon the handednessN LH vs RHO of0effective  new-physicsZbb

�couplings.;

(4a)ê

LH couplings. TableI showsthattherequiredchangein�

the LH Zb7 ¯b

�couplings; mustbe negativeandcomparable

in�

magnitudeto them^ t_ -dependentloop correctionswithin theSM..

The sign must be negativesincethe predictionfor theZ P bb

� ¯ width} mustbe increasedrelativeto the SM result inorder0 to agreewith experiment.This requires Q g L

bO

to2

havethe2

samesign asthe tree-levelvaluefor g LbO,H which is nega-

tive.2

As we shall see,this sign limits the kinds of modelswhich} can producethe desiredeffect. Comparisonwith theSM.

loop contributionshowsthat the magnituderequiredforRg L

bO

is�

reasonablefor a one-loopcalculation.Sincethe sizeof0 the m^ t_ -dependentpart of the SM loop is enhancedby afactor of m^ t_2Ø /S M W

,H the requirednew-physicseffect must belargerS

than2

a genericelectroweakloop correction.(4b)ê

RH couplings. Since.

the SM tree-levelRH couplingis oppositein sign to its LH counterpartand is somefivetimes2

smaller,the new-physicscontributionrequiredby thedata,K T

g RbO,H is positiveandcomparablein sizeto thetree-level

coupling.; This makesit likely thatanynew-physicsexplana-tion2

of the datawhich relies on changingg RÍbO must ariseat

tree2

level, ratherthanthroughloops.(5)ê

Absence of oblique corrections. AM

final provisois thatany� contributionto g L

bO

or0 g RbO

should6 not be accompaniedbylargecontributionsto otherphysicalquantities.For example,Ref. U 6¨ V finds that the best-fit valuesfor the obliqueparam-eters  S

¯and� T

Ware�

S¯ XZY

0.25V [

0.19,V

T \Z] 0.12V ^

0.21V _

3F `

awith} a relativecorrelationof 0.86b even  when c g L

�,RÍb

Oare� free

to2

float in the fit. SinceT often0 getscontributionssimilar insize6 to d g L

bO

these2

boundscanbe quite restrictive.Noticee

that we neednot worry about the possibility ofhaving¥

large cancellationsbetweenthe new-physicscontri-butions1

to the obliqueparametersand f g L�bO in Rb

O . It is truethat2

sucha partialcancellationactuallyhappensfor g bO in the

SM,.

where the loop contributionsproportionalto m^ t_2 in�

TW

and� h g L�bO exactly  cancelin thelimit thats· w¸2 i 1

4,H andsoendupbeing1

suppressedby a factor s· w¸2Ø j 14. We neverthelessneed

not� considersucha cancellationin R�

bO since6 the obliquepa-

rameters8 k especially  TW l

almost� completelycancelbetweenm bO

and� nhado . Quantitatively,we have p 4q

rbO sut

bOSMv w

1 x 4.57y g L�bO z 0.828

V {g RÍbO | 0.00452

VS¯ }

0.0110V

T ~ ,H�

had�u� hadSMv �

1 � 1.01� g LbO �

0.183V �

g RbO �

0.00518V

S¯ �

0.0114V

TW �

,H

3�Introducingeffective b-quark couplingshaspreciselythe oppo-

site effect, sincethe SM predictionfor � b� is�

low andthat for � c� ishigh relativeto observations,lowering thestrongcouplingconstantto � s� (M Z) � 0.103

& �0.007.&

4�More precisely � 11� , we use � g� L

b� �

(� �

w� /16� �

) � r � 2.88lnr � ,�wherer � m� t 2� /� M¡ W

2�

.

TABLE I. Requiredneutral-currentb

-quark couplings: The last two columnsdisplay the size of theeffective! correctionto the left- andright-handedSM Zbb couplingswhich bestfit thedata.The ‘‘individualfit’’¢

is obtainedusingonly oneeffectivechiral coupling in addition to the SM parametersm t  and� £s� (� M¡ Z

¤ ).�The¥

‘‘fit to both’’ includesboth couplings.Also shown for comparisonare the SM predictionsfor thesecouplings,� both the tree-levelcontribution ¦ ‘‘SM tree’’§ ,� and the dominantm t  -dependentone-loopvertexcorrection,� evaluatedat s¨ w�2� © 0.23

& ª‘‘SM top loop’’ « .

Coupling¬

g ­ SM tree® ¯ g� ° SM±

top loop² ³ g� ´ individual�

fit µ ¶ g� · fit¢

to both

g� Lb� ¹

0.4230 0.0065 º 0.0067» 0.0021& ¼

0.0029½ 0.0037g� R¾b� 0.0770 0 0.034¿ 0.010

&0.022À 0.018

54 4277RÁ

b� AND NEW PHYSICS:A COMPREHENSIVEANALYSIS

Page 4: Rb and new physics: A comprehensive analysiscliff/papers/Rb.pdf · 2002-11-12 · Rb and new physics: A comprehensive analysis P. Bamert,1 C. P. Burgess,1 J. M. Cline,1 D. London,1,2

so6 R�

bO Â R

�bOSMv Ã

1 Ä 3.56F Å

g LbO Æ

0.645V Ç

g RbO È

0.00066V

É 0.0004V

T Ê . Ë 4ÌWeÍ

now turn to a discussionof the circumstancesunderwhich} theaboveconditionsmaybeachievedin a broadclassof0 models.

III.,

TREE-LEVEL EFFECTS: MIXING

At treelevel theZbb�

couplings; canbemodifiedif thereismixing� amongstthechargeÎ 1

3Ï quarks,� or theneutral,colour-

less�

vector bosons.Being a tree-leveleffect it is relativelyeasy  and straightforwardto analyzeand comparedifferentscenarios.6 Also, since mixing effects can be large, mixingcan; providecomparativelylargecorrectionsto the Zb

7 ¯b�

cou-;pling,h suchasis neededto modify R

�bO through2

changesto g RbO.

Note

surprisingly, a numberof recentmodels Ð 9,10,12�

–14Ñuse¦ mixing to try to resolvetheRb

O Ò and� Rck Ó discrepancy.K

Ouraim� hereis to surveythepossibilitiesin a reasonablygeneralway.} We thereforepostponefor themomenta moredetailedphenomenologicalh analysisof the variousoptions.

In generalwe imaginethat all particleshaving the samespin,6 color, andelectricchargecanbe relatedto oneanotherthrough2

massmatricesÔ some6 of whoseentriesmight becon-strained6 to be zero in particularmodelsdue to gaugesym-metries or restrictionson the Higgs-field representationsÕ .WeÍ

denotethe color-triplet, chargeQ× ÖØ× 1

3Ï ,H quarks in the

flavorÙ

basis by BÚ Û

,H and label the correspondingmasseigenstates  5

çby1

b� i. The mass-eigenstatequarks,b

� i,H are ob-tained2

from the B Ü by1

performingindependentunitary rota-tions,2 ÝßÞ

L,R† )f à i amongst� the left- andright-handedfields.The

b�

quark� thathasbeenobservedin experimentsis the lightestof0 themasseigenstates,b

� áb� 1,H andall othersarenecessarily

muchheavierthanthis state.Similar.

considerations also apply for colorless,electrically-neutral  spin-oneparticles.In this casewe imag-ine theweakeigenstates,Z âw¸ ,H to be relatedto themasseigen-states,6 Z ãmä ,H by an orthogonalmatrix, å wm¸ . We take thephysicalh Z

7,H whosepropertiesaremeasuredin suchexquisite

detailK

at LEP andSLC, to be the lightestof the masseigen-states:6 Z æèç Z é1 .

AssumingM

thatall of theb� i and� Zmä ê except  for the lightest

ones,0 the familiar b�

and� Z7

particlesh ë are� too heavy to bedirectlyK

producedat Z-resonanceenergies,we find that theflavor-diagonaleffective neutral-currentcouplings relevantforJ

R�

bO are�

g L�

,RÍb

O ìîíg mä ï 1 ð L

�,RÍ11 ñóòôöõ w¸ ÷ g w¸ ø L

�,RÍùûúýü

L�

,RÍþ 1* ÿ L

�,RÍ� 1 � w¸ 1

��� �w¸ � g w¸ �� L,R � L,R

� 1 2Ø � w¸ 1,H � 5� �

where} the neutral-currentcouplingsaretakento be diagonalin the flavor B � basis.

1 6�

ThisI

expressionbecomesreasonablysimple in the com-mon� situation for which only two particlesare involved inthe2

mixing. In this casewe may write BÚ ���

(d

B �B ),f

b� i � (

dbO �bO

),f

and� Z7 w¸ � (

dZ �Z )f, and take � L

� ,H � RÍ ,H and � to

2be two-by-two

rotationmatricesparametrizedby the mixing angles� L ,H R ,Hand� !

Z . In this caseEq. " 5� # reduces8 to

g L,RbO $&%('

g ZB )

L�

,RÍ c¹ L,R

2 *&+ g ZB ,.-

L�

,RÍ s· L,R

2 / c¹ Ze 02143 g Z 5B 6

L�

,RÍ c¹ L,R

2

7&8g Z 9B :.;

L,Rs· L,R2 < s· Z ,H = 6¨ >

where} s· L denotesK

sin ? L ,H etc. IncreasingR�

bO requires8 increas-

ing�

the combination(g L�bO )f 2 @ (

dg RÍbO )f 2. To seehow this works

we} now specializeto morespecificalternatives.

A.A

ZB

mixing

First considerthecasewheretwo gaugebosonsmix. ThenEq. C 6¨ D reducesto

g L,RbO E&F

g ZBG H

L�

,RÍ c¹ Ze I&J g Z

e KBG L

L�

,RÍ s· Ze ,H M 7± N

where} (g ZeBG )f

L�

,RÍ is the SM couplingin the absenceof Z mix-

ing, and (g Ze OBG

)f

L,R is the b�

-quark coupling to the new fieldZ P Q�R which} might itself be generatedthrough b

�-quark mix-

ing� S

. It is clearthatso long astheZ7 T

b�

b�

coupling; is nonzero,then2

it is alwayspossibleto choosethe angle U Ze to2

ensurethat2

the total effectivecoupling is greaterthan the SM one,(dg Z

B)f

L,R . This is becausethe magnitudeof any function ofthe2

form f»

(d V

Z)f W

Ac~

Z X BsÚ

Z is�

maximized by the angletan2 Y

Ze Z B/

SA,H for which [ f» \ maxö ]_^ A/

Sc¹ Ze `ba&c A d .

The model-buildingchallengeis to ensurethat the sametype2

of modificationsdo not appearin an unacceptablewayin�

theeffectiveZ7

couplings; to otherfermions,or in too largean� M Z

e shift6 due to the mixing. This can be ensuredusingappropriate� choicesfor the transformationpropertiesof thefieldse

underthe new gaugesymmetry,andsufficiently smallZ7

-Z7 f

mixing� angles.Models along theselines have beenrecentlydiscussedin Refs. g 9,

�16h .

B. bi

-quark mixing

The secondnatural choice to consideris pure b�

-quarkmixing, with no newneutralgaugebosons.We consideronlythe2

simplecaseof 2j 2�

mixings,sincewith only onenewBÚ k

quark� mixing with the SM bottom quark,Eq. l 6¨ m simplifies6considerably.; As we will discussbelow,we believethis to besufficient6 to elucidatemost of the featuresof the possibleb�

-mixing solutionsto the R�

bO problem.h

Let:

us first establishsomenotation.We denotethe weakSU. n

2� o

representations8 of the SM BÚ

L,R and� of the BÚ

L,R

pas�

5qWe imaginehavingalreadydiagonalizedthe SM massmatrices

so that in the absenceof this non-standardmixing one of the B rreducesto the usualb

quark,with a diagonalmassmatrix with the

d ands¨ quarks.s

6tEquation u 5� v describesthe mostrelevanteffectsfor the Rb

� prob-�lem, namelythe mixing of Z

�and b

with new states.However,in

generalotherindirecteffectsarealsopresent,suchas,for example,a shift in M Z dueto themixing with theZ w .� For a detailedanalysisof the simultaneouseffectsof mixing with a Z x and� new fermions,seeRef. y 15z .

4278 54BAMERT,�

BURGESS,CLINE, LONDON, AND NARDI

Page 5: Rb and new physics: A comprehensive analysiscliff/papers/Rb.pdf · 2002-11-12 · Rb and new physics: A comprehensive analysis P. Bamert,1 C. P. Burgess,1 J. M. Cline,1 D. London,1,2

R�

L,R and� R�

L,R

{,H respectively,where R

� |(dIÜ,H IÜ 3� )f . The SM

B-quark assignmentsare RL }�~ 12� ,H � 1

2� � ,H and R

�R �_� 0,0

V �. By

definition,K

a BÚ �

quark� must haveelectric chargeQ× �_�

1/3,but1

may in principle have arbitrary weak isospin R�

L,R

�� (dIL�

,R� ,H I3

�L�

,R�).f

In terms of the eigenvaluesI3�

L�� and� I3

�R� of0 the weak-

isospin�

generatorIÜ

3� acting� on B

ÚL� and� B

ÚR� ,H the combination

of0 couplingswhich controls � bO becomes1

�bO ��� g L

�bO � 2 �&� g RÍbO � 2 � � c¹ L

�2Ø2 � s· w¸2Ø

3F � s· L

�2Ø I3�

L�� 2Ø �

s· w¸2Ø3F   s· R

Í2Ø I3�

RÍ¡ 2Ø.¢

8% £

In orderto increase¤ bO using¦ this expression,¥ L

� and� ¦RÍ must

be1

suchasto makeg L�bO morenegative,g R

ÍbO morepositive,orboth.1

Two waysto ensurethis areto choose

I3�

L�§_¨2© 1

2or0 I3

�Rͪ¬« 0.

V ­9� ®

ThereI

arealsotwo otheralternatives,involving largemixingangles� or large B

Ú ¯representations:8 I

Ü3�

L

°_±0V

, with s· L2(dIÜ

3�

L

²´³12� )f µ 1 ¶ 2

�s· w¸2 /3S ·

0.85,V

and IÜ

3�

R

¸¬¹0V

, with s· R2 º IÜ 3�

R

»½¼b¾2�

s· w¸2 /3S¿ 0.15.

VNote that, in the presenceof LH mixing, the

Cabibbo-Kobayoshi-Maskawa�

elementsVqbÀ (dqÁ  uà ,H c¹ ,H t� )f get

rescaledasVqbÀ Ä c¹ L� VqbÀ ,H thus leadingto a decreasein rates

forJ

processesin which theb�

quark� couplesto a W. Thereforecharged-current; datacanin principleput constraintson largeLH mixing. For example,futuremeasurementsof thevarioust� -quark decaysat the Tevatronwill allow the extractionofV tb_ in

�a model-independentway, thus providing a lower

limit�

on c¹ L . At present,however,when the assumptionofthree-generation2

unitarity is relaxed Å as� is implicit in ourcases; Æ the

2current measurementof B(

dt� Ç Wb)/

fB(dt� È Wq )

fimplies�

only the very weaklimit É V tb_ ÊbË 0.022V Ì

at� 95% C.L.ÍÎ17Ï . Henceto date thereare still no strongconstraintson

large LH mixing solutions.Regardingthe RH mixings, asdiscussedK

below there is no correspondingway to deriveconstraints; on c¹ R ,H andso larges· R solutions6 arealwayspos-sible.6

WeÍ

proceednow to classify the modelsin which the SMbottom1

quarkmixeswith othernew Q× Ð_Ñ

1/3 fermions.Al-though2

thereare endlesspossibilitiesfor the kind of exoticquark� onecould consider,the numberof possibilitiescanbedrasticallyK

reduced,and a completeclassificationbecomespossible,h after the following two assumptionsare made: Ò i ÓThereI

areno new Higgs-bosonrepresentationsbeyonddou-blets1

andsinglets; Ô ii� Õ the2

usualBÚ

quark� mixeswith a singleB Ö ,H producingthemasseigenstatesb

�and� b

� ×. This constrains

the2

massmatrix to be 2Ø 2:

ÙB B ÚÜÛ L

M 11

M�

21

M 12

M�

22

BBÚ Ý

R

. Þ 10ßWeÍ

will examineall of the alternativesconsistentwiththese2

assumptions,both of which we believeto be well mo-tivated,2

andindeednot very restrictive.Theresultingmodelsincludethe‘‘standard’’ exoticfermionscenariosà 18áãâ vector�singlets,6 vectordoublets,mirror fermionsä ,H as well asa num-ber1

of others.

Let us first discussassumptionå i æ . From Table I andEq.ç8% è

one0 seesthat the mixing anglesmustbe at leastaslargeas� 10% to explainR

�bO ,H implying that the off-diagonalentries

in�

themassmatrix Eq. é 10ê which} give riseto themixing areof0 orders· L

�,RÍ M 22

Ø ë 10 GeV. If theseentriesaregeneratedbyHiggs fields in higher than doublet representations,suchlarge�

vacuumexpectationvalves ì VEV’s� í

would} badly un-dermineK

the agreementbetweentheory and experimentforthe2

M W/SM Ze massratio.7

îAccording to assumptionï i ð ,H the permittedHiggs repre-

sentations6 are R�

H ñ_ò 12� ,H ó 1

2� ô and� õ 0,0

V ö. It is then possibleto

specify6 which representationsR�

L,R

÷allow� the B

Ú øto2

mix withthe2

B quark� of the SM.ù1ú Since

.the B

Ú ûshould6 be relatively heavy,we require

that2

M�

22ü 0.V

Then the restriction ý i� þ on0 the possibleHiggsrepresentationsimplies that

ÿIÜ

L� � IÜ

R

�����0,V 1

2� � 11�

and�IÜ

3�

L

��IÜ

3�

R

����0,V 1

2� . � 12�

�2� To haveb

�-b� �

mixing, at leastoneof the off-diagonalentries,  M 12 or0 M 21

Ø ,H must be nonzero.Theseterms ariserespectively from the gauge-invariantproducts RH � R

�L�

R�

R� and� R�

H � R�

R � R�

L� so6 that R�

L(àR)á�

must� transformas theconjugate; of the tensorproductRH

� � RRÍ

(àL�

)á :

R�

L�! R�

H� " R

�RÍ #%$ 0,0

V &,H 1

2� ,H ' 1

2� ,H ( 13)

or0

R�

R*,+ R�

H� - R

�L� .%/ 0,0

V 0,H 1

2� ,H 1 1

2� ,H 2 1,3 1 4 ,H 5 1,06 . 7 148

Thus the only possiblerepresentationsfor the B 9 are� thosewith} IR

Í :,; 0,V 1

2,H 1 andIL� <!= 0,

V 12, 1H , 3

Ï2,H subjectto therestrictions>

11? – @ 14A .AsM

for assumptionB ii� C ,H it is of coursepossiblethatseveralspecies6 of B D quarks� mix with theB,H giving rise to anN

÷ EN÷

massmatrix, but it seemsreasonableto study the allowedtypes2

of mixing one at a time. After doing so it is easytoextend  the analysisto the combinedeffectsof simultaneousmixing with multiple B F quarks.� Thus G ii H appears� to be arather8 mild assumption.

7IThe¥

contribution of theserelatively large nonstandardVEV’scannot be effectively compensatedby new loop effects. On theotherhand,beyondHiggs doublets,the next caseof a Higgs mul-tiplet preservingthetree-levelratio is thatof I3

JH K 3,

LY H M 2. We do

not considersuchpossibilities,which would alsorequirethemixedBN O

to�

belongto similarly high-dimensionalrepresentations.We alsoneglect alternativescenariosinvoking, for example,more HiggstripletsandcancellationsbetweendifferentVEV’s, sincethesesuf-fer from severefine-tuningproblems.

54 4279RÁ

b� AND NEW PHYSICS:A COMPREHENSIVEANALYSIS

Page 6: Rb and new physics: A comprehensive analysiscliff/papers/Rb.pdf · 2002-11-12 · Rb and new physics: A comprehensive analysis P. Bamert,1 C. P. Burgess,1 J. M. Cline,1 D. London,1,2

Thereis one sensein which P ii Q might appearto restrictthe2

classof phenomenawe look at in a qualitativeway: itis�

possibleto obtainmixing betweentheBÚ

and� a BÚ R

in�

oneofthe2

higher representationswe haveexcludedby ‘‘bootstrap-ping,’’h thatis, by intermediatemixing with a B

Ú1S in�

oneof theallowed� representations.The ideais that, if the SM B mixeswith} such a B1T ,H but in turn the latter mixes with a B2

Ø U of0larger�

isospin,this would effectively inducea BÚ

-BÚ

2V mixing,�which} is not consideredhere.However,sincemassentriesdirectlyK

coupling B to2

B2Ø W are� forbiddenby assumptionX i Y ,H

the2

resultingBÚ

-BÚ

2Z mixing� will in generalbe proportionaltothe2

-BÚ

1[ mixing,� implying that theseadditionaleffectsaresubleading,6 i.e., of higher order in the mixing angles.Thismeansthat if the dominantB-B1\ mixing effectsare insuffi-cient; to accountfor the measuredvalueof R

�bO ,H addingmore

B ] quarks� with larger isospinwill not qualitatively changethis2

situation.ThereI

is, however,a loophole to this argument.If themassmatrix hassomesymmetrywhich givesriseto a special‘‘texture,’’ thenit is possibleto havelargemixing anglesandthus2

evadethe suppressiondue to productsof small mixingangles� alludedto above.Indeed,we haveconstructedseveralexamples  of 3 3

Fquasidegeneratematriceswith three and

four texture zeros,for which the B-B2Ø _ mixing is not sup-

pressedh and,dueto the degeneracy,canbe maximally large.Foré

example, let us chooseBÚ

1 in�

a vector doublet withIÜ

3�

L,R a�b 1/2 and BÚ

2c in�

a vector triplet with IÜ

3�

L,R d�e 1. Be-cause; of our assumptionof no Higgstriplets,directB mixingwith} sucha B2

Ø f is forbidden,and M 13g M 31� h M 12i 0

V. It is

easy  to checkthat for a genericvaluesof the nonvanishingmassmatrix elements,the induceds· L

�,RÍ13 mixings are indeed

subleading6 with respectto s· L�

,RÍ12 . However,if we insteadsup-

poseh that all the nonzeroelementsare equal to somelargemass� j ,H thentherearetwo nonzeroeigenvaluesm^ b

O kmlon and�m^ bO p!q 2r while} the B-B2

Ø s mixing angless· L�13tvu 1/3 and s· R

Í13

wvx 3/8F

are unsuppressedrelative to s· L�

,RÍ12 .8 Although it may

be1

unnaturalto havenearequalityof themassentriesgener-ated� by singletanddoubletHiggsVEV’s, asis neededin thiscase; and in most of the otherexampleswe found, it is stillpossibleh that someinterestingsolutionscouldbeconstructedalong� theselines.

ApartM

from somespecialcasesanalogousto the oneout-lined�

above,we canthereforeconcludethat neitherdoesas-sumption6 y ii z seriously6 limit the generalityof our results.

WeÍ

can now enumerateall the possibilitiesallowed byassumptions� { i� | and� } ii� ~ .

WithÍ

the permittedvaluesof IÜ

R� and� IÜ

L� listed�

above,andthe2

requirementthat at leastone of the two conditions � 13�and� � 14� is satisfied,thereare19 possibilities,listed in TableII.î

Althoughnot all of themareanomalyfree, theanomaliescan; always be canceledby adding other exotic fermionswhich} haveno effecton R

�bO . Sinceonly thevaluesof I

Ü3�

L

�and�

I3�

R� are� importantfor theb

�neutralcurrentcouplings,for our

purposeh modelswith thesameI3�

L�

,R�

assignments� areequiva-lent,�

regardlessof IÜ

L,R

�or0 differencesin the massmatrix or

mixing pattern.Altogether thereare 12 inequivalentpossi-bilities.1

Equivalentmodelsareindicatedby a prime ����� in�

the‘‘Model’’ columnin TableII.

8�A�

small perturbationof the orderof a few GeV canbe addedtosomeof the nonzeromassentriesto lift the degeneracyandgive anonzerovaluefor mb

� .�

TABLE II. Modelsandchargeassignments.All thepossiblemodelsfor B-B � mixing allowedby theassumptionsthat � i � hereareno newHiggsrepresentationsbeyondsingletsanddoublets,and � ii � only mixing with a singleB � is considered.Thepresenceof LH or RH mixingswhich canaffect the b

neutralcurrentcouplingsis indicatedunder‘‘Mixing.’’ Subleadingmixings, quadraticallysuppressed,aregiven in

parentheses.Equivalentmodels,for the purposesof RÁ

b� , areindicatedby a prime ��� � in

�the ‘‘Model’’ column,while modelssatisfyingEq.�

9� and� which canaccountfor thedeviationsin Rb� with smallmixing angles,arelabeledby anasterisk� * � . LargeRH mixing solutionsare

labeledby a doubleasterisk� **   , while models7, 7 ¡ , and7¢ allow for a solutionwith largeLH mixing.

IL£ ¤ I3

�L£¥ IR

¦ § I3�

R¦¨ Model

©Mixing

0 0 0 0 1 Vector singlet L�

1/2 ª 1/2 2« ** ¬ Mirror family L�

,RÁ

­1/2 3® * ¯ (L),

�R

1/2 ° 1/2 0 0 4 Fourthfamily1/2 ± 1/2 5² ** ³ Vector doublet ´ Iµ ¶ R

Á1 · 1 6 ** ¹ R

0 4º»1/2 0 0 7 L

�1/2 ¼ 1/2 8½ * ¾ Vector doublet ¿ IIµ À (L

�),RÁ

1 0 7 Á LÂ1 9Ã * Ä (L),

�R

1 Å 1 1 Æ 1 10Ç * È Vector triplet É Iµ Ê L�

,(RÁ

)1/2 Ë 1/2 11Ì * Í L

�,(RÁ

)0 1 0&

1 Î Vector triplet Ï II Ð L1/2 Ñ 1/2 2 Ò L,(R)

3/2 Ó 3/2 1 Ô 1 12Õ * Ö L�

,(RÁ

)× 1/2 1 Ø 1 6Ù (�RÁ

)�

0 4ÚÛ1/2 1 0 7Ü L

4280 54BAMERT,�

BURGESS,CLINE, LONDON, AND NARDI

Page 7: Rb and new physics: A comprehensive analysiscliff/papers/Rb.pdf · 2002-11-12 · Rb and new physics: A comprehensive analysis P. Bamert,1 C. P. Burgess,1 J. M. Cline,1 D. London,1,2

Due to gaugeinvarianceand to the restriction Ý i Þ on0 theHiggssector,in severalcasesoneof theoff-diagonalentriesM�

12 or0 M�

21 in�

Eq. ß 10à vanishes,� leadingto a hierarchybe-tween2

the LH and the RH mixing angles.If the b� á

is muchheavierthan the b

�,H M 12â 0

Vyields s· R

Í ã M 21Ø /SM 22Ø ,H while the

LH:

mixing is suppressedby M�

22ä 2Ø. If on the other hand

M�

21å 0,V

thenthe suppressionfor s· R is�

quadratic,leavings· L

as� thedominantmixing angle.For thesecases,thesubdomi-nantmixingsareshownin parenthesesin the ‘‘Mixing’’ col-umn¦ in TableII. Notice that while models2 and6 allow fora� large right-handed mixing angle solution of the R

�bO

anomaly,� the ‘‘equivalent’’ models2 æ and� 6 ç doK

not, pre-cisely; becauseof sucha suppression.

Six.

choicessatisfy one of the two conditionsin Eq. è 9� é ,Hand� hencecan solve the R

�bO problemh using small mixing

angles.� Theyarelabeledby anasteriskê * ë in TableII. Threeof0 thesemodels ì 10,11,12í satisfy6 the first condition for so-lutions�

using small LH mixings. Since for all thesecasesIÜ

3�

R

îðï0V

, a largeRH mixing could alternativelyyield a solu-tion2

but becauses· RÍ is alwayssuppressedwith respectto s· L

� ,Hthis2

latter possibility is theoreticallydisfavored.The otherthree2

choicesñ models� 3,8,9ò ó

satisfy6 the secondconditionforsolutions6 usingsmall RH mixing. It is noteworthythat in allsix6 modelsthe relevantmixing neededto explain Rb

O is au-tomatically2

the dominantone,while the other,which wouldexacerbate  the problem, is quadratically suppressedandhence¥

negligiblein thelargem^ bO ô limit.�

Therearetwo choicesõmodels� 5,6

ö ÷forJ

which IÜ

3�

R

øðù0V

andthereis only RH mixing,and� one ú model2û for which I3

�RÍüðý

0V

ands· RÍ is unsuppressed

with} respectto s· L . Thesethreecasesallow for solutionswithlargeRH mixings,andarelabeledby a doubleasteriskþ ** ÿ .Finally, a solutionwith largeLH mixing is possible� models7,H 7 � ,H and 7��� in which I3

�L����� 1/2, and I3

�R�� 0

Vimplies no

RH7

mixing effects.In the light of TableII we now discussin moredetail the

most popular models,as well as some other more exoticpossibilities.h

Vector singlet. Vector�

fermionsby definition haveidenti-cal; left- andright-handedgaugequantumnumbers.A vectorsinglet6 model1� is one for which IL

� ��� IRÍ ��� 0

V. Inspectionof

Eq.� �

8% �

shows6 that mixing with sucha vector-singletquarkalways� actsto reduceR

�bO .9�

Mirror family. A mirror family � model2� is a fourth fam-ily�

but with the chiralities of the representationsinter-changed.; BecauseI3

�L�� vanishes,� LH mixing actsto reducethe

magnitudeof g L�bO ,H andso tendsto makethepredictionfor Rb

Oworse} than in the SM. For sufficiently large RH mixingangles,� however,this tendencymaybereversed.As wasdis-cussed; immediatelybelow Eq. � 9� � ,H since I

Ü3�

R

�is�

negativeacomparatively; large mixing angle of s· R

Í2Ø � 1/3 is neededtosufficiently6 increaseRb

O . Such a large RH mixing angle isphenomenologicallyh permittedby all off-resonancedetermi-nations� of g R

bO

19! . In fact, the b�

-quark production crosssection6 andasymmetry,asmeasuredin the " -Z interference

region # 21,22$ ,H cannotdistinguishbetweenthe two valuess· RÍ2Ø % 0

Vand 4s· w¸ /3,

Swhich yield exactly the same rates.10

Hence&

this kind of modelcansolvethe R�

bO problem,h though

perhapsh not in the most aestheticallypleasingway. As isshown6 in Fig. 2, the allowedrangeof mixing anglesis lim-ited to a narrowstrip in the s· L

�2Ø -s· RÍ2Ø plane.h

Fourth'

family. AM

fourth family ( model� 4) cannot; resolveR�

bO via� tree-leveleffectsbecausethe new B

Ú *quark� hasthe

same6 isospinassignmentsastheSM b�

quark,� andsotheydonot mix in the neutral current.11 Two other possibilities+models4 , and� 4-�. yield/ the sameI3

�L�

,RÍ0

assignments� as thefourthJ

family model, and are similarly unsuccessfulin ex-plainingh Rb

O since6 they do not modify the b�

quark� neutralcurrent; couplings.

Vector doublets. ThereI

aretwo possibilitieswhich permita� Q× 1�2 1

3Ï quark� to transformas a weak isodoublet,and in

both1

casesmixing with the SM b�

is allowed. They can belabeledby the different hyperchargevalue using the usualconvention; Q

× 3IÜ

3� 4 Y .

WithÍ

the straightforwardchoice IÜ

3�

L

5�6IÜ

3�

R

798:1/2 ; model�

5� <

,H we haveY L=�> Y R?�@ 1/6.This typeof modelis discussedinRef. A 13B ,H wheretheisopartnerof theB C is

�a toplike quarkT

W D

9EA Q FHG 2/3

�vector singlet can howeverbe usedto reduceR

ÁcIJ

10,12,14K , provided that stepsare taken,as suggestedin Sec. IIabove,to avoid the resultingpreferencefor an unacceptablylargevaluefor L s� (� M¡ Z

¤ ).

10The current90% C.L. upperbounds¨ R2� M

0.010 N 20O holdsin thesmall mixing angleregions¨ R

P2 Q 1/3.11Thesemodelshavethe further difficulty that, exceptin certain

cornersof parameterspaceR 23S ,� they producetoo largea contribu-tion to the oblique parameters,S and T,� to be consistentwith thedata.

FIG. 2. The experimentallyallowedmixing anglesfor a mirrorfamily. Thethick line coverstheentireareaof valuesfor s¨ L

� and� sRP

which areneededto agreewith theexperimentalvaluefor RÁ

b� to the

2T levelU

or better.Thethin line representstheone-parameterfamilyof mixing angleswhich reproducethe SM prediction.Notice thatthe small-mixing solution, which passesthrough s¨ L

� V sRP W 0,

&is

ruled out sinceILXZY 0&

implies that any LH mixing will reduce g Lb�

andthusincreasesthe discrepancywith experiment.

54 4281RÁ

b� AND NEW PHYSICS:A COMPREHENSIVEANALYSIS

Page 8: Rb and new physics: A comprehensive analysiscliff/papers/Rb.pdf · 2002-11-12 · Rb and new physics: A comprehensive analysis P. Bamert,1 C. P. Burgess,1 J. M. Cline,1 D. London,1,2

having charge [ 23Ï . Sincetheseare the samechargeassign-

mentsasfor thestandardLH b�

quark,� this leadsto no mixingin�

the neutralcurrentamongstthe LH fields, and thereforeonly0 the right-handedmixing angle s· R

Í is�

relevant for R�

bO .

Since.

3�

R

\is�

negativea comparativelylargemixing angleofs· R

2Ø ]

1/3 is neededto sufficiently increaseR�

bO ,H in much the

same6 way as we found for the mirror-family scenariodis-cussed; above.The requiredmixing anglethat gives the ex-perimentalh value,Rb

O ^ 0.2219V _

0.0017,V

is

s· RÍ2Ø ` 0.367

V a0.014qb 0.013q

. c 15dThe otherway to fit a Q

× e�f1/3 quark into a vectordou-

blet1

correspondsto I3�

L�g�h I3

�RÍi9jk 1/2 l model 8m and� so

Y n�o�p 5/6� q

10r . The partnerof the BÚ s

in�

the doubletis thenan� exoticquark,R

�,H havingQ

× t�u4/3.£

HereIÜ

3�

L

vhas¥

thewrongsign6 for satisfyingEq. w 9� x and� somixing decreases

¡the2

mag-nitude of g L

�bO . On the other hand, I3�

RÍy has the right sign to

increase�

g RbO. Whetherthis type of modelcanwork therefore

dependsK

on which of thetwo competingeffectsin RbO wins.} It

is easyto seethat in this model the M 21 entry  in the B-B zmass� matrix Eq. { 10| vanishes,� which as discussedaboveresults8 in a suppressionof s· L quadratic� in the largemass,butonly0 a linear suppressionfor s· R

Í . Hences· L� becomes1

negli-gible} in the largem^ b

O ~ limit, leavings· R as� thedominantmix-ing�

angle in R�

bO . The mixing angle which reproducesthe

experimental  valuefor R�

bO then2

is

s· RÍ2Ø � 0.059

V �0.015q� 0.013q

. � 16�However,&

in order to accountfor sucha large value of themixing anglein anaturalway, theb

� �cannot; bemuchheavier

than2 �

100 GeV.Similarly.

to the Y ����� 5/6�

vectordoubletcase,models3and� 9 alsoprovidea solutionthroughRH mixings. In model3,F

the subdominantcompetingeffect of s· L� is further sup-

pressedh by a smallerI3�

L�� ,H while in model9 theeffectof s· R

Í isenhanced  by I3

�R

�9��1, andhencea mixing anglea factorof

4 smallerthat in Eq. � 16� is sufficient to explainRbO .

Vector triplets. Thereare threepossibilitiesfor placingavector� B � quark� in an isotriplet representation: I3

�L���� I3

�R���� 1,0,� 1. The last does not allow for b

�mixing, if only

Higgs&

doubletsand singlets are present,and for our pur-poses,h I

Ü3�

L

���I3�

R

�9�0V �

model 1 ��� is equivalentto the vectorsinglet6 case already discussed.Only the assignmentI

Ü3�

L

�  I3

�RÍ¡9¢£ 1 ¤ model 10¥ allows¦ for a resolution of the Rb

§problem,¨ and it was proposedin Ref. © 12ª . If B

Ú «is¬

thelowest-isospin­

memberof the triplet thereis an exotic quarkof® chargeQ

¯ °�±5/3²

in the model.Again in the limit of largeb³ ´

massonecombinationof mixing anglesµ in this cases¶ R · isnegligible,¸ due to the vanishingof M

¹12 in¬

Eq. º 10» . As aresult,¼ s¶ L

½ plays¨ themain role in R¾

b§ . Agreementwith experi-

mentrequires

s¶ L½2¿ À 0.0127

Á Â0.0034.Á Ã

17ÄSinceÅ

the resultingchangeto gÆ L½b§ is so small, sucha slight

mixing anglewould haveescapeddetectionin all other ex-periments¨ to date.

SimilarlyÅ

to this case,models11Ç

and¦ 12Ç

also¦ provide asolutionÈ throughLH mixings. In model11 the

Éunwantedef-

fectsof s¶ RÊ are¦ further suppressed,while for model12 a L¦ H

mixing somewhatsmaller than in Eq. Ë 17Ì is sufficient toexplainÍ the data.

OurÎ

analysisof tree level effectsshowsthat both ZÏ

mix-Ðing and b

³mixing can resolvethe Rb

§ discrepancy.Ñ

-quarkmixing solutionssatisfyingthetwo assumptionsthat Ò i Ó there

Éare¦ no new Higgs representationsbeyondsingletsand dou-blets,Ô

and Õ ii¬ Ö only® mixing with a singleBÚ ×

is¬

relevant,havebeenÔ

completely classified.The list of the exotic new B ØquarksÙ with the right electroweakquantumnumbersis givenin¬

TableII. Solutionswith smalls¶ R and¦ s¶ L mixingÐ anglesarepossible¨ when the B

ÚRÛ is¬

the memberwith highestIÜ

R

Þin¬

anisodoublet¬

or isotriplet,or whenBÚ

L½ ß is thememberwith low-

estÍ I3Ý

L½à in an isotriplet or isoquartet.In all thesecases,new

quarksÙ with exotic electric chargesare also present.Someother® possiblesolutionscorrespondto I

Ü3Ý

R

á9â0Á

andareduetomixing amongsttheRH b

³quarksÙ involving ratherlargemix-

ing angles,while for I3Ý

L½ã�äå 1/2 we find anothersolution

requiring¼ even larger LH mixing. It is intriguing that suchlarge mixing anglesare consistentwith all other b

³-quark

phenomenology.¨ We havenot attemptedto classify modelsin¬

which mixing with new stateswith very large valuesofIÜ

LR

æcanç ariseasa resultof bootstrappingthroughsomein-

termediateÉ

B è mixing. Under special circumstances,theycouldç allow for additionalsolutions.

Foré

someof the modelsconsidered,the contributionstotheÉ

obliqueparameterscould be problematic,yielding addi-tionalÉ

constraints.However,for the particularclassof vec-torlikeÉ

modelsê whichë includestwo of thesmallmixing anglesolutionsÈ ì loop

­effects are sufficiently small to remain

acceptable.¦ 12 This is because,unlike the top quark whichbelongsÔ

to a chiral multiplet, vectorlikeheavyb³ í

quarksÙ tendtoÉ

decouplein the limit that their massesget large.Introduc-ing¬

mixing with other fermions doesproducenonzeroob-lique corrections,but theseremain small enoughto haveevadedÍ detection.Exceptionsto this statementare modelsinvolving¬

a largenumberof new fields, like entirenew gen-erations,Í sincethesetend to accumulatelarge contributionstoÉ

and¦ T.

IV. ONE-LOOP EFFECTS: tï-QUARK MIXING

Weð

now turn to the modificationsto the Zbb³

couplingsçwhichë canariseat oneloop. Recall that this option canonlyexplainÍ R

¾b§ if¬

the LH b³

-quarkcoupling,gÆ Lb§,ñ receivesa nega-

tiveÉ

correctioncomparablein size to the SM mò tó -dependentcontributions.ç As wasarguedin Sec.II, it is theLH couplingweë areinterestedin becausea loop-levelchangein gÆ R

is¬

toosmallÈ to fix thediscrepancybetweentheSM andexperiment.

The fact that the Rb§ problem¨ could be explainedif the

mò tó -dependentone-loopcontributionsof the SM wereabsentnaturally¸ leadsto the idea that perhapsthe tô -quark couplesdifferentlyÑ

to the b³

-quarkthanis supposedin the SM. If thetô quarkÙ mixessignificantlywith a new tô õ quarkÙ onemight beable¦ to significantly reducethe relevantcontributionsbelowtheirÉ

SM values.In this sectionwe show that it is at best

12Vectorlike modelshavethe additionaladvantageof beingauto-matically anomalyfree.

4282 54BAMERT,ö

BURGESS,CLINE, LONDON, AND NARDI

Page 9: Rb and new physics: A comprehensive analysiscliff/papers/Rb.pdf · 2002-11-12 · Rb and new physics: A comprehensive analysis P. Bamert,1 C. P. Burgess,1 J. M. Cline,1 D. London,1,2

possible¨ to reducethe discrepancyto ÷ 2ø in modelsof thistype,É

andso they cannotclaim to completelyexplainthe Rb§

data.Ñ

OurÎ

surveyof tô -quarkmixing is organizedasfollows. Wefirst describethe frameworkof modelswithin which we sys-tematicallyÉ

search,andwe identify all of the possibleexotictô -quarkquantumnumberswhich canpotentiallywork. ThisstudyÈ is carriedout much in the spirit of the analysisof b

³mixing presentedin Sec.III. We thendescribethepossibletô ùloop contributionsto the neutral-currentb

³couplings.ç Since

thisÉ

calculation is very similar to computing themò tó -dependenteffectswithin the SM, we briefly review thelatter. Besidesproviding a usefulcheckon our final expres-sions,È we find that the SM calculationalso hasseveralles-sonsÈ for the moregeneraltô -quarkmixing models.

A. Enumerating the models

Inú

this section we identify a broad class of models inwhichë the SM top quarkmixeswith otherexotic top-quark-like fermions.As in theprevioussectionconcerningb

³-quark

mixing, we denotethe electroweakeigenstatesby capitals,Tû ü

,ñ and the masseigenstatesby lower-caseletters, tý i. Toavoid¦ confusion,quantitieswhich specificallyrefer to the b

³sectorÈ will be labeledwith thesuperscriptB. By definition,aT þ quarkÙ musthaveelectricchargeQ

¯ ÿ2/3, but may in prin-

cipleç have arbitrary weak isospin RL½

,R���

(�IL½

,RÊ� ,ñ I3

ÝL½

,R�)�. Fol-

lowing closely the discussionin the previous section,wemakethreeassumptionswhich allow for a drasticsimplifi-cationç in the analysis,without muchlossof generality.�

i¬ �

First,é

the usualTû

quarkÙ is only allowedto mix with asingleÈ T quarkÙ at a time, producingthe masseigenstatestýand¦ tý .�

ii¬ �

Second,Å

for the Higgs-bosonrepresentations,we as-sumeÈ only one doublet and singlets. Additional doubletswouldë complicatethe analysisof the radiativecorrectionsina¦ model-dependentway dueto the extradiagramsinvolvingchargedç Higgs bosons.

iii¬ �

Finally,é

certainTû �

-quarkrepresentationsalsocontainnew¸ B

Ú �quarks.Ù We denotetheB

ÚL� and¦ B

ÚR� as¦ ‘‘exotic’’ when-

everÍ theyhavenonstandardweakisospinassignments,that isI3Ý

L½� B� ��� 1

2or® I3Ý

R� B� � 0

Á. As we havealreadydiscussed,for exotic

BÚ �

quarksÙ b³

-b³ �

mixingÐ will modify the b³

neutral-current¸couplingsç at tree level, overwhelmingthe loop-suppressedtý -tý � mixing effectsin Rb

§ . We thereforecarry out our analy-sisÈ underthe requirementthatany b

³-b³ �

mixingÐ affectingtheb³

neutral-current¸ couplingsbe absent.OurÎ

purposeis now to examineall of the alternativeswhichë canarisesubjectto thesethreeassumptions.Accord-ing¬

to � i¬ � ,ñ the Tû

-Tû

massÐ matriceswe considerare2! 2,"

andcanç be written in the generalform

#T T $&% L

½ M¹

11

M 21¿ M

¹12

M 22¿ T

ûT '

R

. ( 18)Due*

to our restriction + ii¬ , on® the Higgs sector,certainele-mentsof this massmatrix are nonzeroonly for particularvalues- of the T . weakë isospin.Moreover,wheneverTR

Ê / be-Ô

longs to a multiplet which also contains a Q¯ 021

1/3 BRÊ 3

quark,Ù the M¹

12B and¦ M

¹12 entriesÍ of the B

Ú-BÚ 4

and¦ Tû

-Tû 5

massÐmatricesarethe same.In thosecasesin which the B 6 quarkÙ

is exotic, assumption7 iii 8 thenÉ

forcesus to set M 129 0.Á

Incontrast,ç the M 21 entriesÍ are unrelated—forexample, thechoiceç M

¹21B : 0

Áis alwayspossibleevenif M

¹21; 0

Áfor the T

ûand¦ T

û <quarks.Ù

Inú

order to selectthoserepresentations,R¾

L,R

=,ñ which can

mix with the SM T quark,Ù we requirethe following condi-tionsÉ

to be satisfied.>1? Inú

order to ensurea largemassfor the tý @ ,ñ we requireM 22¿ A 0.

ÁAnalogouslyto Eqs. B 11C and¦ D 12E ,ñ this implies

FIÜ

LGIH IÜ

RJLKNM 0,Á 1

2" O 19P

and¦QIÜ

L

R2SIÜ

R

TVUNW0,Á 1

2" . X 20

" YZ2[ To ensurea nonvanishingtý -tý \ mixing we requireat

leastoneof the two off-diagonalentries,M 12 or® M 21¿ , tñ ob e

nonvanishing.¸ This translatesinto the following conditionson® R

¾L] and¦ R

¾R :

RL½ _I` RH

� a RRÊ bdc 0,0

Á e,ñ 1

2,ñ f 1

2,ñ g 21h

or®

RRÊ ikj RH

� l RL½ mdn 0,0

Á o,ñ 1

2,ñ p 1

2,ñ q 1,0r ,ñ s 1,t 1 u . v 22w

x3y z

Wheneverð

R{ containsç a Q¯ |2}

1/3 quark, and eitherBÚ

L~ or® BÚ

R� have�

nonstandardisospinassignments,we requireM 12� 0.

ÁThis ensuresthat at tree level the neutralcurrentb

³couplingsç are identical to thoseof the SM. Clearly, in thecasesç in which theparticularR

¾L� representation¼ impliesavan-

ishing M 21¿ element,Í imposing the condition M 12� 0

Ácom-

pletely¨ removesall tý -tý � mixing.ÐWeð

now may enumerateall the possibilities.From Eqs.�19� – � 22� ,ñ it is apparentthat as in the B � caseç the only al-

lowed representationsmusthaveIRÊ �k� 0,

Á 12,1ñ and IL

½ ��� 0,Á 1

2,1,ñ 3�2.

Consider�

first IL½ �I� 1 or 3

�2. In this case,from Eq. � 21� ,ñ

21� 0.Á

Thus we need M¹

12� 0Á

if there is to be any tý -tý �mixing.Ð The four possibilitiesfor R

¾R� are¦ shownin Eq. � 22

" �.

OfÎ

these,RRÊ �k� (

�0,0) is not allowed since Eq. � 19� is not

satisfied.È In addition,RR�¡  (� 1

2¢ ,ñ 1

2¢ )� and £ 1,0¤ both

Ôcontainexotic

BÚ ¥

quarksÙ ¦ IÜ 3Ý

R

§ B ¨�© 12¢ or® ª 1« and¦ so M

¹12 is¬

forcedto vanish,leadingto no tý -tý ¬ mixing. This leavesRR

Ê ­k® (�1,1) asa possi-

bility,Ô

sincethe BRÊ ¯ is not exotic (I3

ÝRÊ° B� ± 0

Á). If we chooseRL

½ ²suchÈ that I

Ü3Ý

L

³ B ´�µ 12¢ ,ñ then both B

ÚL¶ and¦ B

ÚR· are¦ SM-like, and

-b³ ¸

mixing is not prohibitedsinceit doesnot affect the b³

neutralcurrentcouplings.Thus the combinationRL½ ¹Iº (

� 3�2,ñ 1

2),�

R»k¼ (�1,1) is allowed.

Next½

considerIÜ

L¾�¿ 0 oÁ

r 12¢ . Here,regardlessof thevalueof

I3Ý

L½À ,ñ M 21

¿ canç benonzero.ThusanyRRÊ Á representationwhich

satisfiesÈ Eqs. Â 19Ã and¦ Ä 20" Å

is¬

permitted.It is straight-forwardtoÉ

showthat thereare11 possibilities.TheÆ

list of theallowedvaluesof IÜ

L

Çand¦ I

Ü3Ý

R

Èwhichë under

our® assumptionslead to tý -tý É mixingÐ is shownin Table III.ThereÆ

are twelve possiblecombinations,including fourth-

54 4283RÊ

bË AND NEW PHYSICS:A COMPREHENSIVEANALYSIS

Page 10: Rb and new physics: A comprehensive analysiscliff/papers/Rb.pdf · 2002-11-12 · Rb and new physics: A comprehensive analysis P. Bamert,1 C. P. Burgess,1 J. M. Cline,1 D. London,1,2

generationÌ fermions, vector singlets, vector doublets,andmirror fermions.Not all of thesepossibilitiesare anomalyfree,Í

but asalreadynotedonecouldalwayscancelanomaliesbyÔ

adding other exotic fermions which give no additionaleffectsÍ in Rb

§ .It is useful to group the twelve possibilities into three

differentÑ

classes,accordingto the particularconstraints,ontheÉ

form of the Tû

-Tû Î

massÐ matrix in Eq. Ï 18Ð .Thefirst two entriesin TableIII, which we haveassigned

toÉ

groupA,ñ correspondto thespecialcasein which the BL½

,RÊ

and¦ BL½

,RÊÑ have the samethird componentof weak isospin,

hence�

leaving the b³

neutral¸ current unaffectedby mixing.BecauseÒ

both BÚ

LÓ and¦ BÚ

RÔ appear¦ in the samemultipletswithTL½ Õ and¦ TR

Ê Ö ,ñ two elementsof the B-quarkandT-quarkmassmatricesÐ areequal:

12× M¹

12B ,ñ M

¹22Ø M

¹22B . Ù 23

" ÚAsÛ

we will see,this conditionis importantsinceit implies arelation¼ betweenthemixingsandthe mÜ tÝ ,ñ mÜ tÝ Þ massÐ eigenval-ues.ß Althoughoutsidethesubjectof this paper,it is notewor-thyÉ

that for thesemodelsthe simultaneouspresenceof bothb³

-b³ à

and¦ tý -tý á RHâ

mixing generatesnew effects in thechargedç currents: right-handedWtb

³chargedç currentsget

induced,proportionalto the productof the T and¦ B quarkÙmixings s¶ Rs¶ R

B�. Comparedto the modificationsin the neutral

currentsç andin the LH chargedcurrents,theseeffectsareofhigher�

order in the mixing angles ã 18,19ä and,¦ most impor-tantly,É

theycanonly changetheRH b³

coupling.ç But asnotedabove,¦ gÆ R

is far too small to accountfor the measuredRb§

value- using loop effectsof this kind. Thereforethe mixing-inducedRH currentsallowedin modelsA1 and¦ A2

¿ are¦ inef-fective for fixing theRb

§ discrepancy,Ñ

andwill not beconsid-eredÍ in the remainderof this paper.

Foré

the modelsin groupBÚ

,ñ the condition

12å 0Á æ

24" ç

holds.�

In the four casescorrespondingto R¾

Rèké (�1,0) ê modelsÐ

B1,ñ B3Ý ë and¦ RR

Ê ì¡í (�1/2,1/2) î modelsB2

¿ ,ñ B6ï ð ,ñ an exotic BR

Ê ñquarkÙ is presentin the sameTR

Ê ò multiplet. HenceM 12 hastobeÔ

set to zero in order to forbid the unwantedtree-levelb³

mixing effects.In theotherthreecasesbelongingto groupB,ñTRÊ ó correspondsç to the lowest componentof nontrivial mul-

tiplets:É

RRÊ ô¡õ (1

�, ö 1) ÷ model B4

ø ù and¦ RRÊ úkû (

�1/2,ü 1/2)ý

modelsÐ BÚ

5þ ,ñ BÚ

7ÿ � . For thesevaluesof I

Ü3Ý

R

�,ñ M¹

12� 0Á

is auto-matically ensured,dueto our restrictionto Higgs singletsordoublets.Ñ

Furthermore,theserepresentationsdo not containaBRÊ � quark,Ù and no BL

½ � quarkÙ appearsin the correspondingR¾

L� . Thereis thereforeno b³

-b³ �

mixing.ÐWeð

should also remark that in model B3Ý no BL

½ � quarkÙappears¦ in RL

½ � . However,a BL½ is neededasthehelicity part-

ner¸ of the BÚ

R present¨ in R¾

R� � (�1,0). Becauseof our restric-

tionÉ

on theallowedHiggsrepresentations,BÚ

L� mustÐ belongtoRL½ ��� (

�1,0) or RL

½ � (� 1/2,1/2), which in turn containa new TL½ ��

L� . While the first choice correspondsto a type of Tû

L

�mixing which we havealreadyexcludedfrom our analysis,theÉ

secondchoiceis allowedandcorrespondsto model B1.Followingé

assumption� i¬ � ,ñ evenin this casewe neglectpos-sibleÈ T

ûL

�mixingsÐ of type B

Ú1,ñ whenanalyzingB

Ú3Ý .

Finally, the remainingthreemodelsconstitutegroup C,ñcorrespondingç to RR

Ê � � (�0,0). In In this group,TR

Ê � is an isos-inglet,¬

asis theSM Tû

R ,ñ implying thatonly LH tý -tý � mixingÐ isrelevant.¼ For C2 and¦ C3

Ý ,ñ R¾

L� doesÑ

not containa BÚ

L� ,ñ while forC1 the

ÉBL½ is not exotic. Hencein all the threecasesthe b

³neutral-current¸ couplingsareunchangedrelative to the SM,and¦ we neednot worry abouttree-levelb

³-mixing effects.

B. t!-quark loops within the standard model

Beforeexaminingthe effect of tý -tý " mixing on the radia-tiveÉ

correctionto Zbb³

,ñ we first review the SM computation.Weð

follow the notationand calculationof Bernabe´u,ß Pich,and¦ Santamarı´a¦ # 11$&% BPS

Ò '. Thecorrectionsaredueto the10

diagramsÑ

of Fig. 3. All diagramsarecalculatedin ’ tý Hooft–Feynmangauge,andwe neglecttheb

³-quarkmassaswell as

theÉ

difference ( V tbÝ ) 2¿ * 1.Due*

to theneglectof theb³

-quarkmass,anddueto theLHcharacterç of the charged-currentcouplings,the tý -quarkcon-tributionÉ

to theZbb³

vertex- correctionpreserveshelicity. Fol-lowing­

BPS we write the helicity-preservingpart of theZÏ +

bb³ ¯ scatteringÈ amplitudeas

TABLE,

III. Modelsandchargeassignments.Valuesof the weakisospinof TL- . andTR

/ 0 which,1 underthe only restrictionsof singletanddoubletHiggs representations,leadto nonzerot2 -t 3 neutralcurrentmixing. The ‘‘Model’’ columnlabelsthe morefamiliar possibilitiesforthe T 4 quarks: vectorsinglets,mirror fermions,fourth family, andvectordoublets.The othermodelsaremoreexotic.

IL5 I36

L

7IR8 I3

6R

9Model:

Group

3/2 ; 1/2 1 < 1 A1

1/2 = 1/2 1 > 1 A?

2@

0 BA

1

1/2 B 1/2 Vector doublet C I D B2

0 0 Fourthfamily C1

1/2 E 1/2 1 0 BA

3FG 1 B4

1/2 H 1/2 Vector doublet I III J B5K

0 0 C2@

0 0 1/2 L 1/2 Mirror fermions BA

6MN 1/2 B7O

0 0 Vector singlet C3F

4284 54BAMERT,P

BURGESS,CLINE, LONDON, AND NARDI

Page 11: Rb and new physics: A comprehensive analysiscliff/papers/Rb.pdf · 2002-11-12 · Rb and new physics: A comprehensive analysis P. Bamert,1 C. P. Burgess,1 J. M. Cline,1 D. London,1,2

QSRUT eVs¶ wW cX wW b

³ Y pZ 1 ,ñ [ 1 \^]`_ b³ a

pZ 2 ,ñ b 2 cedgfih qj ,ñ kml ,ñ n 25o

withëp`qsrut

0v wsxzy|{`} ,ñ ~|�`�s���

2 ����� LI � s¶ ,ñ r � ,ñ � 26�

whereë ����� represents¼ theloop-inducedcorrectionto theZbÏ ¯b³

vertex.- I(�s¶ ,ñ r)

�is a dimensionlessandLorentz-invariantform

factor which depends,a� priori,ñ on the threeindependentra-tios:É

r� � mÜ tÝ2¿ /� M¹ W2¿

,ñ s¶ � M¹

Z2¿/�M¹

W2¿

,ñ andqj 2¿/�M¹

W2¿

. For applicationsat¦ the Z

Ïresonance¼ only two of theseareindependentdueto

theÉ

mass-shellconditionqj 2 � M Z�2. Moreover,for an on-shell

Z,ñ nonresonantbox-diagramcontributionsto eV � eV ��� bb³ ¯ are¦

unimportant,ß and IÜ(�s¶ ,ñ r� )� can be treated as an effectively

gauge-invariantÌ quantity.Thecontributionsdueto the tý quarkÙ maybeisolatedfrom

other® radiativecorrectionsby keepingonly the r-dependentpart¨ of I

Ü(�s¶ ,ñ r� )� . BPSthereforedefinethe difference

F� �

s¶ ,ñ r �g� I   s¶ ,ñ r ¡g¢ I £ s¶ ,0ñ ¤ . ¥ 27¦Given§

this function, the mÜ tÝ dependenceÑ

of the width Z ¨ bb³ ¯

is obtainedusing

©b§SMª «

r� ¬g­u® b§SMª ¯

r� ° 0Á ±

1 ²´³µ gÆ Lb§

¶gÆ L

b§ ·

2¿ ¸º¹

gÆ Rb§ »

2¿ F� SMª ¼

s¶ ,ñ r� ½

¾VP¿ À

s¶ ,ñ r� Á . Â 28" Ã

In this lastequationVP (s¶ ,ñ r)�

denotesthemÜ tÝ -dependentcon-tributionsÉ

which enterÄ b§ throughÉ

the loop correctionsto thegauge-bosonÌ vacuumpolarizations.

TheÆ

functionF� SMª

(�s¶ ,ñ r� )� is straightforwardto compute.Al-

thoughÉ

the resultingexpressionsare somewhatobscure,thespecialÈ cases¶ Å 0

Árevealssomeinterestingfeatureswhich are

also¦ presentin our new-physicscalculations,andsowe showtheÉ

s¶ Æ 0Á

limit explicitly here.For s¶ Ç 0,Á

an evaluationof thegraphsÌ of Fig. 3 givesthe expressions

F1È aÉ ÊÌËÎÍ 1

2"

s¶ wW2gÆ L½tÝ

2

r� Ï r� Ð 2" Ñ

Òr Ó 1 Ô 2 lnr Õ r�

r Ö 1 × gÆ RÊtÝ r�Ø

r Ù 1 Ú 2 ln r

Û r

r� Ü 1,ñ Ý 29

" Þ

F1ß bà áÌâ 3y

cX wW24s¶ wW2¿

r2ãr ä 1 å 2

¿ ln r æ r

r ç 1,ñ è 30

y é

F1ê cë ìÌí 1î dï ðÌñ 1

121 ò 3

y2"

s¶ wW2¿r� 2ó

r ô 1 õ 2¿ ln r ö r�

r ÷ 1,ñ ø 31y ù

F� 1ú eû üÌý 1þ f ÿ�� r�

2" r��

r� � 1 � 2 ln­

r� � 1

r� � 1,ñ � 32

y �

F� 2 aÉ ��� r

4s¶ wW2gÆ R

tÝ2" ��� r � r � 2 ��

r� � 1 � 2 ln­

r� � 2r � 1

r� � 1

�gÆ L½tÝ r�

r � 1 � 2¿ ln r � r

r 1,ñ ! 33

y "

F� 2# bà $�%'& 1

8( 1 ) 1

2s¶ wW2 r� *�+ r� 2¿,

r� - 1 . 2 ln­

r� / 1

r� 0 1,ñ 1

34y 2

F2¿ 3

cë 4�5 2¿ 6

dï 7�8 1

241 9 3

y2s¶ wW2¿ r :�; r2<

r = 1 > 2¿ ln r ? 1

r @ 1,ñ A35y B

withëC�D 2

nE F 4G�HJI

ln K M W2 /4� LNM 2

¿ OQP 3y2

,ñ R 36y S

whereë nE is¬

the spacetimedimensionarising in dimensionalregularization,¼ and

gÆ L½tÝ T 1

2 U 2"3y s¶ wW2¿ ,ñ gÆ R

ÊtÝ V'W 2"3y s¶ wW2¿ . X 37

y YTheÆ

picture becomesmuch simpler after summingthe dia-gramsÌ to obtainthe total SM contribution:

F� SMª Z

s¶ [ 0,Á

r� \Q]_^i ` 1a aÉ b

2¿ c

dd e

F� i f 1

8(

s¶ wW2¿r2¿

r� g 1 h 6i r

r� j 1

k r l 3y r m 2 nor p 1 q 2

¿ ln r . r 38y s

FIG. 3. The Feynmandiagramsthrough which the top quarkcontributesto the Zb

t ¯bu

vertexv within the standardmodel.

54 4285Rw

bx AND NEW PHYSICS:A COMPREHENSIVEANALYSIS

Page 12: Rb and new physics: A comprehensive analysiscliff/papers/Rb.pdf · 2002-11-12 · Rb and new physics: A comprehensive analysis P. Bamert,1 C. P. Burgess,1 J. M. Cline,1 D. London,1,2

Therearetwo pointsof interestin this sum.First, it is ultra-violet- finite sinceall of thedivergencesy 1/z nE { 4| havecan-celled.ç This is requiredon generalgroundssincetherecanbeno¸ r� -dependentdivergencesin I

Ü SMª

(�s¶ ,ñ r� )� , and so thesemust

cancelç in FSMª

(�s¶ ,ñ r)

�. A similar cancellationalsooccurswhen

new physics is included, provided that it respects theSUÅ

L } 2" ~�� U�

Y � 1� gaugeÌ symmetryandthat thecompletesetofnew¸ contributionsis carefully included.

The secondinterestingfeatureof Eq. � 38y �

lies in its de-pendence¨ on the weakmixing angle,s¶ wW . Eachof the contri-butionsÔ

listed in Eqs. � 29" �

– � 35y �

has�

the formF� i � (

�x� i � y� is¶ wW2 )/

�s¶ wW2 ; however,all of the terms involving y� i

havecancelledin thesum,Eq. � 38y �

. This very generalresultalso¦ appliesto all of the new-physicsmodelswe considerinsubsequentÈ sections.As will be provedin Sec.V, the can-cellationç is guaranteedby electromagneticgaugeinvariance,becauseÔ

the termssubleadingin s¶ wW2¿ are¦ proportionalto theelectromagneticV b-quarkvertexat qj 2

¿ �0,Á

which mustvanish.ThisÆ

givesa powerful checkon all of our calculations.Rather than using completeexpressionsfor F(

�s¶ ,ñ r),

�we

find it moreinstructiveto quoteour resultsin the limit r � 1,whereë powersof 1/r� and¦ s¶ /� r� may� be neglected.We do thesameÈ for the ratio of massesof other new particlesto M

¹W2

whenë thesearise in later sections.Besidespermitting com-pact¨ formulas,this approximationalsogivesnumericallyac-curateç expressionsfor mostof the models’parameterrange,as¦ is alreadytrue for the SM, eventhoughr� in

¬this caseis

only� � 4. In the large-r limit FSMª

(�s¶ ,ñ r)

�becomes

F� SMª �

r� �Q� 1

8(

s¶ wW2 r� � 3y � s¶

6i � 1 � 2

"s¶ wW2 � ln

­r� �'����� ,ñ �

39y  

whereë the ellipsis denotesterms which are finite as r ¡�¢ .SeveralÅ

pointsarenoteworthyin this expression.£1¤ TheÆ

s¶ -dependenttermappearingin Eq. ¥ 39y ¦

is¬

numeri-callyç very small, changingthe coefficientof ln r from 3 to2.88.This typeof s¶ dependence

§is of evenlessinterestwhen

weë considernew physics,sinceour goal is then to examinewhetherë the new physics can explain the discrepancybe-tweenÉ

theoryandexperimentin Rb§ . That is, we want to see

if the radiativecorrectionscanhavethe right sign andmag-nitude¸ to change b

§ byÔ

the correctamount.For thesepur-poses,¨ so long as the inclusion of qj 2-dependenttermsonlychangesç thenumericalanalysisby factors © 25% ª as¦ opposedtoÉ

changingits overall sign« theyÉ

may be neglected.¬2" ­

TheÆ

above-mentionedcancellationof the terms pro-portional¨ to s¶ wW2 when® s¶ ¯ 0

Áno longer occursoncethe s¶ de-

§pendence¨ is included.This is as expectedsincethe electro-magnetic Ward identity only enforcesthe cancellationatqj 2 ° 0,

Ácorrespondingto s¶ ± 0

Áin thepresentcase.Notice that

theÉ

leading term, proportional to r� , iñ s s¶ independent,¬

andbecauseÔ

of the cancellationit is completelyattributabletographÌ ² 2a³ of� Fig. 3. All of the other graphscancelin theleading´

term. Due to its intrinsic relationwith the cancella-tionÉ

of the s¶ wW2 -dependentterms,the fact that only onegraphis responsiblefor the leadingcontributionto µ gÆ L

½b§ still¶ holdsonce� new physics is included. This will prove useful foridentifying¬

which featuresof a givenmodelcontrol theover-all· sign of the new contributionto ¸ gÆ L

b¹.

º3y »

SinceÅ

the large-r limit correspondsto particlemasses¼in this casemÜ t

Ý ½ that¾

arelargecomparedto M W and· M Z ,¿ thisisÀ

the limit where the effective-Lagrangiananalysis de-scribed¶ in Sec.II directly applies.ThenthefunctionF

�canç be

interpretedastheeffectiveZbb³

couplingç generatedwhentheheavy particle is integratedout. Quantitatively, Á gÆ L

is re-lated´

to F�

byÂ

ÃgÆ L½b¹ Ä Å

2 Æ F. Ç 40ÈÉ4Ê Ë

TheÆ

vacuumpolarizationcontributionsto Ì b¹ of� Eq.Í

28Î Ï

haveÐ

a similar interpretationin the heavy-particlelimit.In this casethe removalof the heavyparticlescangenerateoblique� parameters,which also contribute to Ñ b

¹ . In theheavy-particleÐ

limit Eq. Ò 28Î Ó

therefore¾

reducesto the first ofEqs.Ô Õ

4Ê Ö

.

C. × gØ LÙbÚ in the t

!-quark mixing models

Weð

maynow computehow mixing in thetop-quarksectorcanÛ affecttheloop contributionsto theprocessZ Ü bb

Ý ¯. As inthe¾

SM analysis,we set mÜ b¹ Þ 0.

ÁIn addition, following the

discussion§

in the previoussubsection,we neglectthe sß de-§

pendenceà in all our expressions.We alsoignoreall vacuum-polarizationà effects,knowing that they essentiallycancelinRb¹ . Finally, in the CKM matrix, we set á V id âäãæå V is çäè 0,

éwhere® i

ê ëtý ,¿ tý ì . Thusthecharged-currentcouplingsof interest

to¾

us are describedby a 2í 2Î

mixing matrix, just as in theneutral-currentsector.In theabsenceof tý -tý î mixing this con-dition§

implies ï V tbÝ ðäñ 1.Foré

tý -tý ò mixing,� independentof the weak isospinof theTû ó

,¿ we write

TûT ô

L,R õö

,Rø tý

tý ùL,R

,¿ ú L÷ û cX L

÷ü sß L

sß L÷

cX L,¿

ýRø þ cX R

sß Rø ÿ sß R

cX Rø ,¿ � 41

� �where® cX L

÷ � cosÛ �L÷ ,¿ etc. The matrices� L

÷,Rø are· analogousto

the¾

-bÝ �

mixing matricesdefinedin Eq. � 5� in our tree-levelanalysis· of b

Ýmixing.�

In

the presenceof tý -tý � mixing,� the diagonal neutral-currentÛ couplingsaremodified:

g� L,Ri �

a� � T,T � g� L,Ra� ���

L,Rai� � 2 � g� L,R

tÝ ,SM � g� L,Ri ,¿ � 42

� �

where® iê �

tý ,¿ tý � ,¿ andg� L,RtÝ ,SM are· theSM couplingsdefinedin Eq.�

37� �

. The new termsg� L÷

,Røi explicitly� read

g� LtÝ I

!3"

L

#%$ 1

2Î sß L

2,¿ g� RtÝ & I!

3"

R

'sß R

2 ,¿ ( 43� )

g� L÷ tÝ *�+ I3

"L÷,%- 1

2cX L÷2. ,¿ g� R

ø tÝ /�0 I3"

Rø1 cX Rø2. . 2 443

In

addition,wheneverthe Tû

L,R

4hasÐ

nonstandardisospinas-signments,¶ I

!3"

L

5%61/2 or I

!3"

R

7980é

, flavor-changing neutral-currentÛ : F.C.N.C.; couplingsÛ arealso induced:

4286 54BAMERT,P

BURGESS,CLINE, LONDON, AND NARDI

Page 13: Rb and new physics: A comprehensive analysiscliff/papers/Rb.pdf · 2002-11-12 · Rb and new physics: A comprehensive analysis P. Bamert,1 C. P. Burgess,1 J. M. Cline,1 D. London,1,2

g� L÷

,Røi j < =

a� > T,T ? g� L÷

,Røa� @

,Røai� A

,Røa j� B

g� L÷

,Røij ,¿ C 45D

where® iê,¿ jE F tý ,¿ tý G ,¿ and i

ê HjE. Here,

g� L÷ ttÝ IKJ 1

2 L I3"

L÷M sß LcX L ,¿ g� R

ø ttÝ N�O I3"

RøP sß RcX R . Q 46R

EquationS T

41� U

determinesV

the effective tý and· tý W neutral-XcurrentÛ couplings Y Eqs.

S Z42� [

– \ 46� ]K^

. However, the charged-currentÛ couplingsdependon thematrix _a`%b L

† cLB . Hencewe

needto consideralso bÝ

mixing, since,as discussedin Sec.IV

A, in thosecasesin which the Bd e

quarkf is not exotic(�I!

3"

L

g B hji 1/2, I!

3"

R

k B l 0é m

,¿ we haveno reasonto require n LB o I

!pi.e., no b

Ý-bÝ q

mixingr . We thendefinethe2s 2 chargedcur-rentt mixing matrix

uavxwL† y

LB ,z { tbÝ | cX L

÷ cX LB } sß L

÷ sß LB ,z ~ tÝ � b� � sß L

÷ cX LB � cX L

÷ sß LB ,z �

47�which� trivially satisfies the orthogonality conditions��� † ��� †�a� I. In the absence of b

�-b� �

mixing, clearly�tbÝ � cX L

÷ ,z � tÝ � b� � s� L÷ . We also note that, by assumption,

whenever� �a���L÷ we� necessarilyhave I3

"L÷�%�j� 1/2 � so� that

I!

3"

L

� B� �j� 1/2  in¡

order to guaranteethat the Bd

L¢ is¡

not exotic.

From£

Eqs. ¤ 43¥ ¦

,z § 44¥ ¨

,z and © 46¥ ª

,z this implies that g� LtÝ « g� L

tÝ ¬­ g� L

® ttÝ ¯K° 0±

, that is, the mixing effects on the LH tý and² tý ³neutral-current´ couplingsvanish.

Theµ

Feynmanrules of relevancefor computingthe Zb¶ ¯b�

vertex· loop correctionsin the presenceof a mixing in thetop-quark¸

sectorcannow be easilywritten down:

Wt ib�

:igê¹»º tÝ i¼ b� ½¿¾À½ L

® ,z

Átý ib� :

igê

ÂM W

ÃtÝ i¼ b� mÜ i Ä L ,z

Zt¶

itý

i :igêcX wW Å¿ÆÈÇÊÉ gË L

tÝ ,SM ÌL® Í gË R

tÝ ,SM ÎRÏ ÐÒÑÔÓ tý LtÝ i¼ Õ

L® Ö gË R

tÝ i¼ ×RÏ ØÚÙ ,z

Zt tÛ Ü : igêcX wW Ý¿ÞÈß gË L

® ttÝ à�áL â gË R

Ï ttÝ ãåäR æ ,z ç 48è

where� é are² theunphysicalchargedscalars,andtÛ i ê tÛ ,z tÛ ë . Thevertices· listedin Eq. ì 48

¥ íreducet to theSM Feynmanrulesin

the¸

limit of no mixing.As pointedout at theendof Sec.IV A, in somegroupsof

modelsî equalitiescan be found betweensomeelementsofthe¸

-Tï ð

and² Bd

-Bd ñ

massî matrices.Thesehave importantconsequences.ò In particular,onceexpressedin termsof thephysicaló massesandmixing angles,theequalitiesof Eq. ô 23õöwhich� hold in the modelsof groupA

÷ øcanò be written

ù�úLMû

diagd ü

R† ý

aþ 2 ÿ���� LBMû

diagdB �

RB†�

aþ 2 ���� LB

aþ 2mÜ b� � cX R

B

�a� 1,2� ,z � 49�

where� Mû

diagd � diag[

VmÜ tÝ ,z mÜ tÝ � ]� , and we have used M

ûdiagdB���

i2� mÜ b� ��� recallt thatwe takemÜ b

� � 0± �

. Multiplying now on theleft�

by ( � LB†)�

1aþ and² summingover a� we� obtain

�! †M diagd "

Rφ #

12$ mÜ tÝ % tbÝ s� R & mÜ tÝ ')( tÝ * b� cX R + 0.± ,

50- .

For/

the modelsin groupBd

,z the vanishingof Mû

12 implies¡

nob�

mixing. Then 02143 L® ,z andEq. 5 496 still� holds in the limit7

tbÝ 8 cX L ,z 9 t

Ý : b� ; s� L . For themodelsin groupC no particularrelationt betweenmassesand mixing anglescan be derived.For/

example,it is clear that in the fourth family model C1,zEq.< =

50- >

doesV

not hold. However, for all thesemodelsI?

3@

R

AB 0±

. Hence,noting that all the gË R couplingsò in Eqs. C 43¥ D

,zE44¥ F

,z and G 46¥ H

are² proportional to I?

3@

R

I,z and defining r� JK mÜ tÝ L2 /

MM W

2 ,z squaringEq. N 50- O

yieldsP a relation which holdsforQ

all modelsin TableIII:

RtbÝ 2gË R

tÝr� SUT t

Ý Vb�2 gË R

tÝ W

r� XZY�[U\tbÝ ] tÝ ^ b� gË R

ttÝ _�`

rr� a . b 51c d

This relation is used extensivelyin the calculation whichfollows.Q

Howe

do we generalizethe SM radiativecorrectionto in-cludeò tf -tf g mixing?First notethat for eachof thediagramsinFig. 3, thereis alsoa diagramin which all the tf quarksh arereplacedt by tf i quarks.h Second,thereare two new diagramsjFig.k

4l dueV

to theFCNCcouplingof theZ¶

to¸

thetf and² tf m . Soto¸

generalizethe SM result to the caseof mixing, threethings¸

haveto be done: n i o multiply Eqs. p 29q – r 35s t

byu v

tbw2xfor the tf contributionò and y tw z b�2x for tf {}| with� r ~ r �)� ,z � ii � replacegË L,R

tw byu

themodifiedcouplingsin Eq. � 42¥ �

,z addingEqs. � 43¥ �

and² � 44� respectivelyfor tf and² tf � ,z and, � iii � includediagrams3s �

a² � and² 3� bu ��� Fig. 4� correspondingò to the FCNC couplings�Eqs.� �

45¥ �

and² � 46¥ ���

.A�

glanceat the Feynmanrules in Eq. � 48¥ �

shows� that inthe¸

first step � i¡ � ,z a correctionproportionalto gË L,Rtw ,SM ,z and in-

dependentV

of the gË L®

,R  couplings,ò is generated.This correc-

tion¸

is commonto all modelsin TableIII—it appearseveninthe¸

casein which the tf NC¡

couplingsarenot affected¢ fourthQ

family£ . In contrast,steps ¤ ii ¥ and² ¦ iii § generate¨ correctionswhich� differ for different models.It is useful to recasttheminto¡

two types,one proportionalto the LH neutral currentcouplingsò ©�ª�«

ib ¬ jb­ gË L ® ,z andthe otherproportionalto the RH

neutralcurrentcouplings ¯�°�± ib ² jb­ gË R

  ³ . The LH andRH cor-

FIG. 4. The additionalFeynmandiagramswhich are requiredfor models in which the t quark´ mixes with an exotic, heavy tµ ¶quark.

54 4287R·

b¸ AND NEW PHYSICS:A COMPREHENSIVEANALYSIS

Page 14: Rb and new physics: A comprehensive analysiscliff/papers/Rb.pdf · 2002-11-12 · Rb and new physics: A comprehensive analysis P. Bamert,1 C. P. Burgess,1 J. M. Cline,1 D. London,1,2

rectionst vanishrespectivelyfor I¹

L

»U¼�½1/2 andI

¹3º

R

¾À¿0±

, whenthe¸

correspondingneutral-currentcouplingsarenot affectedbyu

the mixing.InÁ

the presenceof mixing, the correctiondue to the dia-grams¨ of Fig. 3 involving internal tf quarksh becomes

Âi à 1Ä aÉ Å

2x Æ

dd Ç

F i ÈUÉtbw 2 Ê FSM

Ë Ìr ÍÏÎ F Ð gË L

®,R tw ,z r ÑÓÒ ,z Ô 52

c Õ

whereÖ F× SMË

(ØrÙ )Ú is given by Eq. Û 38

s ÜandÝ

F Þ gË L®

,R tw ,z r ßÏà 1

s� wâ2x gË L® tw r 2 ã 4

¥r ä 1

ln r å gË R  tw r æUç 2

èrÙ é 5

cr ê 1

ë rÙ 2 ì 2è

rÙ í 4¥

îr ï 1 ð 2 ln r . ñ 53

c ò

Theµ

third step ó iii¡ ô gives¨ rise to a new contribution

F× 3º õ

aÉ öø÷ F× 3º ù

bú ûýüUþ

tbw ÿ

tw �

b� Fט � gË L,R

ttw � ,� r٠,� r٠��� . � 54c �

Evaluating�

diagrams3 aÝ andÝ 3� bu �� Fig.k

4� weÖ find

F3º �

aÉ ����� 1

s� wâ2x � tbw � tw � b� 1

2gË L® ttw � 1

r ��� r

r � 2r ��� 1

ln r ! rÙ 2

r " 1ln r # gË R

  ttw $�% rr & 1

r '�( r

rÙ )r *,+ 1

ln r -. r

rÙ / 1ln0

rÙ ,� 1 55c 2

F3º 3

bú 4�5 1

s� wâ2x 6 tbw 7 tw 8 b9 2gË L® ttw : rrÙ ;

r <�= r

rÙ >r ?,@ 1

ln r A�B rÙr C 1

ln r

D gË R  ttw EGF rr H IKJ 1 L 1

r M�N r

rÙ O 2r P�Q 1

ln r RS r2

xrÙ T 1

ln0

rÙ . U 56c V

Puttingall thecontributionstogether,for thegeneralcaseweÖ find

F WYXi Z 1[ aÉ \

3º ]

bú ^

F i _a`j­ b

1,2 c tw jd b92x e

FSMË f

r j­ gih F j gË L

k,R tw jd ,� r j

­ l�mnpo

tbw q tw r b9 F s gË L,Rttw t ,� r,� r u�v ,� w 57

c xwhereÖ tf j

­ y tf ,� tf z andÝ r j­ { r,� r | . We notethatdueto Eq. } 51

c ~allÝ

the�

divergenttermsproportionalto g� R � cancel� in the sum.Now,¡

the correction� � g� Lb9 �

(Ø �

/2M �

)ÚX�

corr� to�

theSM resultcanbe�

explicitly extractedfrom Eq. � 57c �

by�

meansof therelation�tbw 2 � 1 ��� tw � b92 .

Moreover,asanticipatedit is possibleto divide the vari-ous� contributionsto X

�corr� into�

threedifferent pieces: a uni-versal� correction,a correctiondueto LH mixing only, andacorrection� dueto the RH mixing. Hencewe write

Xcorr� � F � FSMË �

Xcorr�univ� �Xcorr�LH� �

Xcorr�RH�

,� � 58c �

whereÖXcorr�univ�  p¡

tw ¢ b92x £ FSMË ¤

r ¥�¦i§ FSMË ¨

r ©�ª ,� « 59c ¬

Xcorr�LH� ­p®

tbw 2x F ¯ g� Lk tw ,� r °i±p² tw ³ b92x F ´ g� L

k tw µ ,� r ¶�·i¸p¹ tbw º tw » b9 F ¼ g� L½ ttw ¾ ,� r,� r ¿�À ,�Á

60Â Ã

X�

corr�RH ÄpÅtbw 2Fט Æ gÇ R

tw ,� rÙ ÈiÉpÊ tw Ë b92 Fט Ì gÇ R

tw Í ,� rÙ Î�ÏiÐpÑ tbw Ò tw Ó b9 Fט Ô gÇ Rttw Õ ,� rÙ ,� rÙ Ö�× .Ø

61Â Ù

UsingÚ

theexplicit expressionsfor gÇ L,Rtw ,� gÇ L

½,RÛtw Ü ,� andgÇ L

½,RÛttw Ý asÝ

givenÞ in Eqs. ß 43à ,� á 44â ,� and ã 46ä above,Ý togetherwith rela-tion� å

51c æ

for the RH piece,theseread

Xcorrçunivè épêtw ë b92x fì

1corrç í

r,� r î�ï ,� ð 62 ñ

corrçLH óõô 1 ö 2÷

L

øúù,ûtbw ü tw ý b9 sþ L

½ cÿ L½ fì

2corrç �

r٠,� r٠��� ,� � 63 �

XcorrçRH� ��

2I3º

R���

tbw 2x sþ RÛ2x fì

3ºcorrç �

r,� r ��� ,� � 64 �

withÖ

1corrç �

r,� r ����� 1

sþ wâ2rÙ ��� rÙ ��� 6

� �r �! 1

" rÙ #%$ 3s rÙ &�' 2÷ (

)r *�+ 1 , 2

x ln r -�. rÙ / rÙ 0 61 2

r 3 1 4rÙ 5 3s rÙ 6 2

÷ 78r 9 1 : 2

x ln r ,� ; 65< =

2corrç >

rÙ ,� rÙ ?�@�A 1

sþ wâ2xcÿ L B tw C b9sþ L D tb

w E rÙ F�G 2r HrÙ I�J 1

ln0

rÙ K L sþ L M tbwcN L O t

w Pb9 Q rÙ R 2r

rÙ S 1ln0

rÙ T 2r U 2x V r W 1 XYrÙ Z�[ 1 \^] rÙ _�` rÙ a ln

0rÙ b�c 2r2

x dr e�f 1 gh

rÙ i 1 j^k rÙ l�m rÙ n ln0

rÙ ,�o66< p

fq

3ºcorrr s

rÙ ,� rÙ t�u�v 1

sw wâ2 rÙ x 1

2y 2y

rÙ z 5c

rÙ { 1| r2

x }2y

rÙ ~ 4�

�rÙ � 1 � 2 ln

0rÙ � 1

2y 2y

r٠��� 5c

r٠��� 1� r � 2x � 2

yr٠��� 4

��r٠��� 1 � 2 ln

0rÙ � � 4

� 1

rÙ ��� rÙrÙ �

r٠��� 1ln0

rÙ ��� rÙrÙ � 1

ln0

�1 � 1

rÙ ��� rÙr � 2x

rÙ  �¡ 1ln0

rÙ ¢�£ r2x

rÙ ¤ 1ln0

rÙ . ¥ 67< ¦

4288 54BAMERT,§

BURGESS,CLINE, LONDON, AND NARDI

Page 15: Rb and new physics: A comprehensive analysiscliff/papers/Rb.pdf · 2002-11-12 · Rb and new physics: A comprehensive analysis P. Bamert,1 C. P. Burgess,1 J. M. Cline,1 D. London,1,2

Note¡

that a value of V tbw differentV

from unity can be easilyaccountedÝ for by using the unitary condition¨ ©

tbw ª 2 «­¬ ® tw ¯ b9 ° 2 ±³² V tbw ´ 2 µ 1 ¶¸· V tsw ¹ 2 º¸» V tdw ¼ 2 in�

Eqs. ½ 62< ¾

– ¿ 67< À

.As�

we havealreadypointedout, becauseof our require-ment of no B-B Á mixing when the B Â is exotic, only whenI3º

LÃÄÆÅÈÇ 1/2 canwe havecN L

à ÉËÊtbw ,Ì sÍ L

à ÎÐÏtw Ñ bÒ . However,in this

caseÓ XÔ

corrÕLH vanishes.Ö Hence,without lossof generality,we canset× the LH neutralcurrentmixing equalto the chargedcur-rent mixing in X corrÕLH

Ø,Ì obtaining

XcorrÕLHØ Ù¸Ú

1 Û 2I3º

LÃÜÞÝ�ß

tbw 2x à tw á bâ2x

2xcorrÕ ä

r,Ì r å�æ ,Ì ç 68< è

2xcorrÕ é

r,Ì r ê�ë�ì 1

sÍ wí2x î¸ï r ð r ñ�ò�ó 2rr ôr õ�ö r

lnr ÷r

. ø 69< ù

FromEqs. ú 62< û

,Ì ü 64< ý

,Ì and þ 68< ÿ

we� seethat thereareonly twoindependentmixing parametersrelevant for the completeanalysis� of our problem: theLH matrix element

�tbw and� the

RH�

mixing sÍ R� . Furthermore,notethat asr� ��� r� ,Ì all the cor-

rectionsin Eqs. 65<

,Ì � 67< �

,Ì and 69< �

vanish,Ö independentofthe�

mixing angles.This comesaboutbecauseof a GIM-likemechanismfor all the pieceswhich do not dependon I3

ºR�� .

The�

R

�-dependentcontributionfrom the RH fermionscou-

pling� to the Z vanishesÖ in the limit r ��� r as� a consequenceof� Eq. � 50

c �.

In�

thelimit r� ,Ì r� ��� 1, for thefunctionsfã

icorrÕ

(�r� ,Ì r� � )� we obtain

1corrÕ �

r,Ì r "!$# 1

sÍ wí2 r %�& r ' 3 ls

nr� (r

,Ì ) 70* +

2corrÕ ,

r� ,Ì r� -�.$/ 1

sÍ wí2 021 r� 3 r� 4�5$6 2y

rr� 7r� 8�9 r� ln

0 r� :r� ,Ì ; 71

* <

3ºcorrÕ =

r� ,Ì r� >"?$@ 1

sÍ wí2x A r� B 1

2y 1 C r

r� Drr E

r� F�G r� ln0 r H

r�I 3

sr

r� JLK r� lnr Mr� N 3

s2y 1 O r

r� P . Q 72* R

Let usnow considerthenumericalvaluesof thesecorrec-tions�

in moredetail.UsingmS tw T 180GeV, MU

W V 80á

GeV,andsÍ wí2 W 0.23,

XEq. Y 38

s Zgives[ a SM radiativecorrectionof

F\ SM] ^

4.01.� _

73* `

The questionis whetherit is possibleto cancelthis correc-tion,�

thuseliminatingtheRbâ problem,� by choosingparticular

valuesÖ of mS tw a and� the mixing angles.For variousvaluesofmS tw b ,Ì the valueof X

ÔcorrÕ c Eq.

� d58c e�f

isg

shownin TableIV.Weh

seethat even for mS tw i�j mS tw , iÌ t isk

possible� to chooseI3º

LÃl ,Ì I3

ºR�m ,Ì and the LH and RH mixing anglessuchthat the

correctionÓ is negative.So the discrepancyin Rn

bâ betweeno

theory�

and experimentcan indeedbe reducedvia tf -tf p mix-qing.

Referring to the modelslisted in Table IV, the optimalchoiceÓ for the weak isospinof the T r is I3

ºLÃsutwv

1/2 and I3º

R�xywz 1, regardlessof thevalueof mS tw { . Furthermore,maximal

RH�

mixing, sÍ R2 | 1, is also preferred.However, even with

these�

choices,it is evidently impossibleto completely re-movetheRb

â problem.� Fromtheabovetable,thebestwe candoV

is to takemS tw }�~ 75*

GeV and � tw � bâ2 � sÍ LÃ2 � 0.6,

Xin which case

the�

total correctionis XcorrÕ �u� 3.68.s

This leavesa 1.5� dis-V

crepancyÓ in Rn

bâ ,Ì which would put it in the categoryof the

other� marginal disagreementsbetweenexperimentand theSM.�

However,sucha light tf � quark� hasother phenomeno-logical0

problems.In particular,CDF hasput a lower limit of91�

GeV on charge2/3 quarkswhich decayprimarily to Wb�24� . Unlessoneaddsothernew physicsto evadethis con-

straint,× the lightesttf � allowed� is aboutmS tw ��� 100GeV. In thiscase,Ó maximal LH mixing ( � tw � bâ2

x �sÍ LÃ2x � 1) gives the largest

effect:� XÔ

corrÕ �u� 2.7.y

The predictedvalueof Rn

bâ isg

thenstillsome× 2� below

othe measurednumber.

Anotherpossibility is that the charge2/3 quarkobservedbyo

CDF is in fact the tf � ,Ì while the real tf quark� is muchlighter,0

saymS tw � 100 GeV. Assumingsmall tf -tf � mixing,q andthat�

the tf � isg

the lightestmemberof thenewmultiplet, the tf �will� then decayto Wb,Ì as observedby CDF, but the SMradiativecorrectionwill be reduced.This situationis essen-tially�

identicalto that discussedabove,in which the LH tf -tf �mixingq is maximal, and mS tw �"� 100 GeV: the SM value ofRbâ will� still differ from the experimentalmeasurementby

about� 2� . The only way for such a scenarioto work is ifmS tw   M

UW . However,newphysicsis thenonceagainrequired

to�

evadethe constraintfrom Ref. ¡ 24y ¢

.For all the possibilitiesof this sectionour conclusionis

TABLE IV. Dependenceof the tµ -tµ £ mixing resultson m¤ t¥ ¦ :§ This table indicatesthe dependenceon themassof the tµ ¨ quarkof the correctionsto g L

dueto tµ -tµ © mixing, with the t massfixed at 180 GeV.

m¤ t¥ ª«GeV¬ ­

Xcorr®75 ¯ 3.31° t

± ²b³2´ µ 1.21(1 ¶ 2

·I¸

L

º)» ¼

t± ½

b³2´ ¾

tb± 2´ ¿ 1.39(2I

¸3¹

R

À)» Á

tb± 2´ sR

100 Â 2.70Ã t± Ä b³2 Å 0.71(1Æ Ç

LÈÉ )» Ê

t± Ë b³2 Ì tb± 2 Í 0.59(2I¸

RÎÏ )» Ð

tb± 2sRÎ 2

125 Ñ 1.97Ò t± Ó b³2´ Ô 0.34(1Æ Õ

2I3¹

L

Ö)» ×

t± Ø b³2´ Ù tb± 2´ Ú 0.22(2I3¹

R

Û)» Ü

tb± 2´ sR2´

150 Ý 1.14Þ t± ß b³2 à 0.10(1Æ á

LÈâ )» ã

t± ä b³2 å tb± 2 æ 0.05(Æ

2I¸

RÎç )» è

tb± 2sRÎ 2

175 é 0.20ê t± ë b³2 ì 0.003(1Æ í

2I3¹

LÈî )» ï

t± ð b³2 ñ tb± 2´ ò 0.001(2I3¹

RÎó )» ô

tb± 2´ sRÎ 2´

200 0.84õ t± ö

b÷2ø ù 0.04(

ú1 û 2I

ü3ý

L

þ) ÿ t� �

b÷2ø �

tb� 2ø � 0.02(2Iü

R

�) � tb� 2ø s� R

225 1.97� t� � b÷2 0.23(ú

1 2Iü

L�� ) t� � b÷2 � tb� 2 � 0.07(2I

ü3ý

R�� ) � tb� 2s� R

� 2250 3.20� t� � b÷2ø � 0.55(1 � 2I3

ýL

�) � t� � b÷2ø � tb� 2ø � 0.15(

ú2I3ý

R

�) � tb� 2ø sR

275 4.52� t� b÷2 ! 1.01(1 " 2Iü

L

#) $ t� % b÷2 & tb

� 2 ' 0.24(ú

2Iü

R

() ) tb� 2sR

2

300 5.93* t� + b÷2 , 1.61(1 - 2I3ý

L�. ) / t� 0 b÷2 1 tb� 2 2 0.34(

ú2I3ý

R�3 ) 4 tb� 2sR

� 2

54 4289R·

b¸ AND NEW PHYSICS:A COMPREHENSIVEANALYSIS

Page 16: Rb and new physics: A comprehensive analysiscliff/papers/Rb.pdf · 2002-11-12 · Rb and new physics: A comprehensive analysis P. Bamert,1 C. P. Burgess,1 J. M. Cline,1 D. London,1,2

therefore�

the same: it is not possibleto completelyexplainRbâ through�

tf -tf 5 mixing. The bestwe can do is reducethediscrepancyV

betweentheory and experimentto about 26 ,Ìwhich� might turn out to be sufficient, dependingon futuremeasurements.

V.7

ONE-LOOP EFFECTS: OTHER MODELS

Another way to changeg8 L9bâ at� the one-loop level is to

introduceexotic new particlesthat coupleto both the Z and�the�

quark.� One-loopgraphsinvolving such particlescanthen�

modify the Zb: ¯bÒ

vertexÖ as measuredat LEP and SLC.Recall oncemore the conclusionfrom Sec.II: agreementwith� experimentrequiresthe LH b

Ò-quark coupling, g8 L

9bâ , tÌ oget[ a negative correction comparablein size to the SMm; tw -dependentcontributionssince loop-level changesto g8 R

are� too small to be detectable.In this sectionwe first exhibit the generalone-loopcor-

rection< dueto exoticnewscalarandspin-halfparticles,withthe�

goalof identifying thefeaturesresponsiblefor theoverallsign× and magnitudeof the result. We then usethis generalresult to investigatea numberof morespecificcases.

The=

answer is qualitatively different depending onwhether� or not the new scalarsand fermionscan mix, andthus�

haveoff-diagonalcouplingsto the Z boson.o

We there-fore treatthesetwo alternativesseparately.Thesimplestcaseisg

when all Z:

couplingsÓ are diagonal,so that the one-loopresults< dependonly upon two masses,thoseof the fermionand� the scalarin the loop. Then the correctionto the ZbbvertexÖ is given by a very simple analytic formula, whichenables� us to easilyexplainwhy a numberof modelsin thiscategoryÓ give the ‘‘wrong’’ sign, reducing > b

â rather thanincreasingit.

More?

generallyhowever,the new particlesin the loopshave@

couplingsto theZ:

which� arediagonalonly in theflavorbasiso

but not the masseigenstatebasis,so the expressionsbecomeo

significantly more complicated.This occursin su-persymmetric� extensionsof thestandardmodel,for example.After�

proposingseveralsamplemodelswhich can resolvethe�

Rbâ problem,� we useour resultsto identify which features

of� supersymmetricmodelsare instrumentalin so doing.

A. Diagonal couplings to the ZA

:B General results

WeC

now presentformulas for the correctionto the ZbbÒ

vertexÖ due to a loop involving genericscalarand spin-halfparticles.� In this sectionwe makethesimplifying assumptionthat�

all of the Z:

-bosoncouplingsare flavor diagonal.ThisconditionÓ is relaxedin later sectionswhere the completelygeneral[ expressionis derived.Theresultingformulasmakeitpossible� to seeat a glancewhethera given modelgives theright< signfor alleviatingthediscrepancybetweenexperimentand� the SM predictionfor Rb

â .The one-loopdiagramscontributingto the decayZ D bb

Ò ¯canÓ begroupedaccordingto whetherthe loop attachesto thebÒ

quark� E i.e., thevertexcorrectionandself-energygraphsofFig. 5F or� whether the loop appearsas part of the gaugebosono

vacuumpolarization G Fig.k

6H . For the typesof modelswe� considerthesetwo classesof graphsareseparatelygaugeinvariant and finite, and so they can be understoodsepa-rately.This is particularlyclearin the limit that the particleswithin� the loop are heavycomparedto M

IZ ,Ì sincethen the

vacuumÖ polarizationgraphsrepresentthe contributionof theoblique� parameters,S

Jand� T

K,Ì while the self-energyand

vertex-correctionÖ graphsdescribeloop-inducedshifts to thebÒ

-quarkneutralcurrentcouplings, L g8 L9

,RMb

â.

Furthermore,althoughwe must ensurethat the obliqueparameters� do not becomelarger thanthe boundof Eq. N 3s O ,ÌEq.� P

4� Q

shows× that they largely cancelin the ratio RR

bâ . We

therefore�

restrictour attentionin this sectionto thediagramsof� Fig. 5 by themselves.Thesumof thecontributionsof Fig.5c

is also finite as a result of the Ward identity which wasalluded� to in Sec.III. This Ward identity relatesthe vertex-part� graphsof Figs.5S a� T and� 5U bo V to

�theself-energygraphsof

Figs. 5W cÓ X and� 5Y dV Z . Since this cancellationis an importantcheckÓ of our results,let us explainhow it comesabout.

WeC

first consideran unbrokenU [ 1\ gauge[ bosonwith atree-level�

couplingof g8 bâ to�

thebÒ

quark.� This givesriseto thefamiliar Ward identity from quantumelectrodynamics:forexternal� fermionswith four-momentap] and� p] ^ ,Ì

_p] ` p] acbedgf dih g8 effj k SJ Fl 1 m p] npo S

JFq 1 r p] sctvu ,Ì w 74

* x

where� y{z isg

the one-particle-irreduciblevertex part andSJ

F| (}p] )~

is thefermionpropagator.If we denotethevertex-partcontributionsÓ � Figs. 5� a� � and� 5� bo �c� to

�the effectivevertexat

zero momentum transfer by � g8 bâ ,Ì and the self-energy-

inducedg

wavefunctionrenormalizationof thebÒ

quark� by Z:

bâ ,Ì

FIG. 5. The one-loopvertex correctionand self-energycontri-butionsto the Zbb

� ¯ vertexdueto fermion-scalarloops.

FIG. 6. Theone-loopcontributionsto theZbb�

vertexdueto thegauge-bosonvacuumpolarizations.

4290 54BAMERT,�

BURGESS,CLINE, LONDON, AND NARDI

Page 17: Rb and new physics: A comprehensive analysiscliff/papers/Rb.pdf · 2002-11-12 · Rb and new physics: A comprehensive analysis P. Bamert,1 C. P. Burgess,1 J. M. Cline,1 D. London,1,2

then�

at one loop the Ward identity � 74* �

reduces tog8 bâ (1} �

Zbâ )(~ p] ��� p] ��� )~ � (

}g8 bâ ��� g8 b

â )(~ p] ��� p] ��� ),~ or

�g8 bâ � g8 b

â Zbâ � 0.

�  75* ¡

This=

lastequationis themoregeneralcontextfor thecancel-lation which we found in Sec. III; it statesthat the self-energy� graphs¢ Figs.5£ cÓ ¤ and� 5¥ dV ¦¨§ mustpreciselycancelthevertexÖ part © Figs.

k5ª a� « and� 5¬ bo ­c® in

gthe limit of zeromomen-

tum�

transfer.Another way of understandingEq. ¯ 75* °

isg

toimaginecomputingthe effectiveb

Ò-photonvertexdue to in-

tegrating�

out a heavyparticle.Equation ± 75* ²

is theconditionthat�

the two effectiveoperatorsbÒ ³ ´ bÒ and� b

ÒAµ ¶

have@

the rightrelative< normalizationto begroupedinto thegauge-covariantderivative:V

D· bÒ

.But for the externalZ boson,

othe Ward identity only ap-

plies� to thosepartsof the diagramswhich are insensitivetothe�

fact thattheU ¸ 1¹ symmetry× is now broken.Theseincludethe�

1/º nE » 4¼ poles� from dimensionalregularization,andalsothe�

contributionsto the bÒ

neutral-currentcoupling propor-tional�

to sÍ wí2 ,Ì sincethe latter ariseonly throughmixing fromthe�

couplingsof the photon.WeC

now returnto thediagramsof Fig. 5. Thefirst stepisto�

establishtheFeynmanrulesfor thevariousverticeswhichappear.� Sincewe careonly abouttheLH neutral-currentcou-plings,� it sufficesto considercouplingsof the new particlesto�

L9 :

½scalar¾ y¿ f

À Ái fã Ã

LbÒ Ä

H.c. Å 76* Æ

and� we write the Z:

couplingÓ to fã

and� Ç as�È

NCÉ Ê eË

sÍ wí cÌ wí Z Í�Î fã ÏÑÐ�Ò g8 L9 fÀ Ó

L9 Ô g8 R

M fÀ ÕRM Ö fã ×

igØ

SÙ Ú † Û ÜÞÝißáà . â

77* ã

The couplings, g8 äæå g8 L9 fÀ ,Ì g8 R

M fÀ ,Ì g8 SÙ ç ,Ì are normalized so that

g8 è I3º é Qs

êwí2x for all fields, f

ãL9

,RMaë and� ì mí .

Inî

the exampleswhich follow, the field fã

canÓ representeither� anordinaryspinor ï e.g.,� tf ð or� a conjugateÌ spinor× ñ e.g.,�tf cò ó . This differencemustbe kept in mind wheninferring thecorrespondingÓ charge assignmentsfor the neutral-currentcouplingsÓ of the f

ã. For example,the left-handedtop quark

has@

L ôöõ 12÷ , sÌ o g8 L

fÀ ø

12÷ ù 2

÷3ú sÍ wí2 and� I

¹3º

R û 0,�

so g8 RfÀ üþý

2÷3ú sÍ wí2 . If

the�

internal fermion were a top antiÿ quark,� however, wewould� insteadhaveg8 R

M fÀ � � 12� 2

3ú sÍ wí2x and� g8 L

9 fÀ ��� 23ú sÍ wí2x . Thelat-

ter�

couplingsfollow from the former using the transforma-tion�

of the neutral current under chargeconjugation:Ó ����

L9 ��� ��

RM .

WeC

quotetheresultsfor evaluatingthegraphsof Fig. 5 inthe�

limit whereMI

Z

�and� of coursem; b

â � are� negligiblecom-pared� to m; f

À and� MI �

,Ì sincetheyarequitesimpleandillumi-natingin this approximation.It will be shownthat the addi-tional�

correctionsdueto thenonzeromassof theZ bosono

aretypically�

lessthan10% of this leadingcontribution.WeC

find that

�g8 L9bâ � 1

32s � 2

�fÀ � nE cò � y¿ f

À ��� 2 � 2 � g8 L9 fÀ � g8 R

M fÀ "!$# r %&('*)

g8 RfÀ +

g8 Lbâ ,

g8 SÙ -/.10325476˜ 8 r� 9/:<; ,Ì = 78

* >

where� ? (}r� )~ and @˜(

}r� )~ are functions of the mass ratio

r� A m; fÀ2/BMI C2 ,Ì

D3Er� FHG r�I

r� J 1 K 2 L r� M 1 N ln0

r� O ,Ì P 79* Q

R˜ S r THU r�Vr W 1 X 2

Y Z r [ 1 \ r ln r ] . ^ 80á _

`badenotesV

the divergent combinationcbdfe2/g nE h 4ikjml�n ln o M p2Y /4B q5r 2

Y skt12,Ì and nE cò is a color factor

that�

dependson theSUcò u 3v w quantum� numbersof thefields xand� f

ã. For example,nE cò y 1 if z|{ 1

}or� fã ~

1} �

colorÓ singlets� ;nE cò � 2 i

�f fã �

3�

and� ��� 3�

or� 6�; nE cò � 16

3ú ifg

fã �

3�

and� �|� 8�.

The cancellationof divergenceswe expectedon generalgrounds[ is now evidentin thepresentexample,becauseelec-troweak�

gaugeinvarianceof the scalarinteraction � 76* �

im-g

plies� that the neutral-currentcouplingsarerelatedby

g8 SÙ � g8 L

9bâ � g8 RM fÀ � 0.

� �81� �

This forcesthe term proportionalto �˜ to�

vanishin Eq. � 78* �

.As advertisedthe remaining term is both ultraviolet finiteand� independentof sÍ wí2 ,Ì which cancelsin the combinationg8 L9 fÀ � g8 R

M fÀ .WeC

are left with the compactexpression

�g8 L

bâ � 1

16� 2Y �

fÀ � nE cò   y¿ f

À ¡�¢ 2Y £ g8 LfÀ ¤

g8 RfÀ ¥"¦3§

m; fÀ2/BMI 2 © . ª 82

� «

Interestingly,î

it dependsonly on theaxial-vectorcouplingofthe�

internal fermion to the gaugebosonW3º and� not on the

vectorÖ coupling.The functionof themasses¬ (}r)~

is positiveand� monotonically increasing,with ­ (

}r)~ ®

r as� r ¯ 0�

and°3±³²�´kµ1, ascanbe seenin Fig. 7.

Itî

is straightforwardto generalizeEq. ¶ 82� ·

to�

includetheeffect� of thenonzeroZ boson

omass.Expandingto first order

in M Z2Y,Ì oneobtainsan additionalcorrectionto the effective

vertex,Ö

FIG. 7. From top to bottom, the functions ¸º¹ m» f¼2/½M¾ ¿2 ),

ÀFRÁ (r) Â FS

à (r), andFLÄ (rÅ )À which appearin the loop contributionto

the left-handedZbb vertex,Æ Secs.V A andV C.

54 4291RÇ

bÈ AND NEW PHYSICS:A COMPREHENSIVEANALYSIS

Page 18: Rb and new physics: A comprehensive analysiscliff/papers/Rb.pdf · 2002-11-12 · Rb and new physics: A comprehensive analysis P. Bamert,1 C. P. Burgess,1 J. M. Cline,1 D. London,1,2

ÉZg8 L

bâ ÊÌË

fÀ Í

Îy¿ fÀ Ï�Ð 2Y nE cò96Ñ Ò 2

Y MI

Z2

m; fÀ2Y 0

Ó1

dxÔ

Õ xÖ 3º ×

g8 Lbâ Ø

g8 RfÀ ÙHÚ

2� Û

1 Ü xÖ Ý 3ºg8 R

xÖ Þ M ß2Y /Bm; fÀ2Y à 1 áHâ 1

ã ä 1 å xÖ æ 3ºg8 L

çxÖ è M é2Y /

Bm; fÀ2Y ê 1 ëHì 1 í 2

Y . î 83ï ð

To=

seethat this is typically an unimportantcorrection,con-sider× the limit in which the scalarand fermion massesareequal,� r ñ 1. Thenthe total correction ò 82

ï ókô7õ83ï ö

is

÷g8 L

bâ øúù

Zû g8 L

bâ üÌý

fÀ þ

ÿy¿ fÀ ��� 2Y nE c�32� � 2 g8 L

fÀ �

g8 RfÀ � M

IZ2

12m; fÀ2 � g8 L

g8 LfÀ

g8 R

fÀ �

. � 84ï

Although�

the MI

Z2 correctionÓ can be significant if g8 L

fÀ �

g8 RfÀ

,Ìthe�

total correctionwould thenbetoo small to explaintheRbâ

discrepancy,V

andwould thusbe irrelevant.

B. Why many models do not work

WhatC

is importantfor applicationsis the relativesign be-tween�

the tree and one-loopcontributionsof Eq. � 82ï �

. Inorder� to increaseR

Rbâ so× as to agreewith the experimental

observation,� oneneedsfor themboth to havethe samesign,and� so � g8 L

9bâ � (}g8 L9 fÀ � g8 R

M fÀ )~ �

0�

in Eq. � 82ï �

. Thus an internalfermion�

with the quantum numbersof the bÒ

quark� hasg8 L

fÀ �

g8 RfÀ ���

12� and� would increaseR

Rbâ . Conversely,a fermion

like the t� -quarkhasg8 L9 fÀ � g8 R

M fÀ �! 12 and� so causesa decrease.

Moreover, becausethe combination(g8 L9 " g8 R

M )~

is invariantunder# chargeconjugation,the samestatementshold true forthe�

antiparticles: a bÒ

running< in theloop would increaseRR

whereas� a t� would� decreaseit.It thus becomesquite easyto understandwhich models

with� diagonalcouplingsto theZ:

bosono

canimprovethepre-dictionV

for RR

bâ . Multi-Higgs-doubletmodelshavea hardtime

explaining� an Rbâ excess� becausetypically it is the top quark

that�

makesthe dominantcontribution to the loop diagram,since× it has the largestYukawa coupling, y¿ f

À $&% 1, and thelargest0

mass,to which thefunction ' isg

very sensitive.How-ever,� for very largetan (*) the

�ratio of thetwo HiggsVEV’s + ,Ì

the�

Yukawa coupling of the t� quark� to the chargedHiggsbosono

can be madesmall and that of the bÒ

quark� can bemadeq large,asin Ref. , 25

� -. Figure7 showsthat, in fact, one

must go to extremevaluesof theseparameters,becauseinaddition� to needingto invert thenaturalhierarchybetweeny¿ t

.and� y¿ b

â ,Ì one must overcomethe big suppressionfor smallfermion�

massescomingfrom the function / .Preciselythe sameargumentappliesto a broadclassof

Zee-typemodels,where the SM is supplementedby scalarmultipletsq whose weak isospin and hyperchargepermit aYukawa0

coupling to the bÒ

quark� and one of the other SMfermions.So long asthescalarsdo not mix andtherearenonew fermionsto circulatein the loop, all suchmodelshavethe�

samedifficulty in explainingthe RR

bâ discrepancy.V

Belowwe� will give someexamplesof modelswhich, in contrast,are1 able� to explainRb

â .

C. Generalization to nondiagonal Z couplings2WeC

now turn to themorecomplicatedcasewheremixingintroducesg

off-diagonalcouplingsamongthe new particles.Becauseof mixing the couplingsof the fermions to the Zwill� be matricesin the massbasis.Similar to Eqs. 3 424 and�5456 7

we� write

8g8 L,R 9 f f

À :<;>=a? @BA<C L

9,RMa f? D

* E L9

,RMaf? F I3º

L9

,RMa? GIH f f

À JQK a? sÍ wí2 L ,Ì M 85

ï N

where� OL,Ra f? are� the mixing matrices.An analogousexpres-

sion× givesthe off-diagonalscalarZ:

couplingÓ in termsof thescalar× mixing matrix, P S

Qa? R . Of courseif all of the mixingparticles� sharethe samevalue for I3

º ,Ì then unitarity of themixingq matrices guaranteesthat the couplings retain thisform�

in any basis.This modificationof theneutral-currentcouplingshastwo

importanteffectson the calculationof S g8 L9bâ . One is that the

off-diagonal� Z:

couplingsÓ introducethe additionalgraphsofthe�

typeshownin Figs.5T a� U and� 5V bo W ,Ì wherethe fermionsorscalars× on eithersideof the Z vertexÖ havedifferent masses.The other is that the mixing matricesspoil the relationship,Eq.X Y

81ï Z

,Ì wherebythetermproportionalto [˜ canceledÓ in Eq.\78* ]

. But this is only becauseof the massdependenceof ^and� _a` . Thereforethe cancellationstill occursif all of theparticles� that mix with each other are degenerate,as onewould� expect. Moreover, the ultraviolet divergencesstillcancelÓ sincethey aremassindependent.

Evaluation of the graphsgives the following result atqb 2c d

M Z2c e

0:�f

g8 Lbâ g 1

32� h 2

c i Gj diagd k G

jf fÀ lnm G

j oporqts,Ì u 86

ï v

where� Gj

diagd /32B w 2

crepresentsthe contributioninvolving only

the�

diagonalZ:

couplings,Ó and so is identical to the previ-ously� derivedEq. x 78

* y. It is convenientto write it as

Gj

diagd z>{

fÀ | nE c� } y¿ f

À ~�� 2� 2 � g8 L9 � g8 R

M � f fÀ ���

r ��������� g8 RM � ffÀ �

g8 L9bâ

���g8 SQ ���p���������&���˜ � r�  ¢¡¤£ . ¥ 87

ï ¦Here and in the following expressionswe use the notationr § m; f

À2c /B M ¨2 and� r ©«ª m; fÀ ¬2c

/BM ­2 . As before®a¯ denotes

VtheUV-

divergentV

quantity °a±³² 2/� ´

nE µ 46 ¶B·¹¸»º

ln0 ¼

MI ½2/4

B ¾À¿ 2ÁB 12� .

The=

remaining terms in Eq. Ã 86ï Ä

comeÓ from the newgraphs[ of Figs. 5Å a� Æ and� 5Ç bo È ,Ì wherethe scalarsor fermionson� eithersideof the Z vertexÖ havedifferent masses,due tomixing:q

Gj

f fÀ ÉnÊÌËÍ

, fÀ Î

fÀ Ï nE c� y¿ f

À Ð y¿ fÀ Ñ�Ò* Ó 2 Ô g8 L Õ f f

À ÖØ×L Ù r,Ì r Ú<Û�Ü�Ý g8 R Þ f f

À ßáàáâ�ãä�å

RM æ r� ,Ì r� ç<è¢éëê ,Ì ì 88

ï í

Gî ïpïrð<ñ ò

, ó&ôõórö nE c� y¿ fÀ ÷ y¿ f

À ørù* ú g8 Sû ü�ýþýrÿ���������

Sû xÖ ,Ì xÖ ���� ,Ì �

89ï �

where� �L(}r� ,Ì r� � ),~ �

R(}r� ,Ì r� � )~ , and � S

û (} xÖ ,Ì xÖ � )~ aregiven by

4292 54BAMERT,�

BURGESS,CLINE, LONDON, AND NARDI

Page 19: Rb and new physics: A comprehensive analysiscliff/papers/Rb.pdf · 2002-11-12 · Rb and new physics: A comprehensive analysis P. Bamert,1 C. P. Burgess,1 J. M. Cline,1 D. London,1,2

�L � r,Ì r �����

�rr� �

r r !r�

r " 1ln r # r� $

r %'& 1ln r ( ,Ì ) 90

Ñ *

+RM , r� ,Ì r� -�.�/ 1

r� 0 r� 1r22

r� 3 1ln0

r� 4 r 5 22r� 6'7 1

ln0

r� 8 ,Ì 9 91Ñ :

;S< = xÖ ,Ì xÖ >�?�@ 1A

xÖ B 1 CED xÖ FHG 1 IKJ 1 L ln xÖ M�N xÖ O 2PxÖ Q'R 1 SUT xÖ VHW xÖ X

Y1 Z ln

0 xÖxÖ [ \ xÖ 2

2]xÖ ^ 1 _E` xÖ a xÖ b�c ,Ì d 92

Ñ e

and� xÖ ,Ì xÖ f are� the massratios xÖ g MI h2 /

Bm; fÀ2 and� xÖ iHj M

I kml2 /Bm; fÀ2.

Theseexpressionshave severalsalient featureswhich wenown discuss.First, Eqs. o 87

ï p,Ì q 88ï r

,Ì and s 89ï t

are� obviouslymuchq morecomplicatedthanEq. u 82

ï v. In particular,it is no

longerstraightforwardto simply readoff the sign of the re-sult.×

Second,w

thesumof theUV divergencesin Eqs. x 87ï y

,Ì z 88ï {

,Ìand� | 89

ï },Ì

Gî ~����

f fÀ �����m� y¿ f

À � y¿ fÀ ���m�* ����� g8 R

M � f fÀ �������m���

g8 L9bâ � f fÀ ���������

�� g8 S< ¡£¢�¢�¤�¥ f f

À ¦¨§,Ì © 93

Ñ ªis basisÒ

independent since× a unitary transformationof thefields«

cancelsbetweenthe Yukawaandneutral-currentcou-plings.� Thus it can be evaluatedin the electroweakbasiswhere� theneutral-currentcouplingsarediagonalandpropor-tional�

to ¬ g8 RM fÀ ­ g8 L

9bâ ® g8 S< ,Ì which vanishesdueto conservation

of� weak isospinand hyperchargeat the scalar-fermionver-tex.�

Wearethereforefreeto choosetherenormalizationscale¯ 22

in ln ° M ±22 /B ² 22 ³

to�

takeany convenientvalue.The M ´ de-V

pendence� of µ·¶ makesGî ¸�¸�¹

look unsymmetricunder theinterchangeg

of º and� »½¼ ,Ì but this is only an artifact of theway� it is expressed.For examplewhen thereare only twoscalars,× G

î ¾�¾�¿is indeedsymmetricunderthe interchangeof

their�

masses.Third,=

all the contributionsexceptÀ those�

of Gî Á�ÁÃÂ

are� sup-pressed� by powersof m; f

À /B MI Äing

the limit that thescalarsaremuchheavierthanthe fermions.Thusto get a largeenoughcorrectionÓ to g8 L

requiresthat Å i Æ not all of the scalarsbemuchq heavierthanthe fermionswhich circulatein the loop,or� Ç iig È the

�scalarsmix significantlyandhavetheright charges

so× that Gî ÉUÉÃÊ

is nonnegligibleand negative.We useoptionËii Ì in what follows to constructanothermechanismfor in-

creasingÓ RR

bâ .

Finally,Í

evenif thetwo fermionsaredegenerate,onedoesnot generallyrecoverthe previousexpressionÎ 78

* Ïthat�

ap-plied� in the absenceof mixing. This is becauseDirac massmatricesq are diagonalizedby a similarity transformation,MI ÐÒÑ

L†MI Ó

RM ,Ì not a unitary transformation.The left- and

right-handed mixing angles can differ even when thediagonalizedV

mass matrix is proportional to the identity.Thus,=

in contrast to Eq. Ô 93Ñ Õ

,Ì the expressionÖf fÀ ×ÙØÃØmÚ y¿ f

À Û y¿ fÀ Ü�ÝmÞ* ß (} g8 L)

~ f fÀ à�á

(}g8 R)~ f fÀ â¨ã

is not invariant undertransformations�

of the fields, becausey¿ fÀ ä is

grotatedby å Ræ

recall thaty¿ fÀ ç is theYukawacouplingonly for theRH f

ã’sè ,Ì

whereas� g8 L9 is rotatedby é L

9 .

WeC

cangetsomeinsightinto Eqs. ê 88ï ë

– ì 92Ñ í

byo

looking atspecial× valuesof the parameters.Let us assumethere is adominantV

Yukawa coupling y¿ betweeno

the left-handedbÒ

quark� anda singlespeciesof scalarandfermion, fã

1 and� î1 ing

the�

weakbasis,

ïscalarð y¿ ñ 1f

ã1 ò L9 bÒ ó

H.c. ô 94Ñ õ

In the massbasisthe couplingswill thereforebe

y¿ fÀ ö�÷ y¿ ø S

<1 ùûú�üRM1 fÀ ý

* . þ 95Ñ ÿ

Now�

gaugeinvarianceonly relatesthe � 1,1� elements� of theneutral-currentcouplingmatricesin the weakbasis:

�g8 S< � 11� g8 L

bâ ���

g8 RfÀ

11 0.� �

96Ñ �

Thereare three limiting casesin which the resultsbecomeeasier� to interpret.

1� Ifî

all the scalarsare degeneratewith eachother,andlikewise for the fermions,then the nonmixing result of Eq.�82ï �

holds,exceptonemustmakethe replacement

g8 LfÀ �

g8 RfÀ �����

RM SJ

m� � L†g8 L9 �

L9 SJ

m� � R† � g8 R

M � 11,Ì � 97Ñ �

where� SJ

m� isg

the diagonalmatrix of the signsof the fermionmasses.q �

2 If there are only two scalarsand if they are muchheavierthan all of the fermions,only the term G

î !"!$#is sig-

nificant.n Let % 1 and� &2 denoteV

the weak-eigenstatescalars,and� ' and� (*) the

�masseigenstates;then

+g8 L

bâ , y¿ 2

2n- c�

16. 2 / I¹ 3º 0 1 1 I

¹3º 2 23 4

c5 S<22 sÍ S<22 F6 S< 7 MI 822 /

BMI 9;:22 <

; = 98Ñ >

FS< ? r @"A r B 1

2 C r D 1 E ln r F 1, G 99Ñ H

where� c5 S< and� sÍ S

< are� thecosineandsineof thescalarmixingangle.� ThefunctionF

6S< (} r� )~ is positiveexceptat r� I 1 whereit

is zero,and so the sign of J g8 L9bâ is completelycontrolledby

the�

factor (I3º K 1 L I3

º M 23)~. We seethat to increaseRb

â it is neces-

sary× that I¹

3º N 1 O I

¹3º P 2.Q

3� R

WhenC

thereareonly two fermions,with weakeigen-states× f

ã1,Ì fã

2 and� masseigenstatesfã

,Ì fã S

botho

much heavierthan�

any of the scalars,then

Tg8 L9bâ U y¿ 2n- c�

16V 2 W g8 L911c5 LR

922 Xg8 L9222

sÍ LR922 Y

g8 RM11

Z�[g8 RM222 \

g8 RM11] c5 R

M22 sÍ RM22 FR

M ^ m; fÀ22 /B m; f

À _22 `

a 2� b

g8 L22c g8 L

11d c5 L9 sÍ L9 c5 RM sÍ RM F6

L9 e m; f

À2/Bm; fÀ f2 gih ; j 100k

where� sÍ LR and� c5 LR are� the sineandcosineof the differenceorl sumof theLH andRH mixing angles,m L n sÍ m� o R ,Ì depend-ing on the relative sign sÍ m� ofl the two fermion masseigen-values,p and

54 4293Rq

br AND NEW PHYSICS:A COMPREHENSIVEANALYSIS

Page 20: Rb and new physics: A comprehensive analysiscliff/papers/Rb.pdf · 2002-11-12 · Rb and new physics: A comprehensive analysis P. Bamert,1 C. P. Burgess,1 J. M. Cline,1 D. London,1,2

FR s r t"u FS< v r w"x r y 1

2 z r { 1 | lnr } 1

and~ F6

L9 � r� �"�

�r�

r� � 1ln�

r� � 1. � 101�ThefunctionFL

9 hassomeof thesamepropertiesasFS< � FR

M ,�including invarianceunderr � 1/r,� being positive semidefi-nite� andvanishingat r� � 1. Plotsof thesefunctionsareshownin�

Fig. 7. Note that the first line of Eq. � 100� is�

the sameasEq. � 97

Ñ �.

To getsomeideaof theerrorwe havemadeby neglectingthe�

massof the Z:

boson�

one can computethe lowest ordercorrection� as in Sec.V C. The answeris more complicatedthan�

for the caseof diagonalZ couplings,� exceptwhen thefermionsaredegeneratewith eachotherandlikewise for thebosons.�

In that casethe answeris given againby Eq. � 84ï �

except� that g8 RfÀ �

(}g8 R

)~ 11 and~ g8 L

fÀ �

(} �

RM SJ

m� � L†g8 L9 �

L9 SJ

m� � R†� g8 R

M )~ 11,� preciselyasin Eq.   97

Ñ ¡. Thuswe would still expect

it�

to be a small correctionevenwhenthereis mixing of theparticles¢ in the loop.

These simplifying assumptionscan be used to gain asemianalytic£ understandingof why certainregionsof param-eter� spacearefavoredin complicatedmodels,which is oftenmissingin analysesthattreattheresultsfor theloop integralsas~ a black box. The observationswe makeheremay be use-ful¤

whensearchingfor modificationsto a model that wouldhelp¥

to explainR¦

b§ . The next two sectionsexemplify this by

creating� somenew models that take advantageof our in-sights,£ and by elucidatingprevious findings in an alreadyexisting� model,supersymmetry.

D. Examples of models that work

In addition to ruling out certain classesof models,ourgeneral¨ considerationsalsosuggestwhat is

©requiredª in order

to�

explain R¦

b§ . Obviously new fermionsand scalarsare re-

quired,« whoseYukawacouplingsallow themto circulatein-side£ the loop. We give two examples,onewith diagonalandonel with nondiagonalcouplingsof thenewparticlesto theZ

¬boson.�

For our first examplewe introduceseveralexotic quarksF6

,� P­

,� and N®

,� and a new Higgs doublet ¯ ,� whosequantumnumbers� are listed in Table V. The unorthodox electriccharge� assignmentsdo not ensure cancellation of elec-troweak�

anomalies,but this canbefixed by addingadditionalfermions,¤

like mirrors of thosegiven, which do not contrib-ute° to R

¦b§ .

Thehyperchargesin TableV allow thefollowing Yukawainteractions:

±y² ³ yN¿ ¯RQ

´Li µ j¶ ·

i j ¸ g¹ pº P­

RF6

Li H» j¶ ¼

ij ½ g¹ n¾ N®¯RF6

Li H»˜ j¶ ¿

ij À H.c.,Á Â

102ÃwhereÄ Å

i j is the 2Æ 2 antisymmetrictensor,H is the usualSMÇ

Higgs doublet,andQ´

LÈ É (

Êb§

L

tË LÌ )Í

is the SM doubletof third

generation¨ LH quarks.When H gets¨ its VEV, Î H ÏÑÐÓÒ , w� efindÔ

two fermion masseigenstates,pÕ and~ n- ,� whosemassesare~ mÖ pº × g¹ pº Ø and~ mÖ n¾ Ù g¹ n¾ Ú and~ whoseelectric chargesareQ´

pº Û qÜ Ý 1 and Q´

n¾ Þ qÜ . Thereare also two new scalarmasseigenstates,� ß;à ,� whoseelectric chargesare Q

´ á;âqÜ ã 1

3ä and~

Q´ å;æ

qÜ ç 23ä .

Inè

themasseigenstatebasis,theYukawainteractionswiththe�

new scalarsare

éy² ê yn¿ ¯Rb

ëL ìîíðï yn¿ ¯Rtñ L òîóðô H.c.,

Á õ103ö

from which we seethatthen- couples� to thebë

quark« asin Eq.÷76ø ù

.Theú

weak isospinassignmentsof the n- are~ Iû

LNý þ ÿ

123 and~

I 3ü

R�n¾ �

0,�

so that g¹ LÈn¾ � g¹ R

�n¾ ��� 12. Therefore,from Eq. � 82

ï ,� one

obtainsl g¹ Lb§ �

0.�

Thecentralvalueof Rb§ can� bereproducedif�

g¹ Lb§ ��

0.0067,�

which is easilyobtainedby taking y� � 1 andr� � 1, so that � (

Êr� )Í � 1. The Yukawacouplingcould be made

smaller£ by putting the new scalarsin a higher color repre-sentation£ like the adjoint.

We�

havenot exploredthedetailedphenomenologyof thismodel,� but it is clearly not ruled out since we are free tomakethenewfermionsandscalarsasheavyaswewish.Andsince£ we canalwaystake mÖ pº � mÖ n¾ ,� thereis no contributionto�

the obliqueparameterT�

. The contributionto R¦

b§ does�

notvanish� evenas the massesbecomeinfinite, but this is con-sistent£ with decouplingin the sameway asa heavytñ quark,«since£ thenewfermionsget their massesthroughelectroweaksymmetry£ breaking.Thepricewe haveto pay for suchlargemassesis correspondinglylargecouplingconstants.

Next�

we build a modelthatusesour resultsfor nondiago-nal� couplingsto the Z

¬. It is a simplemodificationof theSM

that�

goesin the right directionfor fixing the R¦

b§ discrepancy�

but�

not quite far enoughin magnitude.Variations on thesame£ themecancompletelyexplainRb

§ at~ thecostof makingthe�

modelsomewhatmorebaroque.Our�

startingpoint is a two-Higgs-doubletextensionof theSM.Ç

We takethe two Higgs fields,

Hd� � Hd

�0ÓH»

d� � and~ Hu� � Hu�

u�0Ó ,�to�

transformin theusualway undertheSM gaugesymmetry.Itè

was explainedearlier why this model doesnot by itselfproduce¢ the desiredeffect, but Eq. ! 98

Ñ "suggests£ how to fix

this�

problem by introducing a third scalar doublet, #$ (Ê %'&(')*)

)Í, which mixes with the other Higgs fields. The

charge� assignmentsof thesefields, listed in TableVI, ensurethat�

the two fields H u� + and~ ,.- can� mix even though theyhavedifferent eigenvaluesfor I3

ü .Inè

this model the new scalar field cannot have anyYukawa/

couplingsto ordinaryquarkssincetheseareforbid-den�

by hyperchargeconservation.The only Yukawa cou-plings¢ involving theLH b

ëquark« arethosewhich alsogener-

ate~ the massof the tñ quark:«

TABLE V. Field content and charge assignments: Elec-troweakquantumnumbersfor thenewfieldswhich areaddedto theSM to producethe observedvaluefor R

0b1 .2

Field Spin SUc3 4 35 6 SUL 7 28 UY 9 1:;0<

1 2 q = 16>

FL? 1

23 3 2 q @ 1

23

PR12 3 1 q A 1

NB

R12 3 1 qC

4294 54BAMERT,D

BURGESS,CLINE, LONDON, AND NARDI

Page 21: Rb and new physics: A comprehensive analysiscliff/papers/Rb.pdf · 2002-11-12 · Rb and new physics: A comprehensive analysis P. Bamert,1 C. P. Burgess,1 J. M. Cline,1 D. London,1,2

EYukF G y� tË H tñ I L

È bHë

u� JLK H.c.,Á M

104NwhereÄ y� tË O mÖ tË /P Q u� is the conventionally-normalizedYukawacoupling.� We imagine R u� to

�be of the sameorder as the

single-Higgs£ SM value, and so we expecty� tË to�

be compa-rableto its SM size.

The scalarpotential for sucha model very naturally in-corporates� H

»u� S -TVU mixing.� Gaugeinvariancepermitsquartic

scalar£ interactionsof theform W (ÊH»

u�†X )(Í

u�†H»

d� )Í Y H.c.,

Áwhich

generate¨ the desiredoff-diagonal terms: Z (Ê [�\

u� ] * ^ u�* _ d�`�acb

Hd� dfe

u�22 * )Í g

H.c.

SinceÇ

the weak isospin assignmentsare I3ü hji.kml 1

2

and~ Iû

3üHn uo prqms 1

23 ,� the color factor is n- ct u 1, and the relevant

Yukawa/

coupling is y� v y� tË w ,� we seethat Eq. x 98y z

predicts¢the�

following contribution due to singly-chargedHiggsloops:

{g¹ LÈb§ |m} y� tË22

16~ 2 2c5 S<22 s� S<22 FS< � r � ,� � 105�

withÄ r being�

the ratio of the scalar mass eigenstates,r� M �2 /PM �f�22

. Taking optimistic values for the parameters13���S< ��� /4,

P2c5 S

<2s� S<2 � 1

23 ,� y� tË � 1, and M

� �/PM� �f���

10� , w� efi nd�g¹ L

b§ ���

0.0043,�

which is two-thirds of what isrequired: ( � g¹ L

Èb§ )Í expt� ��� 0.0067� �

0.0021.�

In additionto thecontributionof thesingly-chargedscalarloops,�

one should considerthoseof the other nonstandardscalar£ fields we introduced.Sinceall of the scalarsthat mixhavethe sameeigenvaluefor I3

ü ,� their contributionis givenby�

Eq.   82¡ ¢

,� which is small if the scalarsare much heavierthan�

thelight fermions.Thenonly the tñ -quarkcontributionisimportant.�

In this limit there are appreciablecontributionsonlyl from the three chargedscalarfields, one of which iseaten� by thephysicalW boson

�andsois incorporatedinto the

SMÇ

tñ -quarkcalculation,andtheothertwo of which we havejust£

computed.So,Ç

for an admittedlyspecialregion of parameterspace,this�

simple model considerablyamelioratesthe Rb§ discrep-�

ancy,~ reducingit to a 1¤ effect.� It is easyto adaptit soastofurther¤

increase¥ g¹ LÈb§ and~ alsoenlargethe allowedregionof

the�

model’sparameterspace.Thesimplestway is by increas-ing the size of the color factor n- ct orl the isospindifference

I3ü ¦f§�¨ I3

ü © . For instancethe new scalar,ª ,� could be put into a4«

ofl SUL ¬ 2­ ® ratherª thana doublet,andbegivenweakhyper-

charge� Y ¯�° 5±23 . Then the singly-chargedstate²´³ has

¥Iû

3ü µ'¶.·

¸ 3ä2,� making I

û3ü ¹fº�» I

û3ü ¼¾½m¿ 2

­, which is twice asbig as for the

doublet.�

More newscalarsmustbeaddedto generatemixingamongst~ the singly-chargedscalarstates.

A secondvariationwould be let the two new Higgs dou-blets�

be color octetssince this gives more than a fivefoldenhancement� of À g¹ L

due�

to thecolor factor n- ct Á 163ä . It is still

possible¢ to write downquarticscalarinteractionswhich gen-erate� the desiredscalarmixings. Either of thesemodelshasmuch more room to relax the previouslytight requirementsfor¤

optimal scalarmassesandmixings.

E. The supersymmetric case

LetÂ

us now apply the aboveresultsto gain someinsightinto�

whatwould benecessaryto explainR¦

b§ in�

supersymmet-ric extensionsof the standardmodel.Therearetwo kinds ofcontributions� involving the top-quark Yukawa coupling,whichÄ oneexpectsto give thedominanteffect.Thesearethecouplings� of the left-handedb

ëquark« to the secondHiggs

doublet�

andthetop quark,or to thecorrespondingHiggsinosand~ top squarks,

y� tË bë LÈ hÃ

2,2

R�Ä tñ R� Å y� tË bë L

È tñ R� hÃ

22 Æ . Ç 106È

Of�

these,the secondone gives a loop contribution likethat�

of thetwo-Higgsdoubletmodelsdiscussedabove:it hasthe�

wrong sign for explaining R¦

b§ . Since the massof the

charged� Higgs bosonis a free parameterin supersymmetricmodels,we canimaginemakingit largeenoughcomparedtomÖ tË so£ that,accordingto Eq. É 82

¡ Ê,� it hasonly a smalleffecton

b§ . We thereforeconcentrateon the Higgsino-squarkpart.

The chargedHiggsinomixeswith the W-ino, and the right-handedtop squarkmixeswith its chiral counterpart,soin thenotation� of Eq. Ë 94

y Ì,� we havef

Í1 Î h

Ã2Ï ,� fÍ

2 Ð W Ñ ,� Ò 1 Ó tñ R ,� andÔ2 Õ tñ L . Thecorrespondingchargematricesfor thecouplings

to�

the W3ü are~

g¹ S< Ö 2

3ä s� w×22

0� 0

�12 Ø 2

3ä s� w×22 ;

g¹ LÈ Ù g¹ R

� Ú Û 123 Ü s� w×20� 0

�Ý

1 Þ s� w×22 . ß 107àBecauseá

there are two possiblecolor combinationsfor theinternal lines of the loops diagram,the color factor in Eqs.â87¡ ã

– ä 89¡ å

is n- ct æ 2.Beforeá

exploringthe full expressionfor ç g¹ Lb§

weÄ candis-cover� whatparameterrangesarethemostpromisingby look-ing at the limiting casesdescribedby Eqs. è 97

y é– ê 100ë . The

most important lessonsfrom theseapproximationsfollowfrom¤

the chargematricesì 107í . We do not want the squarksto�

bemuchheavierthanthecharginosbecausethenEq. î 98y ï

wouldÄ apply andgive the wrong sign for the correctiondueto�

thesignof theisospindifferencebetweenthesquarks.Theotherl two cases,wherethesquarksarenot muchheavierthanthe�

charginos,manifest a strong suppressionof the resultunless° the chargino mixing angles are such thatsin(£ ð

L ñ s� m� ò R)Í

is large,wheres� m� is the sign of the determi-nant� of the chargino mass matrix. If on the other handsin(£ ó

LÈ ô s� m� õ R

� )Í ö

0,�

there is exact cancellationbetweeng¹ LÈ

13Note÷

that the charged-scalarmixing in this modelis suppressedif one of the scalarmassesgetsvery large comparedto the weakscale.

TABLE VI. Field content and charge assignments: Elec-troweakquantumnumbersfor all of the scalars,including the SMHiggs doublet,of the three-doubletmodel.

Field Spin SUc3 ø 35 ù SUL ú 2û UY ü 1ýHþ

dÿ 0

<1 2 � 1

23

u� 0<

12� 1

23�

0<

12 � 3ä2

54 4295R0

b1 AND NEW PHYSICS:A COMPREHENSIVEANALYSIS

Page 22: Rb and new physics: A comprehensive analysiscliff/papers/Rb.pdf · 2002-11-12 · Rb and new physics: A comprehensive analysis P. Bamert,1 C. P. Burgess,1 J. M. Cline,1 D. London,1,2

and~ g¹ R� in�

theseequationsbecauseof thefact thatg¹ LÈ � g¹ R

� for¤

the�

charginos.In summary,our analytic formulas indicatethat�

thefavoredregionsof parameterspacefor increasingRb§

are~ where

tan� �

R� tan� �

LÈ � s� m� �� sgn£ mÖ f

� mÖ f� ��� ,� � 108�

and~ at leastoneof the squarksis not muchheavierthanthecharginos.�

In supersymmetricmodelstheYukawacouplingthatcon-trols�

the largestcontributionto R¦

b§ is�

that of the top quark,

and~ it dependson the ratio of the two Higgs VEV’s,tan� �����

2� /P � 1, b� y

y� f� ��� mÖ tË� sin£ � ,� 109!

whereÄ "$# (Ê %

12� &('

22�)Í 1/2) 174 GeV. Thereforeit is important

to�

find tan * in�

terms of the charginomassesand mixingangles.~ The charginomassmatrix is given by

+ g¹ , 2

g¹ - 1 M�

2� .0/ L

† mÖ f�

0� 0

�mÖ f� 1 2 R

� 3 c5 Lc5 RmÖ f� 4 s� Ls� RmÖ f

� 5c5 R� s� LÈ mÖ f� 6 s� R

� c5 LÈ mÖ f� 7 s� Rc5 LmÖ f

� 8 c5 Rs� LmÖ f� 9

s� LÈ s� R� mÖ f� : c5 L

È c5 R� mÖ f� ; ,� < 110=

whereÄ > is�

the coefficientof H»

1H»

2 in�

the superpotentialandM 2� is the soft-supersymmetry-breakingmassterm for the

W-ino. It follows that

tan� ?A@ mÖ f

� tan� B

R C mÖ f� D tan� E

L

mÖ f� tan� F

L G mÖ f� H tan� I

R. J 111K

Theaboveconsiderationsallow usto understandwhy val-ues° of tan L near� unity are necessaryfor a supersymmetricsolution£ to theR

¦b§ problem.¢ FromEq. M 111N and~ themaximi-

zation condition O 108P weÄ seethat tan Q is restrictedto liebetween� R

mÖ f� /P mÖ f

� SUT and~ V mÖ f� W /P mÖ f

� X . Equation Y 108Z together�

withÄ Eq. [ 110\ also~ implies

c5 L2 ] mÖ f

� ^`_ s� L2 a mÖ f

� bUc`dfe M�

W sin£ g ;

c5 LÈ2� h mÖ f

� ikjkl s� LÈ2� m mÖ f

� n`ofp M W cos� q . r 112s

Thisú

means that the averagevalue of the two charginomassescanbeno greaterthanM wt ,� so that theratio u mÖ f

� /P mÖ f� vkw

cannot� differ muchfrom unity unlessoneof thecharginosismuch� lighter than the W boson.

�Using the LEP 1.5 limit of

65x

GeV for thelightestcharginoy 26­ z

this�

would thenrequirethat�

tan {�| 1.5.In the casethat noneof our simplifying limits apply, we

have¥

searchedthe parameterspaceof the threeindependentratiosª betweenthe two scalarmassesand the two fermionmasses,andthe threemixing angles} R

� ,� ~ LÈ ,� � S

� to�

find whichregionsare favorablefor increasingRb

§ . Figures8� a~ � –8� d� �show£ the shift in g¹ L

as~ a function of pairs of theseparam-eters,� using the Yukawa coupling � 109� corresponding� to atop�

quarkmassof 174GeV andthetheoreticalpreferencefortan� ���

1 � weÄ implementthe latter by settingg¹ Lb§ �

0�

for pa-rametersª thatwould give tan ��� 1� . As shownin TableI, oneneeds� � g¹ L

Èb§ �0� 0.0067�

in orderto explainthe observedvalue

FIG. 8. The dependenceof g Lb1

on the varioussupersymmetric� parameters.Since g� L

?b1 dependsonly� on massratios in our approximation,theunits� of massarearbitrary,with themassesof allthe�

charginosand squarkswhich are not beingvaried� set to unity.

4296 54BAMERT,D

BURGESS,CLINE, LONDON, AND NARDI

Page 23: Rb and new physics: A comprehensive analysiscliff/papers/Rb.pdf · 2002-11-12 · Rb and new physics: A comprehensive analysis P. Bamert,1 C. P. Burgess,1 J. M. Cline,1 D. London,1,2

ofl Rb§ . The values of the masses are taken to be

M ��� M ����� mÖ f� � mÖ f

� ��  1 ¡ in arbitraryunits¢ ,� exceptfor thosethat�

areexplicitly variedin eachfigure. In Fig. 8£ a~ ¤ weÄ lookat~ the situationin which tan ¥ L ¦ tan

� §R ¨ 1, in contradiction

to�

condition © 108ª ,� andvary the scalarmixing angleandthemassof mostly tñ R� scalar£ in the limit of zerosquarkmixing.Theú

signof g¹ Lb§

has¥

thewrongvalue,aspredictedby Eq. « 98y ¬

.Figure­

8® b� ¯ shows£ the same situation except that nowtan� °

LÈ ±0² tan

� ³R� ´ 1, in accordancewith Eq. µ 108¶ . Thenthe

sign£ of g¹ LÈb§ is negative,asdesired,andhasthe right size for

substantial£ rangesof · S¸ and~ M

� ¹. In Fig. 8º c� » weÄ keepall the

masses� nearly degenerateand set ¼ S¸ ½ 0�

to show the depen-dence�

on tan ¾ LÈ and~ tan ¿ R

� . It is easyto seethat g¹ LÈb§ hasthe

correct� sign and largestmagnitudeÀ whichÄ is also almostaslarge�

asneededÁ whenÄ condition  108à is�

satisfied.Finally inFig.­

8Ä d� Å weÄ show the dependenceon the massesof themostly W-ino fermion and on Æ R

� whenÄ ÇS¸ È 0�

andtan� É

LÈ Ê0Ë 1, showingagainthe preferencefor mixing angles

obeyingl Eq. Ì 108Í ,� as well assomeenhancementwhenthereis�

a hierarchybetweenthe two charginomasses.One�

might thereforeget the impressionthat it is easytoexplain� Rb

§ using° supersymmetriccontributionsto the Zbbvertex.� Theproblemis thatto geta largeenoughcontributiononel is driven to a ratherspecialregion of parameterspace,whichÄ comesclose to satisfying condition Î 108Ï . As men-tioned�

above,the consequentcondition Ð 112Ñ prevents¢ onefrom¤

making the chargino massesarbitrarily heavy. This,coupled� with the suppressionin R

¦b§ whenÄ the squarksare

heavierthanthecharginos,meansthatall therelevantsuper-symmetric£ particles must be relatively light, except thecharged� Higgs bosonwhich hasto be heavyto suppressthewrong-signÄ contributionfrom H

» Ò$Ótñ loops.�

Thus in the ex-ample~ of Fig. 8Ô c� Õ ,� the preferredvaluesof c5 R

� Ö 1, s� LÈ ×0Ø 1,

s� R� Ù c5 L

È Ú 0�

imply that mÖ f� Û(Ü sin£ Ý and~ mÖ f

� Þ�ß(à cos� á ,� whileâäã M�

2 å 0,�

which arepreciselythe circumstancesof the su-persymmetric¢ modelsconsideredin Refs. æ 27

­ çand~ è 28

­ é. Fig-

ure° 8ê d� ë ,� on the otherhand,hasits maximumvalueof Rb§ at~

c5 R� ì s� R

� í c5 LÈ îï s� L

È ð 1, implying tan ñ�ò 1 andthus from Eq.ó112ô that

� õmÖ f� öU÷`øúù mÖ f

� û`ü 2­

M�

W . Becausethe lightestcharginomass� is constrainedby experimentallower limits, there islittle parameterspacefor getting a large hierarchybetweenthe�

two charginomasses,as one would want in the presentexample� in orderto getthefull shift14 ofl ý 0.0067

�in g¹ L

b§. Our

analysis~ allows one to pinpoint just wherethe favorablere-gions¨ arefor solving the Rb

§ problem.¢We�

thusseethat it is possibleto understandmanyof theconclusions� in the literature þ 2­ 7–31ÿ onl supersymmetryandR¦

b§ using° somerather simple analytic formulas. Thesein-

clude� thepreferencefor smallvaluesof tan � as~ well aslightHiggsinosandsquarks.

VI. FUTURE TESTS

If we excludethe possibility that the experimentalvalueofl Rb

§ is simply a 3.7� statistical£ fluctuation,we canexpect

that,�

once the LEP Collaborationshave completed theiranalyses~ of all the data collectedduring the five yearsofrunningª at the Z

¬pole,¢ the ‘‘ R

¦b§ crisis’’� will becomean even

more� seriousproblemfor the standardmodel. � Of�

course,itis wise to keepin mind that theremay be a simpleexplana-tion,�

namelythatsomesystematicuncertaintiesin theanaly-sis£ of the experimentaldataarestill not well understoodorhave¥

been underestimated.� Inè

Secs. III –V we have dis-cussed� a variety of modelsof new physicswhich could ac-count� for the experimentalmeasurementof Rb

§ . The nextobviousl stepis to considerwhich other measurementsmaybe�

usedto revealthe presenceof this new physics.The most direct method of finding the new physics is

clearly� the discoveryof new particleswith the correctcou-plings¢ to the Z

¬and~ the b

ëquark.« However,failing that, there

are~ some indirect tests. For example,many of the new-physics¢ mechanismswhich havebeenanalyzedin this paperwillÄ affect the rate for somerare B

�decays�

in a predictableway.Ä Theratesfor theraredecaysB

� �X�

s� l� l�

and~ B� �

X�

s� � �are~ essentiallycontrolled by the Zbs� effective� vertex � bs

§ � ,�since£ additionalcontributions� such£ asbox diagramsandZ- �interference� �

are~ largely subleading.15 Inè

the SM, in the ap-proximation¢ made throughoutthis paperof neglectingthebë

-quark massand momentum,a simple relation holds be-tween�

thedominantmÖ tË vertex� effectsin Rb§ and~ in theeffec-

tive�

Zb¬ ¯s� vertex� �

bs§ � :

�bs§ � ,SM � V tbË* V tsË�

V tbË � 2 ����� ,SM,� � 113 whereÄ !#"%$ ,SM is definedas in Eq. & 26' withÄ the SM formfactor as given in Eqs. ( 27) and~ * 38

+ ,. The meaningof Eq.-

113. is�

that, within the SM, the Zb¬ ¯s� effective� vertex mea-

surable£ in Z¬

-mediatedB�

decays�

representsa direct/

measure-�mentof the mÖ tË -dependentvertexcorrectionscontributingtoRb§ ,� moduloa ratio of the relevantCKM matrix elements.In

particular,¢ both correctionsvanish in the mÖ tË 0 0�

limit. Thequestion« is now the following: how is this relationaffectedby�

the new physicsinvokedin Secs.III –V to explainRb§ ?

Consider1

first the tree-levelbë

-bë 2

mixing effectsanalyzedin�

Sec.III. It is straightforwardto relate the correctionsofthe�

LH and RH Zb¬ ¯bë

couplings� to new tree-levelmixing-inducedFCNC couplingsg¹ L

È,R�bs

§. In this caseEq. 3 54 5 reads

g¹ L,Rbs§ 687 9

w: ; g¹ w: <>= LÈ

,R� ?

L,R

@ b§* A L,R

B s� C w: 1. D 114EHenceg¹ L

È,R�bs

§involve the samegaugecouplingsand mixing

matrices� that determinethe deviation/

from¤

the SM of theflavor-diagonalF

couplings.�It is also true that, for manymodelsof new physics,the

loop correctionsto the Zbbë

vertex� would changethe effec-tive�

Zb¬ ¯s� vertex� in much the sameway, thereforeinducing

computable� modificationsto the SM electroweakpenguindiagrams.G

In thesemodels,for eachloop diagraminvolving14An additionalconstraintis that the lightest Higgs bosonmass

mhH 0 vanishes� at treelevel whentan IKJ 1, anda very largesplitting

betweenthe top squarkmassesis neededfor the one-loopcorrec-tions to mh

H 0 to�

be large enough.This is why Ref. L 29M finds lessthanthedesiredshift in Rb

1 in theminimal supersymmetricstandardmodel.We thankJ. Lopezfor clarifying this point.

15Due to the absence of ZN

- O interferenceP

and of largerenormalization-group-inducedQCD corrections, the processB Q XsR S�S representstheoreticallythecleanestproof of theeffectiveZbN ¯sT vertex� U 32

5 V.

54 4297R0

b1 AND NEW PHYSICS:A COMPREHENSIVEANALYSIS

Page 24: Rb and new physics: A comprehensive analysiscliff/papers/Rb.pdf · 2002-11-12 · Rb and new physics: A comprehensive analysis P. Bamert,1 C. P. Burgess,1 J. M. Cline,1 D. London,1,2

the�

new statesfW

,� fW X and~ their coupling to the bë

quark« g¹ f f� Y

b§ ,�

there�

will be a similar diagramcontributingto Z bs§ [ that�

canbe�

obtainedby the simplereplacementg¹ f f� \

b§ ] g¹ f f

� ^s� . For ex-

ample,_ the generalanalysisof tñ -quark mixing effects pre-sented` in Sec. IV can be straightforwardly applied toZ-mediatedB decays.

aDeviationsfrom the SM predictions

for the B b Xs� l� c l� d

and_ B e Xs� fgf decaya

ratescanbe easilyevaluatedh by means of a few simple replacementslikei j

tbË k 2 l V tbË* V tsË and_ m n

tË o bp q 2 r V tË s bp* V tË t s� inu

all our equations.16

To a largeextent,this is alsotruefor supersymmetryv SUSYw x

models.Indeed,the analysisof the SUSY contributionstothey

Zb¬ ¯s� form

zfactor { 34

+ |can} teachmuchaboutSUSYeffects

inu

R~

bp . And oncea particularregionof parameterspacesuit-

able_ to explainthe Rbp problem� is chosen,a definitenumeri-�

cal� prediction� for the B � Xs� l� � l� �

and_ B � Xs� �g� decaya

ratescan} be made.

Thisú

brief discussionshowsthat, for a largeclassof new-physics� models,the new contributionsto Rb

p and_ to the ef-fective � bs

p � vertex� arecomputablein termsof thesamesetofnew-physics� parameters.Therefore,for all thesemodels,theassumption_ that somenew physicsis responsiblefor the de-viations� of Rb

p from the SM predictionwill imply a quanti-tativey

prediction of the corresponding deviations forZ¬

-mediatedB�

decays.a

However,�

this statementcannot be applied to all new-physics� possibilities.For example,if a new Z � boson

�is re-

sponsible` for the measuredvalueof Rbp ,� thenno signal can

be�

expectedin B�

decays,a

sincein this casethe new physicsrespects� the GIM mechanism.This would also be true ifmÖ bp -dependenteffectsareresponsiblefor theobserveddevia-

tionsy

in Rbp as_ could happen,for example,in the very large

tany �

region� of multi-Higgs-doubletor SUSY models.Moregenerally,� the loop contributionsof the new statesf

W,� fW � can}

be�

different, since g¹ f f� �

s� is not necessarilyrelatedto g¹ f f� �

bp ,�

and_ in particular,wheneverthe new physicsinvolved in Rbp

couples} principally to thethird generation,it is quitepossiblethaty

no sizeableeffect will show up in B�

decays.a

Still, thestudy` of B � Xs� l� � l

� �and_ B � Xs� �g� could} help to distinguish

between�

modelsthat do or do not significantly affect thesedecays.a

Unfortunately,�

at presentonly upperlimits havebeenseton� the branching ratios for B � Xs� l� � l

� � �3+5–37¡ and_

B ¢ Xs� £g£ ¤ 32+ ¥

. Sincetheselimits area few timeslargerthanthey

SM predictions,theycannothelp to pin downthecorrectsolution` to the R

~bp problem.� However,future measurements

of� theserare decaysat B factoriescould well confirm thatnew physicsis affectingthe rateof b

ë-quarkproductionin Z

decays,a

as well as give somehints as to its identity. If nosignificant` deviationsfrom theSM expectationsaredetected,thisy

would alsohelp to restrict the remainingpossibilities.

VII.¦

CONCLUSIONS

Until�

recently,the SM hasenjoyedenormoussuccessinexplainingh all electroweakphenomena.However,a number

of� chinks have startedto appearin its armour. There arecurrently} severaldisagreementsbetweentheory and experi-ment§ at the2 level

©or greater.TheyareR

~bp ª¬«

bp /­ ® had ¯ 3.7

+ °g±,�

R~

ct ²¬³ ct /­ ´ hadµ ¶ 2.5

· ¸g¹,� the inconsistencybetweenA

ºe»0¼ as_ mea-

sured` at LEP with thatdeterminedat SLC ½ 2.4¾g¿ ,� andA FBÀ0¼ ÁÃÂÅÄÁ

2.0ÆgÇ . Takentogether,the datanow excludethe SM at the98.8%y

confidencelevel.OfÈ

theabovediscrepancies,it is essentiallyonly R~

bp whichÉ

causes} problems.If Rbp by�

itself is assumedto be accountedforz

by new physics,thenthe fit to the datadespitethe otherdiscrepanciesa

is reasonableÊÌË min2 /­NÍ

DF Î 15.5/11Ï —the othermeasurementscould thus be regardedsimply as statisticalfluctuations.

InÐ

this paperwe haveperformeda systematicsurveyofnew-physics� models in order to determinewhich featuresgive� correctionsto Rb

p of� the right sign andmagnitude.Themodelsconsideredcan be separatedinto two broad class-es:h thosein which new Zb

Ñ ¯bÒ

couplings} appearat treelevel,by�

or� bÒ

-quarkmixing with newparticles,andthosewhichgive� loop correctionsto the Zbb

Òvertex.� The latter type in-

cludes} tÓ -quarkmixing andmodelswith new scalarsandfer-mions.§ We did not considertechnicolormodelsor newgaugebosons�

appearingin loopssincethesecasesaremuchmoremodeldependent.

The new physics can modify either the left-handedorright-handed� Zb

Ñ ¯bÒ

couplings,} gÔ Lbp

or� gÔ Rbp. To increaseR

~bp toy

itsexperimentalh value, Õ gÔ L

bp

must§ be negativeandhavea mag-nitude typical of a loop correctionwith large Yukawa cou-plings.� Thus Ö gÔ L

×bp could} eitherbea small tree-leveleffect,ora_ largeone-loopeffect. On the otherhand,the SM valueofgÔ R

bp

isu

oppositein sign to its LH counterpartandis aboutfivetimesy

smaller.Thereforeone would needa large tree-levelmodificationto gÔ R

Øbp toy

explainfor Rbp .

Here�

areour results.(1)Ù

Tree-level effects. ItÐ

is straightforwardto explainR~

bp ifu

they

Z or� bÒ

mix with newparticles.With Z Ú Z Û mixing thereare_ constraintsfrom neutral-currentmeasurements,but thesedoa

not excludeall models.UsingbÒ

-bÒ Ü

mixing§ is easiersincethey

experimentalvalue of R~

bp can} be accommodatedby

L× -bÒ

LÝ or� bÒ

RØ -bÒ

RÞ mixing.§ If the mixing is in the LB bÒ

sector,`theny

solutionsare possibleso long as I3ß

L×à¬áãâ 1/2. An addi-

tionaly

possibility with I3ß

L×ä¬å 0

æand very large LH mixing,

thoughy

perhapsunappealing,is still viable.For RH bÒ

mixing,§ifu

R

èêé0æ

then small mixing is permitted,while if Iç

R

ëíì0,æ

large©

mixing is necessary.Interestingly,the requiredlargebÒ

-mixing anglesare still not ruled out phenomenologically.A numberof papersin the literaturehaveappealedto b

Ò-bÒ î

mixing to explainRbp . Our ‘‘master formula’’ ï 8¡ ð and_ Table

IIÐ

includeall of thesemodels,aswell asmanyothers.(2)Ù

Loops: t-tÓ ñ mixing.ò InÐ

the presenceof tÓ -tÓ ó mixing,§they

SM radiative correctioncan be reduced,dependingonthey

weakisospinquantumnumbersof the tÓ ô as_ well ason theLHõ

andRH mixing angles.However,we found that it is notpossible� to completelyexplainR

~bp via� this method.The best

weÉ cando is to decreasethediscrepancybetweentheoryandexperimenth to about2ö . Sucha scenariopredictsthe exist-enceh of a light ÷ùø 100 GeVú charge} 2/3 quark,decayingpri-marily§ to Wb.

(3)Ù

Loops: Diagonal couplings to the Z. We consideredmodelswith exotic fermionsandscalarcouplingto both theZÑ

and_ bÒ

quark.û We assumedthat the couplingsto the ZÑ

are_16Forü

example,theparticularcaseof mixing of thetop quarkwitha new isosingletT ý , and the correspondingeffectsinducedon theZbs vertex,þ wasstudiedin Ref. ÿ 33

� �through�

ananalysisvery simi-lar to that of Sec.IV.

4298 54BAMERT,�

BURGESS,CLINE, LONDON, AND NARDI

Page 25: Rb and new physics: A comprehensive analysiscliff/papers/Rb.pdf · 2002-11-12 · Rb and new physics: A comprehensive analysis P. Bamert,1 C. P. Burgess,1 J. M. Cline,1 D. London,1,2

diagonal,a

i.e., thereare no flavor-changingneutral currents�FCNC’s� . The correction � gÔ L

bp

can} then be written in asimple` form, Eq. � 82

� . The key point is that gÔ L

bp

isu

propor-tionaly

to Iç

L×f� � I

ç3ß

RØf� ,� where I

ç3ß

,RØf

�isu

the third componentofweakÉ isospinof the fermion field f

WL×

,RØ in the loop. This ex-

plains� at a glancewhy many models,suchas multi-Higgs-doubleta

models and Zee-type models, have difficulty ex-plaining� R

~bp . Since the dominant contributions in these

models§ typically havetop-typequarks Iç 3ß

L

��� 12� ,� Iç

R

���0æ �

cir-}culating} in the loop, they give correctionsof the wrong signtoy

R~

bp . However,theseconsiderationsdid permit us to con-

struct` viable modelsof this type which do explainRbp . Two

such` examplesaregiven in Sec.V D, andmany otherscanbe�

invented.(4)Ù

Loops: Nondiagonal couplings to the Z. We alsoex-amined_ modelswith exotic fermionsandscalarswhich wereallowed_ to havenondiagonalcouplingsto the Z. SuchFC-NC’s�

canoccurwhenparticlesof differentweakisospinmix.The�

correction � gÔ Lbp

isu

much more complicated � Eq.� �

86� ���

thany

in the previouscase;evenits sign is not obvious.However, there are several interesting limiting cases

whereÉ it againbecomestransparent.The contributionsto R~

bp

of� supersymmetryfall into this category,which we discussedin somedetail.

NoteÍ

added. After completingthis work we becameawareof� Ref. � 38

+ ,� which discussesa different regionof parameter

space` in SUSYmodelsthantheonewe focusedon. Becauseof� our criterionof explainingtheentireRb

p discrepancya

ratherthany

only reducingits statisticalsignificance,we excludetheregion� in question.

ACKNOWLEDGMENTS

This�

researchwas financially supportedby NSERC ofCanada!

andFCAR of Quebec.�

E.N. wishesto acknowledgethey

pleasanthospitalityof thePhysicsDepartmentat McGillUniversity,�

during the final stageof this work. D.L. wouldlike©

to thankKen Raganfor helpful conversations.

"1# LEP electroweakworking groupand the LEP Collaborations,

‘‘A Combinationof PreliminaryLEP ElectroweakResultsandConstraints$

on the StandardModel,’’ preparedfrom summer1995conferencetalks % unpublished& .'

2( SLC)

Collaboration,aspresentedat CERN by C. Baltay, June1995 * unpublished+ .,

3- CDF$

Collaboration,F. Abe et al., Phys.Rev. Lett. 74, 2626.1995/ ; D0 Collaboration,S. Abachi et al.,0 ibid.

174, 26322

19953 .445 C.$

P. Burgess,S. Godfrey, H. Konig,6 D. London, and I.Maksymyk,Phys.Rev.D 49,0 6115 7 19948 .9:

5; P.<

Bamert,C. P.Burgess,andI. Maksymyk,Phys.Lett. B 356=

,0282> ?

1995@ .9A6B P. Bamert,McGill University ReportNo. McGill/95-64, hep-

ph/9512445,C 1995 D unpublishedE F .G7H M.I

E. PeskinandT. Takeuchi,Phys.Rev.Lett. 65, 964 J 1990K ;Phys.<

Rev.D 46,0 381 L 1992M ;N W. J. MarcianoandJ.L. Rosner,Phys.Rev.Lett. 65,0 2963 O 1990P ;N D. C. KennedyandP. Lan-gacker,Q ibid. 65, 2967 R 1990S ;N B. HoldomandJ.Terning,Phys.Lett.T

B 247U

, 88 V 1990W .9X8Y Z

M.I

Shifman,Mod. Phys.Lett. A 10[

,0 605 \ 1995] ;N UniversityofMinnesotaReport No. TPI-MINN-95/32-T, hep-ph/9511469,1995 ^ unpublished_ .`

9a P.<

Chiappetta,J. Layssac,F. M. Renard,andC. Verzegnassi,Phys.<

Rev. D 54, 789 b 1996c .9 G. Altarelli, N. Di Bartolomeo,F. Feruglio,R. Gatto,andM. L. Mangano,CERN ReportNo.CERN-TH/96-20,$

hep-ph/9601324,1996 d unpublishede .9f10g C.-H.

$V. Chang,D. Chang,andW.-Y. Keung,NationalTsing-

Huah

University Report No. NHCU-HEP-96-1, hep-ph/9601326,i

1996 j unpublishedE k .9l11m J.

nBernabe´u, A. Pich, and A. Santamarı´a,o Phys.Lett. B 200,0

569p q

1988r .9 For other discussionsof the radiativecorrectionsto�

Zbs ¯b, seeA. A. Akhundov,D. Yu. Bardin,andT. Riemann,

Nucl.t

Phys.B276, 10 u 1986v ; W. BeenakkerandW. Hollik, Z.Phys.C 40, 141 w 1988x ;N B. W. Lynn and R. G. Stuart,Phys.Lett. B 252, 676 y 1990z ;N J. Bernabe´u, A. Pich, and A. San-tamarı� ´a, Nucl. Phys.B363,0 326 { 1991| .

}12~ E. Ma, Phys.Rev.D 53

�,0 2276 � 1996� .9�

13� T. Yoshikawa,HiroshimaUniversityReportNo. HUPD-9528,hep-ph/9512251,�

1995 � unpublished� .�14� G.

�Bhattacharyya,G. C. Branco,andW.-S.Hou,Phys.Rev.D�

to�

be published� .�15� E. Nardi,E. Roulet,andD. Tommasini,Phys.Rev.D 46, 3040�

1992� .�16� B.

�Holdom,Phys.Lett. B 351

=, 279 � 1995� .9�

17� CDF$

Collaboration,T. J. LeCompte,Fermilab Report No.FERMILAB-CONF-96/021-E� unpublished� .�

18� See,for example,P. LangackerandD. London,Phys.Rev.D38,0 886 � 1988� .�

19� E. Nardi,E. Roulet,andD. Tommasini,Nucl. Phys.B386, 239�1992  .¡

20¢ E.£

Nardi,E. Roulet,andD. Tommasini,Phys.Lett. B 344,0 225¤1995¥ .¦

21§ JADEn

Collaboration, E. Elsen et al., Z. Phys. C 46,0 349¨1990© .ª

22« CELLO$

Collaboration,H. J.Behrendet¬ al.,0 Z. Phys.C 47, 333­1990® .¯

23° P. BamertandC. P. Burgess,Z. Phys.C 66,0 495 ± 1995² .³24 CDF

$Collaboration,F. Abe et¬ al.,0 Phys. Rev. D 45,0 3921µ

1992¶ .·25 E.

£Ma andD. Ng, Phys.Rev.D 53,0 255 ¹ 1996º .»

26¼ L. Rolandi,H. Dijkstra, D. Strickland,and G. Wilson, repre-senting½ the ALEPH, DELPHI, L3, andOPAL Collaborations,Jointn

Seminaron theFirst Resultsfrom LEP 1.5,CERN,1995¾unpublishedE ¿ .9À

27Á J.n

L. Feng,N. Polonsky,andS. Thomas,Phys.Lett. B 370, 90 5Â1996Ã .Ä

28Å E. H. SimmonsandY. Su,Phys.Rev.D Æ to� be publishedÇ .È29É J.

nEllis, J. L. Lopez,andD. V. Nanopoulos,CERNReportNo.

CERN-TH/95-314,$

hep-ph/9512288Ê unpublishedË .9Ì30Í J.

nD. Wells andG. L. Kane,Phys.Rev.Lett. 76, 869 Î 1996Ï .Ð

31Ñ Y. Yamada,K. Hagiwara,andS.Matsumoto,KEK ReportNo.KEK-TH-459,Ò

hep-ph/9512227Ó unpublishedÔ .Õ32Ö Y.

×Grossmann,Z. Ligeti, andE. Nardi, Nucl. Phys.B465

Ø,0 369Ù

1996Ú .

54 4299RÛ

bÜ AND NEW PHYSICS:A COMPREHENSIVEANALYSIS

Page 26: Rb and new physics: A comprehensive analysiscliff/papers/Rb.pdf · 2002-11-12 · Rb and new physics: A comprehensive analysis P. Bamert,1 C. P. Burgess,1 J. M. Cline,1 D. London,1,2

Ý33Þ E. Nardi, Phys.Lett. B 365

=, 327 ß 1996à .9á

34â See,)

for example,S. Bertolini, F. Borzumati,A. Masiero,andG.�

Ridolfi, Nucl. Phys.B353ã

, 591 ä 1991å .9æ35ç UA1

èCollaboration,C. Albajar et¬ al.,0 Phys.Lett. B 262

U, 60 3é

1991ê .ë36ì CLEO

$Collaboration, R. Balest et al., Cornell Report No.

CLEO-CONF-94-4$ í

unpublishedE î .9ï37ð CDF

$Collaboration,C. Anway-Wieseet al., FermilabReport

No.t

Fermilab-Conf-95/201-Eñ unpublishedE ò .ó38ô P.

<H. ChankowskiandS. Pokorski,ReportNo. IFT-96/6,hep-

ph/9603310C õ unpublishedE ö .9

4300 54BAMERT,�

BURGESS,CLINE, LONDON, AND NARDI