arX

iv:0

708.

3717

v1 [

hep-

ex]

28

Aug

200

7

Evidence for the Production ofSlow A ntiprotonic H ydrogen in Vacuum

N.Zurlo,1,2 M .Am oretti,3 C.Am sler,4 G .Bonom i,5,6 C.Carraro,3,7 C.L.Cesar,8 M .Charlton,9

M .Doser,10 A.Fontana,6,11 R.Funakoshi,12 P.G enova,6,11 R.S.Hayano,12 L.V.J�rgensen,9

A.K ellerbauer,10 V.Lagom arsino,3,7 R.Landua,10 E.LodiRizzini,1,2 M .M acr��,3 N.M adsen,9

G .M anuzio,3,7 D.M itchard,9 P.M ontagna,6,11 L.G .Posada,12 H.Pruys,4 C.Regenfus,4

A.Rotondi,6,11 G .Testera,3 D.P.Van derW erf,9 A.Variola,3 L.Venturelli,1,2 and Y.Yam azaki13

(ATHENA Collaboration)1Dipartim ento diChim ica e Fisica per l’Ingegneria e per iM ateriali,Universit�a diBrescia,25123 Brescia,Italy

2Istituto Nazionale diFisica Nucleare, G ruppo Collegato diBrescia, 25123 Brescia, Italy

3Istituto Nazionale di Fisica Nucleare, Sezione diG enova, 16146 G enova Italy

4Physik-Institut, Z�urich University, CH-8057 Z�urich, Switzerland

5Dipartim ento diIngegneria M eccanica, Universit�a diBrescia, 25123 Brescia, Italy6Istituto Nazionale di Fisica Nucleare, Universit�a diPavia, 27100 Pavia, Italy

7Dipartim ento di Fisica, Universit�a di G enova, 16146 G enova, Italy

8Instituto de Fisica, Univesidade Federaldo Rio de Janeiro, Rio de Janeiro 21945-970

9Departm ent ofPhysics, University of W ales Swansea, Swansea SA2 8PP,UK

10EP Division, CERN, CH-1211 G eneva 23, Switzerland

11Dipartim ento diFisica Nucleare e Teorica, Universit�a diPavia, 27100, Pavia, Italy

12Departm ent of Physics, University of Tokyo, Tokyo 113-0033, Japan13Atom ic Physics Laboratory, RIK EN, Saitam a 351-0198, Japan

(D ated:June 2,2013)

W e presentevidence showing how antiprotonic hydrogen,the quasi-stable antiproton (�p)-proton

bound system ,hasbeen synthesized following theinteraction ofantiprotonswith them olecularion

H+

2in a nested Penning trap environm ent. From a carefulanalysis ofthe spatialdistributions of

antiproton annihilation events,evidenceispresented forantiprotonichydrogen production with sub-

eV kineticenergiesin statesaround n = 70,and with low angularm om enta.The slow antiprotonic

hydrogen m ay be studied using laserspectroscopic techniques.

PACS num bers:36.10-k,34.80.Lx,52.20.H v

Studies ofthe properties ofthe two-body hydrogenic

bound statesofthestableleptonsand baryonshavepro-

duced som e ofthe m ostprecise m easurem entsofphysi-

calquantitiesand provided powerfultestsofourunder-

standing ofthe laws ofnature. Interest in this area is

stillstrong,followingtherecentproduction ofantihydro-

gen, �H,at low energies [1,2]. Accurate com parisonsof

thetransitionsin hydrogen and antihydrogen areeagerly

awaited asa stringenttestofCPT sym m etry.

Antiprotonichydrogen (�pp)isalso ofinterest.Itslevel

structure issim ilarto thatofhydrogen,butwith m uch

larger binding energies. Precision m easurem ents ofits

spectroscopicpropertiesm ay allow determ ination ofthe

so-called antiprotonic Rydberg constantand/orthe an-

tiproton/electron m assratio.

Although �pp hasbeen studied extensively in the past,

this has exclusively been achieved by stopping antipro-

tonsin liquid orgaseoustargetsforX-ray spectroscopy

ofinnershellcascades,orfortheproduction ofnew light

m esonsand baryons(see e.g. [3,4]). Here we reporta

radically new m ethod of�pp production resulting in em is-

sion alm ost at rest in vacuum . This has been achieved

usingachem icalreactionbetween antiprotonsand m olec-

ularhydrogen ions(H +

2)in the ATHENA Penning trap

apparatus [1,5]. This advance has opened the way for

laser spectroscopic studies of �pp,or other antiprotonic

system sform ed by �p interactionswith HD + orD +

2 ,akin

tothosesuccessfullydeployedin thestudyofantiprotonic

helium (seee.g.[6,7]).

The experim entswere m ade possible by the availabil-

ity of a high-quality low energy �p beam delivered by

the CERN Antiproton Decelerator to the ATHENA �H

apparatus. The latter contained a m ulti-electrode sys-

tem ofcylindricalPenning traps,2.5 cm in diam eterand

� 90 cm in length im m ersed in an axialm agnetic �eld

of3 T.The residualpressure of� 10� 12 Torr,in the 15

K cryogenicenvironm entofthe trap,wasdue to hydro-

gen and helium gases. The centralregion contained the

m ixing trap: a nested Penning trap,approxim ately 10

cm long,thatallowed positrons,e+ ,and antiprotonsto

be con�ned sim ultaneously. For �H production the m ix-

ing trap contained a spheroidalcloud of� 3:5� 107 e+ .

Around 104 antiprotons were injected into this plasm a,

with the resulting �p annihilationsm onitored for60 sby

position sensitive detectors [5,8]. These registered the

passageofthecharged pions,to localizeannihilation ver-

tices which were due,not only to �H form ation followed

by annihilation on the electrode surface [1,9],but also

�p annihilation following transportto the electrode walls

[10],and annihilation followinginteractionswith residual

gas atom s or ions present in the trap. It was shown in

[1,9]that the vertex data were predom inantly �H anni-

2

r (cm)0 0.5 1 1.5 2

z (c

m)

-2

-1

0

1

2

(a)

r (cm)0 0.5 1 1.5 2

z (c

m)

-2

-1

0

1

2

(b)

r (cm)0 0.5 1 1.5 2

radi

al d

ensi

ty (

arb.

uni

ts)

0

10

20

30

40

50

r (cm)0 0.5 1 1.5 2

radi

al d

ensi

ty (

arb.

uni

ts)

0

10

20

30

40

50

60

70

Figure 1: r � z scatter plot and radialdensities�1

r

dN

dr

�of

the annihilation vertices for: (a) hot m ixing;(b) cold m ix-

ing. The dashed black line indicatesthe position ofthe trap

wall;thered sem i-ellipse showsthe section ofthe e+plasm a.

The green radialdensities are for the corresponding central

z-region events(insidethegreen lines)whilstthedata in blue

are forthe 2 lateralz-regions,norm alized forr> 1:25 cm .

hilationsduring so-called \cold m ixing" (CM ),when the

e+ cloud washeld atthe trap am bientof15 K .In con-

trastfor \hotm ixing" (HM ),when the e+ were heated

to a tem perature,Te,ofseveralthousand K (here 8000

K )[11,12],�H form ation wassuppressed and the �p anni-

hilationswere m ainly a resultofcollisionswith trapped

positive ions. It is this e�ect (which is also present for

CM )thatisaddressed here.

Fig.1 showsr� z scatterplotsforannihilation vertices

taken under HM and CM conditions. Here the radial

positions, r, (i.e. the distance from the trap axis) of

the events are plotted versustheir axialcoordinates,z.

Alsoshown aretheattendantradialdensity distributions�1

r

dN

dr

�. The distributions are broadened by the uncer-

tainties in the vertex determ ination (around 1.8 m m in

thez-direction and 3.5m m in thetransversedim ensions)

caused m ainly by theinability oftheATHENA detector

to reconstructthecurved trajectoriesofthepionsin the

3 T m agnetic �eld. The presentHM resultsare forthe

highestTe achieved by ATHENA.

There are striking di�erences between the two r� z

plots. Besidesthe �H annihilationson the trap wallcen-

tred around r= 1:25 cm (CM only),thereareeventslo-

calized atsm allerradiiwhich dom inate in HM (Fig.1a),

but which are also evident in CM (Fig.1b). Exam in-

Counts

r (cm)

120

100

80

60

40

20

00 0.5 1.5 2.51 2

τ = 0.8µsτ = 1.1µsτ = 1.4µs

Figure 2: Experim ental radial distribution of �p annihila-

tion vertices (black histogram ) for HM (� 1.5 cm < z <

1:5 cm ) with a M onte Carlo sim ulation (Te = 8000 K ,vth

= 5600 m s� 1,generation on the surface ofa spheroid with

zp = 16 m m and rp = 1 m m rotating with a frequency of

300 kHz,i.e. vtang = 2000 m s� 1);see text for details. Re-

sults ofsim ulations with di�erentm ean lifetim es are shown:

green,� = 0:8�s (�2

red = 2:78);red,� = 1:1�s(�2

red = 1:48);

blue,� = 1:4�s(�2

red = 2:14).

ing Fig.1b, the radialdensity distributions for CM at

sm all radii (r <� 0:5 cm ) behave quite di�erently be-

tween the central(jzj< 0:5 cm )and itsadjacentregions

(0:5 cm < jzj< 1:5 cm ).M oreover,the shape ofthe dis-

tribution for the events in the centralregion resem bles

thatforHM .

In Fig.2 and Figs.3a,b theradial�dN

dr

�and axial

�dN

dz

�

annihilation distributions are plotted for the HM case.

Fig.3cshowstheaxialdistribution forCM foreventswith

r< 0:5 cm being notably narrowerthan the HM casein

Fig.3b.

In orderto determ ine the characteristicsofthe near-

axis events the possibility that they are due to �H has

been investigated. In ATHENA, �H wasdetected by the

coincidence in space and in tim e (within 5 �s)of�p and

e+ annihilations. Thiswasachieved foreventshaving a

charged pion vertex accom pained by a pair of511 keV

photonswith theanglebetween them denoted � [1,9].

Exam ination ofthedistributionsofcos� indicatesthat

thetrap centreeventsarenotdueto �H,exceptforasm all

fraction in CM where the long tailsin Fig.3c are due to

poorreconstructionsof�H annihilationson thetrap wall.

No �H annihilationsoccurforHM .M oreover,cos� dis-

tributionsforCM in thethreez-regionsin Fig.1b suggest

thatthefraction ofnon-�H annihilationson thewallin the

centralz-region isinsigni�cant.

Thedistributionsin Figs.1{3alsoshow thatin HM the

annihilation distribution isextended and hasa conspic-

uous fraction out to the trap wall. W e interpret these

featuresas evidence forthe production ofslow antipro-

tonichydrogen asfollows:

i) the lim ited axialrange ofthe vertices,when com -

pared to the totallength ofthe nested trap,indi-

catesthat �p do notannihilate in- ighton residual

gaswhich ispresentthroughout;

3

Counts

0-2 -1 0 1 2

50

100

150

2000

20

40

60

80

0

20

40

60

80C

ounts

Counts

r > 1.0 cm

r < 0.5 cm

r < 0.5 cm

z (cm)

(c)

(b)

(a)

Figure 3: Experim entalaxialdistributions: hot m ixing for

eventsnearthe trap wall(a),and nearthe trap axis(b)and

forcold m ixing foreventsnearthetrap axis(c).In (a,b):the

red line is the sim ulation with the param eters ofFig.2 and

with lifetim e � = 1:1�s. In (c)the red line is the sim ulation

with the param etersofFig.4a.

ii) the eventscannotcorrespond to in- ightannihila-

tion on trapped positive ions since the latter are

only presentnearthee+ cloud whilsttheannihila-

tion eventsareradially di�use;

iii) positive ions can capture �p in the centralpart of

the recom bination trap. In the case of �pHe+ the

residualelectron would be rapidly ejected,in the

m ajority ofcasesin lessthan 10 ns[13],leaving a

charged system thatwould annihilatevery nearits

pointofform ation. Since thisisinconsistentwith

ourobservations,�p capture by helium ionscan be

excluded.

Further inform ation on the inferred antiprotonic sys-

tem form ed from capture by a positive ion can be ob-

tained by exploiting thedi�erentcharged pion m ultiplic-

itiesexpected for �p annihilation on a proton orneutron.

Tab.Ishowsthe ratios,R 23,ofthe num berofthe recon-

structed annihilation vertices having two charged pion

tracksto those with three tracksfordi�erentdata sam -

ples.

D ata set Ratio R 23 on wallRatio R 23 atcentre

Cold m ixing 1.35� 0.01 1.22� 0.04

Hotm ixing 1.38� 0.10 1.17� 0.04

antiprotonsonly 1.40� 0.03

M onte Carlo �pp 1:19� 0:01 1:19� 0:01

TableI:Experim entaland M onteCarloresultsforthenum ber

ofcharged pion tracksdue to �p annihilations.

From Tab.I,theR 23 valuesforannihilation on thetrap

wallforallthe sam plesagree,within uncertainties,but

di�er from those for the trap centre by 4 standard de-

viations. This is not due to geom etry,since the M onte

Carlo sim ulationsassum ing the �pp system givethe sam e

resultboth forthe \wall" and \centre" annihilations.It

islikely thatthetrap centreeventsaredueto �pp annihi-

lation,since the �pp M onte Carlo resultagreeswellwith

theexperim ent.This,togetherwith thethreeconstraints

described above,suggestthat�ppisresponsiblefortheob-

served annihilation distributions.Thus,the data can be

furtheranalysed to search forconsistency with the con-

ditions pertaining to the production region at the two

di�erent positron tem peratures,the �pp lifetim e and �-

nally the energeticsofthe form ation reaction.

First,�pp isnotcon�ned by the electrom agnetic �elds

ofthe traps and ifform ed in a m etastable state [3,14,

15,16]itcan decay in ightfarfrom itspointofform a-

tion,perhapseven on the trap wall. The convolution of

the �pp lifetim esand velocitiesgovernstheobserved anni-

hilation distributions. M oreover,Figs. 3b and c clearly

indicate thatthe tem peratureofthe e+ cloud in uences

the spatialorigin ofthe �pp.

W e have attem pted to generate the initialposition of

the �pp using a M onte Carlo sim ulation which takesinto

accountthe radius,rp,and the axialhalf-length,zp,of

the e+ spheroid, both deduced by m eans of the non-

destructive technique described in [11,12]. Param eters

obtained using thistechnique allow the plasm a rotation

frequency to be extracted.

In this sim ulation,the spheroid was characterized by

rp = 1 m m and zp = 16 m m . For the CM case the �pp

wasgenerated in a region with a �xed radialposition at

r = rp = 1 m m and with a G aussian distribution along

the axis centred at the sym m etry plane ofthe plasm a

with � = 2:5 m m .Thisgavethebest�tto thedata.For

theHM case,however,�ppwasgeneratedwith � = 10m m ,

though lim ited to the length ofe+ plasm a.Itisnotable

that,forthe HM case,the sim ulated annihilation distri-

butionswere notstrongly dependentupon the assum ed

startingconditions,takingintoaccountourexperim ental

resolution.

The velocity ofthe �pp wasgenerated from the sum of

a therm alM axwellian distribution,vth,and the tangen-

tialvelocity,vtang,induced by the~E � ~B plasm arotation

as vtang = ~E � ~B=jBj2. Following this prescription,the

m ean radialkineticenergy ofthe �pp isabout40 m eV for

CM (15K ),and dom inated bythee�ectoftheplasm aro-

tation,and about700m eV in theHM case(8000K ),and

dom inated by the plasm a tem perature. An exponential

decay law for the �pp lifetim e distribution was assum ed

such thatitsm ean lifetim ewasdeterm ined by �tting the

sim ulationsto the observed data.

The sim ulated radialand axialannihilation distribu-

tionsareplotted with theHM datain Fig.2and Fig.3a,b.

Theagreem entisgood and thebest�twasobtained with

4

co

un

ts

r (cm) r (cm)

250

200

150

100

50

0

-50

-100

200

150

100

50

0

-50

-100

0 0.5 1 1.5 2 2.5 0 0.5 1 1.5 2 2.5

(a) (b)

Figure 4: Experim entalradialdistribution for cold m ixing

2003 (a) and cold m ixing 2002 (b),obtained by subtracting

the �H contribution (see text). The red lines are the sim ula-

tion results. In (a): T= 15 K ,vth = 250 m s� 1,generation

from a spheroid with zp = 16 m m and rp = 1 m m rotating

with a frequency of300 kHz,i.e. vtang = 2000 m s� 1;m ean

lifetim e is �= 1.1 �s. In (b): sam e param eters as in (a) ex-

cept rp = 2:5 m m and the rotation frequency is 80 kHz,i.e.

vtang = 1300 m s� 1. In (b) the green line corresponding to

theparam etersofthered linein (a)isshown forcom parison.

am ean lifetim eof(1.1� 0.1)�s.Thesensitivity ofthe�t

to thislifetim eisillustrated in Fig.2 by thecleardiscord

between theexperim entaldataand thesim ulationswhen

lifetim esof0.8�sand 1.4�swereused in thelatter.The

�ts im ply thatabout25% ofthe antiprotonic hydrogen

atom sannihilateon the trap wallin HM .

To isolate the radialdistribution ofthe non-�H anni-

hilations in the centralz-region for the CM sam ple the

norm alized distribution,asevaluated from Fig.1b forthe

two lateralz-regions,was subtracted from the central

one. Since,as noted above,non-�H annihilations on the

wallin the centralz-region areinsigni�cantforCM ,the

radialdistributionswerenorm alised forr> 1:25 cm .

The results are plotted in Fig.4a. The M onte Carlo

sim ulated events,assum ing the sam e lifetim e extracted

from �tsto the HM data,also show good agreem entfor

CM .In this case,the sim ulation indicates that < 0:5%

ofthe �pp reachesthe trap surface.

A furthertestofourm odelhasbeen obtained byexam -

iningasam pleofCM dataacquired by ATHENA in 2002

with a e+ plasm a radiusof2.5 m m (and hence a di�er-

ent rotation frequency). Fig.4b showsthe radialvertex

distribution of this sam ple together with M onte Carlo

sim ulationsfortwo valuesoftheradiusofthe �pp source.

Thesim ulated eventsforthe2.5 m m sourcearein m uch

betteragreem entwith the experim entaldata than those

for1.0 m m . This isillustrative ofthe sensitivity ofthe

m odelto the spatialorigin ofthe �pp forCM sam ples.

A possibleexplanation oftheexperim entaldi�erences

between the �pp distributions in CM and HM liesin the

nature ofthe therm alequilibrium state ofthe com bina-

tion of a e+ plasm a with an adm ixture of ions. The

physicsoftheradialseparation ofthedi�erentspeciesin

two-com ponentplasm ashasbeen given elsewhere[17,18]

and has been experim entally observed for a m ixture of

positronsand 9Be+ in [19].Forourexperim entalcondi-

tions,assum ing therm alequilibrium ,the centrifugalpo-

tentialbarrierisoftheorderof10 m eV.Thus,forCM at

15K ,thetherm alenergy oftheionsm eansthatthey will

bepartially separated from thee+ and con�ned nearthe

equatorialregion ofthe plasm a. However,ata e+ tem -

perature of8000 K the barrierisnegligible and the ions

willbe presentthroughoutthe plasm a.

W enoteherethattheexperim entaldatacannotbere-

produced by sim ulation ifitisassum ed thatthe �pp gains

a recoilenergy ofthe orderof1 eV orhigher. Thus,it

is contended that �pp is being produced in a recoil-free

collision ofan antiproton with a positive ion such that

itsvelocity isdeterm ined predom inantly by the therm al

and plasm a environm ent.Thisinsight,togetherwith the

constraintsi)-iii)noted above,suggestthattheonly pos-

siblecollision partnerforthe �p isthem olecularion,H+

2 .

Thus,the inferred �pp production m echanism is,

�p+ H +

2 ! �pp(n;l)+ H: (1)

Hence we believe thatATHENA hasobserved around

100 antiprotonic hydrogen annihilationsevery 60 s �p in-

jection cyclefortheCM and HM conditions.In an exper-

im entin which the num berofionspresentin ournested

wellwascounted by m easuring the charge collected fol-

lowingem ptying ofthetrap ithasbeen deduced thatthe

trap contained H+

2 ions. These probably arose as a re-

sultofthe positron loading procedure [20],in which the

positronscould collidewith H 2 residualgasasthey were

slowly squeezed into the m ixing region. Ions m ay also

beproduced and trapped during �p loading (seealso [21])

and weestim atethataround 104 � 105 ionswerepresent

undertypicalam bientconditions.Itisstraightforwardto

show thatthision density,togetherwith theobserved �pp

production rate and the �p speedsused in the sim ulation

areconsistentwith calculated crosssectionsforreaction

(1)[16].

Agreem ent with [16]does not,unfortunately,extend

to them ostlikely principalquantum num bersofthean-

tiprotonichydrogen atom sproduced in thereaction.The

calculation [16]�ndsproduction peaked around n = 34,

in the presence of substantial �pp recoil. The latter is

contrary to observation. Sim ple kinem atics relating to

near zero-energy �p� H+

2 collisions suggest that n = 68

should dom inate,with theliberated hydrogen atom in its

ground state.In thiscasethelifetim eof1.1 �sextracted

from the sim ulationsim pliesthatproduction in low an-

gular m om entum states (l< 10) is favoured,since the

radiativelifetim e to an l= 0 state(which isfollowed by

prom ptannihilation)weighted by thestatistical(2l+ 1)

distribution isaround 15 �sforn = 68.The dom inance

oflow lisintuitive forsuch a slow collision in which the

m olecularion willbe severely polarised by the incom ing

�p,resulting in an alm ostcollinearcollision system .

5

In conclusion we have presented evidence forthe pro-

duction ofantiprotonic hydrogen,in vacuum ,with sub-

eV kineticenergiesand in a m etastablestate.G iven the

capability ofaccum ulating 108 H+

2 ions in tens ofsec-

ondsand storing � 5� 106 antiprotonsin som em inutes

[22,23],our resultopens up the possibility ofperform -

ing detailed spectroscopicm easurem entson antiprotonic

hydrogen asa probeoffundam entalconstantsand sym -

m etries.

This work was supported by CNPq, FAPERJ,

CCM N/UFRJ (Brazil),INFN (Italy),M EXT (Japan),

SNF (Switzerland),SNF (Denm ark)and EPSRC (UK ).

[1]M .Am orettietal.,Nature 419,456 (2002).

[2]G .G abrielse etal.,Phys.Rev.Lett.89,213401 (2002).

[3]C.J.Batty,Rept.Prog.Phys.52,1165 (1989).

[4]L.M ontanet,Nucl.Phys.A 692,383c (2001).

[5]M .Am orettiet al., Nucl.Instrum .M eth.A 518, 679

(2004).

[6]T.Yam azakietal.,Phys.Rep.366,183 (2002).

[7]M .Horietal.,Phys.Rev.Lett.96,243401 (2006).

[8]C.Regenfus,Nucl.Instrum .M eth.A 501,65 (2003).

[9]M .Am orettietal.,Phys.Lett.B 578,23 (2004).

[10]M .C.Fujiwaraetal.,Phys.Rev.Lett.92,065005(2004).

[11]M .Am orettietal.,Phys.Rev.Lett.91,055001 (2003).

[12]M .Am orettietal.,Phys.Plasm as10,3056 (2003).

[13]V.I.K orobov Phys.Rev.A 67,062501 (2003).

[14]G .Reifenrother & E.K lem pt Nucl.Phys.A 503, 885

(1989).

[15]J.S.Cohen,Rept.Prog.Phys.67,1769 (2004).

[16]K .Sakim oto,J.Phys.B At.M ol.O pt.Phys.37,2255

(2004).

[17]D .H.E.D ubin & T.M .O ’Neil,Rev.M od.Phys.71,87

(1999).

[18]T.M .O ’Neil,Phys.Fluids24,1447 (1981).

[19]B. M . Jelenkovic, A. S. Newbury, J. J. Bollinger,

W .M .Itano & T.B.M itchell,Phys.Rev.A 67 063406

(2003).

[20]L.V.J�rgensen etal.Phys.Rev.Lett.95 025002 (2005).

[21]G .G abrielse etal.,Phys.Lett.B 455,311 (1999).

[22]N.K uroda etal.,Phys.Rev.Lett.94,023401 (2005).

[23]N.O shim a etal.,Phys.Rev.Lett.93,195001 (2004).