HAL Id: jpa-00209747
https://hal.archives-ouvertes.fr/jpa-00209747
Submitted on 1 Jan 1984
HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
On tests of time reversal invariance in nucleon nucleon scattering
J. Bystricky, F. Lehar, P. Winternitz
To cite this version:
J. Bystricky, F. Lehar, P. Winternitz. On tests of time reversal invariance in nucleon nucleon scattering.
Journal de Physique, 1984, 45 (2), pp.207-224. �10.1051/jphys:01984004502020700�. �jpa-00209747�
On
testsof time reversal invariance in nucleon nucleon scattering
J.
Bystricky (1),
F. Lehar(1)
and P. Winternitz(2,3)
(1) D.Ph.P.E., CEN-Saclay, 91191 Gif sur Yvette, France
(2) Centre de recherche de mathématiques appliquées,Université de Montréal, Montréal,Québec,Canada H3C3J7 (Refu le 22 juin 1983, accepté le 14 septembre 1983)
Résumé. - On développe le formalisme de la diffusion nucléon-nucléon en assumant l’invariance de Lorentz et la conservation de la parité mais hon l’invariance dans le renversement du temps. La matrice de diffusion pp contient alors 6 amplitudes (une d’elles, g(E, 03B8) ne satisfait pas à I’IRT); celle de diffusion np, 8 amplitudes
(g(E,
0) ne satisfaitpas à I’IRT, f(E, 03B8) ne satisfait pas à l’invariance isotopique et h(E, 03B8) ne satisfait aucune de ces deux symétries).
Une analyse préliminaire de l’etat expérimental actuel sur 1’IRT en diffusion pp mène à des conclusions non encore définitives.
Abstract. 2014 The nucleon-nucleon scattering formalism is developed assuming Lorentz invariance and parity
conservation but allowing for a violation of time reversal invariance. The pp scattering matrix involves six ampli-
tudes (one, g(E, 0) violates TRI) the np matrix eight
(g(E,
0) violates TRI, h(E, 0) violates TRI and isospin invariance, f(E, 03B8) violates isospin invariance). A preliminary analysis of the experimental status of TRI in pp scattering is performed and the results are surprisingly inconclusive.Classification
Physics Abstracts
11.80 -11.30 -13.75C -13.85C
1. Introductioa
The purpose of this article is to
provide
a detailedframework for
performing
tests of time reversal(TRI)
in elastic nucleon nucleon
scattering
and simulta-neously
to review the present status of TRI in theelementary
nucleon nucleon interaction.A recent experimental comparison
[1]
of the protonpolarization
P in the reactions ’He + 7Li -+ p + 9Be and 3He + 9Be -+ p + 11B with the ’He asymmetry Ain the inverse reactions
using
apolarized
proton beam reports alarge
P-A difference (for 14 MeV 3He ions).This P-A difference, if confirmed,
directly implies
aserious breakdown of TRI.
In the strong interactions the consequences of this breakdown would be
extremely
dramatic since all current notions of the strong interactions assumeTRI, at least
implicitly
via the CPT theorem. Inparticular it is hard to
imagine
that QCD couldsurvive in
anything
like its present form if TRI does not hold.The authors of reference 1
point
out that theirs is the firstcomparison
of P and A made in an inelastic reaction, and thatprevious
P-A tests in elastic scat-tering
on 3He or 13C nuclei wereperformed
in unfor-tunately
chosen kinematicregions (where
P A issmall
indepently
ofTRI).
To this one can add thattests of TRI via detailed balance are not sensitive to
possible
spindependent
violations(e.g.
if differentspin amplitudes
in an inelastic reactionacquire
different
phases
under time reversal).Our motivation for
reconsidering
thequestion
ofTRI in NN
scattering
is that we find itextremely unlikely
that TRI could be violated in nuclear reac-tions, without
being
violated in the interaction of two nucleons.Independently
of the outcome of the present contraversy over the P-A tests in inelasticscattering,
we find that the situation in NN
scattering
is far from conclusive and that further tests of TRI would be most valuable.We
systematically develop
a 6amplitude
formalismfor pp
scattering
and an 8amplitude
formalism for the np case. Our notations and conventions are the same as in twoprevious
articles[2,
3]. The 8amplitude
formalism is also
appropriate
for thestudy
of inelasticreactions of the
type 2 + .1 -+ .1 + I (e.g. Ap -+ I + n).
In our
analysis
we make use of a recentcompilation
of the world nucleon nucleon
scattering
data[4]
and also of a
recently performed phase
shiftanalysis [5].
Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:01984004502020700
For a
general
review of theexperimental
status ofTRI in weak,
electromagnetic
and strong interactionswe refer to Richter
[6].
An earlieranalysis
of the ppscattering
data wasperformed by
Thorndike [7].A violation of TRI in NN
scattering
isexpected
incertain models. Sudarshan [8]
proposed
a unifiedmodel of strong, weak and electromagnetic inter-
actions in which CP invariance, and hence TRI is violated in a
specific
manner. The consequence of this model for TRI in NNscattering [9, 10]
have beenstudied and
predictions
have been madeassuming
that TRI violation occurs in the lowest
possible angular
momentum stateonly [11].
The formalism
presented
in this article should be of use in theplanning
ofexperiments directly testing
TRI in NN
scattering
on one hand and in the recons-truction of
scattering amplitudes
from data, without theassumption
of TRI, on the other. Such a recons-truction
would provide
the bestpossible
test of TRI.If
performed
viaphase
shiftanalysis
it wouldprovide
the
possible
TRI violatingamplitudes
as functions of the scatteringangle
andpossibly
also of the energy.A model
independent
test of TRI is bestprovided by
areconstruction of the NN
scattering
matrix from acomplete
experiment
[12]. For ppscattering
this wasperformed recently by
the Geneva group[13]
at579 MeV for six different
scattering angles (use
was.made of some formulas from a
preliminary
versionof this paper
[14]).
Their conclusion is that at the considered energy the TRIviolating amplitude
doesnot contribute more than 1
%
to the differentialcross section.
2. The
scattering
matrix.Assuming only
Lorentz invariance(or
Galilei inva-riance)
andparity
conservation we can write thescattering
matrix for the elasticscattering
of twoparticles
ofspin 1/2
in terms of 8complex
functionsof the
energy E
andscattering angle
0.Using
invariantamplitudes a,
b, c, d,e, f, g
and h we can writewhere
and
k;
andkf
are unit vectors in the direction of the incident and scatteredparticles
in the centre-of-mass system(c.m.s.).
For identical
particles
the Pauliprinciple requires
Thus
(2 . 3)
must hold for pp and nnscattering
whereas for npscattering
this reation would be a consequence of an additionalrequirement, namely isotopic spin
invariance.Electromagnetic
corrections, for instance due to onephoton exchange taking
the nucleonelectromagnetic
form-factors into account, will cause theamplitude
f todiffer from zero
[15-19J.
Time reversal invariance, on the other hand,
requires
that g and h vanishThus in proton proton
scattering precisely
oneamplitude, namely g
=g(E, 0),
violates TRI, whilerespecting parity
conservation. In neutron protonscattering
there are two suchamplitudes;
the second one h =h(E, 0)
also violates
isospin
invariance. Forf = g = h = 0 we
return to the fiveamplitudes
introducedoriginally by
Wolfenstein
[20].
Alternatively,
thescattering
matrix can be described in termsof eight helicity amplitudes [21]. Using
standardnotations and conventions we express the
helicity amplitudes
in terms of the invariant ones as(0
is the c.m.s.scattering angle).
For identical
particles
we haveTime reversal invariance
implies
Finally, nucleon nucleon
phase
shift analysis is bestperformed using
thesinglet-triplet representation.
Generalizing
the notations ofStapp et
al. [22] we putwhere we set ms
and ms equal
to S for thesinglet
spin state and to 1, 0, - 1 for thecorresponding
three pro-jections
in thetriplet
state.We then have
For identical
particles
we haveTime reversal invariance
implies
A
phase
shiftexpansion
of thesinglet triplet amplitudes (2 . 8)
can be written asThe
phases
and normalization are so chosen that the formulas reduce to thoseof Stapp et
al.[22]
for pp scatter-ing
with TRI conserved; the Clebsch-Gordan coefficients andspherical
harmonics are as in reference 23.For
fixed j we parametrize
thepartial
waveamplitudes R¿,I’S’
asand
All other elements of the matrix
Rjis,rs,
vanish because ofangular
momentum orparity
conservation.If the
particles
are identical, thesinglet-triplet mixing
is forbidden, i.e. the newmixing
angle y j vanishes :For np scattering yj is a measure of isospin violation.
If TRI is assumed, then the matrices (2.13) and
(2.14)
aresymmetric,
i.e.If both
(2.15)
and(2.16)
hold, then formulas(2.13)
and (2.14) reduce to those of Stapp et al. [22]. The notation2 yj
for thesinglet-triplet mixing
angle was chosen to coincide with Gersten[17].
Note that the
parametrization (2.13)
and(2.14)
of a 2 x 2unitary
matrix is somewhatarbitrary.
E.g. in(2.13)
the TRIbreaking phase
shift T. was introducedby
a rotation of the usual [22]phase
shift parameters :and
similarly
for (2.14)Binstock et al.
[ 11 introduced
a different TRIbreaking phase
shift parameter. Instead of a TRIbreaking
rotation
t j they
introduced a TRImixing angle Àj’ putting
Similarly
we couldparametrize
the 1 = j matrixS 2 = R4
+ I asIn the
purely
elasticregion,
below thepion production
threshold in NNscattering,
allphases
andmixing angles
are real. Above this threshold inelasticities must be introduced, e.g.
by allowing
the parameters to become complex.Finally,
let usgive
thepartial
waveexpansions
of the TRI andisospin violating amplitudes explicity :
For np
scattering,
theexpansions
stand as written. The summation is over all indicated valuesof l or j.
Theelectromagnetic amplitude f EM
can be taken from the onephoton approximation
andb,il 61
and gj parametersare to be
interpreted
as the « nucleon bar »phase
shifts ofStapp et
al.[22].
For pp
scattering
we havef
= h = 0. In terms of thephase
shift parameters this means 7j = 0 in equa- tion 2.14and yj
= pj = 0 inequation
2.20. Theamplitude g
must be odd under theexchange
0 - n - 0,hence the summation is over even values
of j only.
Theelectromagnetic
corrections can at this stageonly
betreated in a somewhat
hybrid
way. Thus, we haveThe
phase
shifts are approximated aswhere
0l,
is the usual Coulomb phase shift [22] and6fl 87
areagain
« nucleon bar »phase
parameters[22].
3. Experimental
quantities.
Using
the same notations as in twoprevious
articles [2, 3] we write ageneral experimental quantity
aswhere Q is the differential cross section for
unpolarized particles.
The order of labels from left toright
is scatteredparticle,
recoilparticle,
beam and target. For « pure »experiments
in the c.m.s. all labels take the values 0, n, I,or m. In the I.s. we introduce the usual three sets of basis vectors
where k, k’ and k" are unit vectors in the directions of the initial scattered and recoil
particle
l.s. momenta,respectively.
The Bohr rule
for 2 + -1 2 -+ 2 I + 2
scatteringis a consequence of
parity
conservation alone and hence holds for thescattering
matrix(2 .1 )
under considera- tion. Relation(3.3)
allows us to reduce the number of observables to be calculatedby
a factor of two. Indeed,in the c.m.s. we have
where
[li], [mi], [1f]
and[m f]
denote the numbers of I and m labels in the initial and final states,respectively
and alabel a’ is
equal
to n, 0, m, or 1, when a isequal
to o, n, I or m. In the I.s. we havewhere
[kj, [s;], [k f]
and[s f]
denote the number of k and s type labels in the initial and final statesrespectively,
and a label x’ is
equal
to n, o, k(k’, k"), ors(s’,
s") when x isequal
to o, n, s(s’, s") or k(k’, k"),respectively.
In tables I and II we
give expressions
for all « pure » c.m.s. and l.s.experiments
in terms of theamplitudes
a, ..., h. The observables 0,
Cnnoo, Dnono, Knoon, Pnooo, Ponoo’ Aoono
andAooon
are the same in both systems and are omitted from table II.Altogether
64linearly independent experiments
existThe tables
(and
all results of thispaper)
are valid for the elasticscattering
of two nonidenticalspin 1 /2 particles
for which TRI is not assumed
(e.g.
elastic np --> np scattering in which both isospin and TRI violation isallowed).
They
can also beinterpreted
to describe an inelastictwo-body
reaction of thetype i + L --, -L
+-L (e.g.
A + p - E + + n)independently
of TRI or its violation.2 2 2 21
Formulas for elastic pp or nn
scattering,
in which TRI violation is allowed but the Pauliprinciple
is assumed,are obtained
by
puttingFormulas of reference 3 are obtained
by
putting g = h = 0; those of reference 2by putting f
= g = h = 0.Table I. - Centre
of
mass experimental quantities in termsof
scatteringamplitudes.
Table I (continued)
Table II. -
Laboratory
system experiments in termsof scattering amplitudes.
Table II (continued)
Table II
(continued)
All vectors and
angles
involved in transformations between the c.m.s. and l.s. are given onfigure
1 of refe-rence 2. We have
where
R1
andR2
denote a relativistic spin rotation. Theangles
a andfl
for elastic scattering are related to therelativistic spin rotation for the scattered and recoil particles
where 0 is the c.m.s.
scattering angle, 0,
and02
are the l.s.scattering
and recoilangles.
In the nonrelativistic limit we have a =0, p = n/2.
For an inelastic twobody
reaction theangles
aand P depend
in aquite compli-
cated way on the energy and
scattering angle.
We do notgive
theexpression
here but refer e.g. to [24] for ageneral
treatment of the Wigner
spin
rotation.4. Direct tests of time reversal invariance.
All
possible
direct tests of TRI can be read off from tables I and II. If we restrict ourselves to observables involv-ing
at most twospin
components, we find 12experiments measuring
the interference between g or h and one ofthe
amplitudes
a,..., f
A furtherexperiment
is sensitive to the interference between g and h. In the c.m.s. we have for npscattering :
Further tests of TRI would involve 3 and 4 component tensors, e:g. the square moduli of g and hare :
For pp
scattering
we havef
= h = 0 soonly
5 of the relations 4.1 survive and areindependent.
Translating
the relations into the I.s. we obtain(for
npscattering) :
For identical
particles
the situation is muchsimpler
andonly
5 independentexperimental
tests of TRIinvolving
at most two component tensors exist. For pp scattering wesimplify
relations 4. 3 to obtain;No tests of TRI have so far been
performed
in npscattering.
In ppscattering
thepolarization
asymmetry relation has been tested atquite
a few energies(see below).
Theonly
other testperformed
[24, 25] was based onthe
depolarization
tensor relation 4.4b.Although
four components of the Wolfenstein tensorDpoio (or Doqok)
are involved in relation 4.4b. Handler et al. used an
ingenious experimental
setup which made itpossible
toverify
the relation whileperforming only
twoscattering experiments.
Since the same is true for some of the other TRI
testing relations,
let us discuss these setups in some detail,performing
all considerationsdirectly
in thelaboratory systeni.
4.1 DEPOLARIZATION TENSOR FOR POLARIZED BEAM IN pp OR np SCATTERING (see
Fig.
la). - Consider aninitial beam of nucleons with
polarization
vectorPB
in thescattering plane
and anunpolarized
target :Let the
polarization
of the scattered nucleon beanalysed by rescattering
on a zerospin
target withanalysing
power 1’1 (the polarization-asymmetry equality
for this target follows fromparity
andangular
momentumconservation).
Let the normal n, to theanalysing plane
lie in the studiedscattering plane,
i.e. we aredetecting
anup-down
asymmetry. Place amagnetic
fieldparallel
to the normal n between the target nucleon and thespinless analyser.
Themagnetic
field will rotate the scatteredparticles
linear momentum k’through
some .fixedangle
and its
polarization through
some additionalangle 4/ (about
the normal n). After this rotation we havewhere +
( - )
refers to down(up)
scattering. Theup-down
asymmetry on theanalyser
will begiven
as[2,
3]Now consider two such asymmetry measurements, I and II
performed
for the same value of the l.s. scatter-ing
angle81,
but for different values of the initialpolarization angles x,
and xI, and themagnetic
field rotationFig. 1. - Experimental set-ups for tests of TRI involving the depolarization D or polarization transfer K tensor. In all four
cases T denotes the target M a magnetic field parallel to the normal n, A1 and A2 denote spinless analysers with analysing
powers
Pi
andP2,
respectively. The unit vectors kD, SD k; and sD point to the directions to which the vectors k’, s’, k", s"were rotated by the magnetic field M. The unit vectors
kp, sp, kp,
sp point to the directions to which the corresponding polari-zation components were rotated, 4/ is the angle of the additional spin rotation. P, and PT are the beam and target polarizations lying in the scattering plane, x is the angle between the initial particle polarization and the beam direction k. Figures 1 a-I d
show the set-ups for
Dpoio(np
and pp),Dpqok(np
and pp),KOqiO(PP),
andKpook(pp),
respectively.angles 4/,
and1/111’
ChooseWe then obtain
If TRI holds, then
AI + All
= 0. In different notations thiscorresponds
to theexperiment performed by
Handler4.2 DEPOLARIZATION TENSOR FOR POLARIZED TARGET IN pp OR np SCATTERING
(Fig. lb).
- Consider anunpolarized
beam and a target withpolarization
vectorPT
in thescattering plane
Let the
polarization
of the recoil nucleon beanalysed
on aspin
zeroanalyser
withanalysing power P2
and letthe normal n2 to the
analysing plane
lie in thescattering plane.
Rotate thepolarization
of the recoilparticle through
anangle 4/
about the normal nby
means of amagnetic
field. After this rotation we havewhere +
(-) again
refers to down(up) rescattering.
Theup-down
asymmetry on theanalyser
will bewhere
P2
is theanalysing
power on theanalyser.
Let two such
experiments
beperformed
for the same recoilangle 92
and for Xiand qli satisfying
We then have
Again,
theright
hand side vanishes if TRI holds(see
(4.3d)
and(4 . 4b)).
4.3 POLARIZATION TRANSFER FOR POLARIZED BEAM IN pp SCATTERING
(Fig.
1 c). - Consider a beam of pro- tons withpolarization
vectorPB
as in(4.5)
and anunpolarized
proton targetPT
= 0. Let thepolarization
ofthe recoil proton be
analysed
on aspin
zeroanalyser
and let the normal n2 to theanalysing
plane lie in thescattering plane.
Rotate thepolarization
of the recoilparticle through
anangle t/J
about the normal nby
meansof a
magnetic
field After this rotation weagain
have relations 4.11. Theup-down
asymmetry on theanalyser
is
[2,
3]Let two such
experiments
beperformed
for the same recoilangle 02
and Xiand Oi satisfying (4.13).
We then have
which
provides
a test of the first of relations 4. 4c.4.4 POLARIZATION TRANSFER FOR POLARIZED TARGET IN pp SCATTERING
(Fig.
1 d). - Consider an unpo- larized proton beamPB
= 0 and a proton targetpolarized
as in(4 .1 U).
Let thepolarization
of the scattered proton beanalysed
on aspin
zeroanalyser
and let the normal n, to theanalysing plane
lie in the scattering plane. Rotate thepolarization
of the scatteredparticle through
theangle 0
about the normal nby
means of amagnetic
field. After this rotation we have relations 4.6. Theup-down
asymmetry on theanalyser
is[2,
3]Let two such
experiments
beperformed
for the samescattering angle 81 and xi and 4/, satisfying (4.8).
Wethen have
providing
a test of the second of relations 4.4c.The
polarization
transfer tests(4. 3e)-(4. 3h)
for npscattering
are morecomplicated They
involve 8 quan- tities each; these can be obtainedby performing
4experiments
in each case. The same holds for the npdepola-
rization test
(4. 3m).
5. Discussion of time reversal invariance tests.
Some conclusions can be drawn
directly
from theformalism
presented
above.1) TRI tests should be
performed
inkinematically
favorable regions. For pp scattering such
regions
arecharacterized
by
the fact that the quantitiesare all
large.
Inparticular,
P-A tests should beperform-
ed where the
amplitude d
is known to belarge (from
other
experiments
and fromphase
shiftanalysis).
Similarly,
the D and K tensor tests should beperform-
ed where c and b,
respectively,
are known to belarge.
2)
Only
two of the five possible TRI tests in ppscattering
have even beenperformed, namely (4.4a)
and
(4.4b),
the second one only at onespecific
energyand
angle (E
= 430 MeV, 0 = 300). Theseprovide
limits on
rather than
directly
onI g 1.
3) It would be desirable to perform a new
phase
shift
analysis
of all pp scattering data without assum-ing
TRI. This would inprinciple provide
the am-plitude
g(E,0)
as a function ofscattering angle
andenergy. This is
unfortunately
very hard to do,mainly
because TRI is
usually
built into the manner that dataare presented
(e.g. Aoono
=Pnooo
isexplicitly imposed).
A less ambitious
approach
is to assume that the TRIconserving phases
are known and then to obtain values and limits for the additional TRIviolating phases.
This has beenattempted
before[7, 11]
and wedo this below in section 6, using a
considerably
more complete set of data.4) The most convincing test of TRI would involve
a direct reconstruction of all scattering
amplitudes
from a
complete experiment
at different energies andangles.
As mentioned in the introduction, this hasrecently
beenperformed
for one energy [13] in ppscattering.
5)
It is quite concievable that TRI could be violated in np scattering and hence in nuclear reactions, withoutbeing
violated in pp scattering. Thus thegeneral
nucleon nucleon
scattering
matrix could contain a term(where tiz
is the third component of theisospin
matrixfor the ith nucleon), while
g(E, 0)
could beidentically
zero. Very little is known
experimentally
about TRIin np
scattering.
Tests should beperformed
inregions
where the
following quantities
arelarge :
For each of
the
tests in(4. 3)
theregion
must be sochosen that the
amplitude
with which g or h is inter-fering
islarge.
6.
Preliminary analysis
ofexperimental
data.In order to illustrate the status of TRI in pp scatter-
ing
we have taken data on the A-P difference fromexperiments specifically designed
to test TRI[26-34]
at 142, 213 and 635 MeV. These are
reproduced
intable III. We have fitted the asymmetry
Aaono
andpolarization Pnooo by
aLegendre polynomial
expan- sionThe
resulting
curves arepresented
onfigures
2a, 3a and 4a. The corridor of errors isprovided by
an errormatrix calculated from the
experimental
errors. Wesee that close to 900 the asymmetry approaches zero
somewhat faster than the
polarization (this
is consist-enly
true for manyenergies
up to 1GeV). Using
the-formula
we have determined the
angular dependence
ofI g I sin ogd.
The results are shown onfigures
2b, 3band 4b, for the three
energies
concerned. The diffe- rential cross sectiona(0)
and theamplitude d(O)
werecalculated from the
Saclay-Geneva phase
shifts [5].(I.e.
modifications ofa(0)
andd(O)
due to the presence of TRIviolating phase
shifts wereignored).
The quantities(5 .1 )
are allreasonably
large for the consi- deredenergies.
Indeed the minimal valuesof (1- Dnono
are
approximately equal
to 0.8, 0.6 and 0.3 at 142,213 and 635 MeV
respectively,
for8cM
> 30°. Thequantity (1 - Knoon) is
consistentlylarger
than(1
-Dnono)
in the considered energyregion
and (1 +Cnnoo)
is evenlarger,
sinceCnnoo
ispositive
in thisregion
of energies andangles.
At lowenergies
thecondition on (5 .1 c) becomes the
limiting
on sincethere we have
Cnnoo = Aoonn ’" -
1.Table III. -