www.elsevier.com/locate/physletb
Instructions to authors
Aims and Scope Physics Letters B ensures the rapid publication of letter-type communications in the fields of Nuclear Physics, Particle Physics and Astrophysics. Articles should influence the physics community significantly. Submission Electronic submission is strongly encouraged. The electronic file, accompanied by a covering message, should be e-mailed to one of the Editors indicated below. Easy submission via the LANL-preprint server is certainly possible; please visit http://www.elsevier.com/locate/plbsubmission. If electronic submission is not feasible, submission in print is possible, but it will delay publication. In the latter case manuscripts (one original + two copies), accompanied by a covering letter, should be sent to one of the following Editors: L. Alvarez-Gaumé, Theory Division, CERN, CH-1211 Geneva 23, Switzerland, E-mail address: Luis.Alvarez-Gaume@CERN. CH Theoretical High Energy Physics (General Theory) J.-P. Blaizot, ECT*, Strada delle Tabarelle, 266, I-38050 Villazzano (Trento), Italy, E-mail address: [email protected]. FR Theoretical Nuclear Physics M. Cvetic,ˇ David Rittenhouse Laboratory, Department of Physics, University of Pennsylvania, 209 S, 33rd Street, Philadelphia, PA 19104-6396, USA, E-mail address: [email protected] Theoretical High Energy Physics M. Doser, EP Division, CERN, CH-1211 Geneva 23, Switzerland, E-mail address: [email protected] Experimental High Energy Physics D.F. Geesaman, Argonne National Laboratory, 9700 South Cass Avenue, Argonne, IL 60439, USA, E-mail address: [email protected] Experimental Nuclear Physics H. Georgi, Department of Physics, Harvard University, Cambridge, MA 02138, USA, E-mail address: Georgi@PHYSICS. HARVARD.EDU Theoretical High Energy Physics G.F. Giudice, CERN, CH-1211 Geneva 23, Switzerland, E-mail address: [email protected] Theoretical High Energy Physics N. Glover, Institute for Particle Physics Phenomenology, Department of Physics, Science Laboratories, University of Durham, South Road, Durham DH1 3LE, UK, E-mail address: [email protected] Theoretical High Energy Physics W. Haxton, Institute for Nuclear Theory, Box 351550, University of Washington, Seattle, WA 98195-1550, USA, E-mail address: [email protected] Theoretical Nuclear Physics and Nuclear Astrophysics V. Metag, II. Physikalisches Institut, Universität Giessen, Heinrich-Buff-Ring 16, D-35392 Giessen, Germany, E-mail address: [email protected]. UNI-GIESSEN.DE Experimental Nuclear Physics L. Rolandi, EP Division, CERN, CH-1211 Geneva 23, Switzerland, E-mail address: [email protected] Experimental High Energy Physics W.-D. Schlatter, CERN, CH-1211 Geneva 23, Switzerland, E-mail address: [email protected] Experimental High Energy Physics
0370-2693/2005 Published by Elsevier B.V. doi:10.1016/S0370-2693(05)00827-0 vi Instructions to authors
H. Weerts, 3247 Biomedical and Physical Sciences Building, Department of Physics and Astronomy, Michigan State University, East Lansing, MI 48824-1111, USA, E-mail address: [email protected] Experimental High Energy Physics T. Yanagida, Department of Physics, Faculty of Science, University of Tokyo, Tokyo 113-0033, Japan, E-mail address: [email protected] Theoretical High Energy Physics The authors should indicate in which of the following four sections they would like to see their article published: Astrophysics & Cosmology, covered by L. Alvarez-Gaumé, M. Cveticˇ and W. Haxton. Experiments, covered by M. Doser, D.F. Geesaman, V. Metag, L. Rolandi, W.-D. Schlatter and H. Weerts. Phenomenology, covered by J.-P. Blaizot, M. Cvetic,ˇ H. Georgi, G.F. Giudice, N. Glover, W. Haxton and T. Yanagida. Theory, covered by L. Alvarez-Gaumé, J.-P. Blaizot, M. Cvetic,ˇ H. Georgi, G.F. Giudice, N. Glover, W. Haxton and T. Yanagida. For submissions in High Energy Physics authors are encouraged to provide the number of their Los Alamos preprint to the Physics Letters B Editor at the moment of submission. Original material. By submitting a paper for publication in Physics Letters B the authors imply that the material has not been published previously nor has been submitted for publication elsewhere and that the authors have obtained the necessary authority for publication. Refereeing. Submitted papers will be refereed and, if necessary, authors may be invited to revise their manuscript. If a submitted paper relies heavily on unpublished material, it would be helpful to have a copy of that material for the use of the referee. Publication speed The Editors and Publisher cooperate closely to ensure minimal publication delays. All proofreading will be done by the Publisher and proofs are not sent to the author(s). In order to keep delays to a minimum it is of utter importance for the author(s) closely to observe the guidelines given in the “Preparation of Manuscripts” below before submission of the manuscript and to supply an E-mail address and/or telefax number of the corresponding author. Preparation of manuscripts The following requirements as regards presentation of the manuscript should be met: (1) Manuscript: All manuscripts should be written in good English. The original typescript should be typed on one side of the paper, with double spacing and a wide margin. Instead of the original typescript, a copy of high quality (not a carbon copy) is also acceptable. Please adhere to the following order of presentation: Article title, Author(s), Affiliation(s), Abstract, Classification codes and keywords, Main text, Acknowledgements, Appendices, References, Figure captions, Tables. The name, complete postal address, telephone and fax numbers and the E-mail address of at least the corresponding author should be given on the first page of the manuscript. The title page should contain title, author(s), address(es) and abstract. The main text should start on a new page. All pages should be numbered. (2) Length: The total length of the paper should preferably not exceed six journal pages, equivalent to ten typewritten pages with double spacing, including the list of authors, abstract, references, figure captions and three figures. In the case that more figures are required, the text should be shortened accordingly. (3) Title: The title should be brief and such that it conveys to the informed reader the particular nature of the contents of the paper. (4) Address: The name(s) of the author(s) and the name and address of the institute where the research work was done should be indicated on the manuscript. The name of the author to whom correspondence is to be addressed should be underlined and an E-mail address and/or a telefax number supplied. (5) Abstract: An abstract of less than 60 words is required. It should contain the keywords of the paper as well as the essence of the results achieved. (6) Classification codes and keywords: Supply one to four classification codes (PACS and/or MSC) and up to six keywords of your own choice that describe the content of your article in more detail. (7) Formulae: Displayed formulae should be numbered and typed or clearly and unambiguously written by hand. Symbols requiring bold-face type, like vectors, etc., should be identified properly in the margin of the manuscript. Long equa- tions should be avoided as much as possible by introduction of suitable abbreviations of component expressions. Special attention should be paid to symbols that can easily be misread, such as i (lower case), I (cap.), 1 (el), 1 (one), (prime), ◦ o (lower case), O (cap.), 0 (zero), (degree), u,v (vee), ν (Greek nu), V (cap.), x, ×,X,z,Z,p,P,ρ (Greek rho), etc. (8) Footnotes: The footnotes may be typed at the end of the page on which they are alluded to, or at the end of the paper, or on a separate sheet. Please do not mix footnotes and references. Instructions to authors vii
(9) References: In the text, reference to other parts of the paper should be made by section (or equation) number, not by page number. References to other papers should be consecutively numbered in the text using square brackets and should be listed by number on a separate sheet at the end of the paper. Please do not combine multiple references to different papers into one numbered reference. The references should be as complete as possible and be presented as follows: For a book: B. de Wit and J. Smith, Field theory in particle physics, Vol. 1 (North-Holland, Amsterdam, 1986). For a paper in a journal: UAl Collab., G. Arnison et al., Phys. Lett. B 177 (1986) 244. For a paper in a contributed volume: R. Jackiw, in: Progress in quantum field theory, eds. H. Ezawa and S. Kamefuchi (North-Holland, Amsterdam, 1986) p. 83. For an unpublished paper: J.F. Gunion and H.E. Haber, UCD Report 86-12 (1986), unpublished. For a preprint: A. Lahanas and D.V. Nanopoulos, CERN preprint CERN-TH 4400/86 (1986). For a conference report: M.B. Green, Superstrings and the unification of forces and particles, in: Proc. fourth Marcel Gross- mann Meeting on General relativity (Rome, June 1985), Vol. 1, ed. R. Ruffini (North-Holland, Amsterdam, 1986) p. 203. (10) Figures: Each figure should also be submitted in triplicate: one master figure and two copies, the figure must be referred to in the text, be numbered and have a caption. The captions should be collected on a separate sheet. The appropriate place of each figure should be indicated in the margin. Axes of figures must be labelled properly. The (line) drawings for the figures must be submitted on separate sheets, drawn in black India ink and carefully lettered (with the use of stencils). The lettering as well as the essential details should have proportionate dimensions so as not to become illegible or unclear after the usual reduction by the printers (ideal lettering size after reduction of the drawing to one-column width is 1.8 mm). The drawings should preferably be of the same size as the typescript and designed for a reduction factor in print of two to three. The photographs should be originals, with somewhat more contrast than is required in the printed version. They should be unmounted unless part of a composite figure. Any scale markers should be inserted on the photograph, not drawn below it. The figures should be identified by the name of the first author, the journal name and the figure number. Instead of original drawings, sharp and contrasty glossy prints of about typescript size or high quality laserprints are also acceptable. If requested, original drawings will be returned to the author(s) upon publication of the paper. For detailed instructions on the preparation of electronic artwork, consult the Author Gateway from Elsevier at http:// authors.elsevier.com. (11) Colour illustrations: Illustrations in colour will be accepted in cases when the use of colours is judged by the Editor to be essential for the presentation. The Publisher and the author will each bear part of the extra costs involved. The costs charged to the authors of articles containing colour figures will be € 635 (approximately US$ 760) for the first page containing colour and € 318 for each additional page with colour, independent of the number of colour figures on each page. These prices are exclusive of Value Added Tax (VAT). Authors will be billed in Euros; the dollar price is for guidance only. The author receives 200 reprints of an article with colour illustrations free of charge. More reprints can be ordered at the usual rates for (black and white) reprints, there will be no additional charge for reprints containing colour illustrations. Colour illustrations should be submitted in the form of good quality colour photographs, transparencies, colour printer output, or 35 mm slides. Polaroid colour prints should be avoided. When supplying colour photographs and transparencies they should be close to the final size expected for publication. It should be noted that, in some cases, printing requirements will prevent figures from being located in the most preferred position in the text. (12) Tables: Tables should be typed on separate sheets and each table should have a number and a title. The appropriate places for the insertion of the tables should be indicated in the margin.
After acceptance Notification. You will be notified by the Editor of the journal of the acceptance of your article and invited to supply an electronic version of the accepted text, if this is not already available. Copyright transfer. You will be asked to transfer the copyright of the article to the Publisher. This transfer will ensure the widest possible dissemination of information. No proofs. In order to speed up publication, all proofreading will be done by the Publisher and proofs are not sent to the author(s). Enquiries. Visit the Author Gateway from Elsevier (http://authors.elsevier.com) for the facility to track accepted articles and set up e-mail alerts to inform you of when an article’s status has changed. The Author Gateway also provides detailed artwork guidelines, copyright information, frequently asked questions and more. Contact details for questions arising after acceptance of an article, especially those relating to proofs, are provided when an article is accepted for publication. viii Instructions to authors
Instructions for LATEX manuscripts The Publisher welcomes the receipt of an electronic version of your accepted manuscript (preferably encoded in LATEX). If you have not already supplied the final, accepted version of your article to the journal Editor, you are requested herewith to send a file with the text of the accepted manuscript directly to the Publisher by E-mail to the address given below. If the electronic file is suitable for processing by the Publisher, the article will be published without rekeying the full text. The article should be encoded in LATEX, preferably using the Elsevier document class ‘elsart’, or alternatively the standard document class ‘article’ or the document style ‘revtex’. The Elsevier LATEX package (including detailed instructions for LATEX preparation) can be obtained from Elsevier’s web site: www.elsevier.com/locate/latex, or from the Comprehensive TEX Archive Network (CTAN). The Elsevier package consists of the files: elsart.cls (use this file if you are using LaTeX2e, the current version of LATEX), elsart.sty and elsart12.sty (use these two files if you are using LaTeX2.09, the previous version of LATEX), instraut.dvi and/or instraut.ps and/or instraut.pdf (instruction booklet), model-harv.tex or model-num.tex (model files with instructions), template-harv.tex or template-num.tex (template files). To obtain the package from CTAN, use direct access via FTP at ftp.dante.de (Germany), ftp.tex.ac.uk (UK), or ctan.tug.org (Massachussets, USA) and go to the directory /tex-archive/macros/latex/contrib/supported/elsevier, or search for Elsevier with one of the CTAN search engines (http://ctan.tug.org/CTANfind.html, http://www.tex.ac.uk/CTANfind.html or http://www.dante.de/cgi-bin/ctan-index). CTAN is a mirrored network of ftp.tex.ac.uk, ftp.dante.de and ctan.tug.org, which are widely mirrored (see ftp://ctan.tug.org/ tex-archive/README.mirrors) and hold up-to-date copies of all the public-domain versions of TEX, LATEX, Metafont and ancillary programs. Questions concerning the LATEX author-prepared article and requests for the booklet with instructions to authors should be directed to the address given below. No changes from the version accepted by the Editor of the journal are permissible, without the prior and explicit approval by the Editor. Such changes should be clearly indicated on an accompanying printout of the file. The Publisher reserves the right to decide whether to process a manuscript from authors’ files or not. Articles coded in a simple manner with no user-defined macros are most likely to be handled this way. If sent via electronic mail, files should be accompanied by a clear identification of the article (name of journal, Editor’s reference number) in the “subject field” of your electronic-mail message. Authors should include an ASCII table (available from the Publisher) in their files, to enable any transmission errors to be detected. Addresses Ð Mail: Drs. E.J. van der Wolk, Physics Letters B, Editorial-Production Journals Department, Elsevier B.V., P.O. Box 2759, 1000 CT Amsterdam, The Netherlands. E-mail: [email protected]. Author benefits No page charges. Publishing in Physics Letters B is free. Free offprints. The corresponding author will receive 25 offprints free of charge. An offprint order form will be supplied by the Publisher for ordering any additional paid offprints. Discount. Contributors to Elsevier journals are entitled to a 30% discount on all Elsevier books. Contents Alert. Physics Letters B is included in Elsevier’s pre-publication service Contents Alert (for information, please contact: [email protected]). ContentsDirect. Physics Letters B is included in Elseviers free E-mail service ContentsDirect. Please register at http://www. elsevier.com/locate/ContentsDirect. Publication scheme Physics Letters B will be published weekly. All correspondence with the Editors or Publisher should contain full reference to the paper concerned, i.e., the names of all the contributors, the full title, as well as the reference number. Contributors are reminded that, once their contribution has been accepted for publication, all further correspondence should be sent directly to the Publisher and not to the Editors (Drs. E.J. van der Wolk, Physics Letters B, Editorial-Production Journals Department, Elsevier B.V., P.O. Box 2759, 1000 CT Amsterdam, The Netherlands; telephone number +31 20 485 2634; telefax number +31 20 485 2431; electronic mail: [email protected] (the receipt of an E-mail message will be acknowledged; in the absence of this acknowledgement, the mailer should assume that his message was never received by the Publisher)). A prompt reply may be expected whenever appropriate. If no reaction is received within three weeks, authors are kindly requested to seek contact again. PHYSICS LETTERS B
EDITORS
L. ALVAREZ-GAUMÉ J.-P. BLAIZOT M. CVETICˇ GENEVA VILLAZZANO (TRENTO) PHILADELPHIA, PA
M. DOSER D.F. GEESAMAN H. GEORGI GENEVA ARGONNE, IL CAMBRIDGE, MA
G.F. GIUDICE N. GLOVER W. HAXTON GENEVA DURHAM SEATTLE, WA
V. METAG L. ROLANDI W.-D. SCHLATTER GIESSEN GENEVA GENEVA
H. WEERTS T. YANAGIDA EAST LANSING, MI TOKYO
VOLUME 619, 2005
Amsterdam – Boston – Jena – London – New York – Oxford Paris – Philadelphia – San Diego – St. Louis Physics Letters B 619 (2005) 1–4 www.elsevier.com/locate/physletb
Future universe with w < −1 without big smash
S.K. Srivastava
Department of Mathematics, North Eastern Hill University, Shillong 793022, India Received 17 February 2005; received in revised form 12 May 2005; accepted 20 May 2005 Available online 1 June 2005 Editor: T. Yanagida
Abstract It is demonstrated that if cosmic dark energy behaves like a fluid with equation of state p = wρ (p and ρ being pressure and energy density respectively) as well as generalized Chaplygin gas simultaneously, big rip or big smash problem does not arise even for equation of state parameter w < −1. Unlike other phantom models, here, the scale factor for the future universe is found regular for all time. 2005 Elsevier B.V. All rights reserved.
PACS: 98.80.Cq
Keywords: Dark energy; Phantom model; Big rip and accelerated universe
Experimental probes, during last few years sug- state (EOS) p = wρ with p as isotropic pressure, ρ as gest that the present universe is spatially flat as well energy density and −1 w < −1/3. as it is dominated by yet unknown form of dark en- In the recent past, it was pointed out that the current ergy [1,2]. Moreover, studies of Ia Supernova [3,4] data also allowed w < −1 [7]. Rather, in Refs. [8–10], and WMAP [5,6] show accelerated expansion of the it is discussed that these data favor w < −1 being EOS present universe such that a>¨ 0 with a(t) being the parameter for phantom dark energy. Analysis of recent scale factor of the Friedmann–Robertson–Walker line- Ia Supernova data also support w < −1 strongly [11– element 13]. Soon after, Caldwell [8] proposed the phantom dS2 = dt2 − a2(t) dx2 + dy2 + dz2 . (1) dark energy model exhibiting cosmic doomsday of the future universe, cosmologists started making efforts Theoretically accelerated expansion of the universe to avoid this problem using w < −1 [14,15].Inthe is obtained when the cosmological model is supposed braneworld scenario, Sahni and Shtanov has obtained to be dominated by a fluid obeying the equation of well-behaved expansion of the future universe without big rip problem with w < −1. They have shown that E-mail address: [email protected] (S.K. Srivastava). acceleration is a transient phenomenon in the current
0370-2693/$ – see front matter 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.physletb.2005.05.056 2 S.K. Srivastava / Physics Letters B 619 (2005) 1–4 universe and the future universe will re-enter matter- and integrating, it is obtained that dominated decelerated phase [16]. + ρ(1 α)/α(t) It is found that GR (general relativity)-based phan- tom model encounters “sudden future singularity” = + (1+α)/α − 3(1+α)/α A ρ0 A a0/a(t) (6) leading to divergent scale factor a(t), energy density with ρ0 = ρ(t0) and a0 = a(t0), where t0 is the present and pressure at finite time t = ts . Thus the classical ap- proach to phantom model yields big-smash problem. time. For models with “sudden future singularity” Elizalde Eqs. (3) and (4) yield w as et al. [17] argued that, near t = ts , curvature invariants A w(t) =− . (7a) become very strong and energy density is very high. ρ(1+α)/α(t) So, quantum effects should be dominant for |t − t| < s = one unit of time, like early universe. This idea is pur- So, evaluation of Eq. (7a) at t t0 leads to sued in Refs. [18–20] and it is shown that an escape (1+α)/α A =−w0ρ , (7b) from the big-smash is possible on making quantum corrections to energy density ρ and pressure p in with w0 = w(t0).FromEqs.(6) and (7a), (7b),itis Friedmann equations. obtained that In the framework of Robertson–Walker cosmol- + + = − + + 3(1 α)/α α/(1 α) ogy, Chaplygin gas (CG) is also considered as a good ρ ρ0 w0 (1 w0) a0/a(t) source of dark energy for having negative pressure, (8) given as with w0 < −1. In the homogeneous model of the universe, a scalar A p =− (2) field φ(t) with potential V(φ)has energy density ρ 1 ˙2 with A>0. Moreover, it is the only gas having super- ρφ = φ + V(φ) (9a) symmetry generalization [21,22]. Bertolami et al. [12] 2 have found that generalized Chaplygin gas (GCG) is and pressure better fit for latest Supernova data. In the case of GCG, 1 ˙2 Eq. (2) looks like pφ = φ − V(φ). (9b) 2 A Using Eqs. (3), (4), (7a), (7b) and (8), it is obtained p =− , (3) ρ1/α that ∞ = (1+α)/α (1+α)/α where 1 α< .α 1 corresponds to Eq. (2). ρ + ρ w0 φ˙2 = 0 . (10) In this Letter, a different prescription for GR-based ρ future universe, dominated by the dark energy with w < −1, is proposed which is not leading to the Connecting Eqs. (8) and (10), it is obtained that catastrophic situations mentioned above. The scale (1+α)/α 3(1+α)/α (1 + w0)ρ (a0/a) factor, obtained here,does not possess future singular- φ˙2 = 0 . [− + + 3(1+α)/α]α/(1+α) ity. In the present model, it is assumed that the dark w0 (1 w0)(a0/a) (11) energy behaves like GCG, obeying Eq. (3) as well as ˙ fluid with equation of state This equation shows that φ2 > 0 (giving positive kinetic energy) for w0 > −1, which is the case p = wρ with w < −1 (4) of quintessence and φ˙2 < 0 (giving negative ki- netic energy) for w < −1, being the case of super- simultaneously. 0 quintessence (phantom). As a reference, it is rele- Connecting Eq. (3) with the hydrodynamic equa- vant to mention that, long back, Hoyle and Narlikar tion used C-field (a scalar called creation field) with neg- a˙ ρ˙ =−3 (ρ + p) (5) ative kinetic energy for steady-state theory of the uni- a verse [23]. S.K. Srivastava / Physics Letters B 619 (2005) 1–4 3
Thus, it is shown that dual behaviour of dark energy The horizon distance for this case (a(t) given by fluid, obeying Eqs. (3) and (4) is possible for scalars, Eq. (16)) is obtained as frequently used for cosmological dynamics. So, this + 3(1 + α)a(t) 2(1 + α)|w | α/3(1 α) assumption is not unrealistic. d (t) 0 H αa α + (α + 2)|w | Now the Friedmann equation, with dominance of 0 0 α/2(1+α) dark energy having double fluid behaviour, is × exp 6H0|w0| Ω0 αt/3(1 + α) a˙ 2 (15a) = H 2Ω |w | a 0 0 0 showing that 3(1+α)/α α/(1+α) + 1 −|w0| a0/a(t) , dH (t) > a(t). (15b) (12a) So, horizon grows more rapidly than the scale factor impling colder and darker universe. It is like flat or where |w0| > 1. H0 is the present value of Hubble’s = = 2 open universe without dominance of dark energy. constant and Ω0 ρ0/ρcr,0 with ρcr,0 3H0 /8πG (G being the Newtonian gravitational constant). In this case, Hubble’s distance is −| | 3(1+α) 1 w0 3(1 + α) Neglecting higher powers of (a0/a(t)) α , −1 |w0| H = √ | |α/2(1+α) Eq. (12a) is written as αH 0 Ω0 w0 α(1 −|w0|) a˙ + × 1 − H Ω |w |α/2(1 α) + + | | 0 0 0 α (α 2) w0 a −| | α/2(1+α) α(1 w0 ) 3(1+α)/α × exp −H0|w0| Ω0(t − t0) (16) × 1 + a0/a(t) . (12b) 2(1 + α)|w0| −1 → Eq. (12b) is integrated to showing its growth with time such that H 3(1+√α) −α/2(1+α) −1 |w0| = 0ast →∞. Here, H∞ is a αH0 Ω0 a(t) = 0 found large and finite. It means that, in the present α/3(1+α) [2(1 + α)|w0|] case, galaxies will not disappear when t →∞.Itis √ | |α/2(1+α) − × + + | | 6H0 w0 Ω0(t t0) unlike phantom models with future singularity ex- α 2(1 α) w0 e n panding as |t − ts| for n<0, where galaxies are α/3(1+α) − α 1 −|w0| , (13) expected to vanish near future singularity time ts [8] as H −1 → 0fort → t . In Barrow’s model [24] yielding accelerated expansion of the universe with s q n →∞ →∞ − B + Ct + D(t − t) a(t) as t , supporting observational evi- H 1 = s , q−1 n−1 (17) dences of Ia Supernova [3,4] and WMAP [5,6].Itis qCt − Dn(ts − t) interesting to see that expansion, obtained here, is free where B,C,D are positive constants and q>0. from “finite time future singularity” unlike other GR- −1 Eq. (17) shows that, for n<1,H → 0ast → ts based phantom models. It is due to GCG behaviour of −1 and at t = ts,H is finite for n>1. In the model, phantom dark energy. taken by Nojiri and Odintsov [18] Moreover, Eqs. (8) and (13) that energy density −1 ˜ n −1 grows with time for w0 < −1 and decreases for H = H(t)+ A |ts − t| , (18) − → | |α/3(1+α) → w0 > 1. Also ρ ρ0 w0 (finite) and p ˜ −p /|w |α/3(1+α) as t →∞.Eqs.(7a), (7b) and (8) where H(t)is a regular function of t and A > 0. This 0 0 −1 → → imply time-dependence of EOS parameter equation shows that, for n<0, H 0ast ts and it is finite at t = t for n>0. s 3(1+α)/α −1 Thus, it is found that if phantom fluid behaves like w =−|w0| |w0|− |w0|−1 a0/a(t) , GCG and fluid with p = wρ, it is possible to get (14) accelerated growth of scale factor of the future uni- with a(t), given by Eq. (13). This equation shows that verse for time t0
Viscous dark energy and phantom evolution
Mauricio Cataldo a, Norman Cruz b, Samuel Lepe c
a Departamento de Física, Facultad de Ciencias, Universidad del Bío–Bío, Avenida Collao 1202, Casilla 5-C, Concepción, Chile b Departamento de Física, Facultad de Ciencia, Universidad de Santiago, Casilla 307, Santiago, Chile c Instituto de Física, Facultad de Ciencias Básicas y Matemáticas, Pontificia Universidad Católica de Valparaíso, Avenida Brasil 2950, Valparaíso, Chile Received 29 April 2005; received in revised form 10 May 2005; accepted 11 May 2005 Available online 23 May 2005 Editor: M. Cveticˇ
Abstract In order to study if the bulk viscosity may induce a big rip singularity on the flat FRW cosmologies, we investigate dissipative processes in the universe within the framework of the standard Eckart theory of relativistic irreversible thermodynamics, and in the full causal Israel–Stewart–Hiscock theory. We have found cosmological solutions which exhibit, under certain constraints, a big rip singularity. We show that the negative pressure generated by the bulk viscosity cannot avoid that the dark energy of the universe to be phantom energy. 2005 Elsevier B.V. All rights reserved.
PACS: 98.80.Jk; 04.20.Jb
1. Introduction The dark energy is considered a fluid characterized by a negative pressure and usually represented by the The existence of an exotic cosmic fluid with nega- equation of state w = p/ρ, where w lies very close tive pressure, which constitutes about the 70 percent of to −1, most probably being below −1. Dark energy the total energy of the universe, has been perhaps the with w<−1, the phantom component of the uni- most surprising discovery made in cosmology. This verse, leads to uncommon cosmological scenarios as dark energy is supported by the astrophysical data ob- it was pointed out in [1]. First of all, there is a vio- tained from Wilkinson Microwave Anisotropy Probe lation of the dominant energy condition (DEC), since (WMAP) (Map) and high redshift surveys of super- ρ + p<0. The energy density grows up to infinity novae. in a finite time, which leads to a big rip, character- ized by a scale factor blowing up in this finite time. E-mail addresses: [email protected] (M. Cataldo), These sudden future singularities are, nevertheless, not [email protected] (N. Cruz), [email protected] (S. Lepe). necessarily produced by a fluids violating DEC. Bar-
0370-2693/$ – see front matter 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.physletb.2005.05.029 6 M. Cataldo et al. / Physics Letters B 619 (2005) 5–10 row [2] has shown, with explicit examples, that exist The organization of the Letter is as follows. In Sec- solutions which develop a big rip singularity at a finite tion 2 we present the field equations for a flat FRW time even if the matter fields satisfy the strong-energy universe filled with a bulk viscous fluid within the conditions ρ>0 and ρ + 3p>0. A generalization framework of the Eckart theory. We indicate that un- of Barrow’s model has been realized in [3], giving its der a constraint for the parameters of the fluid, one of Lagrangian description in terms of scalar tensor the- the Barrow’s solutions presents a future singularity in ory. It was also proved by Chimento et al. [4] that a finite time. In Section 3 we obtain big rip solutions exists a duality between phantom and flat Friedmann– using the approach of the full Israel–Stewart–Hiscock Robertson–Walker (FRW) cosmologies with nonex- causal thermodynamics. In Section 4 we discuss our otic fluids. This duality is a form-invariance transfor- results in relation to the nature of the dark energy of mation which can be used for constructing phantom the universe. cosmologies from standard scalar field universes. Cos- mological solutions for phantom matter which violates the weak energy condition were found in [5]. 2. Eckart theory The role of the dissipative processes in the evolu- tion of the early universe also has been extensively The FRW metric for an homogeneous and isotropic studied. In the case of isotropic and homogeneous cos- flat universe is given by mologies, any dissipation process in a FRW cosmol- 2 =− 2 + 2 2 + 2 2 + 2 2 ogy is scalar, and therefore may be modelled as a bulk ds dt a(t) dr r dθ sin θdφ , (3) viscosity within a thermodynamical approach. where a(t) is the scale factor and t represents the cos- A well-known result of the FRW cosmological so- mic time. In the following we use the units 8πG= 1. lutions, corresponding to universes filled with perfect In the first-order thermodynamic theory of Eckart [8] fluid and bulk viscous stresses, is the possibility of the field equations in the presence of bulk viscous violating DEC [6]. The bulk viscosity introduces dis- stresses are sipation by only redefining the effective pressure, Peff, a˙ 2 ρ according to = H 2 = , (4) a 3 Peff = p + Π = p − 3ξH, (1) ¨ a ˙ 2 1 where Π is the bulk viscous pressure, ξ is the bulk = H + H =− (ρ + 3Peff), (5) viscosity coefficient and H is the Hubble parameter. a 6 Since the equation of energy balance is with = + ρ˙ + 3H(ρ+ p + π)= 0, (2) Peff p Π, (6) and the violation of DEC, i.e., ρ + p + Π<0 implies an increasing energy density of the fluid that fills the Π =−3Hξ. (7) universe, for a positive bulk viscosity coefficient. The condition ξ>0 guaranties a positive entropy produc- The conservation equation is tion and, in consequence, no violation of the second ρ˙ + 3H(ρ+ p + Π)= 0. (8) law of the thermodynamics [7]. In the present Letter we show that the above results Assuming that the dark component obey the state are straightforward to obtain from the exact cosmolog- equation ical solutions already found by Barrow in [6]. These = − solutions were obtained using non-causal thermody- p (γ 1)ρ, (9) namics. Nevertheless, we consider a more physical where 0 γ 2, we can obtain from Eqs. (4)–(9) a approach like the full Israel–Stewart–Hiscock causal single evolution equation for H : thermodynamics, showing that it is also possible to ob- ˙ 2 tain big rip type solutions. 2H + 3γH = 3ξH. (10) M. Cataldo et al. / Physics Letters B 619 (2005) 5–10 7
This equation may be integrated directly as a func- ξ>0, and if γ<0Eq.(12) implies that we can have tion of the bulk viscosity. For γ = 0 the solution has a big rip singularity at a finite value of cosmic time. the form Let us consider some examples to see this more clearly. From Eq. (12) the well-known standard case exp 3 ξ(t)dt H(t)= 2 , (11) for a perfect fluid, i.e., ξ = 0, takes the form a(t) = + 3 3 2/(3γ) C 2 γ exp 2 ξ(t)dt dt D(C + (3/2)γ t) . This scale factor may be rewritten as where C is an integration constant. From this equation we find the following expression for the scale factor: 3 2/(3γ) a(t) = a 1 + H γt , (15) 0 2 0 3 3 2/(3γ) a(t) = D C + γ exp ξ(t)dt dt , 2 2 and the energy density is given by (12) ρ0 ρ = , (16) + 3 2 where D is a new integration constant. Thus for a (1 2 H0γt) given ξ(t) we have the expressions for a(t), ρ(t) and where ρ = 3H 2, in order to have H(t = 0) = p(t). 0 0 0 H > 0. If γ<0 we have a big rip singularity at a For the case γ = 0wehavefromEq.(10) 0 finite value of cosmic time tbr =−2/(3H0γ)>t0 = 0. ˙ 2 H In the special case of ξ(t) = ξ0 = const we have ξ = , (13) 3 H from Eq. (12) for the scale factor and substituting this expression into Eq. (8) we have γ 2/(3γ) a(t) = D C + e(3/2)ξ0t . ξ ρ˙ = 6HH,˙ (14) 0 We can rewrite it into the form from which we conclude that ρ = 3H 2 + const. Com- H 2/(3γ) paring this expression with (4) we have that the inte- 0 3ξ0t/2 a(t) = a0 1 + γ e − 1 , (17) gration constant is zero. ξ0 =− Thus we have that for the state equation p ρ, from which we obtain for the energy density i.e., for γ = 0, the scale factor is not defined by the field equations. So for a given a(t) we can write H e3ξ0t ρ(t)= ρ , (18) 0 H 2 and then obtain the expressions for the energy den- 1 + 0 γ(e3ξ0t/2 − 1) sity from Eq. (4) and the bulk viscosity from Eq. (13). ξ0 = = 2 Clearly, if ξ 0, the well-known de Sitter scale factor where ρ0 3H0 . As before, for γ<0wehaveabig a(t) = eH0t is obtained, where p =−ρ and both are rip singularity at a finite value of cosmic time constants. 2 ξ0 Notice that the solution of the field equations may tbr = ln 1 − >t0 = 0. be written through ξ(t) or a(t) because there are three 3ξ0 H0γ independent equations for the four unknown func- Note that any additional condition on the system tions a(t), ρ(t), ξ(t) and p(t). of the field equations will fix the unknown functions. Now we are interested in the possibility that there So for instance, for a variable ξ(t) we can take the are cosmological models with viscous matter which condition ξ(t)= ξ(ρ(t)). present in its development a big rip singularity. Another example in this line is given by the so- lution obtained by Barrow [6] for the case ξ ∼ ρ1/2. 2.1. The case for γ = 0 Effectively, Barrow [6] assumed that the viscosity has a power-law dependence upon the density Firstly, let us consider the case γ = 0. If the viscous ξ = αρ s,α 0, (19) fluid satisfies DEC, then the condition 0 γ 2must be satisfied. Thus for γ<0 we have a phantom cos- where α and s are constant parameters, and exact cos- mology. Now from the thermodynamics we know that mological solutions for a variety of ξ(ρ) in the form 8 M. Cataldo et al. / Physics Letters B 619 (2005) 5–10 givenbyEq.(19). In particular, for the case s = 1/2, 2.2. The case for γ = 0 i.e., ξ = αρ 1/2, yields a power-law expansion for the scale factor. Nevertheless, none condition was im- Notice that the structure of Eq. (10) changes if posed upon the parameters α and γ in order to obtain γ = 0 and ξ is an arbitrary function of the density, solutions with big rip. since the quadratic term in H disappears. Neverthe- For the case s = 1/2, the integration of Eq. (10) less, if ξ ∼ ρ1/2, the structure of Eq. (10) is the same yields the following expression for H(t) for any value√ of γ in the range 0 γ 2, except in the case 3 α = γ , where Eq. (10) becomes H˙ = 0. √ 1 1 3 Then, the solution with γ = 0 can be obtained directly = − 3 α − γ (t − t0), (20) H H0 2 from the general solution given by Eq. (21).Inthis case there is a big rip singularity at a finite value of where H = H(t = t ) and t correspond to the time 0 0 0 cosmic time where dark component begins to become dominant. The scale factor becomes 2 tbr = √ >t0 = 0. 3 3 H α 2√ 0 t − t0 3(γ − 3 α) a(t) = a0 1 − , (21) tbr 3. Israel–Stewart–Hiscock theory where a0 = a(t = t0). If we demand to have the oc- currence of a big rip in the future cosmic time then We now consider the dissipative process in the uni- we have the following constraint on the parameters α verse within the framework of the full causal theory and γ of Israel–Stewart–Hiscock. In this case we have the √ same Friedmann equations but instead of Eq. (7),we 3 α>γ, (22) have an equation for the causal evolution of the bulk viscous pressure, which is given by leading the scale factor blow up to infinity at a finite ˙ ˙ time tbr >t0, which expression is 1 τ˙ ξ T τΠ˙ + Π =−3ξH − τΠ 3H + − − , 2 τ ξ T 2 −1 tbr = √ H . (23) (27) 3( 3 α − γ) 0 where T is the temperature and τ the relaxation time. In terms of time tbr, the Hubble parameter is given by In order to close the system we have to give the equa- tion specifying T − −1 t t0 r H(t)= H0 1 − . (24) T = βρ . (28) tbr The relaxation time is defined by the expression From Eq. (4) and the parameterized equations (21) and (24) for the scale factor and Hubble parameter, re- ξ − τ = = αρ s 1, (29) spectively, we obtain the expression for the increasing ρ density of the dark component in terms of scale factor where β 0. This model imposes the constraint √ 3( 3 α−γ) γ − 1 = 2 a r = , (30) ρ(a) 3H0 . (25) γ a0 in order to have the entropy as a state function. Notice We reproduce completely this solution if we put into that the above constraint exclude the range 0 <γ <1, the Eq. (12) the bulk viscosity given by which implies that quintessence fluids are not allowed √ −1 in this approach. With the above assumptions the field t − t0 ξ(t)= 3 αH0 1 − . (26) equations and the causal evolution equation for the tbr bulk viscosity lead to the following evolution equation M. Cataldo et al. / Physics Letters B 619 (2005) 5–10 9 for H [9]: where the discriminant ∆ has the expression: 3 − − − 2 H¨ + 1 + (1 − r)γ HH˙ + 31 sα 1H 2 2sH˙ b 1 2 ∆ ≡ − 4 . (40) a aγ − 9 − (1 + r)H 1H˙ 2 + (γ − 2)H 3 4 Since we are interested only in positive solutions 1 − − − for A, the coefficient a must be negative. We have two + 32 sα 1γH4 2s = 0. (31) 2 cases of interest. As in the non-causal case we will choose s = 1/2 and ; the above equation becomes Case 1. a<0 γ>0. In this case only A+ correspond to a solution with big rip. The parameters α and γ sat- 1 − isfy the following constraint: H¨ + bHH˙ − 2 − H 1H˙ 2 + aH3 = 0, (32) γ − √ γ 1 where a is defined by 3 α>γ 1 − . (41) 2 9 2 a ≡ 1 + √ γ − 2 , (33) Notice that there is no big rip solution if the cosmic 4 3 α fluid representing the dark component is stiff matter and b by (γ = 2). The factor (1−γ/2)−1 is the correction intro- duced by the causal thermodynamics to the constraint 1 b ≡ 3 1 + √ . (34) given by Eq. (22). The solution for A+ is given by 3 α + √1 + 1 + 2 1/2 Solutions of Eq. (32) were obtained in [10].Inthis 1 1 2 γ = 3α 3α work only was considered γ in the range 1 γ 2. A+ γ , (42) 3 1 − 1 + √2 Some of these solutions presents an increasing energy 3α 2 density and accelerated expansion. which implies that a big rip will occurs at a time Inspired in the solution for the Hubble parameter given by Eq. (24) in the non-causal scheme, we use the = −1 τbr A+H0 . (43) following ansatz, where for simplicity we take t0 = 0 The expressions for a = a(t) and ρ = ρ(a) can be eas- −1 H(t)= A(τbr − t) , (35) ily evaluate from Eqs. (36) and (37), respectively. where A ≡ H τ . With this ansatz the scale factor a(t) 0 br ; evolutes as Case 2. a<0 γ<0. Since we need ∆ 0 in order to have real solutions, the parameters α and γ must −A a(t) ∼ (τbr − t) , (36) satisfy the following constraint: and the energy density, ρ, of the dark component as a √ |γ | function of the scale factor becomes 3 α . (44) 2 √ ∼ 2/A √ ρ(a) a . (37) If ∆ = 0, i.e., 3α = |γ |/2, the solution for A, Using the ansatz (35) in Eq. (32) we obtain a second which we shall call A0, has the following expression: grade equation for A 1 + √1 2 3α 1 A0 = . (45) aA2 + bA + = 0. (38) 3 2 + 1 + √2 |γ | γ 3α The solutions for A are given by If ∆>0, the solutions for A can be written as b √ 1√ 2A± =− ± ∆, (39) A± = A ± ∆. (46) a 0 2 10 M. Cataldo et al. / Physics Letters B 619 (2005) 5–10
4. Discussion also reobtained the Barrow’s solution, for γ = 0, con- sidered here. In the framework of the standard Eckart Within the framework of the non-causal thermo- theory [8], the authors show that fluids which lie in the dynamics we have showed that the power law solu- quintessence region (w>−1) can reduce its thermo- tion, found by Barrow in [6] for dissipative universes dynamical pressure and cross the barrier w =−1, and with ξ = αρ 1/2, yields cosmologies which present big behave like a phantom fluid (w<−1) with the inclu- rip singularity when the constraint given in Eq. (22) sion of a sufficiently large bulk viscosity. The case for holds. If we consider that the dark component is γ = 0 was not considered by these authors. quintessence, i.e., 0 γ 2/3, with a sufficiently large bulk viscosity will make this quintessence be- haves like a phantom energy. In the range 2/3 >γ 2 Acknowledgements it is possible, at least from the mathematical point of view, to obtain solutions with big rip even with a mat- N.C. and S.L. acknowledge the hospitality of the ter fluid. It is not clear for us how can be interpreted a Physics Department of Universidad de Concepción radiation fluid, for example, with a large bulk viscos- where an important part of this work was done last Jan- ity leading to high negative pressures and increasing uary. S.L. acknowledges the hospitality of the Physics densities. Department of Universidad de Santiago de Chile. We At the boundary between the quintessence sector thank the suggestion of a new reference given by the = =− and the phantom sector, i.e., γ 0orp ρ,also referee, in order to improve the presentation of this there exist cosmologies with a big rip singularity. Letter. We acknowledge the partial support to this re- Using a more accurate approach like the full causal search by CONICYT through grants Nos. 1051086, theory of Israel–Stewart–Hiscock, we have also found 1030469 and 1040624 (M.C.); No. 1040229 (N.C. and cosmological solutions with big rip. S.L.); grant MECESUP USA0108 (N.C.). It also was If 1 γ 2 the parameters α and γ satisfy the supported by the Direccion de Investigación de la Uni- constraint given in Eq. (41). Due to the constraint stiff versidad del Bío–Bío (M.C.), PUCV Grant 123.771/04 matter is not allowed. As we mentioned above, this (S.L.). correspond to matter fluids that can lead to a phan- tom behavior. Quintessence region are not allowed. If γ<0 the cosmological solutions can be computed di- References rectly from Eqs. (45) and (46). The main conclusion, in the context of the full causal thermodynamics, is [1] R.R. Cadwell, M. Kamionkowski, N.N. Weinberg, Phys. Rev. that in order to obtain physically reasonable big rip so- Lett. 91 (2003) 071301. lutions, the dark component must be phantom energy. [2] J.D. Barrow, Class. Quantum Grav. 21 (2004) L79; J.D. Barrow, Class. Quantum Grav. 21 (2004) 5619. [3] S. Nojiri, S. Odintsov, Phys. Rev. D 70 (2004) 103522. Note added [4] L. Chimento, R. Lazkoz, Phys. Rev. Lett. 91 (2003) 211301. [5] M.P. Dabrowski, T. Stachowiak, M. Szydlowski, hep-th/ 0307128. While this manuscript was being written we no- [6] J.D. Barrow, Phys. Lett. B 180 (1987) 335; ticed about the work of Brevik and Gorbunova [11]. J.D. Barrow, Nucl. Phys. B 310 (1988) 743. The authors also consider the possibility of big rip [7] W. Zimdahl, D. Pavón, Phys. Rev. D 61 (2000) 108301. in viscous fluids with p = wρ, by a different formal- [8] C. Eckart, Phys. Rev. 58 (1940) 919. [9] R. Maartens, Class. Quantum Grav. 12 (1995) 1455. ism. They consider the case where the bulk viscosity [10] M.K. Mak, T. Harko, Gen. Relativ. Gravit. 30 (1998) 1171. is proportional to the scalar expansion.√ This is equiv- [11] I. Brevik, O. Gorbunova, gr-qc/0504001. alent to the Barrow’s choice ξ(t)∝ ρ, and then they Physics Letters B 619 (2005) 11–16 www.elsevier.com/locate/physletb
Energy conditions and Segre classification of phantom fields
Janilo Santos a,J.S.Alcanizb
a Universidade Federal do Rio Grande do Norte, Departamento de Física, C.P. 1641, 59072-970 Natal, RN, Brazil b Observatório Nacional, Rua Gal. José Cristino 77, 20921-400 Rio de Janeiro, RJ, Brazil Received 20 April 2005; accepted 24 May 2005 Available online 1 June 2005 Editor: N. Glover
Abstract Recent discoveries in the field of observational cosmology have provided increasing evidence that the Universe is undergoing a late time acceleration, which has also stimulated speculations on the nature of the dark component responsible for such a phenomenon. Among several candidates discussed in the current literature, phantom fields, an exotic scalar field with a negative kinetic term and that violates most of the classical energy conditions, appear as a real possibility according to recent observational analysis. In this Letter we examine the invariant characterization for the energy–momentum tensor of phantom fields through the Segre algebraic classification in the framework of general relativity. We also discuss some constraints which are imposed on the values of V(φ)from the classical energy conditions. 2005 Elsevier B.V. All rights reserved.
PACS: 98.80.Jk; 98.80.-k; 04.20.Cv
1. Introduction criteria one chooses to group them into equivalence classes. There are, however, criteria that prove to be The algebraic classification of symmetric second- more important than others. The great appeal of Segre order tensors locally defined on a 4-dimensional classification in general relativity is that it incorpo- Lorentzian manifold, such as the Ricci tensor Rab, rates, ab initio, the Lorentzian character of space–time. the Einstein tensor Gab and the energy–momentum It is of interest in several contexts such as, for example, tensor Tab, is known as Segre classification [1].The in understanding purely geometrical features of space– major idea underlying most of classifications in sci- times [2], in classifying and interpreting matter fields ence is the concept of equivalence. Clearly the objects in general relativity [3] and in higher-dimensional the- may be classified in different ways according to the ories [4] (e.g., 5D brane-worlds [5]) or still as part of the procedure for checking whether apparently dif- ferent space–times are in fact locally the same up to E-mail addresses: [email protected] (J. Santos), [email protected] (J.S. Alcaniz). coordinate transformations (equivalence problem [6]).
0370-2693/$ – see front matter 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.physletb.2005.05.059 12 J. Santos, J.S. Alcaniz / Physics Letters B 619 (2005) 11–16
Because of Einstein’s equations Gab = κTab the Ein- by strange properties such as, for instance, the fact stein tensor (and so Rab) has the same algebraic clas- that its energy density increases with the expansion sification as the energy–momentum tensor. This can of the Universe (in contrast with quintessence fields); be used to decide, on base of generic features such as the possibility of a rip-off of large and small scale the Segre class, which energy–momentum tensors do structures of matter; a possible occurrence of future couple to a given geometry. curvature singularity, etc. [12]. Although having these A great deal of difficulty in Segre classifying sec- unusual characteristics, a phantom type behavior is ond order tensors in the context of general relativity is predicted by several scenarios, e.g., kinetically driven that while Gab and Rab are universal functions of the models [14] and some versions of brane world cos- space–time geometry, Tab depends on the symmetries mologies [15]. From the observational point of view, of the model as well as on the particular type or com- phantom dark energy is found to be compatible with bination of matter fields present in the cosmological most of the classical cosmological tests and provide a scenario. In this regard, a powerful way of imposing better fit to type Ia supernovae observations than do physical constraints and searching realistic forms for ΛCDM or quintessence scenarios (w > −1). There- the energy–momentum tensor is through cosmologi- fore this means that, although exotic, phantom fields cal observations. The high degree of isotropy observed may be the dominant form of energy in our Universe. in the cosmic microwave background radiation [7],for In this Letter we examine the invariant character- instance, restricts the general form of Tab while results ization for the energy–momentum tensor of phantom from distance measurements using type Ia supernovae fields through the Segre algebraic classification in the (which constitute the most direct evidence for the cur- framework of general relativity. In Section 2,byusing rent cosmic acceleration) [8] impose important con- real null tetrad technique, we show that phantom fields straints on the physical quantities of the cosmic fluid. can be classified in two different subclasses of equiva- In light of an impressive convergence of observa- lence which are represented by Segre types [1,(111)] tional results, several groups have recently tested the and [(1, 111)]. In Section 3 we present the so-called viability of different matter fields (or, equivalently, energy conditions of general relativity for the whole different forms of Tab) as a realistic description for Segre class [1, 111] and also examine the further re- the matter content of our Universe. The two favorite strictions they impose on the values of the phantom candidates are the energy density stored on the true field potential V(φ). We end this Letter by summariz- vacuum state of all existing fields in the Universe, ing the main results in the conclusion section. Λ = i.e., the cosmological constant (Tab Λgab) and the potential energy density V(φ) associated with a dy- namical scalar field φ, usually called dynamical dark 2. Segre classification energy (see, e.g., [9] for a recent review on this topic). Physically, the role of such a dynamical dark energy A classification of a generic symmetric second or- can be played by any scalar field violating the so- der tensor Tab can be cast in terms of the eigenvalue called strong energy condition (see Section 3 for a problem discussion). Two possibilities, however, have been ex- T a − λδa vb = 0, (1) haustively explored in the current literature, namely, b b quintessence scalar fields with positive kinetic term where λ are eigenvalues, va are eigenvectors, and the a and an equation-of-state parameter w −1 [10], and mixed tensor T b may be thought of as a linear map phantom fields with negative kinetic term and w<−1 Tp(M) → Tp(M). M is a real 4-dimensional space– [11–13]. In terms of the parameter w, the former case time manifold locally endowed with a Lorentzian met- can be seen as a very general scenario, which includes ric of signature (−+++), Tp(M) denotes the tangent cold dark matter models with a cosmological constant space to M at a point p ∈ M and Latin indices range (ΛCDM), w =−1, and cosmologies dominated by from 0 to 3. Because of the Lorentzian character of the topological defects (e.g., domain walls, strings and metric the mixed form of the energy–momentum ten- =−n a textures) for which w 3 , n being the dimension sor is no more symmetric and T b may not have a di- of the defect. The latter case in turn is characterized agonal matrix representation, i.e., it need not have four J. Santos, J.S. Alcaniz / Physics Letters B 619 (2005) 11–16 13 linearly independent eigenvectors. However, using the taken into account by Segre class [zz¯11], where the zz¯ Jordan canonical forms of the matrix of an operator, refers to a pair of complex eigenvalues. Degeneracy and imposing the Lorentzian character of the metric on amongst eigenvalues in different Jordan blocks will be M, it has been shown [1,3] (for detailed calculations indicated by enclosing the corresponding digits inside in n 5 dimensional space–times and a review on this round brackets, as in [1,(111)], which indicate that topic see [20]) that any energy–momentum tensor de- three out of the four eigenvalues are degenerate. So, fined on Tp(M) reduces to one of the four canonical to each Segre class there may be several subclasses forms: depending on the degeneracies of the eigenvalues. [ ] = + + 1, 111 Tab 2σ1l(amb) σ2(lalb mamb) 2.1. Segre classification of phantom fields + σ3xaxb + σ4yayb, The energy–momentum tensor for the phantom [211] Tab = 2σ1l(amb) ± lalb + σ2xaxb field has the form [13] + σ3yayb, 1 cd [31] Tab = 2σ1l(amb) + 2l(axb) + σ1xaxb T =−φ φ + g g φ φ − V(φ) , (4) ab a b ab 2 c d + σ2yayb, where V(φ) is the phantom potential, φ ≡ φ; and [ ¯ ] = + − a a zz11 Tab 2σ1l(amb) σ2(lalb mamb) the semicolon denotes covariant derivative. Here, we + σ3xaxb + σ4yayb, (2) assume that φ = φ(t)is a function of time alone evolv- ing in an isotropic and homogeneous space–time, so where σ ,...,σ ∈ R, having different values for dif- 1 4 g φaφb =−φ˙2, that is, φa is a time-like vector (in- ferent formulae, and in the Segre class [zz¯11] σ = 0. ab 2 deed an eigenvector of T ). In this case it is always In the above “catalog” the first column is the notation ab possible to find out two null vectors la and ma such commonly used for indicating the Segre class, which a that l ma = 1, and φa can be written as is a list [r1r2 ...rn] of the dimensions of the Jordan blocks of the corresponding Jordan canonical matrix. ˙ = √φ − {l,m,x,y} φa (la ma). (5) The basis vectors form a real null tetrad 2 basis such that the only non-vanishing inner products are Besides, we choose two spacelike vectors xa and ya, belonging to the 2-space orthogonal to the 2-space a a a l ma = x xa = y ya = 1. (3) generated by la and ma, so as to form the real null tetrad basis {l,m,x,y} defined by (3). In terms of this This basis is constructed from the preferred directions basis the metric tensor is written as intrinsically defined by the tensor, i.e., from the Jor- [ ] dan basis. The Segre class 1, 111 distinguishes ten- gab = 2l(amb) + xaxb + yayb, (6) sors that have diagonal matrix representation and is the unique that admits a timelike eigenvector. The comma where the round brackets indicate symmetrization. in this case is used to separate timelike from spacelike Taking into account Eqs. (5) and (6), the canonical eigenvectors. Regardless of the dimension of a Jordan form for the energy–momentum tensor (4) is then writ- block, there is only one eigenvector associated to each ten as block, and the eigenvector associated to a block of di- T = 2σ l m + σ (l l + m m ) mension r>1 is a null vector [4]. Energy–momentum ab 1 (a b) 2 a b a b tensors which belong to Segre class [211], for exam- + σ3(xaxb + yayb), (7) ple, have only three linearly independent eigenvectors where (one of which is a null vector), and its characteris- tic polynomial obtained from (1) has three roots: one 1 ˙2 σ1 =−V(φ), σ2 =− φ , of multiplicity 2 and the others of multiplicity 1. So, 2 1 the digits inside the brackets give also the multiplic- σ =− φ˙2 − V(φ), (8) ity of the real eigenvalues, the complex one being 3 2 14 J. Santos, J.S. Alcaniz / Physics Letters B 619 (2005) 11–16 indicating that phantom fields belong to Segre class here only on the Segre class [1, 111] of Eq. (2).The [1, 111] (see Eq. (2))forσ3 = σ4. In order to find the most common energy conditions are [17,18] (see also subclasses we now determine its eigenvalues and cor- [19] for a recent discussion on this topic): responding eigenvectors. For the sake of brevity we present our results without going into details of calcu- (i) The null energy condition (NEC). NEC states that a b a lations, which can be easily verified from Eq. (7) and Tabn n 0 for null vectors n ∈ Tp(M) which, the expressions below. We find that the set of linearly for Segre class [1, 111], is equivalent to require independent eigenvectors and associated eigenvalues that σ2 − σ1 + σα 0(α = 3, 4). a of T b is given by (ii) The weak energy condition (WEC). WEC states that T tatb 0 for timelike vectors ta ∈ T (M). 1 ab p la − ma → φ˙2 − V(φ) , This will also imply, by continuity, the NEC. The 2 WEC for Segre class [1, 111] means that σ2 − 1 la + ma →− φ˙2 + V(φ) , σ1 0. 2 (iii) The strong energy condition (SEC). SEC is the assertion that for any timelike vector (T − a 1 ab x →− φ˙2 + V(φ) , a b T/2gab)t t 0, where T is the trace of Tab.If 2 Tab belongs to the Segre class [1, 111] then we a 1 ˙2 y →− φ + V(φ) . (9) must have 2σ2 + σ3 + σ4 0. 2 (iv) The dominant energy condition (DEC). DEC a b Note that three out of the four eigenvalues are degen- requires that Tabt t 0 for timelike vectors a erate, making it apparent that the corresponding Segre t ∈ Tp(M) and the additional requirement that b subclass is [1,(111)]. This is the same Segre type as Tabt be a non-spacelike vector. By continuity a that for a perfect fluid [16] (we shall comment this this will also hold for null vectors n ∈ Tp(M). latter). We emphasize that this classification is inde- For energy–momentum tensors of Segre class pendent of the functional form of the potential V(φ) [1, 111] this requires that σ2 − σ1 0, and ˙ as well as the time derivative φ except for φ = const. σ1 − σ2 σα σ2 − σ1 (α = 3, 4). In this latter case the Segre subclass for the phantom field is [(1, 111)] with −V(φ) a fourfold-degenerate When imposed on the energy–momentum tensor of the eigenvalue. This is the same Segre subclass of energy– phantom field these energy conditions require: momentum tensors of Λ-term type (cosmological con- 1 stant) [16]. Since a Λ-term can always be incorporated NEC ⇒− φ˙2 0, into an energy–momentum tensor of the perfect fluid, 2 1 we shall banish this subclass from further considera- WEC ⇒− φ˙2 + V(φ) 0, tion. 2 SEC ⇒ φ˙2 + V(φ) 0, DEC ⇒−φ˙2 0 and V(φ) 0. (10) 3. Classical energy conditions for phantom fields From the above conditions, it should be noticed that In this section we investigate possible constraints although the Segre subclass [1,(111)] for the phan- that the classical energy conditions may impose on tom field is consistent with the Lorentzian signature ˙ the values of the potential V(φ) as well as on φ for of the metric tensor gab, imposition of NEC, as well phantom fields. Restrictions imposed by energy con- as DEC, generates at least one contradiction in the ˙2 a b ditions on energy–momentum tensors of general rel- sense that the timelike character (−φ = gabφ φ ) of ativity theory for matter fields like those represented the vector φa is violated. Maintaining the Lorentzian by Segre classes [211], [31] and [zz¯11] in Eq. (2), timelike character of φa WEC can be preserved only have been presented in the literature [1,17].Aswe for positive potentials since that V(φ) φ˙2/2. SEC, have found the Segre specializations [1,(111)] and on the other hand, cannot be satisfied unless V(φ) is [(1, 111)] for phantom fields, we focus our attention negative and V(φ) −φ˙2. There is however an inter- J. Santos, J.S. Alcaniz / Physics Letters B 619 (2005) 11–16 15
phantom Tab belongs to Segre subclasses [1,(111)] or [(1, 111)]. As is well known, all energy–momentum tensors belonging to these Segre subclasses (e.g., per- fect fluid and Λ-term type energy–momentum tensors) couple to Friedmann–Robertson–Walker geometries. We also have found some constraints which are im- posed on the values of V(φ)from the classical energy conditions. Although the SEC is being violated right Fig. 1. Energy conditions constraints on phantom fields potential. In now, according recent observational data regarding the interval (I) phantom fields do not violate SEC but do violate WEC. accelerating universe, we see from our analysis that In interval (II) both (SEC and WEC) are violated while in region an evolving potential function V(φ)can, in principle, (III) phantom fields do not violate WEC but do violate SEC. be constructed such that this condition is not violated in the past (interval I of Fig. 1), while in that epoch ˙2 ˙2 val (−φ second-order tensor on 4-dimensional Lorentzian space–times J.A.S. Lima, J.S. Alcaniz, Phys. Lett. B 600 (2004) 191; manifolds. J.M. Aguirregabiria, L.P. Chimento, R. Lazkoz, Phys. Rev. [2] R.V. Churchill, Trans. Amer. Math. Soc. 34 (1932) 784; D 70 (2004) 023509. W.J. Cormack, G.S. Hall, J. Phys. A 12 (1979) 55. [12] R.R. Caldwell, M. Kamionkowski, N.N. Weinberg, Phys. Rev. [3] G.S. Hall, D.A. Negm, Int. J. Theor. Phys. 25 (1986) 405; Lett. 91 (2003) 071301; J. Santos, M.J. Rebouças, A.F.F. Teixeira, J. Math. Phys. 34 P.F. Gonzalez-Diaz, Phys. Rev. D 68 (2003) 021303; (1993) 186. S. Nesseris, L. Perivolaropoulos, Phys. Rev. D 70 (2004) [4] J. Santos, M.J. Rebouças, A.F.F. Teixeira, J. Math. Phys. 36 123529; (1995) 3074; P.F. Gonzalez-Diaz, J.A. Jimenez-Madrid, Phys. Lett. B 596 J. Santos, M.J. Rebouças, A.F.F. Teixeira, Gen. Relativ. (2004) 16. Gravit. 27 (1995) 989. [13] P. Singh, M. Sami, N. Dadhich, Phys. Rev. D 68 (2003) [5] M.J. Rebouças, J. Santos, Mod. Phys. Lett. A 18 (2003) 2807. 023522. [6] A. Karlhede, Gen. Relativ. Gravit. 12 (1980) 693; [14] T. Chiba, T. Okabe, M. Yamaguchi, Phys. Rev. D 62 (2000) M.A.H. MacCallum, Computer-aided classification of exact 023511. solutions in general relativity, in: R. Cianci, R. de Ritis, M. [15] V. Sahni, Y. Shtanov, Int. J. Mod. Phys. D 11 (2002) 1515; Francaviglia, G. Marmo, C. Rubano, P. Scudellaro (Eds.), Gen- V. Sahni, Y. Shtanov, JCAP 0311 (2003) 014. eral Relativity and Gravitational Physics, 9th Italian Confer- [16] D. Kramer, H. Stephani, E. Herlt, M. MacCallum, Exact So- ence, World Scientific, Singapore, 1991. lutions of Einstein’s Field Equations, Cambridge Univ. Press, [7] D.N. Spergel, et al., Astrophys. J. Suppl. 148 (2003) 175. Cambridge, 1980, p. 67. [8] A.G. Riess, et al., Astrophys. J. 607 (2004) 665. [17] S.W. Hawking, G.F.R. Ellis, The Large Scale Structure of [9] V. Sahni, A. Starobinsky, Int. J. Mod. Phys. D 9 (2000) 373; Spacetime, Cambridge Univ. Press, Cambridge, 1973. J.E. Peebles, B. Ratra, Rev. Mod. Phys. 75 (2003) 559; [18] M. Visser, Lorentzian Wormholes, AIP, New York, 1996. T. Padmanabhan, Phys. Rep. 380 (2003) 235; [19] M. Visser, Science 276 (1997) 88; J.A.S. Lima, Braz. J. Phys. 34 (2004) 194. M. Visser, C. Barceló, Energy conditions and their cosmologi- [10] B. Ratra, P.J.E. Peebles, Phys. Rev. D 37 (1988) 3406; cal implications, in: Cosmo99, Proceedings of the Third Inter- R.R. Caldwell, R. Dave, P.J. Steinhardt, Phys. Rev. Lett. 80 national Workshop on Particle Physics and the Early Universe, (1998) 1582; ICTP, Trieste, Italy, September–October 1999, World Scien- P.G. Ferreira, M. Joyce, Phys. Rev. D 58 (1998) 023503. tific, Singapore, 2000, gr-qc/0001099; [11] R.R. Caldwell, Phys. Lett. B 545 (2002) 23; C. Barceló, M. Visser, Int. J. Mod. Phys. D 11 (2002) 1553. B. McInnes, astro-ph/0210321; [20] M.J. Rebouças, J. Santos, A.F.F. Teixeira, Braz. J. Phys. 34 V. Faraoni, Int. J. Mod. Phys. D 11 (2002) 471; (2004) 535. S.M. Carroll, M. Hoffman, M. Trodden, Phys. Rev. D 68 [21] L. Perivolaropoulos, astro-ph/0412308. (2003) 023509; [22] J.D. Barrow, Class. Quantum Grav. 21 (2004) L79; S. Nojiri, S.D. Odintsov, Phys. Lett. B 562 (2003) 147; J.D. Barrow, Class. Quantum Grav. 21 (2004) 5619; J.S. Alcaniz, Phys. Rev. D 69 (2004) 083521; J.D. Barrow, C.G. Tsagas, gr-qc/0411045. Physics Letters B 619 (2005) 17–25 www.elsevier.com/locate/physletb Cosmic acceleration and the string coupling John Ellis a, N.E. Mavromatos b, D.V. Nanopoulos c,d,e a TH Division, Physics Department, CERN, CH-1211 Geneva 23, Switzerland b Theoretical Physics, Physics Department, King’s College London, Strand WC2R 2LS, UK c George P. and Cynthia W. Mitchell Institute for Fundamental Physics, Texas A&M University, College Station, TX 77843, USA d Astroparticle Physics Group, Houston Advanced Research Center (HARC), Mitchell Campus, Woodlands, TX 77381, USA e Academy of Athens, Division of Natural Sciences, 28 Panepistimiou Avenue, Athens 10679, Greece Received 24 January 2005; accepted 20 May 2005 Available online 31 May 2005 Editor: L. Alvarez-Gaumé Abstract In the context of a cosmological string model describing the propagation of strings in a time-dependent Robertson–Walker background space–time, we show that the asymptotic acceleration of the Universe can be identified with the square of the string coupling. This allows for a direct measurement of the ten-dimensional string coupling using cosmological data. We conjecture that this is a generic feature of a class of non-critical string models that approach asymptotically a conformal (critical) σ model whose target space is a four-dimensional space–time with a dilaton background that is linear in σ -model time. The relation between the cosmic acceleration and the string coupling does not apply in critical strings with constant dilaton fields in four dimensions. 2005 Elsevier B.V. All rights reserved. String theory [1,2] was first developed as a the- were found to construct low-energy models that could ory of the strong interactions, but it soon turned out be consistent with the current particle physics phe- that mathematical consistency (world-sheet confor- nomenology, but string models of this type had zero mal invariance) required the theory to live in higher- predictability, in the sense that they were unable to dimensional space times. Even target-space supersym- make predictions for the parameters of the Standard metry was not successful in reducing the number of Model, and there were many string models with indis- space–time dimensions below ten. Thus, enormous ef- tinguishable low-energy limits. fort has been expended on the compactification of the Although in principle string theory has no free pa- extra dimensions, with the eventual aim of accommo- rameters, and the ground state corresponding to the dating the Standard Model at low energies. Many ways observable low-energy world is supposed to be chosen dynamically, a detailed understanding of mechanism for choosing the ground state has not been achieved so E-mail address: [email protected] (N.E. Mavromatos). far. Lacking a microscopic, dynamical mechanism for 0370-2693/$ – see front matter 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.physletb.2005.05.047 18 J. Ellis et al. / Physics Letters B 619 (2005) 17–25 specifying the various string model parameters, such the context of the minimal supersymmetric extension as the compactification radii and the four-dimensional of the Standard Model. gauge couplings, one has had simply to fix them by Modern developments in string theory [2] make hand, so as to match the results with experimental ob- possible consistent quantum treatments of domain- servations in particle physics. In this framework, the wall structures in string theory (D-branes). These have mechanism whereby one particular model is chosen opened up novel ways of looking at both the micro- from among the complicated string ‘landscape’ [3] is cosmos and the macrocosmos, offering new insights still unclear. into both particle phenomenology and the cosmic evo- The most important and fundamental string para- lution of our Universe. In the microcosmos, there are meter is the string coupling, gs , which determines the novel ways of compactification, either via the observa- regime of validity of string perturbation theory, and tion [4] that large (compared to the string scale) extra hence the world-sheet σ -model scheme for low-energy dimensions are consistent both with the foundations computations of the low-energy string effective action. of string theory and phenomenology, or by viewing Since gs is connected to the unified ten-dimensional our four-dimensional world as a brane embedded in gauge coupling of the effective supersymmetric low- a bulk space–time. This would allow for large ex- energy theory, its value is usually inferred from par- tra bulk dimensions, which could even be infinite in ticle phenomenology. The string coupling is not a size [5], offering new ways to analyze the large hier- constant but, like any other dynamical coupling in a archy between the Planck scale and the electroweak supersymmetric field theory, is related to the vacuum symmetry-breaking scale. In this modern approach, expectation value of a field, in this particular case fields in the gravitational (super)multiplet of the (su- the dilaton field Φ, which belongs to the gravitational per)string are allowed to propagate in the bulk, but not multiplet obtained from the string [1] the gauge fields, which are attached to the brane world. In this way, the weakness of gravity as compared to the 2 = 2Φ gs e . (1) rest of the interactions is a result of the large extra dimensions. Their compactification is not necessar- Usually, upon compactification the dilaton field is split ily achieved through conventional means, i.e., closing into a product of two factors, one depending on the up the extra dimensions in compact spatial manifolds, compact six-dimensional space coordinates and the but might also involve shadow brane worlds with spe- other on the four-dimensional space–time coordinates, cial reflecting properties (such as orientifolds), which which are supposed to correspond to the large, un- bound the bulk dimension [6]. In such approaches, compactified coordinates of our observable world. In the string scale Ms is not necessarily identical to the most of the phenomenological approaches to model four-dimensional Planck mass scale MP , but instead building, the four-dimensional dilaton field has been they are related through the large compactification vol- assumed to be constant and therefore trivial, since this ume V6: constant value could be absorbed in an innocuous shift 8M8V in the field. M2 = s 6 . P 2 (2) In this approach, neither the string coupling nor gs the unified gauge coupling are accessible directly to As for the macrocosmos, there are novel ways of experimental measurement. It is consistency of the discussing cosmology in brane worlds, which may available phenomenological model with low-energy revolutionize our way of approaching issues such as observational data that leads to an indirect fixing of inflation [7,8]. 2 the string coupling. A popular value is gs 0.52, Mounting experimental evidence from diverse as- which, upon compactification to small dimensions (of trophysical sources presents important cosmological the order of a tenth of the four-dimensional Planck puzzles that string theory must address if it is to pro- 19 mass, MP ∼ 10 GeV), yields a four-dimensional vide a realistic description of Nature. Observations of 2 ∼ unified gauge coupling strength gU /4π 1/24 at large-scale structures, distant type-1a supernovae [9], 16 scales MU ∼ 10 GeV, as suggested by extrapolat- and the cosmic microwave background fluctuations ing the measured gauge couplings to high energies in (by WMAP [10] in particular) have established that J. Ellis et al. / Physics Letters B 619 (2005) 17–25 19 the Hubble expansion of our Universe is currently ac- positive cosmological constant Λ>0 has a non-zero celerating, and that 70% of its energy density consists Ricci tensor Rµν = Λgµν , where gµν is the metric ten- of unknown dark energy that appears in ‘empty’ space sor. and does not clump with ordinary matter. An interesting proposal for obtaining a non-zero These observations have great potential signifi- cosmological constant in string theory was made in cance for string theory, and may even revolutionize [11]. It was suggested that dilaton tadpoles in higher- the approach to it that has normally been followed so genus world-sheet surfaces, which produce additional far. If the dark energy leads to an asymptotic de Sit- modular infinities whose regularization leads to extra ter horizon, as would occur if it turns out to be a true world-sheet structures in the σ -model not appearing cosmological constant, then the entire concept of the at the world-sheet level, modify the string β-function scattering S-matrix breaks down, and hence the con- in such a way that the Ricci tensor of the space–time ventional approach to string theory. On the other hand, background is now that of a de Sitter Universe, with if there is some quintessential mechanism for relaxing a cosmological constant specified by the dilaton tad- the vacuum energy, so that the vacuum energy density pole graph. The problem with this approach is the vanishes at large cosmic times in a manner consis- above-mentioned existence of an asymptotic horizon tent with the existence of an S-matrix, there is still the in the de Sitter case, which prevents the proper defini- open issue of embedding such models in (perturba- tion of asymptotic states, and hence an S-matrix. Since tive) string theory. One would need, in particular, to the perturbative world-sheet formalism is based on the develop a consistent σ -model formulation of strings existence of such an S-matrix, there is a question of propagating in such time-dependent, relaxing space– consistency in this approach. time backgrounds. It was proposed in [12] that a way out of this dif- We here propose a resolution of this dilemma, ficulty would be to assume specific time-dependent based on string theory in a time-dependent dilaton dilaton backgrounds, with a linear dependence on time background, in which the asymptotic acceleration of in the so-called σ -model frame the Universe is directly related to the string coupling. Φ = const − Qt, (4) The world-sheet conformal-invariance conditions of critical string theory are equivalent to the target- where Q is a constant, and Q2 > 0 is a deficit in space equations of motion for the background fields the σ -model central charge. Such backgrounds, even through which the string propagates. These conditions when the σ -model metric is flat, lead to exact solu- are very restrictive, allowing only for vacuum solu- tions (in all orders in α) of the conformal-invariance tions of (critical) strings to be constructed in this way. conditions of the pertinent stringy σ -model, thereby The main problem may be expressed as follows. Con- constituting acceptable solutions from a perturbative sider the graviton world-sheet β function, which is string viewpoint. The appearance of Q allowed this nothing but the Ricci tensor of the target space–time supercritical string theory [12] to be formulated in background to lowest order in α: spaces with numbers of dimensions different from the critical case. This was actually the first example of a β = α R , (3) µν µν non-critical string, with the target-space coordinates where we ignore the possible presence of other fields, Xi , i = 1,...,D − 1, playing the rôles of the perti- for simplicity. Conformal invariance requires the van- nent σ -model fields. This non-critical string was not ishing condition βµν = 0, which implies that the back- conformally invariant, and hence required Liouville ground must be Ricci flat, which is a solution of the dressing [13]. The Liouville field had time-like signa- vacuum Einstein equations. The issue then arises how ture in target space, since the central charge deficit Q2 to describe in string theory cosmological backgrounds, was positive in the model of [12], and its zero mode which are not vacuum solutions, but require the pres- played the rôle of target time. ence of a matter fluid and hence a non-vanishing Ricci As a result of the existence of a non-trivial dilaton tensor. In this respect, we see that a cosmological con- field, the Einstein term in the effective D-dimensional stant is inconsistent with the conformal invariance of low-energy field theory action is conformally rescaled string since, for instance, a de Sitter Universe with a by e−2Φ . This requires a specific redefinition of tar- 20 J. Ellis et al. / Physics Letters B 619 (2005) 17–25 get time in order that the metric acquires the standard conformal invariance. Such backgrounds were also al- Robertson–Walker (RW) form in the normalized Ein- lowed to be time-dependent, and the target time was stein frame for the effective action identified with the Liouville world-sheet zero mode, thereby not increasing the target space–time dimen- 2 =− 2 + 2 2 + 2 2 dsE dtE aE(tE) dr r dΩ , (5) sionality. We have provided several justifications and where we have only exhibited a spatially-flat RW checks of this identification [14], which is possible metric for definiteness, and aE(tE) is an appropri- only when the initial σ -model is supercritical, so that ate scale factor, which is simply a function of the the Liouville mode has time-like signature [12,13].For Einstein-frame time tE in the homogeneous cosmolog- example, in certain models [15,16], such an identifica- ical backgrounds that we assume throughout. tion was energetically preferable from a target-space The Einstein-frame time is related to the σ -model- viewpoint, since it minimized certain effective poten- frame time [12] by tials in the low-energy field theory corresponding to the string theory at hand. t Such non-critical σ models relax asymptotically in = −Φ → = −Φ(t) dtE e dt tE e dt. (6) cosmic Liouville time to conformal σ models, the lat- ter viewed as equilibrium points in string theory space. The linear dilaton background (4) yields then the In some interesting cases of relevance to cosmology, following relation between the Einstein- and σ -model- which were particularly generic, the asymptotic con- frame times formal field theory was that of [12], with a linear c0 Qt tE = c1 + e , (7) dilaton and a flat Minkowski target-space metric in the Q σ -model frame. where c1,0 are appropriate (positive) constants. Thus, One such model was considered in detail in [17]. a dilaton background that is linear in σ -model-frame The model was originally formulated within a specific time (4) will scale logarithmically with the Einstein- string theory, namely ten-dimensional type-0 [18], frame time tE, which is just the Robertson–Walker which leads to a non-supersymmetric target-space cosmic time spectrum, as a result of a special projection of the su- persymmetric partners out of the spectrum. However, Q Φ(t ) = const − ln t . (8) the basic properties of its cosmology, which are those E c E 0 interest to us in this work, are sufficiently generic that In this regime, the string coupling (1) varies with the they can be extended to the bosonic sector of any other cosmic time tE as effective low-energy supersymmetric field theory of supersymmetric strings, including those relevant to 2 ∝ 1 gs (tE) , (9) unified particle physics phenomenology. t2 E The ten-dimensional metric configuration consid- implying that the effective string coupling vanishes ered in [17] was asymptotically in cosmic time. In the linear-dilaton background of [12], the asymptotic space–time met- (4) gµν 00 ric in the Einstein frame reads 2σ GMN = 0 e 1 0 , (11) 2σ 2 =− 2 + 2 2 2 + 2 2 00e 2 I5×5 ds dtE a0tE dr r dΩ , (10) where a0 a constant, which is a linearly-expanding where lower-case Greek indices are four-dimensional Universe. Clearly, there is no acceleration in the Uni- space–time indices, and I5×5 denotes the 5 × 5 unit verse (10). matrix. We have chosen two different scales for the In [14] we went one step further than the analy- internal space. The field σ1 sets the scale of the fifth di- sis in [12], and considered more complicated σ -model mension, while σ2 parametrizes a flat five-dimensional metric backgrounds, which did not satisfy the σ - space. In the context of the cosmological models we (4) model conformal-invariance conditions, and therefore treat here, the fields gµν , σi , i = 1, 2, are assumed to were in need of Liouville dressing in order to restore depend on the time t only. J. Ellis et al. / Physics Letters B 619 (2005) 17–25 21 Type-0 string theory, as well as its supersymmetric runs with the local world-sheet renormalization group versions appearing in other scenarios including brane scale, namely the zero mode of the Liouville field, models, contains appropriate form fields with non- which is identified [14] with the target time in the σ - trivial gauge fluxes (flux-form fields), which live in the model frame. The supercriticality [12] Q2 > 0ofthe higher-dimensional bulk space. In the specific model underlying σ model is crucial, as already mentioned. of [18], one such field was considered to be non-trivial. Physically, the non-critical string provides a frame- As was demonstrated in [17], a consistent background work for non-equilibrium dynamics, which may be the choice for the flux-form field has the flux parallel to result of some catastrophic cosmic event, such as a the fifth dimension σ1. This implies that the internal collision of two brane worlds [7,15,16], or an initial space is crystallized (stabilized) in such a way that this quantum fluctuation. dimension is much larger than the remaining five di- In the generic class of non-critical string models mensions σ2, demonstrating the physical importance considered in this work, the σ model always asymp- of the flux fields for large radii of compactification. totes, for long enough cosmic times, to the linear- Considering the fields to be time-dependent only, dilaton conformal σ -model field theory of [12].But i.e., considering spherically-symmetric homogeneous it is important to stress that this is only an asymptotic backgrounds, restricting ourselves to the compactifica- limit. In this respect, the current era of our Universe is tion (11), and assuming a Robertson–Walker form of viewed as being close, but still not quite at the relax- the four-dimensional metric with scale factor a(t),the ation (equilibrium) point, in the sense that the dilaton generalized conformal-invariance conditions and the is almost linear in the σ -model time, and hence varies Curci–Paffuti σ -model renormalizability constraint logarithmically with the Einstein-frame time (8).Itis [19] yield a set of differential equations which were expected that this slight non-equilibrium will lead to solved numerically in [17]. The generic form of these a time-dependence of the unified gauge coupling and equations reads [13,14,17] other constants such as the four-dimensional Planck length (2) that characterize the low-energy effective ¨i + ˙i =−˜i g Q(t)g β , (12) field theory, mainly through the time-dependence of where the β˜i are the Weyl-anomaly coefficients of the the string coupling (1) that results from the time- stringy σ -model on the background {gi}. In the model dependent linear dilaton (4). of [17],the{gi} include graviton, dilaton, tachyon, The asymptotic regime of the type-0 cosmological flux and moduli fields σ1,2, whose vacuum expecta- string model of [17] has been obtained analytically, tion values control the sizes of the extra dimensions. by solving the pertinent equations (12) for the various The detailed analysis of [17] indicated that the fields. As already mentioned, at late times the theory moduli fields σi froze quickly to their equilibrium val- becomes four-dimensional, and the only non-trivial ues. Thus, together with the tachyon field which also information is contained in the scale factor and the decays to a constant value rapidly, they decouple from dilaton, given that the topological flux field remains the four-dimensional fields at very early stages in the conformal in this approach, and the moduli and ini- evolution of this string Universe.1 There is an infla- tial tachyon fields decouple very fast during the initial tionary phase in this scenario and a dynamical exit stages after inflation in this model. For times that are from it. The important point to guarantee the exit is long after the initial fluctuations, such as the present the fact that the central-charge deficit Q2 is a time- epoch when the linear approximation is valid, the so- dependent entity in this approach, obeying specific lution for the dilaton in the σ -model frame, as derived relaxation laws determined by the underlying confor- from the equations (12), takes the form mal field theory [15–17]. In fact, the central charge αA Φ(t) =−ln cosh(F1t) , (13) F1 1 The presence of the tachyonic instability in the spectrum is due where F1 is a positive constant, α is a numerical con- to the fact that in type-0 strings there is no target-space supersym- stant of order one, and metry, by construction. From a cosmological viewpoint the tachyon s01 fields are not necessarily bad features, since they may provide the = C√5e initial instability leading to cosmic expansion [17]. A , (14) 2V6 22 J. Ellis et al. / Physics Letters B 619 (2005) 17–25 where s01 is the equilibrium value of the modulus field For large tE, e.g., present or later cosmological time σ1 associated with the large bulk dimension, and C5 is values, one has the corresponding flux of the five-form field. Notice F1 that A is independent of this large bulk dimension. a (t ) 1 + γ 2t2 . (19) E E γ E For very large times F1t 1 (in string units), one therefore approaches a linear solution for the dilaton: At very large (future) times, a(tE) scales linearly Φ ∼ const − F1t.From(13), (1) and (2), we then see with the Einstein-frame cosmological time tE [17], that the asymptotic weakness of gravity in this Uni- and hence the cosmic horizon expands logarithmi- verse [17] is due to the smallness of the internal space cally. From a field-theory viewpoint, this would allow V6 as compared with the flux C5 of the five-form field. for a proper definition of asymptotic states and thus The constant F1 is related to the central-charge deficit a scattering matrix. As we mentioned briefly above, of the underlying the non-conformal σ -model by [17] however, from a stringy point of view, there are restric- tions on the asymptotic values of the central-charge q0 dΦ deficit q , and it is only a discrete spectrum of values Q = q0 + F1 + , (15) 0 F1 dt of q0 which allow for a full stringy S-matrix to be de- where q0 is a constant, the parenthesis vanishes as- fined, respecting modular invariance [12]. ymptotically, and the numerical solution√ of (12) stud- Asymptotically in time, therefore, the Universe re- ied in [17]) requires that q0/F1 = (1+ 17)/2 2.53. laxes to its ground-state equilibrium situation and For this behaviour of Φ, the central-charge deficit (15) the non-criticality of the string, caused by the initial tends to a constant value q0. In this way, F1 is related quantum fluctuation or other initial condition, disap- to the asymptotic constant value of the central-charge pears, giving way to a critical (equilibrium) string deficit, up to an irrelevant proportionality factor of Universe with a Minkowski metric and a linear-dilaton order one, in agreement with the conformal model background. These are the generic features of the mod- of [12], to which this model asymptotes. This value els we consider here, which can include strings with should be, for consistency of the underlying string target-space supersymmetry as well as the explicit theory [12], some discrete value for which the fac- bosonic type-0 string considered here for simplicity. torization property (unitarity) of the string scattering The Hubble parameter of such a Universe becomes amplitudes is valid. Notice that this asymptotic string for large tE theory, with a constant (time-independent) central- 2 γ tE charge deficit, q2 ∝ c∗ − 25 (or c∗ − 9 for superstring) H(t ) . (20) 0 E + 2 2 is considered as an equilibrium situation, and an S ma- 1 γ tE trix can be defined for specific (discrete) values of the On the other hand, the Einstein-frame effective four- central charge c∗. The standard critical (super)string dimensional ‘vacuum energy density’, which is deter- corresponds to a central charge c∗ = 25 (c∗ = 9), but mined by the running central-charge deficit Q2 after ∗ in our case c differs from that critical value. compactifying to four√ dimensions the ten-dimensional 10 −2Φ 2 Defining the Einstein-frame time tE through (6),we expression d x −ge Q (tE),is[17] obtain in the case (13) q2γ 2 2Φ−σ1−5σ2 2 0 αA ΛE(tE) = e Q (tE) , (21) t = (F t). F 2(1 + γ 2t2 ) E 2 sinh 1 (16) 1 E F1 where, for large tE, Q is given in (15), and approaches In terms of the Einstein-frame time, one obtains a log- its equilibrium value q0. Thus, we see explicitly how arithmic time-dependence [12] for the dilaton the dark energy density relaxes to zero for tE →∞. Finally, and most importantly for our purposes ΦE = const − ln(γ tE), (17) here, the deceleration parameter in the same regime where of tE becomes 2 2 2 F1 (d aE/dt )aE 1 γ ≡ . (18) q(t ) =− E − . (22) αA E 2 2 2 (daE/dtE) γ tE J. Ellis et al. / Physics Letters B 619 (2005) 17–25 23 The key point about this expression is that, as is clear for the present era: from (17) and (1), up to irrelevant proportionality con- 1 1 6.56 1 stant factors which by conventional normalization are Φ˙ 2 ∼ ,V(Φ)∼ , (27) 2 2 2 2 set to unity, it can be identified with the square of the 2tE tE string coupling where the numerical factor is a consequence of the nu- merical result of [17]. This implies an equation of state =− − =− 2 q(tE) exp 2(Φ const) gs . (23) (26) This is our central result. w (t 1) −0.74 (28) To guarantee consistency of perturbation theory, Φ E one must have gs < 1, which can be achieved in our for (large) times tE in string units corresponding to the approach if one defines the present era by the time present era (24). Correspondingly, we have a cosmic regime deceleration parameter ∼ −1 1 γ tE (24) q = (1 + 3w ) =−0.61. (29) 2 Φ in the Einstein frame. This is compatible with large This fixes the string coupling to perturbative values, enough times tE (in string units) for consistent with naive scenarios for grand unification. − So far the model did not include ordinary matter, |C |e 5s02 1, (25) 5 as only fields from the gravitational string multiplet as becomes clear from (14) and (18). This condition have been included. The inclusion of ordinary matter can be guaranteed either for small radii of the five ex- is not expected to change qualitatively the result. We tra dimensions or by a large value of the flux |C5| of conjecture that the fundamental relation (23) will con- the five-form of the type-0 string. We recall that the tinue to hold, the only difference being that probably relatively large extra dimension, s01, which extends the inclusion of ordinary matter will tend to reduce the in the direction of the flux, decouples from this con- string acceleration, due to the fact that matter is sub- dition. Therefore, effective five-dimensional models ject to attractive gravity, and resists the acceleration of with a large uncompactified fifth dimension may be the Universe. In such a case, one has constructed consistently with the condition (24). 1 We next turn to the equation of state in such a q = ΩM − ΩΛ, (30) 2 Universe. As discussed in [17], this model resembles quintessence, with the dilaton playing the rôle of the where ΩM (ΩΛ) denote the matter (vacuum) energy quintessence field. Hence the equation of state for our densities, normalized to the critical energy density of type-0 string Universe reads [20] a spatially flat Universe. There is a remarkable coincidence in numbers for p 1 (Φ)˙ 2 − V(Φ) this non-supersymmetric type-0 string Universe with w = Φ = 2 , (26) Φ 1 ˙ the astrophysical observations, which yield also a q ρΦ (Φ)2 + V(Φ) 2 close to this value, since the ordinary matter content where pΦ is the pressure and ρΦ is the energy den- of the Universe (normalized with respect to the energy sity, and V(Φ)is the effective potential for the dilaton, density of a flat Universe) is Ωordinary matter 0.04 and which in our case is provided by the central-charge the dark matter content is estimated to be ΩDM = 0.23, deficit term. Here the dots denote Einstein-frame dif- while the dark energy content is ΩΛ 0.73. This ferentiation. In the Einstein frame, the potential V(Φ) yields q =−0.595, which is only a few per cent away is given by ΛE in (21). In the limit Q → q0, which we from (29). Conversely, if one used naively in the ex- assume characterizes the present era to a good approx- pression (30) the value (29) for q, obtained in our 2 2 −2 imation, V(Φ) is of order (q0 /2F1 )tE , where we case where ordinary matter was ignored, one would recall that q0/F1 is of order one, as discussed above. find ΩΛ 0.74, indicating that the contribution of the The exact normalization of the dilaton field in the Ein- dilaton field to the cosmic acceleration is the dominant stein frame is ΦE = const − ln(γ tE). We then obtain one. 24 J. Ellis et al. / Physics Letters B 619 (2005) 17–25 If the relation (23) were to hold also upon the inclu- The initial state of our cosmos may correspond to a sion of matter, even in a realistic case with (broken) certain ‘random’ Gaussian fixed point in the space supersymmetry, one would arrive at a value of the of string theories, which is then perturbed in the Big 2 string coupling, gs 0.55, which would be quite con- Bang by some ‘random’ relevant (in a world-sheet sistent with the unification prediction of the minimal sense) deformation, making the theory non-critical supersymmetric extension of the Standard Model at and taking it out of equilibrium from a target space– scales ∼ 1016 GeV. The only requirement for the as- time viewpoint. The theory then flows, following a ymptotic condition (23) to hold is that the underlying well-defined renormalization-group trajectory, and as- stringy σ model theory is non-critical and asymp- ymptotes to the specific ground state corresponding to totes for large times to the linear-dilaton conformal the infrared fixed point of this perturbed world-sheet field theory of [12]. It should be understood, though, σ -model theory. This approach allows for many par- that the precise relation of the four-dimensional gauge allel universes to be implemented of course, and our coupling with the ten-dimensional string coupling de- world would be just one of these. Each Universe may pends on the details of compactification, which we did flow between a different pair of fixed points, as it may not discuss in this work. be perturbed by different operators. It seems to us that We close this discussion by stressing once more the this scenario is much more attractive (no pun intended) importance of non-criticality in order to arrive at (23). and specific than the static ‘landscape’ scenario [3], In critical strings, which usually assume the absence which is currently advocated as an attempt to parame- of a four-dimensional dilaton, such a relation cannot trize our ignorance of the true structure of the string/M be obtained, and the string coupling is not directly theory vacuum and its specification. measurable. The logarithmic variation of the dilaton field with the cosmic time at late times implies a slow variation of the string coupling (23), |˙gs/gs|=1/tE ∼ Acknowledgements 10−60 for the present era, and hence a correspondingly slow variation of the gauge couplings. N.E.M. wishes to thank Juan Fuster and IFIC- From a physical point of view, the use of critical University of Valencia (Spain) for their interest and strings to describe the evolution of our Universe seems support, and P. Sodano and INFN-Sezione di Perugia desirable, whilst non-critical strings may be associ- (Italy) for their hospitality and support during the final ated with non-equilibrium situations, as undoubtedly stages of this work. The work of D.V.N. is supported occur in the early Universe. The space of non-critical by DOE grant DE-FG03-95-ER-40917. string theories is much larger that of critical strings. Therefore, it is remarkable that the departure from crit- icality has the potential to enhance the predictability References of string theory to such a point that the string coupling may become accessible to experiment. A similar situa- [1] M.B. Green, J.H. Schwarz, E. Witten, Superstring Theory, vols. I, II, Cambridge Univ. Press, Cambridge, 1987. tion arises in a non-critical string approach to inflation, [2] J. Polchinski, String Theory, vol. 2, Cambridge Univ. Press, in the scenario where the Big Bang is identified with Cambridge, 1998; the collision [7] of two D-branes [16]. In such a sce- J.H. Schwarz, hep-th/9907061. nario, astrophysical observations may place important [3] L. Susskind, hep-th/0302219. bounds on the recoil velocity of the brane worlds after [4] N. Arkani-Hamed, S. Dimopoulos, G.R. Dvali, Phys. Lett. B 429 (1998) 263, hep-ph/9803315; the collision, and lead to an estimate of the separation I. Antoniadis, N. Arkani-Hamed, S. Dimopoulos, G.R. Dvali, of the branes at the end of the inflationary period. Phys. Lett. B 436 (1998) 257, hep-ph/9804398. The approach of identifying target time in such a [5] L. Randall, R. Sundrum, Phys. Rev. Lett. 83 (1999) 4690, hep- framework with a world-sheet renormalization-group th/9906064. scale, the Liouville mode [14], provides a novel way [6] See, for instance: L.E. Ibanez, R. Rabadan, A.M. Uranga, Nucl. Phys. B 576 (2000) 285, hep-th/9905098; of selecting the ground state of the string theory, which L.E. Ibanez, hep-ph/9905349; may not necessarily be associated with minimization C.A. Scrucca, M. Serone, JHEP 9912 (1999) 024, hep- of energy, but could be a matter of cosmic ‘chance’. th/9912108; J. Ellis et al. / Physics Letters B 619 (2005) 17–25 25 D.M. Ghilencea, G.G. Ross, Nucl. Phys. B 595 (2001) 277, [14] J.R. Ellis, N.E. Mavromatos, D.V. Nanopoulos, Phys. Lett. hep-ph/0006318, and references therein. B 293 (1992) 37, hep-th/9207103; [7] J. Khoury, B.A. Ovrut, P.J. Steinhardt, N. Turok, Phys. Rev. C. Castro, M.S. El Naschie (Eds.), Invited Review for the Spe- D 64 (2001) 123522, hep-th/0103239. cial Issue of J. Chaos, Solitons and Fractals, vol. 10, Elsevier, [8] D. Langlois, Prog. Theor. Phys. Suppl. 148 (2003) 181, hep- Amsterdam, 1999, p. 345, hep-th/9805120; th/0209261. J.R. Ellis, N.E. Mavromatos, D.V. Nanopoulos, Phys. Rev. [9] Supernova Search Team Collaboration, B.P. Schmidt, et al., D 63 (2001) 024024, gr-qc/0007044. Astrophys. J. 507 (1998) 46, astro-ph/9805200; [15] E. Gravanis, N.E. Mavromatos, Phys. Lett. B 547 (2002) 117, Supernova Cosmology Project Collaboration, S. Perlmutter, et hep-th/0205298; al., Astrophys. J. 517 (1999) 565, astro-ph/9812133; N.E. Mavromatos, Beyond the Desert, in: H.V. Klapdor- Supernova Search Team Collaboration, J.P. Blakeslee, et al., Kleingrothaus (Ed.), Oulu, Finland, 2002, Institute of Physics, Astrophys. J. 589 (2003) 693, astro-ph/0302402; 2003, p. 3, hep-th/0210079. Supernova Search Team Collaboration, A.G. Riess, et al., As- [16] J. Ellis, N.E. Mavromatos, D.V. Nanopoulos, A. Sakharov, trophys. J. 560 (2001) 49, astro-ph/0104455. New J. Phys. 6 (2004) 171, gr-qc/0407089. [10] WMAP Collaboration, D.N. Spergel, et al., Astrophys. J. [17] G.A. Diamandis, B.C. Georgalas, N.E. Mavromatos, E. Pa- Suppl. 148 (2003) 175, astro-ph/0302209. pantonopoulos, Int. J. Mod. Phys. A 17 (2002) 4567, hep- [11] W. Fischler, L. Susskind, Phys. Lett. B 171 (1986) 383; th/0203241; W. Fischler, L. Susskind, Phys. Lett. B 173 (1986) 262. G.A. Diamandis, B.C. Georgalas, N.E. Mavromatos, E. Pa- [12] I. Antoniadis, C. Bachas, J.R. Ellis, D.V. Nanopoulos, Nucl. pantonopoulos, I. Pappa, Int. J. Mod. Phys. A 17 (2002) 2241, Phys. B 328 (1989) 117; hep-th/0107124. I. Antoniadis, C. Bachas, J.R. Ellis, D.V. Nanopoulos, Phys. [18] I. Klebanov, A.A. Tseytlin, Nucl. Phys. B 546 (1999) 155; Lett. B 257 (1991) 278. I. Klebanov, A.A. Tseytlin, Nucl. Phys. B 547 (1999) 143. [13] F. David, Mod. Phys. Lett. A 3 (1988) 1651; [19] G. Curci, G. Paffuti, Nucl. Phys. B 286 (1987) 399. J. Distler, H. Kawai, Nucl. Phys. B 321 (1989) 509; [20] S.M. Carroll, Living Rev. Relativ. 4 (2001) 1, astro- J. Distler, Z. Hlousek, H. Kawai, Int. J. Mod. Phys. A 5 (1990) ph/0004075. 391; See also: N.E. Mavromatos, J.L. Miramontes, Mod. Phys. Lett. A 4 (1989) 1847; E. D’Hoker, P.S. Kurzepa, Mod. Phys. Lett. A 5 (1990) 1411. Physics Letters B 619 (2005) 26–29 www.elsevier.com/locate/physletb Effective cosmological constant from supergravity arguments and non-minimal coupling Remi Ahmad El-Nabulsi Plasma Application Laboratory, Department of Nuclear and Energy Engineering, and Faculty of Mechanical, Energy and Production Engineering, Cheju National University, Ara-dong 1, Jeju 690-756, South Korea Received 15 April 2005; accepted 1 June 2005 Available online 13 June 2005 Editor: W. Haxton Abstract It was shown that in the case of non-minimal coupling between the scalar curvature and the density of the scalar field, and for a particular scalar potential field inspired from supergravities inflation theories, ultra-light masses are implemented naturally in Einstein field equations, leading to a cosmological constant in accord with recent astrophysical observations. 2005 Published by Elsevier B.V. It is well believed today that the cosmological approaching de Sitter (dS) regime. In most of the mod- constant describes the energy density of the vacuum els of dark energy it is assumed that the cosmological (empty space), and it is a potentially important contrib- constant is equal to zero and the potential energy V(φ) utor to the dynamical history of the Universe. Recent of the scalar field driving the present stage of acceler- observations of type Ia supernovae and the CMB indi- ation, slowly decreases and eventually vanishes as the cates that the Universe is in accelerated regime [1–3]. field rolls to φ =∞[4–8]. In this case, after a tran- The total energy of the universe consists in fact of or- sient dS-like stage, the speed of expansion of the Uni- dinary matter and dark matter. One can interpret the verse decreases, and the Universe reaches Minkowski dark energy as the vacuum energy corresponding to regime. Of course, depending on the choice of the the cosmological Einstein constant or as the slowly model, the flat Universe will become dS space, or changing energy of a certain scalar field with a vac- Minkowski space, or collapse [9–12]. uum φ corresponding to the equation of state p =−ρ. However, it was found that one can describe dark In both cases the Universe is accelerated with time and energy in some extended supergravities that have a dS solutions [13,14]. These dS solutions correspond to the extrema of the effective potentials V(φ)for some scalar fields φ. An interesting result of these solutions E-mail addresses: [email protected], [email protected] (R.A. El-Nabulsi). is that the squared mass of these scalars in all theo- 0370-2693/$ – see front matter 2005 Published by Elsevier B.V. doi:10.1016/j.physletb.2005.06.002 R.A. El-Nabulsi / Physics Letters B 619 (2005) 26–29 27 √ ries with N = 2(extended supergravities with unstable We set R˜ = gR in what follows where g is the dS vacua) is quantized in units of the Hubble constant metric scalar. The stress constrained tensor in the non- 2 = 2 H0. That is m nH0 where n are integers of order minimal coupling is then of unity (in units of unity Planck Mass). ˜ ∗ − 2ξ δ(Rφφ ) In extended supergravities with a positive cosmo- T n c =−√ 2 = µν g δgµν logical constant, one always has 3m nΛ where Λ being the cosmological constant. For the N = 8 = µ ∗ − ∗ supergravity, dS vacuum corresponds to an unsta- 2ξ gµν∂µ∂ φφ Dν∂µφφ ble maximum m2 =−6H 2 at |φ|1 and V(φ)= 0 1 2 − 2 = − ∗ − 3H0 (1 φ ). Meanwhile for N 2 gauged super- φφ Rµν gµνR , (1) 2 = 2 2 gravity with stable dS vacuum, one has m 6H0 for = 2 − one of the scalars and in this case V(φ) 3H0 (1 where Dν is the covariant derivative. In this non- φ2) [15–18]. minimal coupling, the field equations read In application to the cosmological constant prob- 1 − lem, this leads to the conclusion that there are ultra R − g R + Λg + T n c + T M + tm = 0, (2) µν 2 µν µν µν µν µν light scalars with the mass of the order m ≈ H ≈ −33 M m 10 eV. The significance of this fact and the pos- where Tµν is the matter stress-energy tensor and tµν sibility to use these supergravity models in modern is the microscopic gauge field stress energy tensor de- cosmology still have to be well studied and under- fined in our model as stood. The existence of such ultra light fields may be ˜ 2 δ(Lm) a desirable feature for the description of the acceler- tm = √ , (3) µν g δgµν ated universe. Their presence signals that the corre- sponding potentials are very shallow. In extended su- where pergravity theories ultra light fields necessarily come √ ˜ 1 µν ∗ ∗ ∗ in a package with too small Λ. Due to the presence Lm = g g ∂µφ ∂νφ + ∂µφ∂νφ − V φφ . 2 of Λ, supersymmetry in dS vacua is broken sponta- neously, the scale of SUSY breaking here is 10−3 eV. (4) Before it is coupled to a ‘visible sector’, both the tiny The trace of (3) is then Λ as well as the ultra light masses of scalars, that ∗ ∗ tm =−2∂ φ ∂ φ + 4V φ φ . (5) is m, are protected from large quantum corrections. µν µ µ Coupling of these theories to real universe is a big We now introduce our quartic potential as [21–28] problem, of course. If they play a role of a ‘hidden ∗ = 2 − 2 ∗2 sector’, one may ask whether the tiny m ≈ H will be V φφ pm 1 ωφ φ , (6) preserved after coupling to the ‘visible sector’. The p = 3/4 for cosmological considerations [29], m ∝ H preservation of the small Λ may imply preservation (the Hubble parameter), the constant or false vacuum of small scalar masses m. Thus, extended supergrav- energy which is the leading term in the potential, and ities suggest a new perspective for investigation of ω is a positive parameter less than unity. the cosmological constant problem, intertwined with The trace of (2) gives ultra-light scalars [19]. From here came our motiva- ∗ ∗ tions. −R + 4Λ + 4V − 2∂µφ ∂µφ + 2ξRφ φ In this Letter, for some scalar field φ, we introduce ∗ µ 2 ∗ + 12ξ∂µφ ∂ φ − 12ξ Rφ φ a non-minimal coupling between the scalar curvature − ∗ ∂V + ∂V = and the density√ of the scalar field in the following form 6ξ φ ∗ φ 0. (7) L =−ξ gRφ∗φ, ξ = 1/6. R is the scalar curvature ∂φ ∂φ and φ∗ is the complex conjugate of φ [20].Froma For conformal coupling, that is ξ = 1/6, implanting view point of quantum field theory in curved space- Eq. (6) into (7), reduces to R = 4Λ − 3m2. time, it is natural to consider such a non-minimal cou- In this particular case and when ω 1, a possible pling. candidate field equations for the scalar curvature R, 28 R.A. El-Nabulsi / Physics Letters B 619 (2005) 26–29 = = 2 will be that in the flat case, for ρ ρc 3H0 /8πG, one has Λ = 3(H 2 − m2)<3H 2 as required by inflation. 1 ¯ 0 0 Rµν − gµνR + Λgµν = 0, (8) ≈ 2 ≈ 2 2 From recent observations, Λ 2H0 so that Λ 6m . We mention that in N = 2 gauged supergravities with ¯ = − 3m2 where Λ Λ 4 . The trace of Eq. (8) in four- stable de Sitter vacua, the effective potential near its = − dimensional spacetime leads certainly to R 4Λ extremum can be represented by the quadratic poten- 2 3m . tial V(φ)= Λ + m2φ2/2. However, in extended su- This case corresponds to a static spacetime. An- pergravities with positive cosmological constant, one other interesting possible candidate, is the dynamical 2 = = 2 | | = has 3m nΛ. When Λ 3H0 and φ 1, V(φ) one, which is Λ(1 + n2φ2/6). In both theories, the universalities of the relation − 1 + =− 2 Rµν gµνR Λgµν 3m uµuν (9) 2 = 2 2 2 2 m nH0 may be attributed to the fact that m /H0 is an eigenvalue of the Casimir operator of the dS group with R = 4Λ − 3m2. In general, this equation is iden- [8–10]. In this case, the ultra-light masses are imple- tical to the Einstein field equations mented naturally in Einstein field equations. This later, 1 as we have seen, could play an important role in cos- R − g R =−8πG (p + ρ)u u + pg (10) µν 2 µν µ ν µν mology. Finally, if we introduce a second mass scale M such with p ≡ p = Λ/8πG and ρ ≡ ρ = (3m2 − Λ)/ √ Λ m that pm2 = M2, p2ω = 1/2, that is m2 = 2ωM2 8πG. Notice that if the cosmological is set equal (m M) and we let x = φφ∗, our potential described to zero, than the pressure vanishes and ρ ≡ ρ = m in Eq. (6) takes the simple form V = M2(1 − x2).The 3m2/8πG.IfΛ<3m2, both the density and the pres- potential has a minimum at x = 0 where it takes a pos- sure remains positive. If 2Λ = 3m2, than the pressure itive value M2. This later plays an important role in and the density are equal. While if 2Λ =−3m2 < 0, new inflation theory, in particular brane-world infla- than p =−ρ/3. tion [28]. Cosmology with this potential shares some We believe that ρ ≡ ρ = 3m2/8πG could play an m common features with the cosmology of the “inverse” important role in standard cosmology. If this latter is harmonic oscillator potential (cosmology with nega- implemented naturally in Einstein field equations in tive potential) [31]. the presence of the cosmological constant, that is It is interesting to investigate in the future how 1 much the presence of the ultra light masses in Einstein Rµν − gµνR + Λgµν 2 field equations will contribute in most of the well- 3m2 known cosmological and astrophysical models. =−8πG p + ρ + u u + pg . (11) 8πG µ ν µν These equations are the same of that of Einstein ones References but there is an additional energy density. In fact, in standard cosmology, the cosmological fluid is in fact [1] A.G. Riess, et al., Astrophys. J. 116 (1998) 1009. not unicomponent; instead mater and radiation in dis- [2] I. Zavadi, et al., Astrophys. J. 403 (1998) 483. equilibrium coexist in many elementary subvolumes of [3] P. Garnavich, Astrophys. J. 493 (1998) 53. the Universe. The new additional term could play the [4] A.D. Dogulov, in: G.W. Gibbons, S.W. Hawking, S. Siklos (Eds.), The Very Early Universe, Cambridge Univ. Press, Cam- role of the density of the dark matter. bridge, 1983. When applying Eq. (9) to the FRW cosmology, the [5] C. Wetterich, Nucl. Phys. B 302 (1998) 668. Friedman equation will be [30] [6] E.J. Copeland, A.R. Liddle, D. Wands, Phys. Rev. D 57 (1998) ˙ 4686. R2 k 8πGρ Λ [7] B. Ratra, P.J.E. Peebles, Phys. Rev. D 37 (1988) 3406. + = + + m2, (12) R2 R2 3 3 [8] C. Armendariz-Picon, V.Mukhanov, J.P. Steinhardt, Phys. Rev. Lett. 85 (2000) 4438. where R is the scale factor and k = (−1, 0, +1) corre- [9] R. Kallosh, A. Linde, S. Prokushkin, M. Shmakova, Phys. Rev. sponding to open, flat or closed spacetime. It is clear D 65 (2002) 105016. R.A. El-Nabulsi / Physics Letters B 619 (2005) 26–29 29 [10] A. Linde, JHEP 0111 (2001) 052. [22] A. Albrecht, P.J. Steinhardt, Phys. Rev. Lett. 48 (1982) 1220. [11] P.J. Steinhardt, N. Turok, Phys. Rev. D 65 (2002) 126003. [23] A.R. Liddle, D.H. Lyth, Phys. Rep. 231 (1993) 1. [12] G.N. Felder, A. Frolov, L. Kolman, A. Linde, Phys. Rev. D 66 [24] A. Linde, Phys. Rev. D 49 (1994) 748. (2002) 023507. [25] S. Mollerach, S. Matarrese, F. Lucchin, Phys. Rev. D 50 (1994) [13] R. Kallosh, hep-th/0205315. 4835. [14] C.M. Hull, N.P. Warner, Class. Quantum Grav. 5 (1988) 1517. [26] E.J. Copeland, A.R. Liddle, D.H. Lyth, E.D. Stewart, [15] M. Dine, W. Fischler, D. Nemeschansky, Phys. Lett. B 136 D. Wands, Phys. Rev. D 49 (1994) 6410. (1984) 169. [27] P.H. Frampton, T. Takahashi, Phys. Rev. D 70 (2004) 083530. [16] G.D. Coughlan, R. Holman, P. Ramond, G.G. Ross, Phys. Lett. [28] A. Lukas, D. Skinner, JHEP 0109 (2001) 020. B 140 (1984) 44. [29] A. Linde, Particle Physics and Inflationary Cosmology, Har- [17] M. Dine, L. Randall, S. Thomas, Nucl. Phys. B 458 (1996) 291. wood, Chur, 1990. [18] A. Linde, Phys. Rev. D 53 (1996) 4129. [30] S. Weinberg, Gravitation and Cosmology, Wiley, New York, [19] S.M. Carroll, Phys. Rev. Lett. 81 (1998) 3067, astro- 1972. ph/9806099. [31] G. Felder, A. Frolov, L. Kofman, A. Linde, Phys. Rev. D 66 [20] E. Elbaz, Cosmologie, Ellipses, 1992. (2002) 023507. [21] A.D. Linde, Phys. Lett. B 108 (1982) 389. Physics Letters B 619 (2005) 30–42 www.elsevier.com/locate/physletb The effects of unstable particles on light-element abundances: Lithium versus deuterium and 3He John Ellis a, Keith A. Olive b, Elisabeth Vangioni c a TH Division, Physics Department, CERN, CH-1211 Geneva 23, Switzerland b Theoretical Physics Institute, University of Minnesota, Minneapolis, MN 55455, USA c Institut d’Astrophysique de Paris, F-75014 Paris, France Received 7 March 2005; received in revised form 21 May 2005; accepted 26 May 2005 Available online 6 June 2005 Editor: G.F. Giudice Abstract We reconsider the effects of the radiation from the decays of unstable particles on the production and destruction of the primordial light elements, with a view to reconciling the high primordial 7Li abundance deduced from big bang nucleosynthe- sis (BBN), as implied by the baryon-to-photon ratio now inferred from the anisotropies of the cosmic microwave background (CMB), with the lower abundance of 7Li observed in halo stars. The potential destruction of 7Li is strongly constrained by observations of deuterium (D), 3He and 6Li. We identify ranges for the unstable particle abundance and lifetime which would deplete 7Li while remaining consistent with the abundance of 6Li. However, in these regions either the D abundance is un- acceptably low or the ratio 3He/D is unacceptably large. We conclude that late particle decay is unable to explain both the discrepancy of the calculated 7Li abundance and the observed 7Li plateau. In the context of supersymmetric theories with neutralino or gravitino dark matter, we display the corresponding light-element constraints on the model parameters. 2005 Elsevier B.V. All rights reserved. 1. Introduction to-photon ratio η [2]. Recent high-precision measure- ments of the cosmic microwave background (CMB) The observed abundances of light elements are radiation by WMAP and other experiments now com- generally in good agreement with the predictions of plement the BBN in important ways [3]. For example, big bang nucleosynthesis (BBN) calculated assum- they impose strong constraints on η and weaker con- ing a homogeneous Robertson–Walker–Friedman cos- straints on the number of light particle species. Of mology [1]. Within this framework, the success of particular interest are the very precise predictions of BBN calculations imposes important constraints on the light element abundances from BBN that are made the number of light particle species and on the baryon- with the CMB value of η [4]. Based on these predictions, there is now tension E-mail address: [email protected] (K.A. Olive). between some observed light-element abundances and 0370-2693/$ – see front matter 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.physletb.2005.05.066 J. Ellis et al. / Physics Letters B 619 (2005) 30–42 31 those that would be calculated using the CMB value in which heavy particles have late decays into radi- of η and assuming no additional light particles beyond ation, which excludes the region of parameter space three light neutrino species. In particular, the predic- where they could have the desired impact on the pri- tion for the primordial abundance of 7Li made using mordial abundance of 7Li.1 We display this and other the CMB value of η [5–9] is somewhat higher (by a constraints on the lifetime and abundance of any mas- factor of 2–3) than the primordial abundance inferred sive unstable relic particle such as a gravitino, pointing from astrophysical observations. The significance of out also the potential impact of weakening the lower this discrepancy should not be over-emphasized, in limit on the primordial deuterium abundance. We also view of the potential systematic errors in the inter- display the effects of the 3He constraint on the parame- pretation of the astrophysical data. However, it has ter spaces of models in which heavier supersymmetric stimulated theoretical explorations of mechanisms for sparticles decay into gravitinos. The overall reductions modifying the CMB/BBN prediction for 7Li, for ex- in the allowed parameter spaces are often small, but ample, via the late decays of massive particles [10,11]. they do exclude the regions where the abundance of The effects of such decays have been studied exten- 7Li could be brought into line with the astrophysi- sively, and the constraints imposed on them by the ob- cal observations. This may motivate a reassessment of served abundances of 4He, deuterium and 6Li are well their interpretation. understood [12–17]. Previous studies had shown no in- compatibilities between these constraints and the sug- gestion that late-decaying particles might have modi- 2. Is there a problem with lithium-7? fied the BBN prediction for the abundance of 7Li so as to agree better with the astrophysical observations. The most direct and accurate estimate of the baryon- Supersymmetric models with conserved R parity nat- to-photon ratio η is currently provided by the acoustic urally predict such a particle, either the gravitino if structures in the CMB perturbations [3], namely, it is not the lightest supersymmetric particle (LSP), −10 or a neutralino or stau slepton if the gravitino is the η = 6.14 ± 0.25 × 10 . (1) LSP [18,19], which may well have the appropriate 6 7 This range may be used an input into homogeneous abundance and lifetime to affect both the Li and Li BBN calculations [5–9], yielding the following abun- abundances. Later in this Letter, we identify these re- dances for the elements of principal interest, which are gions in the parameter space of a constrained version taken from [9]: of the minimal supersymmetric extension of the Stan- dard Model (CMSSM), as well as in very constrained Yp = 0.2485 ± 0.0005, versions of the model [20] motivated by supergravity D +0.21 −5 considerations, which predict a gravitino LSP in parts = 2.55− × 10 , H 0.20 of the parameter space. 3 He − However, we also point out in this Letter that such = 1.01 ± 0.07 × 10 5, scenarios typically yield an abundance of 3He that may H 7 be more than an order of magnitude larger than the Li + − = 4.26 0.73 × 10 10, deuterium abundance. The importance of 3He produc- H −0.60 3 6 tion, and in particular the ratio of He to D, was first Li − = 1.3 ± 0.1 × 10 14, (2) pointed out in this context in [16], and this constraint H has subsequently been used in [14,17]. Since it appears 4 that the abundance of 3He has remained relatively con- where Yp is the He mass fraction, and the other abun- stant in time, whereas deuterium (D) would have been dances are expressed in terms of their numbers relative destroyed in stars, it seems unlikely that the ratio of to H, as shown. the 3He to D abundances could have been significantly These abundances may then be compared with the larger than it present value in the early history of the abundances of the same elements that are inferred Universe [14,16,17]. Imposing this constraint yields a further restriction on the parameter space of models 1 We comment later on the potential impact of hadronic decays. 32 J. Ellis et al. / Physics Letters B 619 (2005) 30–42 from those observed in the most primitive astrophysi- The comparison of (2) and (3) also shows a dis- cal sites [21–25]: crepancy for 6Li. It has generally been assumed that 6Li is produced in post-BBN processes such as galac- Yp = 0.232 to 0.258, tic cosmic-ray nucleosynthesis [33]. Until recently, the abundance of 6Li had been observed only in a few D −5 = 2.78 ± 0.29 × 10 , metal-poor halo stars with metallicity [Fe/H] larger H − 3 than 2.3. However, new observations of this isotope He − = 1.5 ± 0.5 × 10 5, have now been obtained in halo stars. New values of H the ratio 6Li/7Li have been measured with UVES at 7 Li + − the VLT-UT2 Kueyen ESO telescope, in halo stars = 1.23 0.68 × 10 10, H −0.32 with metallicity ranging from −2.7to−0.5 [24,25]. 6 These observations indicate the presence of a plateau Li + − = 6 7 × 10 12. (3) in 6Li/H 10−11, suggesting a pregalactic origin for H −3 the formation of 6Li [34]. These data provide interest- Note that, for 3He, we have at our disposal only lo- ing new constraints in the present context. cal data coming from star-forming HII regions in the The relation of the observed 7Li abundance to its galactic disk [26] or from the proto-solar value [27]. primordial value may be debated, but for the moment Comparing the two sets of abundances, (2) and (3), we take the discrepancy between the observed and we see no significant discrepancies, except in the case calculated 7Li abundances at face value, and explore of 7Li and 6Li. However, whereas 7Li has to be de- its possible theoretical interpretation in terms of late- stroyed, one must produce a factor 1000 more 6Li. decaying massive particles, depleting 7Li and possi- The value quoted above for the 7Li abundance bly producing 6Li, without negative effects on D and assumes that Li depletion is negligible in the stars 3He. observed. Indeed, standard stellar evolution models predict Li depletion factors which are very small (less than 0.05 dex) in very metal-poor turnoff stars [28]. 3. Possible impact of late-decaying particles However, there is no reason to believe that such simple models incorporate all effects which could The decays of massive particles X with lifetimes lead to depletion, such as rotationally-induced mixing > 102 s could, in principle, have modified the BBN and/or diffusion. Including these effects, current esti- predictions in either of two ways. First, their decay mates for possible depletion factors are in the range products would have increased the entropy in the pri- ∼ 0.2–0.4dex[29]. However, the data sample [30] mordial plasma, implying that its value during BBN used in deriving the abundance in (3) shows a negli- was lower than that inferred from the CMB. How- gible intrinsic spread in Li, leading to the conclusion ever, this effect is negligible compared with the second that depletion in these stars is in fact quite low (less effect, which is the modification of the BBN light- than 0.1 dex). element abundances by the interactions of decay prod- Another important source of potential systematic ucts [12,13]. uncertainty is related to the assumed surface tempera- The latter possibility has been explored recently as ture of the star. A recent study [31] found significantly a mechanism for reducing the primordial 7Li abun- and systematically higher temperatures for stars used dance [10,11]. The existence of such late-decaying in 7Li observations, specifically at low metallicity. massive particles X is a generic possibility in super- This result leads to 7Li/H = (2.34 ± 0.32) × 10−10, symmetric models, in particular [18,19]. Examples in- which is still, however, nearly a factor of 2 smaller clude a massive gravitino weighing ∼ 100 GeV, if than the BBN/WMAP prediction. We note, finally, that it is not the lightest supersymmetric particle (LSP), another potential source for theoretical uncertainty lies or some other next-to-lightest supersymmetric parti- in the BBN calculation of the 7Li abundance. How- cle (NSP) if the gravitino is the LSP. Cosmologi- ever, this too has been shown to be incapable of re- cal constraints on such scenarios have been explored solving the 7Li discrepancy [6,32]. previously [12,13,15,17]. However, the potential 7Li J. Ellis et al. / Physics Letters B 619 (2005) 30–42 33 problem motivates a re-examination of the astrophys- 3.1. The D and 4He abundances ical and cosmological constraints on such scenarios. In particular, we wish to determine whether or not a Fig. 1 shows the current constraints in the (τX,ζX) possible solution to this problem can be found in the plane, updating those shown in [15]. The green lines context of motivated and well-studied supersymmetric are the contours models. − D − Specifically, we apply the results of [15], which are (1.3or2.2) × 10 5 < < 5.3 × 10 5. (4) H based on the decays into radiation of a massive par- ticle with a lifetime longer than 104 s. For a given The first of the lower bounds is the higher line to lifetime, τX, the observed abundances provide us with the left of the cleft, and represents the very conserv- constraints on the abundance of relic particles, or more ative lower limit on D/H assumed in [15]. The range × −5 precisely on the quantity ζX ≡ mXnX/nγ , where nγ 1.3–5.3 10 effectively brackets all recent observa- + − is the density of photons after e e annihilation, mX tions of D/H in quasar absorption systems. The second is the mass of the decaying particle, and ζX is re- of the lower bounds is the lower line on the left side, lated to the density of relic particles prior to decay by and represents what now seems a reasonable lower 0 2 = × 7 −1 bound, which is obtained from the 2σ lower limit in ΩXh 3.9 10 GeV ζX. Our treatment of the electromagnetic shower produced by the decay was (3). The upper bound in (4) is the line to the right of the described in detail in [19]. Specifically, we include the cleft, and is the same upper limit as was used in [15]. decays of the NSP into a gravitino and either a pho- A priori, there is also a narrow strip at larger ζX and ton, Z, or a Higgs boson. In the case that the stau τX where the D/H ratio also falls within the range (4), 4 is the NSP, the dominant decay mode is to a grav- but this is excluded by the observed He abundance. itino and tau lepton. Explicit expressions for the de- The solid red lines in the upper right part of Fig. 1 cays widths for each channel considered were given correspond to the limits in [19]. The other NSP decay modes listed above in- Y > 0.227 or 0.232, (5) ject electrons, muons and hadrons into the primordial p medium, as well as photons. Electromagnetic show- ers develop similarly, whether they are initiated by electrons or photons, so we can apply the analysis of [15] directly also to electrons. As in [19], we treat the decays of µ, π and K as if their energies were equipartitioned among their decay products. In this approximation, we estimate that the fractions of par- ticle energies appearing in electromagnetic showers are π 0: 100%, µ:1/3, π ±:1/4, K±:0.3,K0:0.5. Using the measured decay branching ratios of the τ , we then estimate that ∼ 0.3 of its energy also appears in electromagnetic showers. In the case of generic hadronic showers from Z or Higgs decay, we esti- mate that ∼ 0.6 of the energy is electromagnetic, due mainly to π 0 and π ± production. We note that we have not included the effects of hadronic decays which are definitely important for lifetimes shorter than 104 s. The effects of hadronic decays on element abundances have been discussed recently in [10,17], where it was found that, for lifetimes around 103 s and densities of Fig. 1. The constraints imposed by the astrophysical observations of 4He (red lines), D/H (green lines), 6Li (yellow line), 6Li/7Li (blue ∼ −12 7 ζX 10 , some destruction of Li and production lines), 7Li (blue band) and 3He (black lines). (For interpretation of 6 of Li is possible. We comment below on this possi- the references to colour in all figures legends, the reader is referred bility. to the web version of this Letter.) 34 J. Ellis et al. / Physics Letters B 619 (2005) 30–42 where the lower number (corresponding to the higher high as 0.15, and we display that upper limit here. The line) was used in [15], and the higher number (corre- main new effect of this constraint is to disallow a re- sponding to the lower line) is a lower limit that has gion in the near-vertical cleft between the upper and been advocated recently [21]. It is apparent that, for lower limits on D/H, as seen in Fig. 1. our purposes, the third significant figure in the 4He abundance is unimportant: the narrow D/H strip is 3.3. The 7Li abundance in any case excluded, and there are always stronger bounds on ζX at large τX. The main region of interest in Fig. 1 is the blue shaded band that represents the inferred 7Li abun- 3.2. The 6Li abundance dance: 7 − Li − As said above, recent observations of 6Li in halo 0.9 × 10 10 < <(2or3) × 10 10, (7) stars have provided new insight into the origin and the H 6 7 evolution of this isotope. We recall that Li is a pure with the Li abundance decreasing as ζX increases and product of spallation, and many studies have followed the intensity of the shading changing at the interme- the evolution of 6Li in our Galaxy [33]. Of particular diate value. In [15], only the lower bound was used importance in this context is the α + α reaction that due the existing discrepancy between the primordial leads to the synthesis of this isotope as well as 7Li, and observationally determined values. It is apparent and is efficient very early in the evolutionary history that 7Li abundances in the lower part of the range (7) of the Galaxy. The new values of 6Li/7Li that have are possible only high in the deuterium cleft, and even been measured in halo stars with UVES at the VLT- then only if one uses the recent and more relaxed limit UT2 Kueyen ESO telescope indicate the presence of on the 6Li/7Li ratio (6). Values of the 7Li abundance a plateau in 6Li, which suggests a pregalactic origin in the upper part of the range (7) are possible, how- for the formation of this isotope. The evolution of 6Li ever, even if one uses the more stringent constraint on with redshift was calculated [34] following an initial 6Li/7Li. In this case, the allowed region of parame- burst of cosmological cosmic rays up to the formation ter space would also extend to lower τX, if one could − of the Galaxy. This process is able to produce the re- tolerate values of D/H between 1.3 and 2.2 × 10 5. quired abundance of 6Li observed in metal-poor halo For the convenience of the subsequent discussion, 7 stars without the additional over-production of Li. In the region of the (τX,ζX) plane that is of interest for this Letter, we have to consider the new constraint lowering the 7Li abundance is shown alone in panel (a) brought by the existence of this plateau. The poten- of Fig. 2. The blue region in the arc at low τX is the tial destruction of 7Li by unstable particles must not region excluded by the stronger lower limit on the deu- − lead to over-production of 6Li. terium abundance: D/H > 2.2 × 10 5, and the red The constraint imposed by the 6Li abundance is region to its right is the extra domain that is excluded shown as a solid yellow line in Fig. 1, which is the by the 3He/D ratio, as we discuss below. same as that discussed in [15]. Also shown, as solid blue lines, are two contours representing possible up- per limits on the 6Li/7Li ratio: 4. The importance of the 3He abundance 6 Li 3 < 0.07 or 0.15, (6) We now come to the constraint from He, which 7Li was previously discussed in [14,16,17]. We find that with the upper (lower) contour corresponding to the the 3He/D ratio is absurdly high in the deuterium cleft. upper (lower) number in (6). The lower number was Panel (b) of Fig. 2 shows a histogram of the values of used in [15] and represented the upper limit available the 3He/D ratio found in a dense sample of scenar- at the time, which was essentially based on multiple ios in the interesting regions shown in panel (a). Since observations of a single star. The most recent data [24, deuterium is more fragile than 3He, whose abundance 25] includes observations of several stars. The Li iso- is thought to have remained roughly constant since tope ratio for most metal-poor stars in the sample is as primordial nucleosynthesis when comparing the BBN J. Ellis et al. / Physics Letters B 619 (2005) 30–42 35 7 Fig. 2. (a) The region of the (τX,ζX) plane in which a decaying relic particle could have the desired impact on the Li abundance. To derive the blue (darker grey) region, the only abundance cuts applied are: 0.9 < 7Li/H × 1010 < 3.0, 1.3 < D/H × 105 < 5.3, and 6Li/7Li < 0.15. − To obtain the red (lighter grey) region, the lower bound on D/H was increased to 2.2 × 10 5. (b) A histogram of the 3He/D ratios found in scenarios sampling the region displayed in panel (a), with similar colour coding. value to it proto-solar abundance, one would expect, It is interesting to note that in the red region in principle, the 3He/D ratio to have been increased which has acceptable D/H and a 7Li/H abundance by stellar processing. Indeed, there is considerable un- low enough to match the observed values, the 6Li certainty in the evolution of 3He [35]. This uncertainty abundance is relatively high: 7.3 × 10−12 < 6Li/H < is largely associated with the degree to which 3He is 1.6 × 10−11. This matches the new 6Li observations produced or destroyed in stars. Since D is totally de- quite well, and would circumvent the need for an early stroyed in stars, the ratio of 3He/D can only increase period of 6Li production by cosmological cosmic rays. in time or remain constant if 3He is also completely Unfortunately, 3He/D ranges from 17–37 for these pa- destroyed in stars. The present or proto-solar value of rameter values. 3 He/D can therefore be used to set an upper limit on The previous upper limit on ηX [15] corresponded −12 the primordial value. Fig. 1 displays the upper limits to the constraint mXnX/nγ < 5.0 × 10 GeV for 8 3 τX = 10 s. The weaker (stronger) version of the He constraint adopted here corresponds to 3He < 1or2 (8) D n − m X < 2.0(0.8) × 10 12 GeV, (9) X n as solid black lines. Above these contours, the value γ 3 of He/D increases very rapidly, and points high in 8 3 for τX = 10 s. However, the impact of the He con- the deuterium cleft of Fig. 1 have absurdly high val- = 7 3 straint is even stronger for τX 10 s, the location ues of He/D, exceeding the limit (8) by an order of the previous Deuterium cleft. The analysis of [15] of magnitude or more. These are the red points pro- −12 would have given mXnX/nγ < 360 × 10 GeV, ducing the high-end peak of the histogram shown in whereas the weaker (stronger) 3He constraint adopted panel (b) of Fig. 2, whereas the blue points are those here corresponds to excluded by the lower limit D/H > 2.2 × 10−5 that is now preferred. We see that these points mostly have nX −12 3 mX < 9.3(3.8) × 10 GeV, (10) acceptably low values of He/D, though some large nγ values are found near the boundary with the red region 7 in panel (a) of Fig. 2. for τX = 10 s. 36 J. Ellis et al. / Physics Letters B 619 (2005) 30–42 5. Applications to supersymmetric scenarios supersymmetry-breaking scalar masses m0 and gaug- ino masses m1/2 at the GUT scale before renormaliza- We now discuss some examples of the conse- tion. The magnitude of the higgsino mixing parameter 3 quences of the He constraint for various supersym- |µ| and the pseudoscalar Higgs mass mA are fixed by metric scenarios in which R parity is conserved. In the electroweak vacuum conditions. We consider sce- such models, if the gravitino is not the LSP it will narios with µ>0, which are favoured by gµ − 2 and, generically decay gravitationally with a long lifetime. to a lesser extent, b → sγ. The scenarios that we study If the gravitino is the LSP, the next-to-lightest super- differ in their assumptions about the relationship of the symmetric particle (NSP) will decay gravitationally gravitino mass m3/2 to m0, but they all share the com- into the gravitino LSP, again with a long lifetime. mon feature that the LSP is the gravitino in generic domains of parameter space. 5.1. Models with an unstable gravitino In the first set of scenarios, shown in Fig. 3,we fix the ratio of supersymmetric Higgs vacuum expec- = We first consider the possibility that the gravitino is tation values tan β 10, which is among the lower not the lightest supersymmetric particle (LSP), which values consistent with our hypotheses, and assume = = = is instead the lightest neutralino χ. In this case, the (a) m3/2 10 GeV, (b) m3/2 100 GeV, (c) m3/2 = gravitino is unstable, with a lifetime that could well 0.2m0 and (d) m3/2 m0. In each panel of Fig. 3, fall within the range considered here. In such a sce- we display accelerator, astrophysical and cosmologi- nario, the light-element abundances impose an impor- cal constraints in the corresponding (m1/2,m0) planes, tant upper limit on the possible temperature of the concentrating on the regions to the right of the near- Universe, e.g., during reheating after inflation, which vertical black lines, where the gravitino is the LSP. The vertical black dashed and (red) dot-dashed lines we denote by TR [12,13,15,17]. We recall that ther- mal reactions are estimated to produce an abundance represent the lower limits on m1/2 implied for each of gravitinos given by [15]: value of m0 by the non-observation of a chargino and a Higgs boson at LEP, the latter having a theoretical nm − T uncertainty δm ∼ 50 GeV. The (pale green) nar- 3/2 = ( . . ) × 11 × R . 1/2 0 7–2 7 10 10 (11) nγ 10 GeV row diagonal strips represent the regions where the relic density of the NSP would have lain in the range 8 Assuming that m3/2 = 100 GeV and τX = 10 s, and 0.094 Ωh2 0.129 favoured by WMAP and other imposing the constraints (9), we now find measurements of the cold dark matter density, if the gravitino had not been the LSP. In fact, the gravitino is T <(0.8–2.8) × 107 GeV, R always the LSP in the scenarios considered. The NSP (0.3–1.1) × 107 GeV , (12) may be either the lightest neutralino χ or the lighter supersymmetric partner of the τ lepton: this is lighter 3 for the weaker (stronger) version of the He con- than the neutralino χ below the (red) dotted line. straint. This becomes an even more significant con- Below and to the right of the upper (purple) dashed straint on inflationary models, which were already lines, the density of relic gravitinos produced in the somewhat embarrassed by the previous upper limit decays of other supersymmetric particles is always ∼ × 7 2 TR 2 10 GeV. below the WMAP upper limit: Ω3/2h 0.129. To the right of the lower black solid lines, the lifetime 5.2. Models with gravitino dark matter of the next-to-lightest supersymmetric particle (NSP) falls below 104 s, and the analysis of [15] cannot We now consider the possibility that the gravitino is evaluate the astrophysical constraints from the light- itself the LSP, in which case the next-to-lightest super- element abundances, in the absence of a suitably modi- symmetric particle (NSP) would instead be unstable, fied BBN code. The code used in [15], when combined also with a long lifetime. We will work in the con- with the observational constraints used in [15], yielded text of the constrained MSSM (CMSSM) [36–38],so the astrophysical constraint represented by the dashed that all the scenarios we consider have universal soft grey-green lines in the different panels of Fig. 3. J. Ellis et al. / Physics Letters B 619 (2005) 30–42 37 Fig. 3. The (m1/2,m0) planes for µ>0, tanβ = 10 and (a) m3/2 = 10 GeV, (b) m3/2 = 100 GeV, (c) m3/2 = 0.2m0 and (d) m3/2 = m0.We restrict our attention to the regions between the solid black lines, where the gravitino is the LSP and the NSP lifetime exceeds 104 s. In each panel, the near-vertical dashed black (dash-dotted red) line is the constraint mχ± > 104 GeV (mh > 114 GeV), the upper (purple) dashed line 2 2 is the constraint Ω3/2h < 0.129, and the light green shaded region is that where the NSP would have had 0.094 Ωh 0.129 if it had not decayed. The solid red (dashed grey-green) line is the region now (previously) allowed by the light-element abundances: r<1 as described in the text. The red (blue) shaded region is that where the 7Li abundance could have been improved by NSP decays, but which is now excluded by the 3He (D) constraint. These constraints on the CMSSM parameter plane computed. Then for each τX, the limit on ζX is found were computed in [19]. For each point in the (m1/2, from the results shown in Fig. 1. The region to the right ˜ = limit m0), the relic density of either χ or τ is computed and of this curve where r ζX/ζX < 1 is allowed. 2 7 ζX is determined using ΩXh = 3.9 × 10 GeV ζX. The astrophysical constraints obtained with the When X =˜τ , ζX is reduced by a factor of 0.3, as only newer abundance limits used here yields the solid red 30% of stau decays result in electromagnetic showers lines in Fig. 3. The examples where τX and ζX for the which affect the element abundances at these lifetimes. NSP decays fall within the ranges shown in Fig. 2(a), In addition, at each point, the lifetime of the NSP is and hence are suitable for modifying the 7Li abun- 38 J. Ellis et al. / Physics Letters B 619 (2005) 30–42 dance, are shown as red and blue shaded regions in Fig. 4 shows the corresponding (m1/2,m0) planes each panel of Fig. 3. We see that they straddle the erst- for the choice tan β = 57, which is among the larger while WMAP strips.2 If we had been able to allow a values allowed in the context of the constrained deuterium abundance as low as D/H ∼ (1–2) × 10−5, MSSM. Although the shapes of the allowed regions the blue shaded region would have been able to resolve are rather different from the previous tan β = 10 case, the Li discrepancy in the context of the CMSSM with the qualitative conclusions are similar. The 3He con- gravitino dark matter. The blue region that we now re- straint again has relatively modest impact. However, gard as excluded by the lower limit on D/H, which is in all cases the red shaded 7Li regions are excluded 3 stronger than that used in [15], extends to large m1/2. by the He constraint, and the blue regions by the D The red shaded region, which is consistent even with abundance. We note that, in models with m3/2 = m0 3 this limit on D/H, but yields very large He/D, is close as in Fig. 4(d), very little of the (m1/2,m0) plane ad- to the Higgs lower limit on m1/2 for small m3/2,mov- mits gravitino dark matter and, in the viable corner, 7 ing to larger m1/2 for larger m3/2,soastokeepτNSP there were no possibilities for depleting Li. within the desired range. Finally, we consider very constrained models moti- We displayed in Fig. 1 the impact of the improved vated by minimal supergravity (mSUGRA), in which 3 lower limit on D/H and the new He constraint on not only is m3/2 = m0, but also the trilinear soft the abundance of an unstable particle, as a function supersymmetry-breaking parameter A determines the of its lifetime. Interpreting this as a constraint on NSP bilinear soft supersymmetry-breaking Higgs-mixing decay into a gravitino, the panels in Fig. 3 show as parameter: B = A − m3/2. This is compatible with solid red lines the additional restrictions these con- the values of µ and mA specified by the electroweak straints impose on the (m1/2,m0) planes for different vacuum conditions for only one value of tan β for values of m3/2. The effects for small m3/2 = 10 GeV any given pair of values of (m1/2,m0) [20]. Such [in panel (a)] m3/2 = 100 GeV when m0 is large [in mSUGRA models are then specified by a choice of ≡ ˆ ˆ = panel (b)] and 0.2m0 [in panel (c)] are relatively mod- A Am√ 3/2: the panels of Fig. 5 assume (a) A est. This is because the limit occurs in a region where (3 − 3) as found in the simple Polonyi model of the NSP is a neutralino, and the relic density varies supersymmetry breaking√ in mSUGRA [39], and (b) ˆ ˆ relatively rapidly. Hence a small change in the m1/2 or A = 2. For A = (3 − 3), the contours of constant m0 results in a large change in ζX, and the old and new tan β are approximately vertical, and range from about bounds are relatively close. However, they do bite in 10 at low m1/2 to about 30 at high values of m1/2.In the neighbourhood of the shaded 7Li blobs, and have the interesting region of panel (a) where 7Li can be the effect of excluding them entirely. The effects for depleted, tan β ∼ 20–30. As before, we consider here large m3/2 = 100 GeV and small m0 [in panel (b)] and only regions of the (m1/2,m0) planes between the two m3/2 = m0 [in panel (d)] are relatively large, mainly solid black lines: above the higher one, the gravitino is due to the slow variation of ζX near the limit which is no longer the LSP, and below the lower one the life- characteristic of the τ˜ NSP region, as can be seen from time falls below 104 s. In addition to the constraints the relatively wide WMAP strips in this region. Re- discussed earlier, panel (b) also displays a small green flecting this wide separation between the (old) dashed shaded region at low m1/2 that is excluded by b → sγ grey-green lines and the (new) solid red lines, we see decay. large red and blue swaths in panel (d), where the 7Li We see that there is a large difference between the abundance could have been reduced, but the 3He/D effects of implementing the old and new light-element√ and/or D/H ratios are unacceptable.3 constraints in panel (a) of Fig. 5 for Aˆ = (3 − 3), the Polonyi value, whereas the effect in panel (b) for Aˆ = 2 is smaller. In the Polonyi case, there are large 2 This raises the possibility that a discovery of supersymmetry 7Li-friendly regions that are excluded by the 3He and might have an ambiguous interpretation—neutralino LSP or grav- D constraints. This reflects the fact that tan β is rela- itino LSP—in the absence of supplementary information. tively small in this case, so the model is qualitatively 3 We note in passing that in several panels there are parts of the similar to the m = m case for tan β = 10 shown in blue and/or red regions at large m0 that are also excluded by the 3/2 0 WMAP relic density limit. panel (d) of Fig. 3. On the other hand, tan β is typically J. Ellis et al. / Physics Letters B 619 (2005) 30–42 39 Fig. 4. As in Fig. 3, but now for tan β = 57. ˆ larger for A = 2, and when combined with a smaller Fig. 4(a), (b) or (d). When m3/2 = 0.2m0,wedofinda 7 −12 stau mass, we find no visible Li-friendly region, as handful of cases, all with ζX between 4–7 × 10 and a result of small yet significant shifts in the values of all with m0 20 GeV and m1/2 1.7TeV.Amore both ζX and τX. complete treatment of the effects of hadronic decays As noted earlier, the 7Li abundance can be reduced will be given elsewhere. slightly by hadronic decays when the lifetime is ap- 3 ∼ −12 proximately 10 s with a density ζX 10 [10,17]. 6. Conclusions We have searched the parameter spaces of both the CMSSM and mSUGRA models discussed above. For In the absence of a convincing astrophysical ex- lifetimes between 300 and 3000 s, we find no models planation for the apparent discrepancy between the −13 −11 7 with ζX between 10 and 10 when tan β = 10, observed abundance of Li and that calculated on the or in any of the mSUGRA models. When tan β = 57, basis of the baryon-to-photon ratio inferred from CMB we also find no parameters for the models described in observations, it has been natural to explore the possi- 40 J. Ellis et al. / Physics Letters B 619 (2005) 30–42 Fig. 5. As in Fig. 3, but now for very constrained models motivated by mSUGRA. The value of tan β is fixed by the vacuum conditions, and ≡ ˆ varies across the√ (m1/2,m0) planes with values indicated by the steep black contours. These models are specified by the choices of A Am3/2: (a) Aˆ = (3 − 3), the Polonyi model, and (b) Aˆ = 2. ble effects of late-decaying massive particles. Indeed, growth of its abundance between BBN and the for- they could suppress the primordial abundance of 7Li, mation of halo stars could also have been explained but at some price. Either the abundance of deuterium by the decays of the NSP in supersymmetric mod- should be very low compared with the latest avail- els, were it not for either the low resulting abundance able measurements, and/or the primordial 3He/D ratio of D/H or the high ratio of 3He/D. We have also must have been very high. considered the possibility that radiation from decays The latest observations of remote cosmological may be responsible for the observed 6Li abundance clouds along the lines of sight of high-redshift quasars at low metallicity. We find that, indeed, for lifetimes −5 9 −12 suggest that D/H > 2.2 × 10 , ruling out much of longer than 3 × 10 s with ζX 1–3 × 10 , suffi- the parameter space for unstable particles that would cient 6Li production is possible. However, this range 7 otherwise have been suitable for diminishing the Li of τX and ζZ does not correspond to any range of abundance to agree with observations. A significant (m1/2,m0) in the supersymmetric models we have part of this parameter space would have been allowed considered. by the more relaxed limit D/H > 1.3 × 10−5 consid- So what is the interpretation of the apparent dis- ered previously [15]. crepancy between the calculations of the primordial The remaining part of the parameter space for un- 7Li abundance and the Spite plateau? We have ar- stable particles that is consistent with the current lower gued that the origin of the discrepancy cannot be the limit would yield a 3He/D ratio at least an order of possible existence of unstable particles able to de- magnitude higher than the proto-solar value. Since D stroy the primordial nucleus. This leaves the problem has been destroyed by stars, reducing its abundance open. Systematic uncertainties in nuclear effects such from approximately 2.5 to 1.5 ×10−5, while the 3He as higher 7Be + D reaction rates have been consid- abundance is thought to have remained roughly con- ered [6,32], but seem unable to modify substantially stant, a primordial ratio of 3He/D > 1 or 2 is unaccept- the abundance of 7Li. Stellar mechanisms of depletion able. This closes the remaining loophole for suppress- maybethelastresort[29]. Perhaps other new and ex- ing 7Li without running into conflict with the other citing astrophysical or physical effects will have to be light-element abundances. considered. We have also analyzed the potential bounds im- A significant output of this analysis has been the posed by the 6Li abundance. In principle, the high demonstration of the importance of the 3He/D con- J. Ellis et al. / Physics Letters B 619 (2005) 30–42 41 straint on late-decaying massive particles as argued in K.A. Olive, G. Steigman, T.P. Walker, Phys. Rep. 333 (2000) [14,16,17]. Can one make the 3He constraint more pre- 389; cise? This would require considering in more detail the B.D. Fields, S. Sarkar, Phys. Rev. D 66 (2002) 010001. 3 [2] G. Steigman, D.N. Schramm, J. Gunn, Phys. Lett. B 66 (1977) cosmic evolution of D and He. One should allow for 202; the possibility of exotic effects such as large-scale de- For a recent analysis see: R.H. Cyburt, B.D. Fields, K.A. Olive, struction of this isotope in primitive structures such E. Skillman, astro-ph/0408033. as massive Population-III stars, followed by moder- [3] C.L. Bennett, et al., Astrophys. J. Suppl. 148 (2003) 1, astro- ate production by normal galactic evolution. However, ph/0302207; D.N. Spergel, et al., Astrophys. J. Suppl. 148 (2003) 175, astro- any such scenario should consider simultaneously the ph/0302209. 3 cosmic evolution of D. As it is more fragile than He, [4] R.H. Cyburt, B.D. Fields, K.A. Olive, Phys. Lett. B 567 (2003) D would also be destroyed in any Population-III stars. 227, astro-ph/0302431. As we have seen, the 3He constraint sharpens the [5] R.H. Cyburt, B.D. Fields, K.A. Olive, New Astron. 6 (1996) embarrassment of supersymmetric models with heavy 215. ∼ 7 8 [6] A. Coc, E. Vangioni-Flam, P. Descouvemont, A. Adahchour, particles whose lifetimes are 10 to 10 s. This in- C. Angulo, Astrophys. J. 600 (2004) 544, astro-ph/0309480. terest motivates more detailed studies of the cosmic [7] P. Descouvemont, A. Adahchour, C. Angulo, A. Coc, E. evolution of 3He and D, as well as 6Li and 7Li. Vangioni-Flam, At. Data Nucl. Data Tables 88 (2004) 203, Finally, it would be very useful if the observed astro-ph/0407101. abundances of D/H in quasar absorption systems were [8] A. Cuoco, F. Iocco, G. Mangano, G. Miele, O. Pisanti, P.D. Ser- pico, Int. J. Mod. Phys. A 19 (2004) 4431, astro-ph/0307213. improved. There is currently considerable dispersion [9] R.H. Cyburt, Phys. Rev. D 70 (2004) 023505, astro- in the observed abundances. We note that two such ph/0401091. systems in the directions of Q2206-199 at z = 2.0762 [10] K. Jedamzik, Phys. Rev. D 70 (2004) 063524, astro- with D/H = 1.65 ± 0.35 × 10−5 [40] and PKS1937- ph/0402344; +0.25 −5 K. Jedamzik, Phys. Rev. D 70 (2004) 083510, astro- 1009 at z = 3.256, with D/H = 1.6− × 10 [41] 0.30 ph/0405583. have quite low abundances of D, similar to that ob- [11] J.L. Feng, A. Rajaraman, F. Takayama, Phys. Rev. D 68 (2003) served at present in the ISM of our Galaxy [42].If 063504, hep-ph/0306024. these measurements were to represent the correct D/H [12] D. Lindley, Astrophys. J. 294 (1985) 1; abundance in those clouds, 7Li depletion and 6Li pro- J.R. Ellis, D.V. Nanopoulos, S. Sarkar, Nucl. Phys. B 259 duction by sparticle decay would be a viable option, (1985) 175; J. Ellis, et al., Nucl. Phys. B 337 (1992) 399; though by solving one problem we would open two M. Kawasaki, T. Moroi, Prog. Theor. Phys. 93 (1995) 879, hep- new problems. Why is the D/H abundance in most of ph/9403364; the other absorption systems significantly higher, and M. Kawasaki, T. Moroi, Astrophys. J. 452 (1995) 506, astro- how can we account for the D/H abundances in the so- ph/9412055. lar system, which are also in the range 1.5–2.5×10−5? [13] D. Lindley, Phys. Lett. B 171 (1986) 235; M.H. Reno, D. Seckel, Phys. Rev. D 37 (1988) 3441; S. Dimopoulos, R. Esmailzadeh, L.J. Hall, G.D. Starkman, Nucl. Phys. B 311 (1989) 699; Acknowledgements K. Kohri, Phys. Rev. D 64 (2001) 043515, astro-ph/0103411. [14] E. Holtmann, M. Kawasaki, K. Kohri, T. Moroi, Phys. Rev. D 60 (1999) 023506, hep-ph/9805405; We thank R. Cyburt, B. Fields, Y. Santoso and M. Kawasaki, K. Kohri, T. Moroi, Phys. Rev. D 63 (2001) V. Spanos for collaborations on related topics. The 103502, hep-ph/0012279. work of K.A.O. and E.V. was supported by the Project [15] R.H. Cyburt, J.R. Ellis, B.D. Fields, K.A. Olive, Phys. Rev. “CNRS/USA”, and the work of K.A.O. was also sup- D 67 (2003) 103521, astro-ph/0211258. ported partly by DOE grant DE-FG02-94ER-40823. [16] G. Sigl, K. Jedamzik, D.N. Schramm, V.S. Berezinsky, Phys. Rev. D 52 (1995) 6682, astro-ph/9503094. [17] M. Kawasaki, K. Kohri, T. Moroi, astro-ph/0402490; M. Kawasaki, K. Kohri, T. Moroi, astro-ph/0408426. References [18] J.L. Feng, S. Su, F. Takayama, Phys. Rev. D 70 (2004) 063514, hep-ph/0404198; J.L. Feng, S. Su, F. Takayama, Phys. Rev. D 70 (2004) 075019, [1] T.P. Walker, G. Steigman, D.N. Schramm, K.A. Olive, K. Kang, Astrophys. J. 376 (1991) 51; hep-ph/0404231. 42 J. Ellis et al. / Physics Letters B 619 (2005) 30–42 [19] J.R. Ellis, K.A. Olive, Y. Santoso, V.C. Spanos, Phys. Lett. [36] J.R. Ellis, T. Falk, G. Ganis, K.A. Olive, M. Srednicki, Phys. B 588 (2004) 7, hep-ph/0312262. Lett. B 510 (2001) 236, hep-ph/0102098; [20] J.R. Ellis, K.A. Olive, Y. Santoso, V.C. Spanos, Phys. Lett. V.D. Barger, C. Kao, Phys. Lett. B 518 (2001) 117, hep- B 573 (2003) 162, hep-ph/0305212; ph/0106189; J.R. Ellis, K.A. Olive, Y.Santoso, V.C. Spanos, Phys. Rev. D 70 L. Roszkowski, R. Ruiz de Austri, T. Nihei, JHEP 0108 (2001) (2004) 055005, hep-ph/0405110. 024, hep-ph/0106334; [21] K.A. Olive, E.D. Skillman, Astrophys. J. 617 (2004) 29, astro- A.B. Lahanas, V.C. Spanos, Eur. Phys. J. C 23 (2002) 185, hep- ph/0405588. ph/0106345; [22] D. Kirkman, D. Tytler, N. Suzuki, J.M. O’Meara, D. Lubin, A. Djouadi, M. Drees, J.L. Kneur, JHEP 0108 (2001) 055, hep- Astrophys. J. Suppl. 149 (2003) 1, astro-ph/0302006. ph/0107316; [23] S.G. Ryan, T.C. Beers, K.A. Olive, B.D. Fields, J.E. Norris, U. Chattopadhyay, A. Corsetti, P. Nath, Phys. Rev. D 66 (2002) Astrophys. J. 530 (2000) L57, astro-ph/9905211. 035003, hep-ph/0201001; [24] M. Asplund, et al., 2004, in preparation. H. Baer, C. Balazs, A. Belyaev, J.K. Mizukoshi, X. Tata, Y. [25] D.L. Lambert, AIP Conf. Proc. 743 (2005) 206, astro- Wang, JHEP 0207 (2002) 050, hep-ph/0205325; ph/0410418. R. Arnowitt, B. Dutta, hep-ph/0211417; [26] T.M. Bania, R.T. Rood, D.S. Balser, Nature 415 (2002) 54. J.R. Ellis, K.A. Olive, Y. Santoso, V.C. Spanos, Phys. Lett. [27] J. Geiss, G. Gloeckler, Space Sci. Rev. 84 (1998) 239; B 573 (2003) 163, hep-ph/0308075. G. Gloeckler, J. Geiss, Space Sci. Rev. 84 (1998) 275. [37] J.R. Ellis, K.A. Olive, Y. Santoso, V.C. Spanos, Phys. Lett. [28] C.P. Deliyannis, P. Demarque, S.D. Kawaler, Astrophys. J. B 565 (2003) 176, hep-ph/0303043. Suppl. 73 (1990) 21. [38] H. Baer, C. Balazs, JCAP 0305 (2003) 006, hep-ph/0303114; [29] S. Vauclair, C. Charbonnel, Astrophys. J. 502 (1998) 372; A.B. Lahanas, D.V. Nanopoulos, Phys. Lett. B 568 (2003) 55, M.H. Pinsonneault, T.P. Walker, G. Steigman, V.K. Narayanan, hep-ph/0303130; Astrophys. J. 527 (1998) 180, astro-ph/9803073; U. Chattopadhyay, A. Corsetti, P. Nath, Phys. Rev. D 68 (2003) M.H. Pinsonneault, G. Steigman, T.P. Walker, V.K. Narayanan, 035005, hep-ph/0303201; Astrophys. J. 574 (2002) 398, astro-ph/0105439; C. Munoz, hep-ph/0309346; O. Richard, G. Michaud, J. Richer, Astron. Astrophys. 431 R. Arnowitt, B. Dutta, B. Hu, hep-ph/0310103. (2005) 1, astro-ph/0409672. [39] J. Polonyi, Budapest preprint KFKI-1977-93, 1977; [30] S.G. Ryan, J.E. Norris, T.C. Beers, Astrophys. J. 523 (1999) R. Barbieri, S. Ferrara, C.A. Savoy, Phys. Lett. B 119 (1982) 654, astro-ph/9903059. 343. [31] J. Melendez, I. Ramirez, Astrophys. J. 615 (2004) L33, astro- [40] M. Pettini, D.V. Bowen, Astrophys. J. 560 (2001) 41, astro- ph/0409383. ph/0104474. [32] R.H. Cyburt, B.D. Fields, K.A. Olive, Phys. Rev. D 69 (2004) [41] N.H.M. Crighton, J.K. Webb, A. Ortiz-Gill, A. Fernandez- 123519, astro-ph/0312629. Soto, Mon. Not. R. Astron. Soc. 355 (2004) 1042, astro- [33] B.D. Fields, K.A. Olive, New Astron. 4 (1999) 255; ph/0403512. E. Vangioni-Flam, M. Cassé, R. Cayrel, J. Audouze, M. Spite, [42] H.W. Moos, et al., Astrophys. J. Suppl. 140 (2002) 3, astro- F. Spite, New Astron. 4 (1999) 245. ph/0112519. [34] E. Rollinde, E. Vangioni-Flam, K.A. Olive, astro-ph/0412426. [35] E. Vangioni-Flam, K.A. Olive, B.D. Fields, M. Cassé, Astro- phys. J. 585 (2003) 611, astro-ph/0207583. Physics Letters B 619 (2005) 43–49 www.elsevier.com/locate/physletb CP violation in a light Higgs boson decay from τ-spin correlations at a linear collider André Rougé Laboratoire Leprince-Ringuet, Ecole Polytechnique-IN2P3/CNRS, F-91128 Palaiseau cedex, France Received 9 May 2005; received in revised form 30 May 2005; accepted 31 May 2005 Available online 6 June 2005 Editor: L. Rolandi Abstract + − We present a new method to measure the transverse spin correlation in the H → τ τ decay. The method has been devised to be insensitive to the beamstrahlung which affects the definition of the beam energy at a linear collider. In the case of two ± ± τ → π ν¯τ (ντ ) decays, using the anticipated detector performance of the TESLA project, we get a promising estimation of the error expected on the measurement of a CP violating phase. 2005 Elsevier B.V. All rights reserved. 1. Introduction is a CP =+1 state for ξ = 0, a CP =−1 state for ξ = π, and a mixed CP state otherwise. Such a state The possibility to determine the CP properties of a is produced by the decay of a CP =+1 Higgs with a light Higgs boson through the spin correlations in its coupling [5,9] H → τ +τ − decay has been often considered [1–9]. gτ(¯ cos ψ + i sin ψγ )τH. (2) The principle is simple. Let ± denote the projection of 5 O 2 2 the spins of the τ ’s in their respective rest frames on a In this case, neglecting (mτ /mH ), the phase is z-axis oriented in the direction of the τ − line of flight ξ = 2ψ. for the τ + and opposite to the τ + line of flight for the The spin correlations for the state (1) are: τ −.Theτ +τ − spin state Czz =−1, 1 C = C = cos ξ = cos 2ψ, √ |+− + eiξ|−+ (1) xx yy 2 Cxy =−Cyx = sin ξ = sin 2ψ. (3) The way to measure ψ (ξ) is transparent in the case ± ± E-mail address: [email protected] (A. Rougé). of two τ → π ν¯τ (ντ ) decays. Using (1) and the τ 0370-2693/$ – see front matter 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.physletb.2005.05.076 44 A. Rougé / Physics Letters B 619 (2005) 43–49 ± ± decay amplitudes, one gets the correlated decay distri- τ → π ν¯τ (ντ ). This vector can be computed [12,13] bution from the measured momenta in the case of the above ± → ± ¯ + − decay modes. For example, if τ ρ ντ (ντ ) and W3 cos θ , cos θ ,ϕ ρ± → π ±π 0, 1 = + + − i = N · i − · i 1 cos θ cos θ a 2(q pν)q (q q)pν , (6) 8π + − − sin θ sin θ cos(ϕ − 2ψ) , (4) where N is a normalization factor, pν = pτ ± − pρ± is the four-momentum of the neutrino, and q = pπ± − where θ ± is the polar angle in the τ ± rest frame be- p 0 is the difference of the four-momenta of the two ± ˆ π tween the π direction (π) and the above defined z- pions. The distribution (4), where πˆ is replaced by aˆ in = + − − axis. The relative azimuthal angle ϕ ϕ ϕ is the definition of the angles, contains all the available the angle between the two planes defined by the τ − di- + − information on the spin correlation. The three decay rection and the π (π ) direction respectively in the modes (i.e., (55%)2 of the τ +τ − pairs) can therefore Higgs rest frame. The distribution of the azimuthal an- be used in the same way for the measurement of the gle is obtained by integrating out the polar angles: CP violating parameter ψ [5]. 1 π 2 Unfortunately, it has been shown by Monte Carlo W1(ϕ) = 1 − cos(ϕ − 2ψ) . (5) studies [6] that the standard reconstruction (Section 2) 2π 16 of the τ four-momenta is critically impaired by the The merits of the two distributions for the measure- effects of beamstrahlung. A new estimator has been ment of ψ√can be quantified by their sensitivities proposed [7–9], which is less sensitive to the quality of = + − Sψ 1/σψ N, where σψ is the error on ψ expected the τ reconstruction but requires that both τ and τ ± ± from a maximum likelihood fit of the distribution for decay by the process τ → ρ ν¯τ (ντ ). However, the a sample of N events. The sensitivities measure the ideal sensitivity of the new method, when all the four- information per event on ψ, contained in the distribu- momenta are exactly known, is 0.48 to be compared tions [10]; their computation is straightforward when with 0.92 for the standard method. Besides it takes ad- an analytical or numerical expression of the distribu- vantage of (25%)2 only of the τ +τ − pairs. There is 3 = tions is known. They are Sψ 1.15 for the distribu- therefore a possible improvement of the error on ψ by 1 = tion (4) and Sψ 0.92 for (5). The superiority of (4) a factor of at least four, if a method of reconstruction is not very large and it decreases when experimental of the τ ’s insensitive to beamstrahlung is found. effects are introduced. For that reason and for the sake The aim of the present Letter is to devise such a of simplicity, we will only consider the distribution (5) method. To check the sensitivity to beamstrahlung, we in the following. use a simple Monte Carlo procedure. First the energies The best place to study a light Higgs at a linear of the e± after beamstrahlung are generated using the + − + − e e collider is the Higgsstrahlung process e e → program circe [14] with the parameters of the TESLA ZH. We assume that the mass of the Higgs is well project and the cross section for Higgsstrahlung in the measured by the analysis of its dominant decay modes. case of a CP =+1 Higgs [4]; next the production and The four-momentum of the Higgs is determined by the decay angles of the Z are generated according to their measurement of the Z and therefore the four-momenta correlated distribution under the same hypothesis [4]. of the τ + and the τ − can, in principle, be recon- The Higgs decay angles are generated according to an structed in the case of two hadronic decays (Section 2). isotropic distribution. For the correlated decay of the It is known [10] that under such circumstances the τ ’s we consider three cases: two decays into πν (ππ), ± ± three main hadronic decay modes, τ → π ν¯τ (ντ ), a decay into πν and a decay into ρν (ρπ), and two ± → ± ¯ ± → ± ¯ τ ρ ντ (ντ ), and τ a1 ντ (ντ ) have the same decays into ρν (ρρ). For each case, the azimuthal an- analysis power for the measurement of spin effects. gles and the cosines of the polar angles in the decays This is due to the fact [11–13] that, in the τ rest frame, of τ ’s and ρ’s are generated uniformly. From Eq. (1) ± ± all the information on the spin is embodied in the dis- and the decay amplitudes for τ → π ν¯τ (ντ ) and ± ± ± 0 tribution of a unit vector aˆ, the polarization analyser, τ → ρ ν¯τ (ντ ) → π π ν¯τ (ντ ), the correlated decay which is equal to πˆ (pion direction) in the case of probabilities are computed for diverse values of ψ and A. Rougé / Physics Letters B 619 (2005) 43–49 45 the events are accepted or rejected accordingly. The The direction of the τ −, τˆ ≡ˆτ − is on the intersection four-momenta of all the particles are then computed between the cone around hˆ− with angle δ− and the and, finally, the decay lengths of the τ ’s are generated. cone around −hˆ+ with angle δ+. 1/2 The Higgs mass is assumed to be 120 GeV√ and In general there are two solutions τˆ =n ±n⊥, three values√ of the total energy are√ considered: s = where the vectors n and n⊥,showninFig. 1,are 230 GeV, s = 350 GeV and s = 500 GeV. The given by natural energy for a detailed study of the Higgs is the + − − cos δ cos δ cos δ ˆ+ energy of the largest cross section (i.e., near 230 GeV). n =− h 2 The consideration of higher energies allows estimating sin δ − + cos δ − cos δ cos δ − the robustness of the method and possibly the effect of + hˆ , 2 an underestimation of radiative effects. sin δ −2 1 n ˆ+ ˆ− ˆ+ ˆ− n⊥ = h ∧ h , δ =−h · h . 2 with cos 2. Reconstruction of the τ’s in the Higgs rest sin δ (8) frame The ambiguity can be resolved using the information from a vertex detector but, because the main vertex The reconstruction of the τ ’s in the τ +τ − rest is known from the Z decay, the use of the detector is frame for hadronic decays τ ± → h±ν¯ (ν ) has been τ τ rather different and simpler than at LEP [16,17].One known for a long time [15] and was used at LEP to im- needs only to make a chi-square test on the distances prove the measurement of the τ polarization [16].Its in the laboratory between the reconstructed τ lines of principle is sketched in Fig. 1.Bothτ ’s have the same flight and the trajectories of the charged pion’s. energy E± = m /2 and the energies and momenta of τ H The situation is degraded in the presence of beam- the hadrons are measured. The angle δ± between the strahlung and/or other experimental effects because direction of a τ ± displayed by the unit vector τˆ± and the intersection of the two cones is no more granted. the direction of the hadron hˆ± is therefore fixed: As a result the acceptance is reduced and the dis- ± ± − 2 − 2 tribution of ϕ strongly deformed. The reconstruc- ± 2Eτ Eh mτ m ± cos δ = h . (7) tion is especially awkward in the case of two τ ± → 2pτ ± ph± ± π ν¯τ (ντ ) decays, because the polarization analyser is then aˆ± =ˆπ ± = hˆ± and thus ϕ is the angle between the vectors τˆ− ∧ hˆ− and τˆ− ∧ hˆ+.FromEq.(8), one gets 2 2 2 sin δ 1 − cos ϕ = 1 −n , (9) sin2 δ+ sin2 δ− which shows that the two cones are tangent when | cos ϕ|=1, feature that can also√ be observed in Fig. 1. As a consequence, even for s = 230 GeV, where the closeness to the threshold reduces the effect of beamstrahlung, the acceptance becomes very small when cos ϕ is near ±1 and the distribution (5) can hardly be used to test the CP properties of the Higgs. 3. Reconstruction of the τ’s in the laboratory Owing to the observation of the Z decay products, Fig. 1. Reconstruction of the τ direction in the Higgs rest frame. which allows the reconstruction of the main vertex, 46 A. Rougé / Physics Letters B 619 (2005) 43–49 ± ± ± ± 0 Fig. 2. Reconstruction of the τ direction in the laboratory frame: (a) for a τ → π ν¯τ (ντ ) decay mode, (b) for a τ → π π ν¯τ (ντ ) decay mode. one may envisage to perform the reconstruction of the 4. A simplified algorithm τ ’s in the laboratory frame. ± Let us assume that the τ energies in the labora- Implementing the last method by a fit would require L ± tory Eτ ± are known. The angles α between the di- a good knowledge of the errors and their correlations. rections of the τ ’s and the hadrons are given by the This is not possible with our simple simulation. For relation (7), which reads here that reason, we will use a new procedure, which com- bines elements of the approaches followed in the two L L − 2 − 2 ± 2E ± E ± mτ m ± previous sections and gives good results, without the cos α = τ h h . (10) L L intricacies of a fit. 2pτ ± ph± Taking pH =−pZ, we start the reconstruction in The reconstruction of the τ ± direction is then very the Higgs rest frame but use the approximation τˆ = ± ± ± simple in the case of the τ → π ν¯τ (ντ ) decay mode n/|n|. This is always possible if cos δ is replaced (Fig. 2(a)). by 1 when it is found greater than 1. Denoting by ιˆ the unit vector of the perpendicular As the τ energies in the Higgs rest frame are ˆ ± = L from the vertex to the pion trajectory and by π, the unit known: Eτ mH /2, we can compute Eτ ± and per- vector of the pion momentum, the vector τˆ is given by form the τ reconstruction in the laboratory. For that, we replace cos α± by 1 when it is found greater than 1 ± ± τˆ = cos απˆ + sin αι.ˆ (11) and, in the case of a τ → ρ ν¯τ (ντ ) decay, use the projection of hˆ on the (V, π,ˆ ι)ˆ plane to define τˆ when ± ± 0 In the case of a τ → π π ν¯τ (ντ ) decay mode, τˆ is the cone and the plane do not intersect. In the ππ on the intersection of the plane defined by the vertex channel, an event is rejected if both cos α+ and cos α− ˆ and the charged π trajectory with the cone around h are greater than 1. The ambiguities are resolved by with angle α (Fig. 2(b)). There are in general two so- choosing the solution with the smallest missing |p⊥|. lutions. The problem of the reconstruction of the two We can now redefine the Higgs frame as the rest frame L + − τ ’s is therefore a problem with two unknowns: Eτ + of the τ τ pair and compute ϕ. L and Eτ − , but even in the presence of beamstrahlung The reconstructed√ distributions for the three chan- we still have three constraints: the conservation of nels at an energy of s = 350 GeV are shown in Fig. 3 the components of the momentum orthogonal to the and their sensitivities to ψ at the three considered en- beams and the equality of the τ +τ − effective mass ergies are given in Table 1. with the Higgs mass. They are sufficient to deter- Both the curves in Fig. 3 and the numbers in Table 1 mine the two energies and resolve the ambiguities if include the effect of the small loss of acceptance due needed. to the rejection of events in the ππ channel. A. Rougé / Physics Letters B 619 (2005) 43–49 47 For the ππ channel, the sensitivity is nearly the straints can be used, like the τ +τ − effective mass ideal one (0.92) up to 500 GeV. For the ρπ and and the positivity of the decay length. It should also ρρ channels, the sensitivities are slightly reduced at be noted that the conservation of p⊥ is an impor- 230 GeV and decrease more rapidly with the energy tant but not vital point in the method. For example, than for the ππ channel. Two effects contribute to smearing the p⊥ of the Z with σ(px) = σ(py) = = = that. The first is the closing of the τ decay angle 1 GeV√ yields sensitivities Sρπ 0.66 and Sρρ 0.52 when the hadron mass increases. The second is the at s = 350 GeV. imperfect resolution of the ambiguities. The second Finally, a few remarks are in order about the τ ± → ± ¯ point can be improved because the criterion used for a1 ντ (ντ ) decay channel, which is not included in the choice of the solution is not optimal. Other con- our unsophisticated simulation, for the reason that its description is more complex than a simple angular ± → ± 0 0 Table 1 distribution. The a1 π π π decay mode is re- The sensitivities to ψ of the reconstructed distributions when only constructed by the same method that the ρ±,butthe beamstrahlung is taken into account. The effect of the small loss of sensitivity will probably be worsened by the closing acceptance in the ππ channel is included ± → ± + − √ of the τ decay angle. For the a1 π π π de- s(GeV) Sensitivity (Sψ ) cay mode, the determination of τˆ in the laboratory ππ πρ ρρ is in principle possible from vertexing information 230 0.92 0.88 0.83 only, consequently a good sensitivity can certainly be 350 0.91 0.73 0.66 achieved by using this determination of τˆ or by an 500 0.88 0.64 0.55 adaptation of the method used for the ρ±. Fig. 3. The distributions of ϕ: (a) at the generation level, (b), (c), and (d), reconstructed by the method of Section 4 for the three channels ππ, ρπ,andρρ at an energy of 350 GeV. Beamstrahlung effects only are taken into account. The histograms are normalized to the number of generated events and multiplied by the number of bins. With this normalization, the distribution (a) is 2πW1(ϕ). The full lines correspond to ψ = 0, the dotted lines to ψ = π/8. 48 A. Rougé / Physics Letters B 619 (2005) 43–49 Fig. 4. The reconstructed distributions of ϕ for the ππ channel at an energy of 350 GeV, when all the experimental effects are taken into + − account: (a) for the decay of the Z into µ µ , (b) for its decays into two jets. The convention of normalization and the values of ψ are the same as in Fig. 3. . 5. A semi-realistic simulation of the ππ channel Table 2 The sensitivities to ψ of the reconstructed distributions for the ππ channel, when all the experimental effects are taken into account To get a realistic estimation of the sensitivities, it is √ necessary to take into account the performance of the s(GeV) Sensitivity (Sψ ) detector. We use for that the parameters of the TESLA Z → µ+µ− Z → qq¯ 0 project [18,19]. Since the precision of the π measure- 230 0.69 0.71 ment depends not only on the accuracy of the detector 350 0.60 0.61 but also on the quality of the reconstruction algorithm, 500 0.58 0.58 we consider here the ππ channel only. For the charged tracks an independent Gaussian smearing is performed on the five parameters: θ, φ, Because of the key role of the vertex detector in the reconstruction, we have done again the simulations 1/p⊥ and the two components of the impact parame- ter. We use for the widths of the Gaussians the follow- with the vertex detector errors multiplied by two. The ing values [19]: reduction of the sensitivities that results from the in- creased uncertainties is always smaller than 0.1. σ(θ)= σ(φ)= 0.1mrad, −5 −1 σ(1/p⊥) = 5 × 10 GeV , 6. Conclusion σ(rφ)= σ(rz)= 4.2 ⊕ 4.0/(p sin3/2 θ) µm. The energy√ of the jets is smeared according to σ(E)/ We have studied the production of a light Higgs E = 0.3/ E(GeV). The position of the vertex is de- boson by the process of Higgsstrahlung and its sub- termined by the shape of the beam [18] for the x and y sequent decay into τ +τ −, under the conditions of a coordinates and by the charged decay products of the linear collider. Z for the z coordinate. The smearing of these coordi- We have described a method, which by the joint use nates is done accordingly. of kinematics and vertexing allows the measurement The reconstructed distributions at an energy of of the transverse spin correlations of the two τ ’s. This 350 GeV, with all the experimental effects included in method is not impaired by beamstrahlung and can be the simulation, are shown in Fig. 4 for the decays of applied for the main hadronic decay modes of the τ the Z both into µ+µ− and into qq¯. Their sensitivities and most of the visible decay modes of the Z. ± ± to ψ at the three considered energies are given in Ta- In the case of two τ → π ν¯τ (ντ ) decays, a com- ble 2. It is clear from these values that a large part of plete simulation of the detector effects with the para- the sensitivity is retained. meters of the TESLA project [19] has been performed. A. Rougé / Physics Letters B 619 (2005) 43–49 49 A realistic study of the reconstruction of the π 0 is still [8]K.Desch,Z.W¸as, M. Worek, Eur. Phys. J. C 29 (2003) 491. to be done, nevertheless it appears that a reasonable [9]K.Desch,A.Imhof,Z.W¸as, M. Worek, Phys. Lett. B 579 goal for the measurement of the phase ψ that parame- (2004) 157. [10] M. Davier, L. Duflot, F. Le Diberder, A. Rougé, Phys. Lett. trizes a possible CP violation should be to use all the B 306 (1993) 411. above mentioned final states√ and get a sensitivity bet- [11] Y.S. Tsai, Phys. Rev. D 4 (1971) 2821. ter than 0.5, i.e., σψ < 0.6π/ Nevt.. [12] H. Kühn, F. Wagner, Nucl. Phys. B 236 (1984) 16. [13] S. Jadach, J.H. Kühn, Z. W¸as, Comput. Phys. Commun. 64 (1991) 275. References [14] T. Ohl, Comput. Phys. Commun. 101 (1997) 269. [15] Y.-S. Tsai, A.C. Hearn, Phys. Rev. 140 (1965) B721. [16] ALEPH Collaboration, A. Heister, et al., Eur. Phys. J. C 20 [1] J.R. Dell’Aquila, C.A. Nelson, Nucl. Phys. B 320 (1989) 61. (2001) 401. [2] C.A. Nelson, Phys. Rev. D 41 (1990) 2805. [17] J.H. Kühn, Phys. Lett. B 313 (1993) 458. [3] M. Krämer, J. Kühn, M.L. Stong, P.M. Zerwas, Z. Phys. C 64 [18] R. Brinkman, K. Flöttmann, J. Roßbach, P. Schmüser, (1994) 21. N. Walker, H. Weise (Eds.), TESLA Technical Design Report, [4] V. Barger, K. Cheung, A. Djouadi, B.A. Kniehl, P.M. Zerwas, Part II: The Accelerator, DESY 2001-011. Phys. Rev. D 49 (1994) 79. [19] T. Behnke, S. Bertolucci, R.D. Heuer, R. Settles (Eds.), TESLA [5] B. Grz¸adkowski, J.F. Gunion, Phys. Lett. B 350 (1995) 218. Technical Design Report, Part IV: A Detector for TESLA, [6] Z. W¸as, M. Worek, Acta Phys. Pol. B 33 (2002) 1875, hep- DESY 2001-011. ph/0202007. [7] G.R. Bower, T. Pierzchała, Z. W¸as, M. Worek, Phys. Lett. B 543 (2002) 227. Physics Letters B 619 (2005) 50–60 www.elsevier.com/locate/physletb First measurement of the π +π − atom lifetime B. Adeva p,L.Afanasyevl,1, M. Benayoun e, A. Benelli q,Z.Berkab, V. Brekhovskikh o, G. Caragheorgheopol m, T. Cechak b, M. Chiba k, S. Constantinescu m, C. Detraz a, D. Dreossi g, D. Drijard a, A. Dudarev l,I.Evangeloud, M. Ferro-Luzzi a, M.V. Gallas p,a, J. Gerndt b,R.Giacomichg, P. Gianotti f,D.Goldinq,F.Gómezp,A.Gorino, O. Gorchakov l,C.Guaraldof, M. Hansroul a,R.Hosekb, M. Iliescu f,m, V. Karpukhin l, J. Kluson b, M. Kobayashi h, P. Kokkas d, V. Komarov l,V.Kruglovl, L. Kruglova l, A. Kulikov l, A. Kuptsov l, I. Kurochkin o, K.-I. Kuroda l,A.Lambertog,A.Lanaroa,f, V. Lapshin o,R.Lednickyc,P.Lerustee, P. Levi Sandri f, A. Lopez Aguera p, V. Lucherini f,T.Makij,N.Manthosd, I. Manuilov o, L. Montanet a, J.-L. Narjoux e, L. Nemenov a,l, M. Nikitin l, T. Núñez Pardo p,K.Okadai, V. Olchevskii l,A.Pazosp, M. Pentia m,A.Penzog, J.-M. Perreau a,C.Petrascuf,m,M.Plóp, T. Ponta m,D.Popm, G.F. Rappazzo g, A. Rodriguez Fernandez p,A.Romerop, A. Ryazantsev o,V.Rykalino, C. Santamarina p,q,a, J. Saborido p, J. Schacher r, Ch.P. Schuetz q,A.Sidorovo, J. Smolik c,F.Takeutchii, A. Tarasov l,L.Tauscherq, M.J. Tobar p,S.Trusovn, V. Ut k i n l, O. Vázquez Doce p, P. Vázquez p,S.Vlachosq,V.Yazkovn, Y. Yoshimura h, M. Zhabitsky l,P.Zrelovl a CERN, Geneva, Switzerland b Czech Technical University, Prague, Czech Republic c Institute of Physics ACSR, Prague, Czech Republic d Ioannina University, Ioannina, Greece e LPNHE des Universites Paris VI/VII, IN2P3-CNRS, France f INFN, Laboratori Nazionali di Frascati, Frascati, Italy g INFN, Trieste and Trieste University, Trieste, Italy h KEK, Tsukuba, Japan i Kyoto Sangyo University, Kyoto, Japan j UOEH-Kyushu, Japan k Tokyo Metropolitan University, Japan l JINR, Dubna, Russia m IFIN-HH, National Institute for Physics and Nuclear Engineering, Bucharest, Romania n Skobeltsin Institute for Nuclear Physics of Moscow State University, Moscow, Russia o IHEP, Protvino, Russia p Santiago de Compostela University, Spain q Basel University, Switzerland r Bern University, Switzerland 0370-2693/$ – see front matter 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.physletb.2005.05.045 B. Adeva et al. / Physics Letters B 619 (2005) 50–60 51 Received 22 April 2005; accepted 18 May 2005 Available online 31 May 2005 Editor: M. Doser Abstract + − The goal of the DIRAC experiment at CERN (PS212) is to measure the π π atom lifetime with 10% precision. Such a measurement would yield a precision of 5% on the value of the S-wave ππ scattering lengths combination |a0 − a2|. Based on =[ +0.49]× −15 part of the collected data we present a first result on the lifetime, τ 2.91−0.62 10 s, and discuss the major systematic | − |= +0.033 −1 errors. This lifetime corresponds to a0 a2 0.264−0.020mπ . 2005 Elsevier B.V. All rights reserved. PACS: 36.10.-k; 32.70.Cs; 25.80.E; 25.80.Gn; 29.30.Aj Keywords: DIRAC experiment; Elementary atom; Pionium atom; Pion scattering 1. Introduction 0.220 ± 0.005, a2 =−0.0444 ± 0.0010, a0 − a2 = 0.265 ± 0.004 in units of inverse pion mass) and lead −15 The aim of the DIRAC experiment at CERN [1] is to the prediction τ1S = (2.9 ± 0.1) × 10 s. The to measure the lifetime of pionium, an atom consist- generalized chiral perturbation theory though allows + − ing of a π and a π meson (A2π ). The lifetime is for larger a-values [12]. Model independent measure- dominated by the charge-exchange scattering process ments of a0 have been done using Ke4 decays [13,14]. (π +π − → π 0π 0)2 and is thus related to the relevant Oppositely charged pions emerging from a high scattering lengths [4]. The partial decay width of the energy proton–nucleus collision may be either pro- atomic ground state (principal quantum number n = 1, duced directly or stem from strong decays (“short- orbital quantum number l = 0) is [2,5–9] lived” sources) and electromagnetic or weak decays 1 2 (“long-lived” sources) of intermediate hadrons. Pion Γ = = α3p|a − a |2(1 + δ) (1) pairs from “short-lived” sources undergo Coulomb fi- 1S τ 9 0 2 1S nal state interaction and may form atoms. The region with τ1S the lifetime of the atomic ground state, α of production being small as compared to the Bohr 0 the fine-structure constant, p the π momentum in radius of the atom and neglecting strong final state the atomic rest frame, and a0 and a2 the S-wave ππ n interaction, the cross section σA for production of scattering lengths for isospin 0 and 2, respectively. atoms with principal quantum number n is related to The term δ accounts for QED and QCD corrections the inclusive production cross section for pion pairs = ± × −2 [6–9]. It is a known quantity (δ (5.8 1.2) 10 ) from “short lived” sources without Coulomb correla- ensuring a 1% accuracy for Eq. (1) [8]. A measure- 0 tion (σs ) [15] ment of the lifetime therefore allows to obtain in a model-independent way the value of |a −a |.Theππ 0 2 n 2 0 dσ EA ∗ 2 d σ scattering lengths a0, a2 have been calculated within A = ( π)3 Ψ C r = s 2 n 0 (2) the framework of standard chiral perturbation theory dpA MA dp+ dp− p+=p− [10] with a precision better than 2.5% [11] (a0 = with pA, EA and MA the momentum, energy and mass of the atom in the lab frame, respectively, and p+, E-mail addresses: [email protected], p− the momenta of the charged pions. The square of [email protected] (L. Afanasyev). 1 PH Division, CERN, CH 1211 Geneva 23, Switzerland. the Coulomb atomic wave function for zero distance ∗ | C |2 = 2 Annihilation into two photons amounts to ≈ 0.3% [2,3] and is r between them in the c.m. system is Ψn (0) 3 3 = neglected here. pB/πn , where pB mπ α/2 is the Bohr momentum 52 B. Adeva et al. / Physics Letters B 619 (2005) 50–60 of the pions and mπ the pion mass. The production of atoms occurs only in S-states [15]. Final state interaction also transforms the “unphys- 0 ical” cross section σs into a real one for Coulomb correlated pairs, σC [16,17]: 2 2 0 d σC = C ∗ 2 d σs Ψ ∗ r , (3) dp+ dp− −k dp+ dp− where Ψ C (r∗) is the continuum wave function and −k∗ 2k∗ ≡q with q being the relative momentum of the π + and π − in the c.m. system.3 |Ψ C (r∗)|2 de- −k∗ scribes the Coulomb correlation and at r∗ = 0 coin- cides with the Gamov–Sommerfeld factor AC(q) with q =|q| [17]: 2πmπ α/q AC(q) = . (4) 1 − exp(−2πmπ α/q) For low q,0 q q0,Eqs.(2)–(4) relate the num- + − Fig. 1. Relative momentum distributions (q, qL) for atomic π π ber of produced A2π atoms, NA, to the number of pairs at the point of break-up and at the exit of the target. Note that Coulomb correlated pion pairs, NCC [18] qL is almost not affected by multiple scattering in the target. ∞ 1 σ tot 3 = NA = A = (2παmπ ) n 1 n3 q tot| 0 3 NCC σC q q0 π 0 AC(q) d q + − = kth(q0). (5) break up. The π π pairs from break-up (atomic pairs) exhibit specific kinematical features which al- Eq. (5) defines the theoretical k-factor. Throughout the low to identify them experimentally [15], namely very Letter we will use low relative momentum q and qL (the component of + q0 = 2MeV/c and kth(q0) = 0.615. (6) q parallel to the total momentum p+ p−)asshown in Fig. 1. After break-up, the atomic pair traverses the In order to account for the finite size of the pion pro- target and to some extent loses these features by mul- duction region and of the two-pion final state strong tiple scattering, essentially in the transverse direction, interaction, the squares of the Coulomb wave func- while qL is almost not affected. This is one reason for tions in Eqs. (2) and (3) must be substituted by the considering distributions in QL as well as in Q when square of the complete wave functions, averaged over analyzing the data. ∗ the distance r and the additional contributions from Excitation/deexcitation and break-up of the atom 0 0 → 0 0 → + − π π A2π as well as π π π π [17].It are competing with its decay. Solving the transport should be noticed that these corrections essentially equations with the cross sections for excitation and cancel in the k-factor (Eq. (5)) and lead to a correc- break-up, [20–31] leads to a target-specific relation tion of only a fraction of a percent. Thus finite size between break-up probability and lifetime which is corrections can safely be neglected for kth. estimated to be accurate at the 1% level [22,32,33]. Once produced, the A2π atoms propagate with rel- Measuring the break-up probability thus allows to de- ¯ ≈ ativistic velocity (average Lorentz factor γ 17 in termine the lifetime of pionium [15]. our case) and, before they decay, interact with tar- The first observation of the A2π atom [34] has al- get atoms, whereby they become excited/deexcited or lowed to set a lower limit on its lifetime [18,19] of τ>1.8 × 10−15 s (90% CL). In this Letter we present 3 For the sake of clarity we use the symbol Q for the experimen- a determination of the lifetime of the A2π atom, based tally reconstructed and q for the physical relative momentum. on a large sample of data taken in 2001 with Ni targets. B. Adeva et al. / Physics Letters B 619 (2005) 50–60 53 Fig. 2. Schematic top view of the DIRAC spectrometer. Upstream of the magnet: target, microstrip gas chambers (MSGC), scintillating fiber detectors (SFD), ionization hodoscopes (IH) and iron shielding. Downstream of the magnet: drift chambers (DC), vertical and horizontal scintillation hodoscopes (VH, HH), gas Cherenkov counters (Ch), preshower detectors (PSh) and, behind the iron absorber, muon detectors (Mu). 2. The DIRAC experiment binatorials and inefficiencies of the SFD, the distrib- utions for the transverse components have substantial The DIRAC experiment uses a magnetic double- tails, which the longitudinal component does not ex- arm spectrometer at the CERN 24 GeV/c extracted hibit [37]. This is yet another reason for analyzing both proton beam T8. Details on the set-up may be found Q and QL distributions. in [35]. Since its start-up, DIRAC has accumulated Data were analyzed with the help of the DIRAC about 15 000 atomic pairs. The data used for this work analysis software package ARIANE [39]. were taken with two Ni foils, one of 94 µm thickness The tracking procedures require the two tracks ei- (76% of the π +π − data), and one of 98 µm thick- ther to have a common vertex in the target plane ness (24% of the data). An extensive description of the (“V-tracking”) or to originate from the intersect of DIRAC set-up, data selection, tracking, Monte Carlo the beam with the target (“T-tracking”). In the fol- procedures, signal extraction and a first high statistics lowing we limit ourselves to quoting results obtained demonstration of the feasibility of the lifetime mea- with T-tracking. Results obtained with V-tracking do surement, based on the Ni data of 2001, have been not show significant differences, as will be shown published in [36]. later. The set-up and the definitions of detector acronyms The following cuts and conditions are applied (see are shown in Fig. 2. The main selection criteria and [36]): performance parameters [36] are recalled in the fol- lowing. • at least one track candidate per arm with a confi- Pairs of oppositely charged pions are selected by dence level better than 1% and a distance to the beam means of Cherenkov, preshower and muon counters. spot in the target smaller than 1.5 cm in x and y; Through the measurement of the time difference be- • “prompt” events are defined by the time differ- tween the vertical hodoscope signals of the two arms, ence of the vertical hodoscopes in the two arms of the | | time correlated (prompt) events (σ t = 185 ps) can be spectrometer of t 0.5ns; distinguished from accidental events (see [36]). The • “accidental” events are defined by time intervals resolution of the three components of the relative mo- −15 t −5 ns and 7 t 17 ns, determined mentum Q of two tracks, transverse and parallel to by the read-out features of the SFD detector (time de- the c.m. flight direction, Qx , Qy and QL, is about pendent merging of adjacent hits) and exclusion of − 0.5 MeV/c for Q 4MeV/c. Due to charge com- correlated π p pairs. [36]; 54 B. Adeva et al. / Physics Letters B 619 (2005) 50–60 • gen ∝ 2 × protons in “prompt” events are rejected by time- dNCC /dq q AC(q). Processing them with of-flight in the vertical hodoscopes for momenta of GEANT-DIRAC and then analyzing them using the the positive particle below 4 GeV/c. Positive particles full detector and trigger simulation leads to the Cou- MC with higher momenta are rejected; lomb correlated distribution dNCC /dQ. • e± and µ± are rejected by appropriate cuts on Non-correlated π +π − pairs (NC-background). the Cherenkov, the preshower and the muon counter π +π − pairs, where at least one pion originates from information; the decay of a “long-lived” source (e.g., electromag- • cuts in the transverse and longitudinal compo- netically or weakly decaying mesons or baryons) do nents of Q are QT 4MeV/c and |QL| < 15 MeV/c. not undergo any final state interactions. Thus they are gen ∝ 2 The QT cut preserves 98% of the atomic signal. The generated according to dNNC /dq q , using slightly QL cut preserves data outside the signal region for softer momentum distributions than for short-lived defining the background; sources (difference obtained from FRITIOF-6). The • MC only events with at most two preselected hits per Monte Carlo distribution dNNC /dQ is obtained as SFD plane are accepted. This provides the cleanest above. possible event pattern. Accidental π +π − pairs (acc-background). π +π − pairs, where the two pions originate from two different proton–nucleus interactions, are generated according gen 2 3. Analysis to dNacc /dq ∝ q , using measured momentum distri- MC butions. The Monte Carlo distribution dNacc /dQ is The spectrometer including the target is fully simu- obtained as above. lated by GEANT-DIRAC [38], a GEANT3-based sim- All the Monte Carlo distributions are normalized, Qmax MC = MC = ulation code. The detectors, including read-out, ineffi- 0 (dNi /dQ)dQ Ni , i CC, NC, acc, with ciency, noise and digitalization are simulated and im- statistics about 5 to 10 times higher than the experi- MC plemented in the DIRAC analysis code ARIANE [39]. mental data; similarly for atomic pairs (nA ). The triggers are fully simulated as well. The measured prompt distributions are approxi- The simulated data sets for different event types can mated by appropriate shape functions. The functions therefore be reconstructed with exactly the same pro- for atomic pairs, FA(Q), and for the backgrounds, cedures and cuts as used for experimental data. FB(Q), (analogously for QL) are defined as The different event types are generated according to the underlying physics. Atomic pairs. Atoms are generated according to rec MC nA dnA Eq. (2) using measured total momentum distributions FA(Q) = , + − MC dQ for short-lived pairs. The atomic π π pairs are gen- nA erated according to the probabilities and kinematics rec MC rec MC described by the evolution of the atom while propa- NCC dNCC NNC dNNC FB(Q) = + gating through the target and by the break-up process N MC dQ N MC dQ + − CC NC (see [40]). These π π pairs, starting from their spa- MC ωaccNpr dN tial production point, are then propagated through the + acc (7) N MC dQ remaining part of the target and the full spectrome- acc ter using GEANT-DIRAC. Reconstruction of the track rec rec rec pairs using the fully simulated detectors and triggers with nA , NCC, NNC the reconstructed number of leads to the atomic pair distribution dnMC/dQ. atomic pairs, Coulomb- and non-correlated back- + − A Coulomb correlated π π pairs (CC-back- ground, respectively, and ωacc the fraction of acciden- ground). The events are generated according to tal background out of all prompt events Npr. Analyz- Eqs. (3), (4) using measured total momentum dis- ing the time distribution measured with the vertical tributions for short-lived pairs. The generated q-dis- hodoscopes (see [36]) we find ωacc = 7.1% (7.7%) for tributions are assumed to follow phase space modi- the 94 µm (98 µm) data sets [36,37] and keep it fixed fied by the Coulomb correlation function (Eq. (4)), when fitting. The χ2 function for Q (analogously for B. Adeva et al. / Physics Letters B 619 (2005) 50–60 55 Fig. 3. Top: experimental Q and QL distributions after subtraction of the prompt accidental background, and fitted Monte Carlo backgrounds (dotted lines). The peak at Q = 4 MeV/c is due to the cut QT 4 MeV/c. Bottom: residuals after background subtraction. The dotted lines represent the expected atomic signal shape. The bin-width is 0.25 MeV/c. QL) to minimize is meters found, the background is subtracted from the dN 2 measured prompt distribution, resulting in the resid- νmax pr − [ + ] dQ Q ν ( FA(Q) FB(Q) Q)ν ual spectra. For the signal region, defined by the cuts χ2 = dN = = pr Q + (σ )2 + (σ )2 Q 4MeV/c and QL 2MeV/c, we obtain the to- νmin dQ ν A ν B ν residual (8) tal number of atomic pairs, nA and of Coulomb sig with Q the bin width and σA, σB the statistical er- correlated background events, NCC. Results of fits for rors of the Monte Carlo shape functions, which are Q and QL together are shown in Table 1. much smaller than that of the measurement. The fit CC-background and NC- or acc-backgrounds are rec rec rec distinguishable due to their different shapes, most pro- parameters are nA , NCC, NNC (see Eq. (7)). As a constraint the total number of measured prompt events nounced in the QL distributions (see Fig. 3, top). − = rec + Accidental and NC-background shapes are almost is restricted by the condition Npr(1 ωacc) NCC rec + rec identical for Q and fully identical for Q (uniform NNC nA . The measured distributions as well as the L background are shown in Fig. 3 (top). distributions). Thus, the errors in determining the ac- The data taken with 94 and 98 µm thick targets were cidental background ωacc are absorbed in fitting the analyzed separately. The total number of events in the NC background. The correlation coefficient between − prompt window is Npr = 471 290. CC and NC background is 99%. This strong correla- rec rec First, we determine the background composition by tion leads to equal errors for NCC and NNC. The CC- minimizing Eq. (8) outside of the atomic pair signal background is determined with a precision better than region, i.e., for Q>4MeV/c and QL > 2MeV/c. 1%. Note that the difference between all prompt events rec = and the background is N − N rec − N rec − ω N = For this purpose we require nA 0. As a constraint, pr CC NC acc pr the background parameters N rec and N rec represent- 6590, hence very close to the number of residual CC NC residual ing the total number of CC- and NC-events, have to atomic pairs (nA ) as expected. This relation is be the same for Q and QL. Then, with the para- also used as a strict constraint for fits outside of the sig- 56 B. Adeva et al. / Physics Letters B 619 (2005) 50–60 Table 1 rec Fit results (94 and 98 µm targets together, background shapes from Monte Carlo (MC)) for the parameters NCC (total number of CC-events), rec rec residual NNC (total number of NC-events) and nA (atomic pairs) and deduced results for the number of atomic pairs from the residuals (nA )and sig the number of CC-background events in the signal region (NCC). MC-a: background fit excluding the signal region. MC-b: fit of the entire momentum range including Monte Carlo shape for atomic pairs (“shape fit”). The cuts were at Qcut = 4 MeV/c and QL,cut = 2 MeV/c. Q 2 and QL-distributions were fitted together. The normalized χ were 0.9 for MC-a and MC-b rec rec residual rec sig NCC NNC nA nA NCC MC-a Q 374 022 ± 3969 56 538 6518 ± 373 106 500 ± 1130 QL same same 6509 ± 330 82 289 ± 873 MC-b Q 374 282 ± 3561 56213 6530 ± 294 106 549 ± 1014 QL same same same 82 345 ± 783 > − rec> − rec> − > = nal region (>), Npr NCC NNC (ωaccNpr) 0 ber of atomic pairs is obtained from the measured pairs = rec cut and, hence, the fit requires only one free parameter, by nA nA (Q Qcut)/A . rec> NCC . Number of produced A2π atoms.Hereweusethe Second, the atomic pair signal may be directly known relation between produced atoms and Coulomb obtained by minimizing Eq. (8) over the full range correlated π +π − pairs (CC-background) of Eq. (5). gen and including the Monte Carlo shape distribution FA Using the generator for CC pairs, NCC events, of gen (“shape fit”). The signal strength has to be the same which N (q q0) (see Eq. (6))haveq below rec CC for Q and QL. The result for the signal strength nA q0, are generated into the same acceptance window sig Ω as for atomic pairs and processed analogously as well as the CC-background below the cuts, NCC, gen are shown in Table 1. The errors are determined by to the paragraph above to provide the number of MINOS [41]. reconstructed CC-events below the same arbitrary MC-rec The consistency between the analysis in Q with the cut in Q as for atomic pairs, NCC (Q Qcut). one in QL establishes the correctness of the QT recon- These CC-events are related to the originally gener- cut = MC-rec struction. A 2D fit in the variables (QL,QT ) confirms ated CC-events below q0 through CC NCC (Q gen the results of Table 1. Qcut)/NCC (q q0). The number of produced atoms = rec cut thus is NA kth(q0)NCC(Q Qcut)/CC (see Eq. (6)). The break-up probability P thus becomes 4. Break-up probability br rec n n (Q Qcut) P = A = A with In order to deduce the break-up probability, Pbr = br rec NA k(Qcut)NCC(Q Qcut) nA/NA, the total number of atomic pairs nA and the total number of produced A atoms, N ,havetobe cut 2π A k(Q ) = k (q ) A . (9) known. None of the two numbers is directly measured. cut th 0 cut CC The procedure of obtaining the two quantities requires reconstruction efficiencies and is as follows. In Table 2 the k-factors are listed for different cuts Number of atomic pairs. Using the generator for in Q and QL for the two target thicknesses (94 and gen atomic pairs a large number of events, nA , is gener- 98 µm) and the weighted average of the two, corre- ated in a predefined large spatial acceptance window sponding to their relative abundances in the Ni data Ωgen, propagated through GEANT-DIRAC including of 2001. The accuracy is of the order of one part per the target and reconstructed along the standard proce- thousand and is due to Monte Carlo statistics. dures. The total number of reconstructed Monte Carlo With the k-factors of Table 2 and the measurements MC-rec atomic pairs below an arbitrary cut in Q, nA (Q listed in Table 1, the break-up probabilities of Table 3 Qcut) defines the reconstruction efficiency for atomic are obtained. The simultaneous fit of Q and QL with cut = MC-rec gen pairs A nA (Q Qcut)/nA . The total num- the atomic shape results in a single value. B. Adeva et al. / Physics Letters B 619 (2005) 50–60 57 Table 2 k(Qcut) factors as a function of cuts in Q and QL for the 94 and 98 µm thick Ni targets, and the weighted average of the two for a relative abundance of 76% (94 µm) and 24% (98 µm) k94 µm k98 µm kaverage Qcut = 2 MeV/c 0.5535 ± 0.0007 0.5478 ± 0.0007 0.5521 ± 0.0007 Qcut = 3 MeV/c 0.2565 ± 0.0003 0.2556 ± 0.0003 0.2563 ± 0.0003 Qcut = 4 MeV/c 0.1384 ± 0.0002 0.1383 ± 0.0002 0.1384 ± 0.0002 QL,cut = 1 MeV/c 0.3054 ± 0.0004 0.3044 ± 0.0003 0.3050 ± 0.0004 QL,cut = 2 MeV/c 0.1774 ± 0.0002 0.1776 ± 0.0002 0.1774 ± 0.0002 Table 3 The break-up probability has to be corrected for Break-up probabilities for the combined Ni 2001 data, based on the the impurities of the targets. Thus, the 94 µm thick = results of Table 1 and the k-factors of Table 2 for the cuts Qcut target has a purity of only 98.4%, while the 98 µm 4 MeV/c and Q = 2 MeV/c. Errors are statistical L,cut thick target is 99.98% pure. The impurities (C, Mg, residual rec sig nA nA NCC Pbr Si, S, Fe, Cu) being mostly of smaller atomic num- Q 6518 ± 373 106 500 ± 1130 0.442 ± 0.026 ber than Ni lead (for the weighted average of both ± ± ± QL 6509 330 82 289 873 0.445 0.023 targets) to a reduction of the break-up probability of Q & QL 6530 ± 294 106 549 ± 1004 0.447 ± 0.023 1.1% as compared to pure Ni, assuming a lifetime of 3 fs. Therefore, the measured break-up probability has to be increased by 0.005 in order to correspond to pure The break-up probabilities from Q and QL agree Ni. The final result is within a fraction of a percent. The values from shape fit and from background fit are in perfect agreement Pbr = 0.452 ± 0.023stat. (10) (see Table 1). We adopt the atomic shape fit value of Pbr = 0.447 ± 0.023stat, because the fit covers the full rec Q, QL range and includes correlations between nA 5. Systematic errors sig and NCC. Analyzing the data with three allowed hit candi- Systematic errors may occur through the analy- dates in the SFD search window instead of two, re- sis procedures and through physical processes which sults in more atomic pairs (see Ref. [36], T-tracking). are not perfectly under control. We investigate first The break-up probabilities obtained are 0.440 ± 0.024 procedure-induced errors. and 0.430 ± 0.021 for Q and QL, respectively. They The break-up probability will change, if the ratio are not in disagreement with the adopted value of rec rec NCC/NNC depends on the fit range. If so, the Monte 0.447. Despite the larger statistics, the accuracy is not Carlo distributions do not properly reproduce the mea- improved, due to additional background. This back- sured distributions and the amount of CC-background ground originates from additional real hits in the up- may not be constant. In Fig. 4 the dependence is shown stream detectors or from electronic noise and cross- for the fits in Q, QL and both together. The ratio is rea- talk. This has been simulated and leads essentially to sonably constant within errors, with the smallest errors a reduced reconstruction efficiency but not to a dete- for a fit range of Q = QL = 15 MeV/c. At this point rioration of the reconstruction quality. The additional the difference between Q and QL fits leads to a differ- sources of systematic uncertainties lead us not to con- CC = ence in break-up probability of Pbr 0.023. sider this strategy of analysis further on. Consistency of the procedure requires that the V-tracking provides a slightly different data sam- break-up probability does not depend on Qcut.In ple, different k-factors and different signal strengths Fig. 5 the dependence on the cut is shown for break-up and CC-background. The break-up probability, how- residual probabilities deduced from nA . There is a sys- V-tracking = ever, does not change significantly and is Pbr tematic effect which, however, levels off for large cut 0.453±0.025stat, only 0.3σ off from the adopted value momenta. This dependence indicates that the shape of 0.447. the atomic pair signal as obtained from Monte Carlo 58 B. Adeva et al. / Physics Letters B 619 (2005) 50–60 Fig. 4. Ratio of CC-background over NC-background as a function Fig. 5. Pbr as a function of cut momentum for Q and QL. of fit range. In fact we have measured the multiple scattering for (and used for the k-factor determination) is not in per- all scatterers (upstream detectors, vacuum windows, fect agreement with the residual shape. This may be target) and found narrower angular distributions than due to systematics in the atomic pair shape directly expected from the standard GEANT model [42].This, and/or in reconstructed CC-background for small rel- however, may be due also to errors in determining the ative momenta. The more the signal is contained in thickness and material composition of the upstream the cut, the more the Pbr values stabilize. As a con- detectors. Based on these studies we conservatively sequence, we chose a cut that contains the full signal attribute a maximum error of +5% and −10% to mul- (see Eq. (10)). This argument is also true for sharper tiple scattering. cuts in QT than the one from the event selection. Cut Another source of uncertainty may be due to the momenta beyond the maximum cut of Fig. 5 would presence of unrecognized K+K− and pp¯ pairs that only test background, as the signal would not change would fulfill all selection criteria [43]. Such pairs may anymore. be as abundant as 0.5% and 0.15%, respectively, of To investigate whether the atomic pair signal shape π +π − pairs as estimated for K+K− with FRITIOF- is the cause of the above cut dependence, we studied 64 and for pp¯ from time-of-flight measurements in a two extreme models for atom break-up: break-up only narrow momentum interval with DIRAC data. Their from the 1S-state and break-up only from highly ex- mass renders the Coulomb correlation much more cited states. The two extremes result in a difference in peaked at low Q than for pions, which leads to a shape = + − break-up probability of Pbr 0.008. change in effective π π Coulomb background at Sources of systematic errors may also arise from small Q, thus to a smaller atomic pair signal and there- uncertainties in the genuine physical process. We have fore to a decrease of break-up probability. The effect ¯ ¯ investigated possible uncertainties in multiple scatter- leads to a change of P KK,pp =−0.04. We do not ing as simulated by GEANT by changing the scat- br tering angle in the GEANT simulation by ±5%. As a result, the break-up probability changes by 0.002 4 FRITIOF-6 reproduces well production cross sections and mo- per one percent change of multiple scattering angle. mentum distributions for 24 GeV/c proton interactions. B. Adeva et al. / Physics Letters B 619 (2005) 50–60 59 Table 4 Summary of systematic effects on the measurement of the break- up probability Pbr. Extreme values have been transformed into σ assuming uniform distributions Source Extreme values σ CC-background +0.012/−0.012 ±0.007 Signal shape +0.004/−0.004 ±0.002 + − +0.006 Multiple scattering 0.01/ 0.02 −0.013 + − ¯ + − +0 K K and pp 0/ 0.04 −0.023 + − +0 Finite size 0/ 0.03 −0.017 +0.009 Total −0.032 apply this shift but consider it as a maximum sys- tematic error of Pbr. Admixtures from unrecognized e+e− pairs from photon conversion do not contribute because of their different shapes. Finally, the correlation function Eq. (3) used in the Fig. 6. Break-up probability Pbr as a function of the lifetime of the analysis is valid for pointlike production of pions, cor- atomic ground state τ1S for the combined 94 and 98 µm thick Ni related only by the Coulomb final state interaction targets. The experimentally determined Pbr with statistical and to- (Eq. (4)). However, there are corrections due to finite tal errors translates into a value of the lifetime with corresponding errors. size and strong interaction [17]. These have been stud- ied based on the UrQMD transport code simulations − − 6. Lifetime of pionium [44] and DIRAC data on π π correlations. The pa- rameters of the underlying model are statistically fixed The lifetime may be deduced on the basis of the with data up to 200 MeV/c relative momentum. For relation between break-up probability and lifetime for Q 30 MeV/c, the DIRAC data are too scarce to a pure Ni target (Fig. 6). This relation, estimated to serve as a test of the model. The corrections lead to a be accurate at the 1% level, may itself have uncer- finite-size =− change of Pbr 0.02. Due to the uncertain- tainties due to the experimental conditions. Thus the ties we conservatively consider 1.5 times this change target thickness is estimated to be correct to better than as a maximum error, but do not modify Pbr. ±1 µm, which leads to an error in the lifetime (for The systematics are summarized in Table 4.Theex- Pbr = 0.45) smaller than ±0.01 fs, less than 1% of the treme values represent the ranges of the assumed uni- expected lifetime and thus negligible. The result for form probability density function (u.p.d.f.), which, in the lifetime is case of asymmetric errors, were complemented sym- = +0.45 +0.19 × −15 metrically for deducing the corresponding standard τ1S 2.91−0.38 −0.49 10 s stat syst deviations σ . Convoluting the five u.p.d.f. results in + − = 2.91 0.49 × 10 15 s. (12) bell-shaped curves very close to a Gaussian, and the −0.62 ±σ (Table 4, total error) correspond roughly to a The errors are not symmetric because the Pbr–τ rela- 68.5% confidence level and can be added in quadra- tion is not linear, and because finite size corrections ture to the statistical error. and heavy particle admixtures lead to possible smaller The final value of the break-up probability is values of Pbr. The accuracy achieved for the lifetime is about +17%, almost entirely due to statistics and −21%, due to statistics and systematics in roughly = ± +0.009 = +0.025 Pbr 0.452 0.023stat−0.032 syst 0.452−0.039. equal parts. With full statistics (2.3 times more than (11) analysed here) the statistical errors may be reduced 60 B. Adeva et al. / Physics Letters B 619 (2005) 50–60 accordingly. The two main systematic errors (particle J. Gasser, H. Leutwyler, Nucl. Phys. B 250 (1985) 517; admixtures and finite size correction) will be studied J. Gasser, H. Leutwyler, Nucl. Phys. B 250 (1985) 539. in more detail in the future program of DIRAC. [11] G. Colangelo, J. Gasser, H. Leutwyler, Nucl. Phys. B 603 (2001) 125. Using Eq. (1), the above lifetime corresponds to [12] M. Knecht, et al., Nucl. Phys. B 457 (1995) 513. | − |= +0.033 −1 a0 a2 0.264−0.020mπ . [13] L. Rosselet, et al., Phys. Rev. D 15 (1977) 547. [14] S. Pislak, et al., Phys. Rev. Lett. 87 (2001) 221801. [15] L.L. Nemenov, Yad. Fiz. 41 (1985) 980, Sov. J. Nucl. Phys. 41 Acknowledgements (1985) 629. [16] A.D. Sakharov, Zh. Eksp. Teor. Fiz. 18 (1948) 631. [17] R. Lednicky, DIRAC note 2004-06, nucl-th/0501065. We are indebted to the CERN PS crew for pro- [18] L.G. Afanasyev, O.O. Voskresenskaya, V.V. Yazkov, Commu- viding a beam of excellent quality. This work was nication JINR P1-97-306, Dubna, 1997. supported by CERN, the Grant Agency of the Czech [19] L.G. Afanasyev, et al., Phys. Lett. B 338 (1994) 478. Republic, grant No. 202/01/0779 and 202/04/0793, [20] L.S. Dulian, A.M. Kotsinian, Yad. Fiz. 37 (1983) 137, Sov. J. the Greek General Secretariat of Research and Tech- Nucl. Phys. 37 (1983) 78. [21] S. Mrówczynski,´ Phys. Rev. A 33 (1986) 1549; nology (Greece), the University of Ioannina Research S. Mrówczynski,´ Phys. Rev. D 36 (1987) 1520; Committee (Greece), the IN2P3 (France), the Istituto K.G. Denisenko, S. Mrówczynski,´ Phys. Rev. D 36 (1987) Nazionale di Fisica Nucleare (Italy), the Grant-in-Aid 1529. for Scientific Research from Japan Society for the Pro- [22] L.G. Afanasyev, A.V. Tarasov, Yad. Fiz. 59 (1996) 2212, Phys. motion of Science 07454056, 08044098, 09640376, At. Nucl. 59 (1996) 2130. [23] Z. Halabuka, et al., Nucl. Phys. B 554 (1999) 86. 09440012, 11440082, 11640293, 11694099, [24] A.V. Tarasov, I.U. Khristova, JINR-P2-91-10, Dubna, 1991. 12440069, 14340079, and 15340205, the Ministry [25] O.O. Voskresenskaya, S.R. Gevorkyan, A.V. Tarasov, Phys. At. of Education and Research, under project CORINT Nucl. 61 (1998) 1517. No.1/2004 (Romania), the Ministery of Industry, Sci- [26] L. Afanasyev, A. Tarasov, O. Voskresenskaya, J. Phys. G 25 ence and Technologies of the Russian Federation and (1999) B7. [27] D.Yu. Ivanov, L. Szymanowski, Eur. Phys. J. A 5 (1999) 117. the Russian Foundation for Basic Research (Russia), [28] T.A. Heim, et al., J. Phys. B 33 (2000) 3583. under project 01-02-17756, the Swiss National Sci- [29] T.A. Heim, et al., J. Phys. B 34 (2001) 3763. ence Foundation, the Ministerio de Ciencia y Tecnolo- [30] M. Schumann, et al., J. Phys. B 35 (2002) 2683. gia (Spain), under projects AEN96-1671 and AEN99- [31] L. Afanasyev, A. Tarasov, O. Voskresenskaya, Phys. Rev. D 65 0488, the PGIDT of Xunta de Galicia (Spain). (2002) 096001, hep-ph/0109208. [32] C. Santamarina, M. Schumann, L.G. Afanasyev, T. Heim, J. Phys. B 36 (2003) 4273. [33] L. Afanasyev, et al., J. Phys. B 37 (2004) 4749. References [34] L.G. Afanasyev, et al., Phys. Lett. B 308 (1993) 200. [35] B. Adeva, et al., Nucl. Instrum. Methods A 515 (2003) 467. [1] B. Adeva, et al., DIRAC proposal, CERN/SPSLC 95-1, SP- [36] B. Adeva, J. Phys. G 30 (2004) 1929. SLC/P 284 (1995). [37] Ch.P. Schuetz, Measurement of the breakup probability of + − [2] J. Uretsky, J. Palfrey, Phys. Rev. 121 (1961) 1798. π π atoms in a nickel target with the DIRAC spectrom- [3] H.-W. Hammer, J.N. Ng, Eur. Phys. J. A 6 (1999) 115. eter, PhD Thesis, Basel, March 2004, http://cdsweb.cern.ch/ [4] S. Deser, et al., Phys. Rev. 96 (1954) 774. searc.py?recid=732756. [5] S.M. Bilenky, et al., Yad. Phys. 10 (1969) 812, Sov. J. Nucl. [38] P. Zrelov, V. Yazkov, The GEANT-DIRAC Simulation Pro- Phys. 10 (1969) 469. gram, DIRAC note 1998-08, http://zrelov.home.cern.ch/zrelov/ [6] H. Jallouli, H. Sazdjian, Phys. Rev. D 58 (1998) 014011; dirac/montecarlo/instruction/instruct26.html. H. Jallouli, H. Sazdjian, Phys. Rev. D 58 (1998) 099901, Erra- [39] D. Drijard, M. Hansroul, V. Yazkov, The DIRAC offline user’s tum. guide, http://dirac.web.cern.ch/DIRAC/Userguide.html. [7] M.A. Ivanov, Phys. Rev. D 58 (1998) 094024. [40] C. Santamarina, Ch.P. Schuetz, DIRAC note 2003-9. [8] J. Gasser, et al., Phys. Rev. D 64 (2001) 016008, hep- [41] F. James, M. Roos, Function Minimization and Error Analysis, ph/0103157. Minuit, CERN Program Library, D506 MINUIT. [9] A. Gashi, et al., Nucl. Phys. A 699 (2002) 732. [42] A. Dudarev, et al., DIRAC note 2005-02. [10] S. Weinberg, Physica A 96 (1979) 327; [43] O.E. Gortchakov, V.V. Yazkov, DIRAC note 2005-01. J. Gasser, H. Leutwyler, Phys. Lett. B 125 (1983) 325; [44] S.A. Bass, et al., Prog. Part. Nucl. Phys. 41 (1998) 225. J. Gasser, H. Leutwyler, Nucl. Phys. B 250 (1985) 465; Physics Letters B 619 (2005) 61–70 www.elsevier.com/locate/physletb 0 A direct search for the CP-violating decay KS → 3π with the KLOE detector at DANE KLOE Collaboration F. Ambrosino d, A. Antonelli a, M. Antonelli a, C. Bacci i,P.Beltramea,G.Bencivennia, S. Bertolucci a,C.Binig, C. Bloise a, V. Bocci g, F. Bossi a,D.Bowringa,k, P. Branchini i, R. Caloi g, P. Campana a,G.Capona, T. Capussella d,F.Ceradinii,S.Chia, G. Chiefari d, P. Ciambrone a, S. Conetti k,E.DeLuciaa,A.DeSantisg, P. De Simone a, G. De Zorzi g, S. Dell’Agnello a,A.Denigb, A. Di Domenico g,C.DiDonatod, S. Di Falco e, B. Di Micco i,A.Doriad, M. Dreucci a, G. Felici a, A. Ferrari b, M.L. Ferrer a,G.Finocchiaroa,C.Fortia,P.Franzinig, C. Gatti g, P. Gauzzi g, S. Giovannella a, E. Gorini c, E. Graziani i, M. Incagli e,W.Klugeb, V. Kulikov 1, F. Lacava g,G.Lanfranchia, J. Lee-Franzini a,1, D. Leone b, M. Martini a,∗, P. Massarotti d,W.Meia,S.Meolad, S. Miscetti a,∗, M. Moulson a,S.Müllerb, F. Murtas a, M. Napolitano d, F. Nguyen i, M. Palutan a, E. Pasqualucci g,A.Passerii, V. Pat er a a,f,F.Perfettod, L. Pontecorvo g, M. Primavera c, P. Santangelo a, E. Santovetti h, G. Saracino d, B. Sciascia a, A. Sciubba a,f,F.Scurie,I.Sfiligoia, T. Spadaro a,M.Testag,L.Tortorai,P.Valenteg, B. Valeriani b, G. Venanzoni a, S. Veneziano g,A.Venturac, R. Versaci i,G.Xua,2 a Laboratori Nazionali di Frascati dell’INFN, Frascati, Italy b Institut für Experimentelle Kernphysik, Universität Karlsruhe, Germany c Dipartimento di Fisica dell’Università e Sezione INFN, Lecce, Italy d Dipartimento di Scienze Fisiche dell’Università “Federico II” e Sezione INFN, Napoli, Italy e Dipartimento di Fisica dell’Università e Sezione INFN, Pisa, Italy f Dipartimento di Energetica dell’Università “La Sapienza”, Roma, Italy g Dipartimento di Fisica dell’Università “La Sapienza” e Sezione INFN, Roma, Italy h Dipartimento di Fisica dell’Università “Tor Vergata” e Sezione INFN, Roma, Italy i Dipartimento di Fisica dell’Università “Roma Tre” e Sezione INFN, Roma, Italy j Physics Department, State University of New York at Stony Brook, USA k Physics Department, University of Virginia, USA Received 7 May 2005; received in revised form 23 May 2005; accepted 26 May 2005 Available online 2 June 2005 Editor: L. Rolandi 0370-2693/$ – see front matter 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.physletb.2005.05.063 62 KLOE Collaboration / Physics Letters B 619 (2005) 61–70 Abstract 0 + − We have searched for the decay KS → 3π with the KLOE experiment at DANE using data from e e collisions at a 2 −1 center of mass energy W ∼ mφc for an integrated luminosity L = 450 pb . The search has been performed with a pure KS beam obtained by tagging with KL interactions in the calorimeter and detecting six photons. We find an upper limit for the − branching ratio of 1.2 × 10 7 at 90% C.L. 2005 Elsevier B.V. All rights reserved. + − Keywords: e e collisions; DANE; KLOE; Rare KS decays; CP; CPT 0 −7 1. Introduction ing to BR(KS → 3π ) 7.4 × 10 at 90% C.L. The sensitivity to CPT violation via unitarity [5] → 0 The decay KS 3π violates CP invariance. The is now limited by the error in η+− = A(KL → + − + − parameter η000, defined as the ratio of KS to KL de- π π )/A(KS → π π ). cay amplitudes, can be written as: η000 = A(KS → We report in the following an improved limit from 0 → 0 = + 0 3π )/A(KL 3π ) , where quantifies a direct search for the 3π decays of the KS . Apart 000 the KS CP impurity and 000 is due to a direct CP- from the interest in confirming the Standard Model, violating term. Since we expect 000 [1],itfol- knowledge of η000 allows tests of the validity of CPT lows that η000 ∼ . In the Standard Model, therefore, invariance using unitarity. 0 −9 BR(KS → 3π ) ∼ 1.9 × 10 to an accuracy of a few %, making the direct observation of this decay quite a challenge. 2. DANE and KLOE 0 The best upper limit on BR(KS → 3π ) from a search for the decay was obtained by the SND exper- 0 The data were collected with the KLOE detec- iment at Novosibirsk. They find BR(KS → 3π ) − tor [6–9] at DANE [10], the Frascati φ factory. 1.4 × 10 5 at 90% C.L. [2]. CPLEAR has pioneered DANE is an e+e− collider operated at a center-of- the method of searching for interference between K S mass energy W ∼ 1020 MeV, the mass of the φ me- and K decays. Interference results in the appear- L son. Positron and electron beams of equal energy col- ance of a term (η ) cos(mt) −(η ) sin(mt) 000 000 lide at an angle of π − 0.025 rad, producing φ mesons in the decay intensity. (η ) and (η ) are ob- 000 000 nearly at rest (p ∼ 12.5MeV). φ mesons decay 34% tained from a fit, without discriminating between φ of the time into nearly collinear K0K¯ 0 pairs. Because K or K → 3π 0 decays. In this way CPLEAR L S J PC(φ) = 1−−, the kaon pair is in a C-odd antisym- finds η = (0.18 ± 0.15) + i(0.15 ± 0.20) [3]. 000 metric state, so that the final state is always K –K . The NA48 Collaboration [4] has recently reached S L Detection of a K signals the presence of a K of much higher sensitivity. By fitting the K /K → L S S L known momentum and direction. We say that detec- 3π 0 interference pattern at small decay times, they tion of a K “tags” the K . find (η ) =−0.002 ± 0.011 ± 0.015 and L S 000 stat sys The KLOE detector consists of a large cylindrical (η ) =−0.003 ± 0.013 ± 0.017 , correspond- 000 stat sys drift chamber (DC), surrounded by a lead/scintillating- fiber electromagnetic calorimeter (EMC). A supercon- * Corresponding authors. Mailing address: INFN, LNF, Casella ducting coil around the calorimeter provides a 0.52 T postale 13, 00044 Frascati (Roma), Italy. field. The drift chamber, 4 m in diameter and 3.3 m E-mail addresses: [email protected] (M. Martini), long, is described in Ref. [6]. The momentum reso- [email protected] (S. Miscetti). ≈ 1 Permanent address: Institute for Theoretical and Experimental lution is σ(p⊥)/p⊥ 0.4%. Two track vertices are Physics, Moscow, Russia. reconstructed with a spatial resolution of ∼3 mm. The 2 Permanent address: Institute of High Energy Physics of Acad- calorimeter, described in Ref. [7], is divided into a emica Sinica, Beijing, China. barrel and two endcaps, for a total of 88 modules, KLOE Collaboration / Physics Letters B 619 (2005) 61–70 63 and covers 98% of the solid angle. The modules are (IP) to the EMC. Each cluster is required to satisfy read out at both ends by photomultipliers providing the condition |T − R/c| < min(3σT , 2ns), where T is energy deposit and arrival time information. The read- the photon flight time and R the path length; σT also out segmentation provides the coordinates transverse includes a contribution from the finite bunch length to the fiber plane. The coordinate along the fibers is (2–3 cm), which introduces a dispersion in the colli- obtained by the difference between the arrival times of sion time. the signals at either end. Cells close in time and space In order to retain a large control sample for the are grouped into calorimeter clusters.√ The energy and background while preserving high efficiency for the time resolutions√ are σE/E = 5.7%/ E (GeV) and signal, we keep all photons satisfying Eγ > 7MeV σT = 54 ps/ E (GeV) ⊕ 50 ps, respectively. and | cos(θ)| < 0.915. The photon detection efficiency The KLOE trigger, described in Ref. [9],uses is ∼90% for Eγ = 20 MeV, and reaches 100% above calorimeter and chamber information. For this analy- 70 MeV. The signal is searched for by requiring six sis, only the calorimeter signals are used. Two en- prompt photons after tagging. ergy deposits above threshold (E>50 MeV for The normalization is provided by counting the 0 the barrel and E>150 MeV for the endcaps) are KS → 2π events in the same tagged sample. required. Recognition and rejection of cosmic-ray events is also performed at the trigger level. Events with two energy deposits above a 30 MeV threshold 3. Monte Carlo simulation in two of the outermost calorimeter planes are re- jected. The response of the detector to the decay of inter- During 2002 data taking, the maximum luminos- est and the various backgrounds is studied using the ity reached by DANE was 7.5 × 1031 cm−2 s−1, and Monte Carlo (MC) program GEANFI [11]. GEANFI in September 2002, DANE delivered 91.5pb−1.We accounts for changes in machine operation and back- collected data in 2001–2002 for an integrated luminos- ground conditions, following the machine conditions ity L = 450 pb−1. A total of 1.4 billion φ mesons were run by run, and has been calibrated with Bhabha scat- produced, yielding 450 million KS –KL pairs. Assum- tering events and other processes. The response of the 0 −9 ing BR(KS → 3π ) = 1.9 × 10 , ∼1 signal event is EMC to KL interactions is not simulated but has been expected to have been produced. obtained from a large sample of KL-mesons tagged by + − The mean decay lengths of the KS and KL are identifying KS → π π decays. This not only gives λS ∼ 0.6 cm and λL ∼ 340 cm at DANE. About accurate representation of the EMC response to the 50% of KL’s reach the calorimeter before decaying. KL crash, but also results in an effective 40% increase The KL interaction in the calorimeter (“KL crash”) is in MC statistics. The KL-crash efficiency cancels in identified by requiring a cluster with energy greater the final 3π 0/2π 0 ratio to better than 1% and we as- than 100 MeV that is not associated to any track and sign a 0.9% systematic error to the final result due to whose time corresponds to a velocity in the φ rest this source. ∗ frame, β ,of∼0.2. The KL-crash provides a very Backgrounds are obtained from MC φ → KSKL clean KS tag. The average value of the center-of-mass events corresponding to an integrated luminosity L = energy, W , is obtained with a precision of 30 keV for 900 pb−1.WealsouseaMCsampleofφ → K+K− each 100 nb−1 running period (of duration ∼1h)us- events for L = 450 pb−1 and a MC sample of radiative ing large-angle Bhabha events. The value of W and φ decays for L = 2250 pb−1.Asampleof∼340 000 0 the KL-crash cluster position allows us to establish, KS → 3π MC events is used to obtain the signal ef- for each event, the trajectory of the KS with an angu- ficiency. lar resolution of 1◦ and a momentum resolution better than 2 MeV. Because of its very short lifetime, the displacement 4. Photon counting for data and Monte Carlo of the KS from the φ decay position is negligible. We therefore identify as KS decay photons neutral parti- To test how well the MC reproduces the observed cles that travel with β = 1 from the interaction point photon multiplicity after tagging, we determine the 64 KLOE Collaboration / Physics Letters B 619 (2005) 61–70 Table 1 background composition when comparing data and Measured values of F for data and Monte Carlo samples, in percent MC samples. Data 2001 MC 2001 Data 2002 MC 2002 F(3) 30.95 ± 0.16 30.31 ± 0.11 30.79 ± 0.12 30.06 ± 0.08 F(4) 67.35 ± 0.23 67.93 ± 0.17 67.93 ± 0.18 68.15 ± 0.12 F(5) 1.55 ± 0.01 1.80 ± 0.01 1.19 ± 0.01 1.66 ± 0.01 5. Data analysis F(6) 0.15 ± 0.01 0.14 ± 0.01 0.08 ± 0.01 0.13 ± 0.01 0 KS → 3π candidates consist of a KL crash plus six photons. In our data sample of L = 450 pb−1,we Table 2 find 39 538 events, essentially all background. After Measured values of the probabilities PA and PS removing background, we obtain the branching ratio F(K) Data 2001 MC 2001 Data 2002 MC 2002 0 by normalizing to the number of KS → 2π events. 2 10 × PA(1) 0.75 ± 0.30 1.03 ± 0.16 0.38 ± 0.17 0.89 ± 0.08 The latter are found by asking for three to five prompt 2 × ± ± ± ± 10 PA(2) 0.14 0.05 0.16 0.03 0.07 0.02 0.10 0.03 photons plus the K -crash. 3 L 10 × PS (1) 3.6 ± 0.23.8 ± 0.33.7 ± 0.23.3 ± 0.1 4 According to the MC, the six-photon sample is 10 × PS (2) 1.5 ± 0.41.5 ± 0.30.9 ± 0.21.7 ± 0.2 0 dominated (95%) by KS → 2π decays plus two ad- ditional photon clusters. These clusters are due to fragmented or split showers (2S, 1S + 1A, 34%) and = fraction of events of given multiplicity, Nγ k, de- to accidental photons from machine background (2A, = = 6 = fined as F(k) Nev(Nγ k)/ i=3 Nev(Nγ i).As 64%). About 2% of the background events are due to + − 0 shown in Table 1, there is a significant discrepancy false KL-crash tags from φ → KSKL → π π ,3π between data and Monte Carlo for events with mul- events. In such events, charged pions from KS de- tiplicity five and six. These samples are dominated by cays interact in the low-beta insertion quadrupoles,3 → 0 KS 2π decays plus additional clusters due either ultimately simulating the KL-crash signal, while KL to shower fragmentation (split clusters) or the acci- decays close to the IP produce six photons. Similarly, dental coincidence of machine background photons φ → K+K− events give a false signal (∼1%), as well (accidental clusters). To understand this discrepancy, as φ → ηγ → 3π 0γ events (∼0.3%). The cuts de- we have measured the probability, PA(1, 2),ofhav- scribed in the following make the latest two sources ing one, or more than one, accidental cluster passing of contamination negligible. our selection by extrapolating the rates measured in an To reduce the background, we first perform a kine- out-of-time window, (−68 T −14) ns, that is ear- matic fit with 11 constraints: energy and momentum lier than the bunch crossing. In Table 2, we list the conservation, the kaon mass and the velocity of the six average values of these probabilities. The observed photons. The χ2 distribution of the fit to data and MC discrepancy has been traced to an understood prob- background is shown in Fig. 1. In the same plot, we lem with the procedure for the selection of machine- also show the expected shape for signal events. Cutting background clusters. at a reasonable χ2 value (χ2/11 < 3) retains 71% of The MC-true fraction of events with a given mul- the signal while considerably reducing the background tiplicity, fMC, is obtained by ignoring clusters due to from false KL-crash events (33%), in which the direc- machine background and counting at most one cluster tion of the KS and KL are not correlated. However, per simulated particle incident on the calorimeter. Us- this cut is not as effective on the 2S, 2A background, ing the fractions fMC, together with the values of PA due to the soft energy spectrum of the fake clusters. obtained as discussed above, we fit the observed F(k) In order to gain rejection power over the background distribution to get the probability for a cluster to gen- for events with split and accidental clusters, we look erate fragments, PS (see Table 2). This fit accurately at the correlation between the following two χ2-like reproduces the observed fractions in the multiplicity bins five and six. More details on these measurements can be found in Ref. [12]. The results of this study 3 The first quadrupoles are located approximately 45 cm on ei- demonstrate the need for careful calibration of the ther side of the IP. KLOE Collaboration / Physics Letters B 619 (2005) 61–70 65 0 to zero for a KS → 3π event and large for six- photon background events. For each estimator, the photon pairing with smallest ζ value is kept. Fig. 2a shows the distribution of events in the ζ2–ζ3 plane for the MC background. Most of the events are concentrated at low values of ζ2,as 0 expected for KS → 2π events plus some additional isolated energy deposits in the EMC. A clear sig- nal/background separation is achieved as can be seen by comparing the background and signal distributions in Figs. 2a and b. We subdivide the ζ2-ζ3 plane into the six regions B1, B2, B3, B4, B5, and S as indi- cated in Fig. 2a. Region S, with the largest signal-to- background value, is the “signal” box. The scatter plot in the ζ2–ζ3 plane for the data is χ2 Fig. 1. Distribution of for the tagged six-photon sample for data shown in Fig. 3a. Our MC simulation does not accu- (points), MC background (solid line), and 10 000 events of MC sig- nal (dashed line). rately reproduce the absolute number of 2S and 2A background events. This is also true of the predicted number of false KL-crash events. However, the simu- estimators: lation does describe the kinematical properties of these events quite well. The two-dimensional ζ2–ζ3 distri- • ζ2, defined as bution allows us to calibrate the contributions from m2 m2 (E − E )2 the different backgrounds. The MC shapes for each of ζ = 1 + 2 + KS i γi 2 2 2 2 the three categories are shown in Figs. 3b–d. We per- σm σm σ E form a binned likelihood fit of a linear combination (P KS − P γi )2 (P KS − P γi )2 of these shapes to the data, excluding the signal-box + x i x + y i y 2 2 region. From the fit we find the composition of the six- σP σP x y photon sample to be (37.9 ± 1.0)%, (57.4 ± 1.3)%, γ (P KS − P i )2 (π − ϑ∗ )2 and (4.7 ± 0.3)% for the 2S, 2A, and false K -crash + z i z + ππ , L 2 2 categories, respectively. σP σϑ∗ z ππ As a check, we compare data and MC for the pro- selecting the four out of six photons that provide jected distribution in ζ3 for the three bands in ζ2,as → 0 the best kinematic agreement with the KS 2π shown in Figs. 4a, b. Excellent agreement is observed. decay hypothesis. This variable is quite insensitive The large peak at low values of ζ3 in the central band to fake clusters. It is constructed using the two val- is due to the false K -crash events. As a final test, we = − L ues of m mi mπ0 (where mi is the invariant compare data and MC in the signal box and the five mass of a photon pair), the opening angle between surrounding control regions. The agreement is better 0 π ’s in the KS rest frame, and 4-momentum con- than 10% in all regions, as seen from Table 3. servation. The resolutions on these quantities have Although cutting on χ2 substantially suppresses been evaluated using a control sample of events the false KL-crash background, we reduce this back- with a KL-crash and four prompt photons. ground to a negligible level by vetoing events with • ζ3, defined as tracks coming from the IP. This effectively elimi- 2 2 2 nates events in which the false KL-crash is due to m1 m2 m3 + − ζ3 = + + , a K → π π decay with the pion secondaries in- σ 2 σ 2 σ 2 S m m m teracting in the quadrupoles. The effect on the signal where the pairing of the six photons into π 0’s is region can be appreciated by comparison of Figs. 4a performed by minimizing this variable. ζ3 is close and c. Moreover, in order to improve the quality of 66 KLOE Collaboration / Physics Letters B 619 (2005) 61–70 Fig. 2. Scatter plot of ζ2 vs. ζ3 plane for the tagged six-photon sample: (a) MC background, (b) MC signal. Fig. 3. Scatter plots of ζ2 vs. ζ3 for the tagged six-photon sample: data (a), MC sample with two split clusters (b), two accidental clusters (c), and false KL-crash events (d). Table 3 Comparison between data and MC expectations in the different regions of the ζ2–ζ3 plane for the entire sample with a KL-crash and six prompt photons. The boxes are defined as in Fig. 2a B1 B2 S B3 B5 B4 Data 452 ± 21 10132 ± 101 282 ± 17 5037 ± 71 326 ± 18 22309 ± 149 MC 419 ± 19 9978 ± 104 282 ± 13 4816 ± 43 380 ± 10 22682 ± 190 KLOE Collaboration / Physics Letters B 619 (2005) 61–70 67 Fig. 4. Distributions in ζ3 for the tagged six-photon sample. Plots on the left are for events in the central band in ζ2; plots on the right are for events in all other regions of the plane. The plots in the top row are for the entire sample, before any cuts are made. The plots in the bottom row are after the application of the track veto. In all cases, black points represent data; solid line represents MC. Fig. 5. Distributions of ζ3 for the central band 12.1 <ζ2 < 60 (a), the side-bands ζ2 < 12.1, ζ2 > 60 (b), after all cuts. Points represent data, solid line MC. the photon selection using ζ2, we cut on the variable and ∆>1.7. The signal box is defined by 12.1 <ζ2 < 2 ∆ = (mφc /2 − Ei)/σE, where i = 1–4 stands for 60 and ζ3 < 4.6. the four chosen photons in the ζ2 estimator and σE is Figs. 5a and b show the ζ3 distributions for the cen- 0 the appropriate resolution. For KS → 2π decays plus tral band and the sidebands in ζ2. two background clusters, we expect ∆ ∼ 0, while for In Table 4, we also list the number of events ob- → 0 0 0 2 KS π π π , ∆ mπ0 c /σE. tained in each of the six regions of the ζ2–ζ3 plane Before opening the signal box, we refine our cuts at this final stage of the analysis. In Figs. 6a, b we 2 on χ , ζ2, ζ3, and ∆ using the optimization procedure show the ζ2–ζ3 scatter plots for data and Monte Carlo. described in Ref. [13]. We end up choosing χ2 < 40.4 The rectangular region illustrates the boundaries of 68 KLOE Collaboration / Physics Letters B 619 (2005) 61–70 −1 −1 Fig. 6. Distribution of ζ2 vs. ζ3 after cuts: MC background 900 pb (a), data 450 pb (b). Table 4 Same as Table 3, after all cuts. The background in the signal box is expected by MC to be composed of (2.30 ± 0.64), (0.28 ± 0.20) and (0.47 ± 0.47) events for the 2S, 2A and false KL-crash category respectively B1 B2 S B3 B4 B5 Data 0 4 ± 2 2.0 ± 1.4 520 ± 23 3 ± 2 326 ± 18 MC 0 3.2 ± 0.8 3.1 ± 0.8 447 ± 10 2.5 ± 0.8 389 ± 10 the optimized signal box. Seventeen MC events are 6. Systematic uncertainties counted in this region before applying the data-MC scale factors resulting from the calibration procedure Systematics arise from uncertainties in estimation described above. Contributions to the scale factors of the acceptance, backgrounds, and the analysis effi- include the fact that the simulated integrated lumi- ciency. The evaluation of the systematic uncertainties nosity is greater than that for the data set (×2), is described in detail in Ref. [14]. the increased KL-crash efficiency in the simulation Concerning the acceptance of the event selection (×1.4), and the increased probability of having acci- for both the 2π 0 and 3π 0 samples, we estimate the dental or split clusters in the simulation (on average, systematic errors in photon counting by comparing ∼× 1.9). data and MC values for the PA and PS probabilities The selection efficiency at each step of the analysis described above. The photon reconstruction efficiency has been studied using the MC. After tagging, the effi- for both data and MC is evaluated using a large sample + − 0 0 ciency for the six-photon selection is (47.8±0.1stat)%. of φ → π π π , π → γγ events. The momentum Including all cuts, we estimate a total efficiency of of one of the photons is estimated from tracking infor- 3π = (24.4 ± 0.1stat)%. At the end of the analysis mation and position of the other cluster. The candidate chain, we have two candidates with an expected back- photon is searched for within a search cone. The ef- exp ground of B = 3.13 ± 0.82stat. ficiency is parameterized as a function of the photon In the same tagged sample, we also count events energy. Systematics related to this correction are ob- with photon multiplicities of three, four, or five. tained from the variation of the efficiency as a function The corresponding efficiency is (91.8 ± 0.2stat)%for of the width of the search cone. The results are listed 0 0 KS → π π events. The residual background con- in Table 5 under the heading cluster. The total uncer- tamination is estimated to be (0.77 ± 0.24stat+sys)% tainty is smaller for the normalization sample since an and (0.65 ± 0.10stat+sys)% in the 2001 and 2002 run- inclusive selection criterion is used in this case. ning periods, respectively. Subtracting the background The normalization sample also suffers a small and correcting for the efficiency, we count 3.78 × 107 (0.4%) loss due to the use of a filter during data re- 0 0 KS → π π events. We use this number to normal- construction to reject cosmic rays, Bhabha fragments ize the number of signal events when obtaining the from the low-beta quadrupole, and machine back- branching ratio. ground events. This loss is estimated using the MC. KLOE Collaboration / Physics Letters B 619 (2005) 61–70 69 Table 5 Table 6 Systematic acceptance uncertainties, α,forthe2π0 and 3π0 event Systematic uncertainties on the expected background and analysis selection criteria efficiency, ana 0 0 exp exp α/α (KS → 2π ) α/α (KS → 3π ) B /B ana/ana Cluster 0.16% 0.70% Background composition 2.4% − Trigger 0.08% 0.08% Track veto 4.8% 0.2% Background filter 0.20% 0.08% Energy resolution 6.6% 0.5% Energy scale 6.7% 1.0% Total 0.27% 0.71% χ2 5.0% 1.8% Total 11.5% 2.1% We correct for it and add a 0.2% systematic error to the selection efficiency. The trigger and cosmic-ray utions for data and MC. An error of 5% is obtained. veto efficiencies have been estimated with data for A summary of all the systematic errors on the back- the normalization sample and extrapolated by MC to ground estimate is given in Table 6. Adding in quadra- the signal sample. These efficiencies are very close to ture all sources we obtain a total systematic error of unity and the related systematics are negligible. 12% on the background estimate. For the tagged six-photon sample, we have inves- To determine the systematics related to the analysis tigated the uncertainties related to the estimate of the cuts for the signal, we have first evaluated the effect exp background in the signal box after all cuts, B .We of the track veto. Using the MC signal sample, we es- have first considered the error related to the calibration timate a vetoed event fraction of (3.7 ± 0.1)%. The of the MC background composition by propagating data-MC ratio of the cumulative distributions for the the errors on the scale factors obtained from the fit. track-vetoed events in the tagged six-photon sample is exp This corresponds to a relative error of 2.4% on B . RTV = 1.06 ± 0.03, which translates into a 0.2% sys- Moreover, we have investigated the extent to which tematic error on the track-veto efficiency. the track-veto efficiency influences the residual false Because of the similarity of the χ2 distributions KL-crash contamination. To do so, we examine the ∗ for the tagged four- and six-photon samples, as con- data-MC ratio, Rβ , of the sidebands in β for events ∗ firmed by MC studies, an estimate of the systematic rejected by this veto, since for true KL’s β peaks at error associated with the application of the χ2 cut can ∼ 0.2 while false KL-crashes are broadly distributed be obtained from the data-MC comparison of the cu- ∗ = ± in β . We obtain Rβ 1.10 0.01. Knowing that in mulative χ2 distributions for the four-photon sample. the MC only 24% of the fakes survive the veto, we The systematic error arising from data-MC discrepan- find a fractional error of 32% on the fake background. cies in the χ2 distribution is estimated to be 1.8% by Since false KL-crash events account for 15% of the this comparison. exp total background, the error on B from data-MC dif- Moreover, the efficiency changes related to differ- ferences in the track veto efficiency is 4.6%. ences between the calorimeter resolution and energy → 0 A control sample of KS 2π with four prompt scale for data and MC events have been studied in photons has been used to compare the energy scale a manner similar to that previously described for the and resolution of the calorimeter in data and in the evaluation of the systematics on the background. The MC. The distributions of the m and ∆ variables have systematic uncertainties on the analysis efficiency are also been compared by fitting them with Gaussians. summarized in Table 6. Adding all sources in quadra- By varying the mass and energy resolution by ±1σ ture we quote a total systematic error of 2.1% on the in the definitions of ζ2 and ζ3, we observe a relative estimate of the analysis efficiency. change of 6.6% in the background estimate. Similarly, correcting for small differences in the energy scale for data and MC, we derive a systematic uncertainty of 7. Results 6.7% on Bexp. Finally, we have tested the effect of the cut on χ2 by At the end of the analysis, we find 2 events in the constructing the ratio between the cumulative distrib- signal box with an estimated background of Bexp = 70 KLOE Collaboration / Physics Letters B 619 (2005) 61–70 3.13 ± 0.82stat ± 0.37syst. To derive an upper limit detector; A. Balla, M. Gatta, G. Corradi and G. Pa- on the number of signal counts, we build the back- palino for the maintenance of the electronics; M. San- ground probability distribution function, taking into toni, G. Paoluzzi and R. Rosellini for the general sup- account our finite MC statistics and the uncertain- port the detector; C. Piscitelli for his help during major ties on the MC calibration factors. This function is maintenance periods. folded with a Gaussian of width equivalent to the en- This work was supported in part by DOE grant tire systematic uncertainty on the background. Using DE-FG-02-97ER41027; by EURODANE contract the Neyman construction described in Ref. [15],we FMRX-CT98-0169; by the German Federal Ministry 0 limit the number of KS → 3π decays observed to of Education and Research (BMBF) contract 06-KA- 3.45 at 90% C.L., with a total reconstruction efficiency 957; by Graduiertenkolleg ‘H.E. Phys. and Part. Astro- of (24.36 ± 0.11stat ± 0.57sys)%. In the same tagged phys.’ of Deutsche Forschungsgemeinschaft, Contract 7 0 0 sample, we count 3.78×10 KS → π π events. This No. GK 742; by INTAS, contracts 96-624, 99-37; and number is used for normalization. Finally, using the by the EU Integrated Infrastructure Initiative Hadron 0 0 value BR(KS → π π ) = 0.3105 ± 0.0014 [16] we Physics Project under contract number RII3-CT-2004- obtain: 506078. 0 −7 BR KS → 3π 1.2 × 10 at 90% C.L., (1) which represents an improvement by a factor of ∼6 References with respect to the best previous limit [4], and by a factor of 100 with respect to the best limit obtained [1] G. D’Ambrosio, et al., in: L. Maiani, et al. (Eds.), The Second with a direct search [2]. DANE Handbook, vol. 63, Frascati, 1995. [2] SND Collaboration, M.N. Achasov, et al., Phys. Lett. B 459 The limit on the BR can be directly translated into (1999) 674. a limit on |η000|: [3] CPLEAR Collaboration, A. Angelopoulos, et al., Phys. Lett. B 425 (1998) 391. A(K → 3π 0) τ BR(K → 3π 0) [4] NA48 Collaboration, A. Lai, et al., Phys. Lett. B 610 (2005) |η |= S = L S 000 0 0 165. A(KL → 3π ) τS BR(KL → 3π ) [5] G.B. Thomson, Y. Zou, Phys. Rev. D 51 (1995) 1412. < 0.018 at 90% C.L. (2) [6] KLOE Collaboration, M. Adinolfi, et al., Nucl. Instrum. Meth- ods A 488 (2002) 51. This result describes a circle of radius 0.018 cen- [7] KLOE Collaboration, M. Adinolfi, et al., Nucl. Instrum. Meth- teredatzerointhe (η000), (η000) plane and rep- ods A 482 (2002) 364. resents a limit 2.5 times smaller than the result of [8] KLOE Collaboration, M. Adinolfi, et al., Nucl. Instrum. Meth- Ref. [4]. As follows from the discussion in that ref- ods A 483 (2002) 649. erence, our result confirms that the sensitivity of the [9] KLOE Collaboration, M. Adinolfi, et al., Nucl. Instrum. Meth- ods A 492 (2002) 134. CPT test from unitarity is now limited by the uncer- [10] S. Guiducci, Status report on DANE, in: P. Lucas, S. Weber tainty on η+−. (Eds.), Proceedings of the 2001 Particle Accelerator Confer- ence, Chicago, IL, 2001. [11] KLOE Collaboration, F. Ambrosino, et al., Nucl. Instrum. Acknowledgements Methods A 453 (2004) 403. [12] M. Martini, S. Miscetti, Determination of the probability of accidental coincidence between machine background and col- We thank the DANE team for their efforts in lision events and fragmentation of electromagnetic showers, maintaining low-background running conditions and KLOE note 201, 2005, http://www.lnf.infn.it/kloe. their collaboration during all data taking. We would [13] J.F. Grivaz, F. Le Diberder, LAL 92-37, 1992. → 0 like to thank our technical staff: G.F. Fortugno for his [14] M. Martini, S. Miscetti, A direct search for KS 3π decay, dedicated work to ensure efficient operations of the KLOE note 200, 2005, http://www.lnf.infn.it/kloe. [15] G.J. Feldman, R. Cousins, Phys. Rev. D 57 (1998) 57. KLOE computing facilities; M. Anelli for his contin- [16] S. Eidelman, et al., Phys. Lett. B 592 (2004). uous support of the gas system and the safety of the Physics Letters B 619 (2005) 71–81 www.elsevier.com/locate/physletb Measurement of the cross section for open-beauty production in photon–photon collisions at LEP L3 Collaboration P. Achard t, O. Adriani q, M. Aguilar-Benitez y, J. Alcaraz y, G. Alemanni w, J. Allaby r, A. Aloisio ac, M.G. Alviggi ac, H. Anderhub aw, V.P. Andreev f,ah,F.Anselmoh, A. Arefiev ab, T. Azemoon c, T. Aziz i,P.Bagnaiaam,A.Bajoy,G.Baksayz,L.Baksayz, S.V. Baldew b,S.Banerjeei,Sw.Banerjeed, A. Barczyk aw,au, R. Barillère r, P. Bartalini w, M. Basile h,N.Batalovaat, R. Battiston ag,A.Bayw, F. Becattini q, U. Becker m,F.Behneraw, L. Bellucci q, R. Berbeco c, J. Berdugo y,P.Bergesm, B. Bertucci ag,B.L.Betevaw,M.Biasiniag, M. Biglietti ac,A.Bilandaw, J.J. Blaising d, S.C. Blyth ai, G.J. Bobbink b,A.Böhma, L. Boldizsar l,B.Borgiaam,S.Bottaiq, D. Bourilkov aw, M. Bourquin t, S. Braccini t,J.G.Bransonao,F.Brochud,J.D.Burgerm, W.J. Burger ag,X.D.Caim, M. Capell m, G. Cara Romeo h, G. Carlino ac, A. Cartacci q, J. Casaus y, F. Cavallari am, N. Cavallo aj, C. Cecchi ag, M. Cerrada y,M.Chamizot, Y.H. Chang ar,M.Chemarinx,A.Chenar,G.Cheng,G.M.Cheng,H.F.Chenv, H.S. Chen g, G. Chiefari ac, L. Cifarelli an, F. Cindolo h,I.Clarem,R.Clareal, G. Coignet d,N.Colinoy, S. Costantini am,B.delaCruzy, S. Cucciarelli ag, R. de Asmundis ac, P. Déglon t, J. Debreczeni l,A.Degréd,K.Dehmeltz, K. Deiters au, D. della Volpe ac, E. Delmeire t, P. Denes ak, F. DeNotaristefani am,A.DeSalvoaw, M. Diemoz am, M. Dierckxsens b, C. Dionisi am, M. Dittmar aw,A.Doriaac,M.T.Dovaj,5, D. Duchesneau d, M. Duda a, B. Echenard t,A.Eliner,A.ElHagea, H. El Mamouni x, A. Engler ai,F.J.Epplingm, P. Extermann t, M.A. Falagan y, S. Falciano am,A.Favaraaf, J. Fay x,O.Fedinah, M. Felcini aw, T. Ferguson ai, H. Fesefeldt a, E. Fiandrini ag, J.H. Field t, F. Filthaut ae,P.H.Fisherm, W. Fisher ak, I. Fisk ao, G. Forconi m, K. Freudenreich aw,C.Furettaaa, Yu. Galaktionov ab,m, S.N. Ganguli i, P. Garcia-Abia y, M. Gataullin af, S. Gentile am,S.Giaguam, Z.F. Gong v,G.Grenierx,O.Grimmaw, M.W. Gruenewald p, M. Guida an, V.K. Gupta ak,A.Gurtui,L.J.Gutayat, D. Haas e, D. Hatzifotiadou h, T. Hebbeker a,A.Hervér,J.Hirschfelderai,H.Hoferaw, M. Hohlmann z,G.Holzneraw,S.R.Houar,B.N.Jing, P. Jindal n, L.W. Jones c, 0370-2693/$ – see front matter 2005 Published by Elsevier B.V. doi:10.1016/j.physletb.2005.05.072 72 L3 Collaboration / Physics Letters B 619 (2005) 71–81 P. de Jong b, I. Josa-Mutuberría y, M. Kaur n, M.N. Kienzle-Focacci t,J.K.Kimaq, J. Kirkby r, W. Kittel ae, A. Klimentov m,ab, A.C. König ae, M. Kopal at, V. Koutsenko m,ab, M. Kräber aw, R.W. Kraemer ai,A.Krügerav, A. Kunin m, P. Ladron de Guevara y, I. Laktineh x, G. Landi q, M. Lebeau r,A.Lebedevm,P.Lebrunx,P.Lecomteaw, P. Lecoq r,P.LeCoultreaw,J.M.LeGoffr,R.Leisteav,M.Levtchenkoaa, P. Levtchenko ah,C.Liv, S. Likhoded av,C.H.Linar,W.T.Linar,F.L.Lindeb, L. Lista ac, Z.A. Liu g, W. Lohmann av, E. Longo am,Y.S.Lug,C.Luciam, L. Luminari am, W. Lustermann aw,W.G.Mav,L.Malgerir, A. Malinin ab,C.Mañay,J.Mansak, J.P. Martin x, F. Marzano am, K. Mazumdar i, R.R. McNeil f,S.Meler,ac,L.Merolaac, M. Meschini q, W.J. Metzger ae,A.Mihulk,H.Milcentr, G. Mirabelli am,J.Mnicha, G.B. Mohanty i,G.S.Muanzax, A.J.M. Muijs b,B.Musicarao,M.Musyam, S. Nagy o, S. Natale t, M. Napolitano ac, F. Nessi-Tedaldi aw,H.Newmanaf,A.Nisatiam, T. Novak ae,H.Nowakav, R. Ofierzynski aw,G.Organtiniam,I.Palat,C.Palomaresy, P. Paolucci ac, R. Paramatti am,G.Passalevaq,S.Patricelliac,T.Paulj, M. Pauluzzi ag, C. Paus m,F.Paussaw, M. Pedace am,S.Pensottiaa, D. Perret-Gallix d,D.Piccoloac, F. Pierella h, M. Pioppi ag,P.A.Pirouéak, E. Pistolesi aa, V. Plyaskin ab, M. Pohl t, V. Pojidaev q, J. Pothier r,D.Prokofievah, G. Rahal-Callot aw, M.A. Rahaman i, P. Raics o,N.Rajai, R. Ramelli aw,P.G.Rancoitaaa,R.Ranieriq, A. Raspereza av, P. Razis ad,D.Renaw, M. Rescigno am, S. Reucroft j, S. Riemann av,K.Rilesc,B.P.Roec, L. Romero y,A.Roscaav, C. Rosemann a, C. Rosenbleck a,S.Rosier-Leesd,S.Rotha, J.A. Rubio r, G. Ruggiero q, H. Rykaczewski aw,A.Sakharovaw, S. Saremi f,S.Sarkaram, J. Salicio r, E. Sanchez y, C. Schäfer r, V. Schegelsky ah, H. Schopper u, D.J. Schotanus ae, C. Sciacca ac,L.Servoliag, S. Shevchenko af,N.Shivarovap, V. Shoutko m, E. Shumilov ab, A. Shvorob af,D.Sonaq,C.Sougax, P. Spillantini q, M. Steuer m, D.P. Stickland ak,B.Stoyanovap,A.Straessnert, K. Sudhakar i, G. Sultanov ap, L.Z. Sun v,S.Sushkova,H.Suteraw,J.D.Swainj, Z. Szillasi z,3,X.W.Tangg,P.Tarjano, L. Tauscher e, L. Taylor j, B. Tellili x, D. Teyssier x,C.Timmermansae, Samuel C.C. Ting m, S.M. Ting m, S.C. Tonwar i,J.Tóthl,C.Tullyak,K.L.Tungg, J. Ulbricht aw, E. Valente am,R.T.VandeWalleae, R. Vasquez at, V. Veszpremi z, G. Vesztergombi l, I. Vetlitsky ab,G.Viertelaw, S. Villa al, M. Vivargent d,S.Vlachose, I. Vodopianov z,H.Vogelai,H.Vogtav, I. Vorobiev ai,ab, A.A. Vorobyov ah, M. Wadhwa e, Q. Wang ae,X.L.Wangv,Z.M.Wangv, M. Weber r, S. Wynhoff ak,L.Xiaaf,Z.Z.Xuv, J. Yamamoto c,B.Z.Yangv,C.G.Yangg,H.J.Yangc,M.Yangg,S.C.Yehas, An. Zalite ah, Yu. Zalite ah, Z.P. Zhang v,J.Zhaov,G.Y.Zhug,R.Y.Zhuaf, H.L. Zhuang g,A.Zichichih,r,s,B.Zimmermannaw, M. Zöller a a III. Physikalisches Institut, RWTH, D-52056 Aachen, Germany 1 b National Institute for High Energy Physics, NIKHEF, and University of Amsterdam, NL-1009 DB Amsterdam, The Netherlands c University of Michigan, Ann Arbor, MI 48109, USA d Laboratoire d’Annecy-le-Vieux de Physique des Particules, LAPP, IN2P3-CNRS, BP 110, F-74941 Annecy-le-Vieux cedex, France L3 Collaboration / Physics Letters B 619 (2005) 71–81 73 e Institute of Physics, University of Basel, CH-4056 Basel, Switzerland f Louisiana State University, Baton Rouge, LA 70803, USA g Institute of High Energy Physics, IHEP, 100039 Beijing, China 6 h University of Bologna and INFN, Sezione di Bologna, I-40126 Bologna, Italy i Tata Institute of Fundamental Research, Mumbai (Bombay) 400 005, India j Northeastern University, Boston, MA 02115, USA k Institute of Atomic Physics and University of Bucharest, R-76900 Bucharest, Romania l Central Research Institute for Physics of the Hungarian Academy of Sciences, H-1525 Budapest 114, Hungary 2 m Massachusetts Institute of Technology, Cambridge, MA 02139, USA n Panjab University, Chandigarh 160 014, India o KLTE-ATOMKI, H-4010 Debrecen, Hungary 3 p Department of Experimental Physics, University College Dublin, Belfield, Dublin 4, Ireland q INFN, Sezione di Firenze and University of Florence, I-50125 Florence, Italy r European Laboratory for Particle Physics, CERN, CH-1211 Geneva 23, Switzerland s World Laboratory, FBLJA Project, CH-1211 Geneva 23, Switzerland t University of Geneva, CH-1211 Geneva 4, Switzerland u University of Hamburg, D-22761 Hamburg, Germany v Chinese University of Science and Technology, USTC, Hefei, Anhui 230 029, China 6 w University of Lausanne, CH-1015 Lausanne, Switzerland x Institut de Physique Nucléaire de Lyon, IN2P3-CNRS, Université Claude Bernard, F-69622 Villeurbanne, France y Centro de Investigaciones Energéticas, Medioambientales y Tecnológicas, CIEMAT, E-28040 Madrid, Spain 4 z Florida Institute of Technology, Melbourne, FL 32901, USA aa INFN, Sezione di Milano, I-20133 Milan, Italy ab Institute of Theoretical and Experimental Physics, ITEP, Moscow, Russia ac INFN, Sezione di Napoli and University of Naples, I-80125 Naples, Italy ad Department of Physics, University of Cyprus, Nicosia, Cyprus ae Radboud University and NIKHEF, NL-6525 ED Nijmegen, The Netherlands af California Institute of Technology, Pasadena, CA 91125, USA ag INFN, Sezione di Perugia and Università Degli Studi di Perugia, I-06100 Perugia, Italy ah Nuclear Physics Institute, St. Petersburg, Russia ai Carnegie Mellon University, Pittsburgh, PA 15213, USA aj INFN, Sezione di Napoli and University of Potenza, I-85100 Potenza, Italy ak Princeton University, Princeton, NJ 08544, USA al University of Californa, Riverside, CA 92521, USA am INFN, Sezione di Roma and University of Rome, “La Sapienza”, I-00185 Rome, Italy an University and INFN, Salerno, I-84100 Salerno, Italy ao University of California, San Diego, CA 92093, USA ap Bulgarian Academy of Sciences, Central Lab. of Mechatronics and Instrumentation, BU-1113 Sofia, Bulgaria aq The Center for High Energy Physics, Kyungpook National University, 702-701 Taegu, Republic of Korea ar National Central University, Chung-Li, Taiwan, Republic of China as Department of Physics, National Tsing Hua University, Taiwan, Republic of China at Purdue University, West Lafayette, IN 47907, USA au Paul Scherrer Institut, PSI, CH-5232 Villigen, Switzerland av DESY, D-15738 Zeuthen, Germany aw Eidgenössische Technische Hochschule, ETH Zürich, CH-8093 Zürich, Switzerland Received 1 March 2005; received in revised form 13 May 2005; accepted 16 May 2005 Available online 8 June 2005 Editor: L. Rolandi Abstract The cross section for open-beauty production in photon–photon collisions is measured using the whole high-energy and − high-luminosity data sample collected by the L3 detector at LEP. This corresponds to 627 pb 1 of integrated luminosity for 74 L3 Collaboration / Physics Letters B 619 (2005) 71–81 electron–positron centre-of-mass energies from 189 to 209 GeV. Events containing b quarks are identified through their semi- + − + − leptonic decay into electrons or muons. The e e → e e bbX¯ cross section is measured within our fiducial volume and then extrapolated to the full phase space. These results are found to be in significant excess with respect to Monte Carlo predictions and next-to-leading order QCD calculations. 2005 Published by Elsevier B.V. 1. Introduction The production of b quarks through hard processes constitutes a unique environment for the study of per- turbative QCD, as the mass of the b quark, mb, largely exceeds the typical non-perturbative scale of hadronic interactions. High-energy hadron colliders are copi- ous sources of b quarks and therefore extensive ex- perimental studies and QCD calculations have been performed. Much debate has taken place on the appar- ent disagreement between the measured cross section for b-quark production in pp¯ collisions at the Teva- tron [1] and the next-to-leading order (NLO) QCD cal- culations [2]. The first measurements of open beauty production in e±p collisions at HERA were found to be markedly higher than NLO QCD predictions [3]. Some more recent measurements were in better agree- ment [4], while others still showed an excess [5,6]. A comparison of these different measurements with NLO QCD predictions is shown in Ref. [6]. + − Fig. 1. Dominant diagrams contributing to open-beauty production Photon–photon collisions at e e colliders also in photon–photon collisions at LEP. give access to the hard production of√ b quarks. The LEP e+e− centre-of-mass energy, s, was around 200 GeV. In this environment b quarks are expected to be produced with comparable rates by the direct and the direct and the single-resolved process depend on single-resolved processes [7], illustrated in Fig. 1.The mb, while the latter also depends on the gluon density main contribution to the resolved-photon cross section in the photon. is the photon–gluon fusion process. The rates of both The first measurement of the cross section for the e+e− → e+e−bbX¯ process was published by the L3 −1 Collaboration√ using 410 pb of data collected at s = 189–202 GeV [8]. The results were found to 1 Supported by the German Bundesministerium für Bildung, be in excess of the QCD prediction by a factor of Wissenschaft, Forschung und Technologie. 2 Supported by the Hungarian OTKA fund under contract three. Since these first findings, compatible prelimi- No. T019181, F023259 and T037350. nary results were obtained by other LEP collabora- 3 Also supported by the Hungarian OTKA fund under contract tions [9]. In this Letter, we extend our measurement to No. T026178. the whole high-energy and high-luminosity data sam- 4 Supported also by the Comisión Interministerial de Ciencia y ple collected at LEP with√ the L3 detector [10], corre- Tecnología. sponding to 627 pb−1 at s = 189–209 GeV. 5 Also supported by CONICET and Universidad Nacional de La Plata, CC 67, 1900 La Plata, Argentina. Hadronic events from photon–photon interactions 6 Supported by the National Natural Science Foundation of are selected through their specific multiplicity and China. topology. The production of b quarks is then tagged by L3 Collaboration / Physics Letters B 619 (2005) 71–81 75 the detection of electrons7 or muons from their semi- lisions are selected by means of three criteria. First, at leptonic decays. The cross section of the e+e− → least five charged tracks are required, thus suppress- e+e−bbX¯ process is measured in a phase space which ing background from the e+e− → e+e−τ +τ − and + − + − reflects the energy thresholds used in the analyses and e e → τ τ processes. Second, the visible energy√ the fiducial volume for lepton identification: the lep- of the event, Evis, is required to satisfy Evis < s/3, ton momentum must exceed 2 GeV and the angle, in order to reject events from the e+e− → qq¯ annihi- θ, between the leptons and the beam line must sat- lation process and further eliminate events from the isfy |cos θ| < 0.725 for electrons and |cos θ| < 0.8for e+e− → τ +τ − process. Finally, possible instrumental muons, respectively. For the first time the experimen- background and uncertainties in the trigger procedure tal results are compared to Monte Carlo predictions in are reduced by requiring the event visible mass, Wvis, this fiducial volume. An extrapolation factor is then to satisfy Wvis > 3GeV.Wvis is calculated from the applied to compare the measured cross section with four momenta of reconstructed tracks and of isolated the QCD predictions in the full phase space. calorimetric clusters. In this calculation, the pion mass is associated to the tracks while the clusters are treated as massless. Clusters in the low-angle luminosity mon- 2. Monte Carlo simulations itor are included in this calculation. In addition to these cuts, the analysis is restricted The PYTHIA [11] Monte Carlo generator is used to events with small photon virtuality by removing√ to model hadron production in photon–photon colli- events with clusters with energy greater than 0.2 s in the low-angle calorimeter, covering a polar angle sions. Final states without b quarks are generated with ◦ ◦ massless matrix elements [12] while massive matrix from 1.4 to 3.7 . This criteria corresponds to retain- 2 2 elements are used for b-quark production. Resolved ing quasi-real photons with Q 0.015 GeV . processes are described by means of the SaS1d parton About two million photon–photon events are se- density function [13]. The photon–photon luminosity lected by these cuts, with a background contamination function is implemented in the equivalent photon ap- of 0.1%. Events are further analysed if they have an proximation [14] with a cutoff for the virtuality of the identified electron or muon. 2 2 Electrons are identified as clusters in the elec- interacting photons Q Muon candidates are selected from tracks in the muon spectrometer in the range |cos θ| < 0.8. A min- imal muon momentum of 2 GeV is required to en- sure the muons reach the spectrometer after having crossed the calorimeters. The background from anni- hilation processes is suppressed√ by requiring the muon momentum to be less than 0.1 s. Background from cosmic muons is rejected by requiring the muons to be associated with a signal in the scintillator time-of- flight system in time with the beam crossing. After these cuts, 166 events√ with muon candidates are selected in data with s = 202–209 GeV.√ In- cluding the 269 events previously selected at s = 189–202 GeV [8], a total of 435 events with muons are retained. The estimated background from the e+e− → qq,¯ e+e− → τ +τ − and e+e− → e+e−τ +τ − is 5.7% and the signal efficiency is 2.2%. Fig. 2 presents the Wvis spectra of the selected events for the electron and muon samples. 4. Results The cross section for the e+e− → e+e−bbX¯ proc- ess is determined from the distribution of the trans- verse momentum of the lepton with respect to the nearest jet, Pt. As a consequence of the large value of mb, the distribution of this variable is enhanced for high values as compared to the background. The jets are reconstructed using the JADE algorithm [20] with ycut = 0.1. The identified lepton is not included in the jet. Fig. 3 presents the observed distributions of Pt for electrons and muons. The data distributions are fitted using the least- squares method to the sum of four contributions, whose shapes are fixed by Monte Carlo simulations. The first describes the background from annihilation processes and the e+e− → e+e−τ +τ − reaction. Its normalisation, Nbkg, is fixed to the Monte Carlo pre- dictions listed in Table 1. The three other contributions are those from b quarks, c quarks and lighter flavours. + − + − Their normalisations, Nbb¯ , Ncc¯ and Nuds, respectively, Fig. 2. Visible-mass spectra for the selected e e → e e hadrons are the free parameters of the fit. The results of the fits √events containing (a) an electron or (b) a muon candidate at are given in Table 1: a b-quark fraction of 46.2 ± 5.1% s = 189–209 GeV. The points are the data while the dotted line + − + − + − is observed for electrons and 41.2 ± 3.8% for muons, represents the background from the e e → qq,¯ e e → τ τ , + − → + − + − → + − + − where the uncertainties are statistical. The χ2 per de- e e W W and e e e e τ τ processes. The dashed lines are the sum of this background and the light-quark contri- gree of freedom of the fits is acceptable, with values bution, while the solid lines also include b-quark production. The of 13.7/6 for electrons and 6.4/6 for muons. A corre- normalisation follows from the fit discussed in the text. L3 Collaboration / Physics Letters B 619 (2005) 71–81 77 Table 1 Results of the fit to the distribution of the transverse momentum of the lepton with respect to the nearest jet. The fit parameters are constrained to be positive. The correlation between Nbb¯ and Ncc¯ is 75% Electrons Muons Nbkg 4.4 (fixed) 24.8 (fixed) ± ± Nbb¯ 94.3 18.3 172.0 31.0 Ncc¯ 105.4 ± 17.9 220.5 ± 35.4 +12.0 +52.3 Nuds 0.0−0.0 0.0−0.0 χ2/d.o.f. 13.7/66.4/6 lation coefficient of about 75% between Nbb¯ and Ncc¯ is observed. The results of the fits are also graphically shown in Fig. 3. Fig. 4 presents the distributions of the lepton momentum, transverse momentum and cosine of polar angle. The measured fractions of b quarks correspond to observed cross sections√ for the luminosity-averaged centre-of-mass energy s=198 GeV of + − → + − ¯ observed σ e e e e bbX electrons = 0.41 ± 0.08 ± 0.08 pb, + − → + − ¯ observed σ e e e e bbX muons = 0.56 ± 0.10 ± 0.10 pb. The first uncertainties are statistical and the second systematic, and arise from the sources discussed be- low. These cross sections correspond to the phase space of the selected leptons, without any extrapo- lation: lepton momenta above 2 GeV and polar an- gles in the ranges |cos θ| < 0.725 for electrons and |cos θ| < 0.8 for muons, respectively. 5. Systematic uncertainties Several potential sources of systematic uncertainty are considered and their impact on the observed cross section is detailed in Table 2. The largest sources of Fig. 3. Distributions of the transverse momentum of (a) the electron uncertainty arises from the event-selection procedure candidate and (b) the muon candidate with respect to the closest jet and the Monte Carlo modelling of the detector re- for the data and the results of the fit. The points are the data while sponse. Several components contribute to these uncer- + − the dotted line represents the background from the e e → qq,¯ tainties: the event-selection criteria, the lepton identi- + − → + − + − → + − + − → + − + − e e τ τ ,ee W W and e e e e τ τ fication and the detector response and resolution on processes. The dashed lines are the sum of this background and the light-quark contribution, while the solid lines also include b-quark the energy and angular variables which identify the production. The normalisation follows from the fit discussed in the fiducial volume. The effect of these systematic un- text. certainties is estimated by varying the corresponding 78 L3 Collaboration / Physics Letters B 619 (2005) 71–81 Fig. 4. Distribution of (a) the lepton momentum, (c) its transverse momentum and (e) the cosine of its polar angle for events containing electrons + − and (b), (d) and (f) for events containing muons. The points are the data while the dotted line represents the background from the e e → qq,¯ + − + − + − + − + − + − + − e e → τ τ ,e e → W W and e e → e e τ τ processes. The dashed lines are the sum of this background and the light-quark contribution, while the solid lines also include b-quark production. The normalisation follows from the fit discussed in the text. L3 Collaboration / Physics Letters B 619 (2005) 71–81 79 Table 2 CCFM [22] equation. The most important difference Systematic uncertainties on the observed values of the cross section as compared to NLO QCD calculations is the use of + − → + − ¯ of the process e e e e bbX for events tagged by electrons or an unintegrated gluon density function taking explic- muons. An additional uncertainty of 3% affects the extrapolation to the total cross section itly into account the transverse momentum distribution of initial state gluons in hard scattering processes. In Source of uncertainty Uncertainty on cross section (%) NLO QCD, all initial state partons have vanishing Electrons Muons transverse momentum. CASCADE was shown [23] Event selection 6.010.4 to give a consistent description of b-quark production Lepton identification 7.92.2 at the Tevatron, whereas H1 electro-production data Fiducial volume 12.310.0 Jet reconstruction 8.28.2 was found to be in excess by a factor of 2.6. Better Massive/massless charm 3.03.0 agreement was found with ZEUS electro-production Trigger efficiency 2.02.0 data. Monte Carlo statistics 1.61.4 The comparison of measurements and expectations Direct/resolved ratio 0.11.0 in the actual phase space of the selected leptons has Total 18.317.2 the advantage of providing an assessment of the agree- ment before any extrapolation is performed. Summing statistical and systematic uncertainties in quadrature, cuts and repeating the fits for the newly selected event one finds samples. The second most important source of sys- tematic uncertainty is the jet-reconstruction method. It is assessed by varying the value of y used in the + − → + − ¯ observed = ± cut σ e e e e bbX electrons 0.41 0.11 pb, reconstruction of the jets, and performing the fits for + − + − CASCADE the different Pt distributions which are obtained after → ¯ = ± σ e e e e bbX electrons 0.11 0.02 pb, the corresponding variation of the jet direction. This variation also addresses uncertainties in the hadroni- + − → + − ¯ observed = ± σ e e e e bbX muons 0.56 0.14 pb, sation process by excluding or adding soft clusters to the jets. The impact of the modelling of c quarks in the + − → + − ¯ CASCADE = ± σ e e e e bbX muons 0.14 0.02 pb, event generation is estimated by repeating fits by using Monte Carlo events generated with massive matrix el- where the uncertainty on the CASCADE predictions ements. The trigger efficiency is determined from the corresponds to a variation of m in the range 4.75 ± ± b data themselves and found to be (95.6 2.0)%, this 0.25 GeV [24]. A disagreement of about three stan- uncertainty is also propagated to the final results. The dard deviations is observed for both flavours of the limited Monte Carlo statistics has a small impact on final-state leptons. This disagreement is mostly due to the total systematic uncertainty. In the fits, the signal the overall normalisation of the sample rather than to events are produced in two separate samples for the a difference in shape of the most relevant kinematic direct and resolved processes and then combined in a variables, as also shown in Figs. 3 and 4. 1 : 1 ratio [7]. Systematic uncertainties on this predic- The total cross section for open-beauty production tion are estimated by repeating the fits with ratios of in photon–photon collisions is determined by an ex- 1:2and2:1. trapolation of the observed cross section to the full phase space of the process and by correcting for the semi-leptonic branching ratio of b quarks. The ex- 6. Discussion and conclusions trapolation factors are determined with the PYTHIA Monte Carlo program, and similar results are obtained The b-production cross sections measured in the if the CASCADE Monte Carlo is used. Their differ- phase space of the selected leptons are compared ence, which amounts to 3%, is considered as an ad- with the predictions obtained with the CASCADE ditional systematic uncertainty. The experimental un- Monte Carlo program [21]. This generator employs certainties on the semi-leptonic branching ratio of b a backward-evolving parton cascade based on the quarks [25] is also propagated to the measurement. 80 L3 Collaboration / Physics Letters B 619 (2005) 71–81 The results for the electron and muon final states read + − → + − ¯ total σ e e e e bbX electrons = 12.6 ± 2.4 ± 2.3pb, + − → + − ¯ total = ± ± σ e e e e bbX muons 13.0 2.4 2.3pb, where the first uncertainty is statistical and the second systematic. These results are in perfect agreement with each other and their combination gives + − + − σ e e → e e bbX¯ total = 12.8 ± 1.7 ± 2.3pb, where, again, the first uncertainty is statistical and the second systematic. This result is in agreement with our previous measurement performed with just a subset of the data investigated here [8] and has an improved pre- cision. Fig. 5. The open-charm, upper, and open-beauty, lower, produc- As a cross check, the values of Ncc¯ found by the tion cross sections in photon–photon collisions measured with the fit are used to extract the total cross section for the L3 detector. Statistical and systematic uncertainties are added in production of open charm√ at the luminosity-averaged quadrature. The dashed lines correspond to the direct-process con- centre-of-mass energy s=198 GeV as tribution and the solid lines represent the NLO QCD prediction for the sum of the direct and single-resolved processes. The effects of a + − + − σ e e → e e ccX¯ different choice of the values of the quark masses, mc and mb,are electrons illustrated. = (10.4 ± 1.8) × 102 pb, + − → + − ¯ = ± × 2 and 5.0 GeV. The threshold for open-beauty produc- σ e e e e ccX muons (9.8 1.6) 10 pb, tion is set at 10.6 GeV. The theory prediction for the where uncertainties are statistical. These values agree resolved process is calculated with the GRV parton well, and their average density function [26]. The same results are obtained if + − + − the Drees–Grassie parton density function [27] is used. σ e e → e e ccX¯ = (10.0 ± 1.2) × 102 pb For completeness, Fig. 5 also compares the cross sec- agrees with the dedicated measurement of Ref. [8], tions for open-charm production measured in Refs. [8, + − + − 2 σ(√e e → e e ccX¯ ) = (10.2 ± 0.3) × 10 pb for 28] with√ the corresponding predictions. s=194 GeV, where the uncertainties are statis- For s=198 GeV and mb = 4.75 GeV, the cross tical only. section expected from NLO QCD is 4.1 ± 0.6pb, An additional cross check showed that values of the where the uncertainty is dominated by uncertainties open-beauty cross section determined with the fit pro- on the renormalisation scale and on mb. Our measure- cedure discussed above or with a counting method [8] ment is a factor of three, and three standard deviations, are compatible. In the latter case experimental crite- higher than expected. In this respect it is interesting ria were chosen to optimise the charm cross section to remark that the prediction of CASCADE, when ex- measurement yielding a result essentially uncorrelated trapolated to the full phase space, 3.5 pb, agrees with with the b-quark production rate. those from NLO QCD [24], and the excess of our data The total cross section for open-beauty produc- with respect to the expectations is consistent before tion is compared in Fig. 5 to NLO QCD calcula- and after the extrapolation from the fiducial volume to tions [7]. The dashed line corresponds to the direct the full phase-space. process while the solid line shows the prediction for In conclusion, all high-energy data collected by the sum of direct and resolved processes. The cross L3 at LEP is investigated and the e+e− → e+e−bbX¯ section depends on mb, which is varied between 4.5 cross sections are measured within the detector fidu- L3 Collaboration / Physics Letters B 619 (2005) 71–81 81 cial volume and found to be in excess with respect to [10] L3 Collaboration, B. Adeva, et al., Nucl. Instrum. Methods Monte Carlo predictions. The cross sections are ex- A 289 (1990) 35; trapolated to the full phase space and found to be in L3 Collaboration, O. Adriani, et al., Phys. Rep. 236 (1993) 31; M. Acciarri, et al., Nucl. Instrum. Methods A 351 (1994) 300; excess with respect to next-to-leading order QCD cal- M. Chemarin, et al., Nucl. Instrum. Methods A 349 (1994) 345; culations. This confirms our previous findings based G. Basti, et al., Nucl. Instrum. Methods A 374 (1996) 293; on a subset of the full data-sample. I.C. Brock, et al., Nucl. Instrum. Methods A 381 (1996) 236; A. Adam, et al., Nucl. Instrum. Methods A 383 (1996) 342. [11] PYTHIA version 5.722 is used; T. Sjöstrand, preprint CERN-TH/7112/93 (1993), revised in References August 1995; T. Sjöstrand, Comput. Phys. Commun. 82 (1994) 74. [1] CDF Collaboration, F. Abe, et al., Phys. Rev. Lett. 71 (1993) [12] M. Cacciari, et al., Nucl. Phys. B 466 (1996) 173. 500; [13] G.A. Schuler, T. Sjöstrand, Z. Phys. C 68 (1995) 607; CDF Collaboration, F. Abe, et al., Phys. Rev. D 50 (1994) G.A. Schuler, T. Sjöstrand, Phys. Lett. B 376 (1996) 193. 4252; [14] V.M. Budnev, et al., Phys. Rep. 15 (1975) 181. CDF Collaboration, F. Abe, et al., Phys. Rev. Lett. 75 (1995) [15] J.A.M. Vermaseren, Nucl. Phys. B 229 (1983) 347. 1451; [16] KORALZ version 4.02 is used; CDF Collaboration, D. Acosta, et al., Phys. Rev. D 65 (2002) S. Jadach, B.F.L. Ward, Z. W¸as, Comput. Phys. Commun. 79 052005; (1994) 503. CDF Collaboration, D. Acosta, et al., Phys. Rev. D 66 (2002) [17] KORALW version 1.33 is used; 032002; S. Jadach, et al., Comput. Phys. Commun. 94 (1996) 216; D0 Collaboration, S. Abachi, et al., Phys. Rev. Lett. 74 (1995) S. Jadach, et al., Phys. Lett. B 372 (1996) 289. 3548; [18] GEANT version 3.15 is used; D0 Collaboration, B. Abbott, et al., Phys. Lett. B 487 (2000) R. Brun, et al., preprint CERN DD/EE/84-1 (1984), revised 264; 1987. D0 Collaboration, B. Abbott, et al., Phys. Rev. Lett. 85 (2000) [19] H. Fesefeldt, RWTH Aachen Report PITHA 85/2 (1985). 5068. [20] JADE Collaboration, W. Bartel, et al., Z. Phys. C 33 (1986) 23; [2] P. Nason, S. Dawson, K.R. Ellis, Nucl. Phys. B 303 (1988) 607; JADE Collaboration, S. Bethke, et al., Phys. Lett. B 213 (1988) P. Nason, S. Dawson, K.R. Ellis, Nucl. Phys. B 327 (1989) 49; 235. W. Beenakker, et al., Nucl. Phys. B 351 (1991) 507; [21] CASCADE version 1.1 is used; M. Cacciari, et al., JHEP 0407 (2004) 033, and references H. Jung, Comput. Phys. Commun. 143 (2002) 100; therein; H. Jung, G.P. Salam, Eur. Phys. J. C 19 (2002) 351. M.L. Mangano, hep-ph/0411020, and references therein; [22] M. Ciafaloni, Nucl. Phys. B 296 (1988) 49; S. Frixione, in: Proceedings of DIS’2004, hep-ph/0408317. S. Catani, F. Fiorani, G. Marchesini, Nucl. Phys. B 336 (1990) [3] H1 Collaboration, C. Adloff, et al., Phys. Lett. B 467 (1999) 18. 156; [23] H. Jung, Phys. Rev. D 65 (2002) 034015. H1 Collaboration, C. Adloff, et al., Phys. Lett. B 518 (2001) [24] H. Jung, M. Hanson, private communication. 331, Erratum; [25] S. Eidelman, et al., Phys. Lett. B 592 (2004) 1. ZEUS Collaboration, J. Breitweg, et al., Eur. Phys. J. C 18 [26] M. Glück, E. Reya, A. Vogt, Phys. Rev. D 46 (1992) 1973. (2001) 625. [27] M. Drees, K. Grassie, Z. Phys. C 28 (1985) 451. [4] H1 Collaboration, A. Aktas, et al., hep-ex/0411046; [28] L3 Collaboration, M. Acciarri, et al., Phys. Lett. B 453 (1999) ZEUS Collaboration, S. Chekanov, et al., Phys. Rev. D 70 83; (2004) 012008. L3 Collaboration, M. Acciarri, et al., Phys. Lett. B 514 (2001) [5] ZEUS Collaboration, S. Chekanov, et al., Phys. Lett. B 599 19; (2004) 173. L3 Collaboration, M. Acciarri, et al., Phys. Lett. B 536 (2002) [6] H1 Collaboration, A. Aktas, et al., hep-ex/0502010. 217. [7] M. Drees, et al., Phys. Lett. B 306 (1993) 371. [8] L3 Collaboration, M. Acciarri, et al., Phys. Lett. B 503 (2001) 10. [9] DELPHI Collaboration, W. Da Silva, in: Proceedings of PHO- TON2003, Nucl. Phys. B (Proc. Suppl.) 126 (2004) 185; OPAL Collaboration, A. Csilling, in: Proceedings of PHO- TON2000, hep-ex/0010060. Physics Letters B 619 (2005) 82–87 www.elsevier.com/locate/physletb Important pickup coupling effect on 8He(p, p) elastic scattering F. Skaza a,N.Keeleya, V. Lapoux a, N. Alamanos a,F.Augera,D.Beaumelb, E. Becheva b, Y. Blumenfeld b, F. Delaunay b,A.Drouarta, A. Gillibert a, L. Giot c, K.W. Kemper d, R.S. Mackintosh e, L. Nalpas a,A.Pakouf, E.C. Pollacco a,R.Raabea,1, P. Roussel-Chomaz c, J.-A. Scarpaci b, J.-L. Sida a,2,S.Stepantsovg,R.Wolskig,h a CEA-Saclay, DSM/DAPNIA/SPhN, F-91191 Gif-sur-Yvette, France b Institut de Physique Nucléaire, IN2P3-CNRS, F-91406 Orsay, France c GANIL, Bld. Henri Becquerel, BP 5027, F-14021 Caen cedex, France d Department of Physics, Florida State University, Tallahassee, FL 32306-4350, USA e Department of Physics and Astronomy, The Open University, Milton Keynes, MK7 6AA, UK f Department of Physics, University of Ioannina, 45110 Ioannina, Greece g Flerov Laboratory of Nuclear Reactions, JINR, RU-141980 Dubna, Russia h The Henryk Niewodnicza´nski Institute of Nuclear Physics, PL-31342 Kraków, Poland Received 20 April 2005; received in revised form 11 May 2005; accepted 25 May 2005 Available online 2 June 2005 Editor: V. Metag Abstract The 8He(p, p) and (p, d) reactions were measured in inverse kinematics at 15.7 A MeV and analyzed within the coupled reaction channels framework, the (p, d) cross section being particularly large. We find that coupling to 8He(p, d) pickup has a profound effect on the 8He(p, p) elastic scattering, and that these strong coupling effects should be included in analyses of pro- ton elastic and inelastic scattering. Through its modification of the elastic scattering wave functions this coupling will strongly affect the extraction of spectroscopic information such as the relationship between neutron and proton nuclear deformations, with important consequences for our understanding of the structure of exotic nuclei. 2005 Elsevier B.V. All rights reserved. PACS: 25.60.Bx; 25.60.Je; 24.10.Eq Keywords: 8He(p, p); (p, d); Coupled reaction channels calculations; Dynamic polarization potential E-mail address: [email protected] (V. Lapoux). 1 Present address: IKS, University of Leuven, B-3001 Leuven, Belgium. 2 Present address: CEA DIF/DPTA/SPN, B.P. 12, F-91680 Bruyères-le-Châtel, France. 0370-2693/$ – see front matter 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.physletb.2005.05.061 F. Skaza et al. / Physics Letters B 619 (2005) 82–87 83 Strong coupling effects in low-energy nuclear re- detection system consisted of the MUST array [10] actions are well established for heavy-ion collisions, to detect the light charged particles, a plastic wall for and lead to important modifications of the effective the detection of the projectile-like fragment, and two nucleus–nucleus interaction. The 16O + 208Pb system beam tracking detectors (CATS) upstream of the tar- is a well documented example, with coupled reaction get. The position sensitive CATS detectors [11] were channels (CRC) calculations showing how inelastic used to improve the definition of the beam position scattering and transfer channels generate a dynamic and incident angle on target. They provided particle polarization potential (DPP) with a substantial real by particle position and time tracking of the beam. part [1,2], having important consequences for elastic The MUST array consists of eight three-stage tele- scattering and fusion. scopes, each 6 × 6cm2. The first stage is a 300 µm- Important effects on (p, p) elastic scattering due to thick double-sided Si-strip detector which provides coupling to (p, d) pickup have been demonstrated for horizontal and vertical position, time-of-flight (TOF) stable nuclei [3–5]. The effect is particularly large for with respect to the beam detectors, and energy loss of light nuclei [4], reducing with increasing target mass the recoil proton. The second stage is a 3 mm-thick and incident proton energy, although remaining sig- Si(Li) giving the energy for protons up to 25.4 MeV, nificant for 50 MeV protons incident on 64Zn. Pickup and the third stage a 1.5 cm-thick CsI allowing the de- coupling was also found to significantly affect inelas- tection of protons up to 75 MeV in energy. The array tic scattering, mainly through the modification of the was assembled in a wall configuration located 15 cm elastic scattering wave functions [3], leading to signifi- from the target. The wall was placed in two positions, cant changes in the extracted deformation parameters. covering the angular range between 30◦–90◦ (lab.). At However, the possibility of such strong coupling ef- this distance, the 1 mm wide strips result in an angular fects has come to be ignored in analyses of proton resolution of 0.4◦ (lab.) for the detection of the scat- elastic and inelastic scattering, although a recent study tered particle. of 6He(p, p) postulated the existence of a repulsive For the less energetic recoil particles stopped in the real DPP due to breakup that gave improved agree- first stage, e.g., protons with energies below 6 MeV, ment with the data [6], subsequently further investi- mass identification was obtained using the energy ver- gated through coupled discretized continuum channels sus TOF technique. Particles were identified in the (CDCC) calculations [7]. correlation plot of their energy loss, E, in the Si- We report here a measurement of 8He(p, p) scat- strip detector versus their TOF. The TOF was mea- tering at 15.7 A MeV incident energy. Data for sured between the Si-stage and the start signal given 8He(p, d) populating the 3/2− ground state resonance by the passage of the incident particle through the sec- of the unbound 7He measured in the same experi- ond CATS detector. Protons from 6 to 25 MeV were ment have been previously reported [8], and the cross unambiguously identified by the E–E method us- section is found to be very large. This should there- ing energy loss measurements in the Si strip and the fore be a case where (p, d) coupling will have an Si(Li) detectors. The energy resolution obtained var- important influence on the 8He(p, p) scattering. We ied between 600 keV and 1 MeV, depending on target present CRC calculations including 8He(p, d) pickup thickness and the reaction kinematics. Events with a p to the 3/2− ground state of 7He which demonstrate the or d in coincidence with the heavy ejectile, plus a par- profound influence of this coupling on the elastic scat- ticle detected in the two CATS detectors to provide tering and, hence, on the nucleon–nucleus interaction the incident beam trajectory, were retained to build in a way that falls outside the scope of local-density the kinematical spectra and subsequently extract the folding models. (p, p) and (p, d) angular distributions. The 8He beam was produced by the ISOL tech- The elastic data extend from 20◦–110◦, and the nique and accelerated to 15.7 A MeV by the CIME transfer data from 27◦–85◦, in their respective center cyclotron at the SPIRAL facility [9], with no conta- of mass (c.m.) systems. To measure angular distrib- minants. The maximum (average) intensity in the ex- utions from 40◦ down to 20◦ (c.m.) where the en- periment was 14 000 (5000) p/s. The proton target was ergy of the recoiling protons decreases to 1.5 MeV, 2 2 a8.25mg/cm thick polypropylene (CH2)n foil. The a1.48mg/cm polypropylene target was used. To ob- 84 F. Skaza et al. / Physics Letters B 619 (2005) 82–87 tain good statistics at large angles, from 40◦ to 110◦ (c.m.), a 8.25 mg/cm2 target was used. The over- all values for the statistical plus systematic errors in the angular distributions arise from the detection effi- ciency and reconstruction process, which gives ±5% uncertainty, including the effect of background sub- traction (±2%); the target thickness (±5%); and the efficiency in the detection of the incident particles (±2%). This results in a total uncertainty of ±7.5% in the normalization of the data for elastic scattering and transfer to the 7He ground state. In Fig. 1 the measured elastic scattering angular distribution is compared to optical model calculations performed within the framework of the microscopic nucleon–nucleus JLM potential [12], using a no-core shell model 8He density [13]. The JLM potential is complex and the data for well-bound nuclei were found to be well reproduced with slight variations of Fig. 1. Optical model calculations using the JLM potential compared to the 8He + p elastic scattering data. See text for details. the real and imaginary parts, V and W . The required normalization factors, λV and λW , respectively, are found to be close to unity. For well-bound light nuclei in determining which states should be included in the 8 (A 20), the only modification required is λW = 0.8 coupling scheme, or not. In the He(p, p ) experiment [14], adopted as a “standard” normalization. at 72 A MeV reported in Ref. [16], the first excited The standard JLM (dotted curve) does not repro- state of 8He was found to be a 2+ located at 3.6 MeV. ◦ ◦ duce the data. Best agreement was obtained with λV = The cross sections measured between 20 to 50 (c.m.) 1.11, λW = 1.06 (solid curve), but the data at an- were found to lie below 1 mb/sr; a weak excitation gles smaller than 40◦ (c.m.) are significantly underpre- of the 2+ 8He was found [17]. In our experiment, as dicted. It should be emphasized that simply modeling mentioned in Ref. [8], inelastic (p, p) to the 2+ ex- the DPP by a renormalization of the JLM potential is cited state was also selected. These cross sections at unable to reproduce the whole angular range of the 15.7 A MeV will be presented and analyzed in a forth- data. coming article. Compared to the angular distributions Clearly, we need to include explicitly in our calcu- of the (p, d) transfer reaction, they were found to be lations the effect of coupling to other reaction chan- twice up to 5 times lower in the angular range from nels. To investigate the effect of coupling to (p, d) 20◦ to 80◦ (c.m.). We also face the problem of the exit pickup on 8He(p, p) scattering a series of CRC calcu- channel of the (p, d) reaction. It is beyond the scope of lations was carried out using the code FRESCO [15]. present CDCC calculations to include within the cou- The JLM prescription was retained for the p + 8He pling scheme the continuum of the unbound 7He states optical potential. We should include in the coupling and calculate the transfer reaction. The best calcula- scheme, a priori, the following reactions: elastic, in- tion which can be performed, at the present stage, is to elastic scattering and transfer reactions to the ground consider the deuteron states within the continuum. In or excited states of the nuclei produced in the exit Ref. [18], Halderson showed that the recoil corrected channel, either in bound or resonant states. But this re- continuum shell model predictions support a low-lying quires the corresponding inputs, transition strengths to 1/2− excited state for 7He at 1 MeV,as found by Meis- the excited states and spectroscopic factors. To sim- ter et al. [19]. Our recent results [8] also indicated this plify, we limitate the coupling scheme to the main low-lying excited state of 7He; it is weakly excited, channels which may contribute significantly in terms and roughly the cross sections are 10 times lower than 7 of angular distributions in the domain treated in our the (p, d) Hegs ones. In Ref. [20],at50A MeV, a res- analysis. The experimental observations can help us onance at 2.9 MeV was observed in 7He, the cross F. Skaza et al. / Physics Letters B 619 (2005) 82–87 85 Fig. 2. Coupling scheme used in the CRC calculations. sections (from 5◦ to 15◦ (c.m.)) were found to be 5 7 times less than the (p, d) Hegs. Consequently, in our analysis, we did not explicitly include the coupling to the 7He excited states and we considered (p, p) and 7 (p, d) Hegs as the main coupled reactions. The CDCC formalism was employed in the exit channel, as described in Ref. [21]. The bare d + 7He potential was of Watanabe type [22],then and p plus 7He optical potentials being calculated using the global parametrization of Koning and Delaroche [23]. Couplings to deuteron breakup with the neutron and Fig. 3. 8He(p, p) (upper panel) and 8He(p, d) (lower panel) calcu- proton in relative S and D states were explicitly in- lations compared to the data. The solid curves denote the full CRC cluded using the CDCC formalism and the coupling calculation with λV = 1.05, λW = 0.2 and the dotted curve indi- scheme presented in Fig. 2. cates the no-coupling calculation with the same bare potential. The For the transfer step, the neutron–proton over- dashed curves denote the result of a CRC calculation omitting the non-orthogonality correction. lap was calculated using the Reid soft-core potential [24], including the D-state component of the deuteron ground state. The same interaction was used to calcu- In Fig. 3 we present the calculated angular distri- late the exit channel deuteron potentials. The n + 7He butions for 8He(p, p) and 8He(p, d) compared to the binding potential was a Woods–Saxon well with the data. The results shown are for the final calculation 1/3 “standard” geometry of R0 = 1.25 × A fm, a = with JLM normalization factors λV = 1.05, λW = 0.2. 0.65 fm, the well depth being adjusted to give the A 8He(0+)/7He(3/2−) spectroscopic factor of C2S = correct binding energy. The spin–orbit term was omit- 3.3 gave the best agreement with the data, slightly ted as it has no effect on the calculated cross section. smaller than the value (4.1 ± 1.3) obtained in the Transfers to unbound states of the “deuteron” were in- CCBA analysis of Ref. [8], but within the quoted un- cluded in addition to that to the deuteron ground state. certainty. The full complex remnant term and non-orthogonality Excellent agreement between the calculated and correction were also included. measured elastic scattering is obtained over the whole There were three adjustable parameters, the real angular range, which was not possible in the optical and imaginary normalizations of the JLM entrance model calculations shown in Fig. 1.Theverylarge channel potential and the spectroscopic factor for the effect of the (p, d) coupling on the elastic scatter- 8He(0+)/7He(3/2−) overlap. All three were adjusted ing is evident. Note that in the full CRC calculation to obtain the optimum simultaneous agreement with the pickup coupling generates a considerable fraction the elastic scattering and transfer data. The normaliza- of the total absorption; only a small component of tion of the (real) JLM spin–orbit potential was con- the JLM imaginary potential is retained (λW = 0.2), strained to be the same as that of the real central po- which may be mostly attributed to compound nucleus tential. effects. For comparison, the no-coupling calculation 86 F. Skaza et al. / Physics Letters B 619 (2005) 82–87 Table 1 Volume integrals per nucleon pair/(MeV fm3), and rms radii/fm of the bare potential (OM) and the potentials found by inversion for the full CRC calculation and for the CRC calculation in which the non-orthogonality term was omitted (NONO) 21/2 21/2 JR r R JI r I JSOR JSOI OM 704.14 3.092 55.37 3.336 26.60 0.005 CRC 653.94 2.938 307.47 4.138 40.27 1.25 NONO 571.28 2.840 252.62 4.360 33.15 6.55 sion procedure Slj → V(r). The inversion is carried out using the iterative-perturbative inversion method of Kukulin and Mackintosh [25] which can give very reliable potentials, including spin–orbit potentials for the spin-half case, for all relevant radii. The bare di- agonal proton potential (i.e., without coupling) of the CRC calculation is then subtracted from V(r) and the remainder is identified as the DPP. The result is shown in Fig. 4 for two cases, the solid line be- ing the DPP in the case of the full CRC calculation and the dashed line the DPP from the CRC calcu- lation with the non-orthogonality correction omitted. Previous calculations [3–5,26] omitted the latter, but the qualitative finding that pickup leads to substan- tial repulsion as well as absorption is confirmed. We Fig. 4. DPP generated by the 8He(p, d) coupling obtained as ex- find that the non-orthogonality correction changes the plained in the text. shape of the real DPP, in particular, so that for a 8He target it is largely in the nuclear center. For this rea- using the bare JLM potential with λV = 1.05, λW = son, the effect on the real central volume integral, as 0.2 is also shown in Fig. 3. presented in Table 1,isjust7%. The agreement between the calculated and mea- Other features of the DPP are a significant imag- sured (p, d) angular distributions is less good, the inary spin–orbit term and an emissive imaginary cen- calculations overpredicting the data for angles greater tral term at the nuclear center. Emissivity at the nuclear than 50◦ in the c.m. system. This is probably due to center often occurs in local representations of a fun- the use of global potentials as a basis for the exit chan- damentally non-local and, in principle, L-dependent nel bare potential and could be improved by tuning the potential [25]. This emissivity and the other character- potential parameters, although we have chosen not to istics of the radial form of the DPP (accounting for the do so to show the quality of agreement that may be better fit to elastic scattering than renormalized JLM obtained with such potentials. potentials), can be traced to the fact that the contribu- The large change in the elastic scattering induced tion of the pickup coupling to the effective nucleon– by the pickup coupling may be represented as a sub- nucleus potential lies outside the scope of what could stantial DPP. To obtain the local and L-independent be described within the framework of folding models representation of this DPP, we followed the procedure based on an underlying local-density approximation. which was used to obtain the DPP for the 6He + p We therefore conclude that the inclusion of pickup system in Ref. [7]. The elastic scattering S-matrix is coupling is essential for a complete understanding of generated by the full CRC calculations (including cou- proton scattering. pling processes), and the total local optical potential The modification of the elastic scattering wave is obtained by subjecting this S-matrix to an inver- functions by the pickup coupling also has important F. Skaza et al. / Physics Letters B 619 (2005) 82–87 87 implications for proton inelastic scattering and the in- References formation that may be drawn therefrom. If one follows the usual conventions and renormalizes the transition potentials by the same factors as the entrance channel [1] I.J. Thompson, et al., Nucl. Phys. A 505 (1989) 84. optical potential, be they of phenomenological form [2] S.G. Cooper, R.S. Mackintosh, Nucl. Phys. A 513 (1990) 373. or calculated microscopically from theoretical tran- [3] R.S. Mackintosh, Nucl. Phys. A 209 (1973) 91. [4] R.S. Mackintosh, Nucl. Phys. A 230 (1974) 195. sition densities, the effect on the level of agreement [5] R.S. Mackintosh, A.A. Ioannides, I.J. Thompson, Phys. Lett. with data will be important. A full investigation of the B 178 (1986) 1. magnitude of this effect for the L = 2 8He(p, p ) tran- [6] V. Lapoux, et al., Phys. Lett. B 517 (2001) 18. sition to the 2+ first excited state is left for a later [7] R.S. Mackintosh, K. Rusek, Phys. Rev. C 67 (2003) 034607. comprehensive article, but test calculations using col- [8] F. Skaza et al., Phys. Rev. C, submitted for publication. [9] A.C. Villari, et al., Nucl. Phys. A 693 (2001) 465. lective model form-factors show a decrease of 14% in [10] Y. Blumenfeld, et al., Nucl. Instrum. Methods A 421 (1999) the nuclear deformation length extracted from a CRC 471. calculation compared to that obtained from a DWBA [11] S. Ottini, et al., Nucl. Instrum. Methods A 431 (1999) 476. calculation. [12] J.-P. Jeukenne, A. Lejeune, C. Mahaux, Phys. Rev. C 16 (1977) To summarize, we have shown that for a particu- 80. 8 (p, d) [13] P. Navrátil, B.R. Barrett, Phys. Rev. C 57 (1998) 3119. lar case, He, the explicit inclusion of coupling [14] J.S. Petler, et al., Phys. Rev. C 32 (1985) 673. has a profound influence on the calculated elastic scat- [15] I.J. Thompson, Comput. Phys. Rep. 7 (1988) 167. tering. Combined with the results of previous studies [16] A.A. Korsheninnikov, et al., Phys. Lett. B 316 (1993) 38. for stable nuclei [3–5,26], we may infer that this ef- [17] L.V. Chulkov, C.A. Bertulani, A.A. Korsheninnikov, Nucl. fect is probably general throughout the chart of the Phys. A 587 (1995) 291. [18] D. Halderson, Phys. Rev. C 70 (2004) 041603. nuclides. Evidently, an investigation of the systematics [19] M. Meister, et al., Phys. Rev. Lett. 88 (2002) 102501. of the effect to determine under what circumstances it [20] A.A. Korsheninnikov, et al., Phys. Rev. Lett. 82 (1999) 3581. is most pronounced would be of great interest. How- [21] N. Keeley, N. Alamanos, V. Lapoux, Phys. Rev. C 69 (2004) ever, it appears necessary to measure the (p, d) re- 064604. action and include this effect in analyses of proton [22] S. Watanabe, Nucl. Phys. 8 (1958) 484. [23] A.J. Koning, J.P. Delaroche, Nucl. Phys. A 713 (2003) 231. scattering for radioactive beams if correct inferences [24] R.V. Reid Jr., Ann. Phys. (N.Y.) 50 (1968) 441. are to be drawn. Through its modification of the elas- [25] V.I. Kukulin, R.S. Mackintosh, J. Phys. G: Nucl. Part. Phys. 30 tic scattering wave function the pickup coupling will (2004) R1. also have an important influence on the calculated in- [26] S.G. Cooper, R.S. Mackintosh, A.A. Ioannides, Nucl. Phys. elastic proton scattering, with all that this implies for A 472 (1987) 101. the extraction of information such as Mn/Mp ratios from data for this process. Physics Letters B 619 (2005) 88–94 www.elsevier.com/locate/physletb Hindered E4 decay of the 12+ yrast trap in 52Fe A. Gadea a, S.M. Lenzi b,D.R.Napolia, M. Axiotis a,C.A.Urb,c, G. Martínez-Pinedo d, M. Górska e, E. Roeckl e,E.Caurierf,F.Nowackif, G. de Angelis a,L.Batistg, R. Borcea e,F.Brandolinib,D.Cano-Otth, J. Döring e, C. Fahlander i,E.Farneab, H. Grawe e, M. Hellström i,Z.Janase,j, R. Kirchner e, M. La Commara e, C. Mazzocchi e,k, E. Nácher h, C. Plettner l, A. Płochocki j, B. Rubio h, K. Schmidt e, R. Schwengner l,J.L.Tainh,J.Zylicz˙ j a Laboratori Nazionali di Legnaro, I-35020 Legnaro, Italy b Dipartimento di Fisica and INFN, I-35100 Padova, Italy c H. Hulubei NIPNE, Bucharest, PO Box MG-6, Romania d University of Aarhus, DK-8000 Aarhus, Denmark e Gesellschaft für Schwerionenforschung, D-64229 Darmstadt, Germany f Institut Recherches Subatomiques, F-67037 Strasbourg cedex 2, France g PNPI, 188-350 Gatchina, Russia h Instituto de Física Corpuscular, E-46071 Valencia, Spain i Lund University, S-22100 Lund, Sweden j University of Warsaw, PL-00681 Warsaw, Poland k Università degli Studi di Milano, I-20133 Milano, Italy l Institut für Kern- und Hadronenphysik, FZ Rossendorf, 01314 Dresden, Germany Received 31 March 2005; received in revised form 21 May 2005; accepted 26 May 2005 Available online 9 June 2005 Editor: V. Metag Abstract + + The γ decay of the 12 yrast trap in 52Fe has been measured for the first time. The two E4 γ -branches to the 8 states are hindered with respect to other B(E4) reduced transition probabilities measured in the f7/2 shell. The interpretation of the data is given in the full pf shell model framework, comparing the results obtained with different residual interactions. It is shown that measurements of hexadecapole transition probabilities constitute a powerful tool in discriminating the correct configuration of the involved wavefunctions. 2005 Elsevier B.V. All rights reserved. E-mail address: [email protected] (A. Gadea). 0370-2693/$ – see front matter 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.physletb.2005.05.073 A. Gadea et al. / Physics Letters B 619 (2005) 88–94 89 PACS: 21.10.-k; 21.10.Re; 21.60.Cs; 23.20.Lv Keywords: Nuclear structure; Yrast trap; Shell model calculations High multipole moments in nuclei are considered one often encounters isomeric states and even inver- to be a vital source of information in nuclear struc- sion of states in the yrast line creating spin traps that ture studies connected with shape phenomena [1].In decay by high multipolarity transitions. particular electric hexadecapole moments and transi- Recently, Ur et al. [9] have studied the high tion strengths are experimentally accessible and pro- spin structure of 52Fe with the γ -ray detector array vide information on features that are independent of GASP [10]. The level scheme of 52Fe has been ex- the quadrupole structure of the nucleus and therefore tended up to the 10+ state at 7.4 MeV excitation contribute to test the theoretical models and, in partic- energy, lying above the yrast 12+ isomer, thereby con- ular, to reduce the degrees of freedom in the effective firming the predicted inversion [11,12] of the yrast nuclear interaction. 10+ and 12+ states. From a β+-decay end-point As pointed out in Refs. [1,2], the calculation of measurement the excitation of the 12+ state was de- E4 transition strength seems to be more sensitive to termined with an accuracy of the order of hundred model details than the E2 transitions and therefore it keV, and the half-life of the isomer was measured has a higher discrimination power when identifying to be 45.9(6) s [12]. The 12+ isomer mainly decays individual components in the nuclear wave function. (99.98%), by Gamow–Teller transitions, into excited In particular, in the sd shell, the B(E2) values for the states of the daughter nucleus 52Mn. transition to the ground state in even–even nuclei do In this Letter we report on the measurement of the not change very much from nucleus to nucleus while E4 γ -decay of the 12+ yrast trap in 52Fe to the two the B(E4) values show drastic changes. Interestingly, known 8+ states. The experiment was performed at the investigation of E4 transitions has allowed to de- the GSI on-line mass separator, where a 2.5 mg/cm2 termine a significant hexadecapole collectivity of 4+ thick natSi target was bombarded by a 170 MeV 36Ar states in closed shell nuclei (A ∼ 132,A∼ 208) (see beam delivered by the UNILAC accelerator. The esti- Ref. [3] for a systematic survey). mate of the cross section, performed with HIVAP [13], Electric hexadecapole moments in stable N = Z for this reaction, gives a population of ≈ 13 mb for nuclei were studied in the past in sd-shell nuclei by 52Fe, above the 12+ isomer. inelastic scattering with several probes (electrons, pro- The recoiling reaction products were stopped in the tons, α-particles) [2,4]. Such experiments cannot be graphite catcher of a FEBIAD-E type ion source [14]. done for heavier N = Z nuclei without resorting to After ionisation and extraction from the ion source, the radioactive beam facilities. With the recent develop- mass separated A = 52 beam was implanted in a tape ments in detection techniques, information of tran- which moved every 80 s, taking away the undesirable sition strengths can be directly obtained by γ -ray long-lived activity. spectroscopy. Whenever γ -transition probabilities are The implantation position was surrounded by a compatible with the detection sensitivity, these mea- plastic scintillator, with a β-detection efficiency of surements give more complete information than that ∼ 85% (measured with a 24Na source), two composite deduced by using scattering techniques. Moreover, germanium (Ge) detectors of the Cluster [15] and large γ -ray spectroscopy allows to measure hexadecapole Clover [16] type, and a 60% single Ge crystal. The transitions from high spin states. setup included a second single crystal low-energy Ge In the past few years, considerable effort has been detector, as shown in Fig. 1, but was not relevant for put into the study of high spin states in f7/2-shell nu- the present analysis. The photopeak efficiency of the clei. It has been shown that near the middle of the shell Ge setup was 3.9% for a γ -ray energy of 1.33 MeV, (48Cr) nuclei present strong deformation [5–8]. Heav- which improved the detection sensitivity limit by a ier nuclei, like 52Fe, are less deformed due to their factor of 100 compared to that achieved in the previ- proximity to the N,Z = 28 shell closure. In such cases ous study [12]. The large segmentation of the detection 90 A. Gadea et al. / Physics Letters B 619 (2005) 88–94 system (12 independent large volume Ge crystals) was of the setup performed with the GEANT3 library [17], essential to keep the summing losses within a reason- the summing perturbation to all measured quantities able limit, i.e., below 10% for large multiplicity cas- was estimated to be far below the respective experi- cades. By using a complete Monte Carlo simulation mental uncertainties. The total measurement time amounted to 32 hours with a production rate of ≈ 4.5×104 atoms/s. β–γ –γ and γ –γ coincidence events were recorded and af- terwards sorted into 3D-cubes and 2D-matrices. The analysis of the γ –γ coincidences, including the “add- back” of the composite detectors and a veto condition derived from the β counter allowed us for the first time to observe the γ de-excitation of the 12+ isomer + + to the 81 and 82 at 6360 and 6493 keV states via E4 transitions of 597 and 465 keV, respectively. The anti- coincidence with the β counter served to reduce the background contribution from β-delayed γ -rays. The resulting spectrum is shown in Fig. 2. Fig. 1. Sketch of the β–γ detection setup at the GSI on-line mass The new transitions fix the excitation energy of the separator. The A = 52 beam is implanted into a tape which is not 12+ isomer at 6957.5(4) keV (see Fig. 3). This value shown. The implantation position is in the center of the β-detector. is significantly more accurate than the previous result Collimation system, tape and β-detector are mounted in a vacuum chamber, while the Ge detectors are positioned around the chamber. deduced from β-decay measurements [12]. Fig. 2. Spectrum obtained in coincidence with strong transitions in 52Fe and in anti-coincidence with the β-counter. In the spectrum there + is a small “leak” of the large β annihilation peak, due to the background subtraction procedure, and a peak coming from the Compton back-scattering of the 1461 keV background transition (1461–850 keV). The gamma transitions belonging to 52Fe are marked by their energies in keV. A. Gadea et al. / Physics Letters B 619 (2005) 88–94 91 detector. The intensity of the β-decay branch has been obtained from the total spectrum without any condi- tion. All the intensity populating the 2+ in 52Fe is expected to go through the two E4 transitions, and since the relative intensities of these two transitions are easily obtained from the γ –γ coincidence matrix, it is possible to evaluate the intensity of each tran- sition compared to the total isomer decay rate. The second method to determine the intensities is based exclusively on the γ –γ coincidences. Considering a 100% intensity for the E2 850 keV transition to the ground state and the measured absolute efficiencies of the setup for this and the observed E4 transition, the determination of the intensity of the latter is straight- forward. Also in this case the β-decay branch intensity is determined from the total spectrum. Both methods gave the same values. The evaluated intensities reflect very low E4 transi- tion probabilities: 1.1(4) e2 fm8 (4.6(17) × 10−4 W.u.) and 8(3) e2 fm8 (3.5(13)×10−3 W.u.) for the 597 keV and 465 keV transitions, respectively. If one com- pares the 52Fe data with the B(E4) observed in other f7/2-shell nuclei (see Table 1), to obtain the lowest value, corresponding to 52Mn (0.138 W.u.), partial de- excitation branches that are ∼ 300 and ∼ 40 times higher than those observed for the 597.1 keV and 465.0 keV transitions, respectively, would be needed. This explains why these transitions where not ob- + served in previous studies [12]. Fig. 3. Level scheme of the 52Fe 12 isomer decay. Transitions + To interpret these results we have performed cal- from the higher lying 10 state at 7381 keV observed in a in-beam study [9] are shown. culations in the shell model framework with the code ANTOINE [18] in the full pf model space. Three dif- ferent residual interactions have been used, namely the From the spectrum shown in Fig. 2 it is evident that FPD6 [19], the KB3G [20], and the recently intro- the 465 and 597 keV transitions have similar inten- duced GXPF1 [21] interactions. The effective charges sities. Their E4 transition probabilities, however, are used to calculate the B(E4) reduced transition proba- strikingly different due to the strong dependence on bilities are the same as those used to obtain the B(E2) the latter quantities upon the transition energy. The values, i.e., ep = 1.5 and en = 0.5 [9]. A recent mea- + transition intensities per isomer decay have been esti- surement of the 2 → 0+ B(E2) value in 52Fe, using − + 1 mated to be 1.2(4) × 10 4 for the 597 keV (12 → Coulomb excitation techniques [22], is in excellent + × −4 81 ) transition and 0.9(3) 10 for the 465 keV agreement with the calculation performed in Ref. [9]. + → + (12 82 ) transition. These results are based on the The calculated energies and reduced transition combined information of γ –γ coincidence matrices probabilities of the two E4 transitions in 52Fe are with and without β-detector veto. Two methods have confronted with the experimental data in Table 2.All been used to determine the intensities. The first one calculations overestimate the experimental values. The consisted on determining the absolute intensity of the best description is achieved by the FPD6 interaction 850 keV 2+ → 0+ transition in 52Fe starting from while both the KB3G and GXPF1 calculations fail in the γ -ray spectrum in anti-coincidence with the β- reproducing even the order of magnitude of the B(E4) 92 A. Gadea et al. / Physics Letters B 619 (2005) 88–94 Table 1 Experimental E4 systematics for f7/2-shell nuclei Eγ (keV) Ji → Jf T1/2 γ branch B(E4) (W.u.) + + 44Sc 271 6 → 2 58.61 h 0.988 1.42 + + 46Ti 2010 4 → 0 1.62 ps 1.6a + + 52Mn 378 2 → 6 21.1 min 0.0175 0.138 52 + → + × −4 × −4 Fe 597 12 81 45.9 s 1.2(4) 10 4.6(17) 10 + + − − 52Fe 465 12 → 8 45.9 s 9(3) × 10 5 3.5(13) × 10 3 − 2 − 53Fe 701 19/2 → 11/2 2.52 min 0.9866 0.256 + + 54Fe 3578 10 → 6 364 ns 0.019 0.79 + + a The experimental B(E4) value for the 4 → 0 in 46Ti obtained from the γ -intensity measurements reported in Ref. [23] (B(E4) = 400(300) W.u.) is inconsistent with the values expected in the region. This discrepancy is not understood. However, agreement with the expectations is obtained by using the B(E4) value extracted from the hexadecapole deformation measured in Ref. [24] (B(E4) ≈ 1.6W.u.). Table 2 Experimental and calculated energies and reduced transition probabilities of the two E4 transitions in 52Fe and previously known E4 transitions in f7/2-shell nuclei Ji → Jf Eγ (keV) B(E4) (W.u.) Exp FPD6 KB3G GXPF1 Exp FPD6 KB3G GXPF1 52 + → + × −4 × −3 × −1 × −2 Fe 12 81 597 1227 907 888 4.6(17) 10 2.4 10 3.3 10 6.5 10 + + − − − − 52Fe 12 → 8 465 519 700 756 3.5(13) × 10 3 4.7 × 10 3 2.6 × 10 2 2.3 × 10 2 + +2 44Sc 6 → 2 271 674 373 281 1.42 1.96 1.79 1.65 + + 46Ti 4 → 0 2010 1966 1819 2000 1.610.77.97.39 + + 52Mn 2 → 6 378 205 91 213 0.138 0.272 0.422 0.728 − − 53Fe 19 → 11 701 990 883 776 0.256 0.151 1.23 0.84 2 + +2 54Fe 10 → 6 3578 3660 3838 3306 0.79 1.80 0.98 1.25 values. Both interactions yield a higher value for the + → + B(E4) 12 81 transition, in contrast with the ex- perimental findings. It is interesting to see how these interactions re- produce the other B(E4) values known in the f7/2 shell. The results obtained with the three interactions are listed in Table 2 together with the corresponding experimental data. A full pf calculation has been per- formed for 44Sc, 46Ti and 52Mn, whereas for 53Fe nine of the thirteen valence particles have been allowed to 54 be excited to orbitals above the f7/2 one, and for Fe a truncation to eight of the fourteen valence particles has been made. As shown in Table 2, all interactions repro- duce with the same good accuracy the experimental B(E4) data of these nuclei. In Fig. 4, the ratios between the experimental and theoretical reduced transition probabilities are shown Fig. 4. Ratio between experimental and theoretical B(E4) values for nuclei in the f7/2 shell. Results obtained by using the FPD6, GXPF1 for all the measured E4 transitions in the f7/2-shell and KB3G interactions are shown by squares (full line), full circles nuclei. It is evident that both the KB3G and GXPF1 (dashed-line) and triangles (dotted-line), respectively. See Table 2 interactions fail in the case of 52Fe. A possible ori- for details. A. Gadea et al. / Physics Letters B 619 (2005) 88–94 93 + Table 3 laps are obtained between the yrast 8 state and be- + 1 Proton (or neutron) occupation numbers for the states of interest in tween the yrare 8 state. On the contrary, an overlap of 52Fe 2 + ∼ 0.93 is obtained between the 8 FPD6 state and the + 1 + f7/2 p3/2 f5/2 p1/2 8 KB3G state, and vice versa between the 8 FPD6 2 + 2 FPD6 state and the 8 KB3G state. This inversion of the + + 1 81 4.64 0.50 0.71 0.15 + 8 states could explain the fact that the B(E4) values 82 5.27 0.35 0.32 0.06 + obtained with the KB3G interaction, when compared 121 5.43 0.26 0.26 0.05 with experiment, are inverted in strength. In the case + KB3G of the GXPF1 interaction, the two 8 states are simi- + 8 5.70 0.15 0.21 0.04 lar, which translates in similar overlaps (∼ 0.60–0.70) 1 + + + with the 8 and 8 states of FPD6 and KB3G wave 82 5.02 0.30 0.57 0.10 1 2 + functions. 121 5.63 0.14 0.19 0.03 As mentioned above, the B(E4) values have been GXPF1 + obtained with the effective charges used to reproduce 81 5.20 0.30 0.41 0.08 52 + the quadrupole transition probabilities in Fe and 82 5.37 0.30 0.26 0.06 + N = Z B( ) 12 5.67 0.19 0.12 0.02 neighboring nuclei. As for nuclei, the E4 1 2 transition probability is proportional to (ep + en) , it is the square of the sum of the effective charges gin of this failure could arise from the fact that both which enters as a multiplicative factor. In a very re- 8+ states in 52Fe are very close in excitation energy cent work [25], the polarization charges have been de- (see Table 2), and therefore these calculations could duced from B(E2) values measured for the mirror pair = = = mix the configurations of the two levels or invert their A 51, obtaining ep 1.15 and en 0.8. The use of order. A possible way to check the quality of the inter- these effective charges would not change the present 52 actions in describing the two 8+ states is to calculate results for the B(E4) values in Fe. The need of us- the quadrupole transition probabilities B(E2) for their ing very large or even negative polarization charges to + reproduce the systematics of the B(E4) values in this decay to the 61 state and compare the predictions with the measured values [9]. Unfortunately, all the reduced mass region has been discussed by Yokoyama [26]. transition probabilities are of the same order and the In the latter work, however, shell model calculations n experimental uncertainties do not allow a discrimina- were performed in strongly truncated spaces (f7/2 or n + n−1 1 tion. f7/2 f7/2 (p3/2,p1/2,f5/2) ). Even if it is out of To get a better understanding of the wave functions, the scope of the present study to fit the polarization we report in Table 3 the occupation numbers (protons charges, it is interesting to note that enlarging the and neutrons are equivalent in an N = Z nucleus) of model space has allowed us to reproduce on the same the different orbits for the states involved in the E4 footing all the hexadecapole transition probabilities + decay. While the 12 states have similar occupations known for f7/2-shell nuclei by using the same polar- with any of the interactions, the 8+ states are different. ization charges. + The 81 level obtained with FPD6 is the most collective Finally, the origin of the hindrance of the B(E4) + 52 one, followed by the 82 state obtained with KB3G. values in Fe, can be understood from the hexadeca- To elucidate these discrepancies, we have com- pole strength distribution. Using the different residual puted the wave function overlaps between the 8+ interactions, we have calculated the E4 strength from states and between the 12+ states calculated by us- the 12+ isomer to all the I π = 8+ states in the pf ing different interactions. As expected, the overlap be- shell model space. As expected, the results indicate tween the yrast 12+ wave functions obtained with any that most of the E4 strength is located at excitation en- of the interactions are 0.96, which implies that the ergies higher than the 12+ state. In fact, only up to few 12+ states described by all the Hamiltonians almost per cent (10% for KB3G, 2% for GXPF1 and 0.2% for coincide. When comparing the 8+ states calculated FPD6) of the E4 strength is predicted to feed the first with the FPD6 and KB3G interactions, vanishing over- and second experimentally observed 8+ states. 94 A. Gadea et al. / Physics Letters B 619 (2005) 88–94 In conclusion, the combination of in-beam and off- [7] E. Caurier, et al., Phys. Rev. Lett. 75 (1995) 2466. beam (ISOL) experiments has allowed us for the first [8] G. Martínez-Pinedo, et al., Phys. Rev. C 54 (1996) R2150. time to observe the γ -decay of the 52Fe 12+ yrast trap [9] C.A. Ur, et al., Phys. Rev. C 58 (1998) 3163. [10] D. Bazzacco, in: Proceedings of the International Conference and to establish its excitation energy. Two γ -rays of on Nuclear Structure at High Angular Momentum, Ottawa, 597 keV and 465 keV have been assigned to be the E4 AECL Report 10613, vol. II, 1992, p. 376. + + transitions feeding the 81 and 82 states, respectively. [11] D.F. Geesaman, et al., Phys. Rev. Lett. 34 (1975) 326. These two transitions are strongly hindered compared [12] D.F. Geesaman, et al., Phys. Rev. C 19 (1979) 1938. to any other E4 transition in the region. From large [13] W. Reisdorf, et al., Z. Phys. A 343 (1992) 47. [14] R. Kirchner, et al., Nucl. Instrum. Methods 186 (1981) 295. scale shell model calculations performed in the full pf [15] J. Eberth, Prog. Part. Nucl. Phys. 28 (1992) 495; space it has been shown that the hexadecapole transi- J. Eberth, Nucl. Instrum. Methods A 369 (1996) 135. tions can give vital information to distinguish the more [16] J. Gerl, et al., in: Proceedings of the Conference on Physics realistic wave function from those predicted by differ- from Large γ -ray Detector Arrays, Berkeley, LBL 35687, ent interactions, which might be indistinguishable on CONF 940888, UC 413, 1994, p. 159. [17] GEANT—Detector Description and Simulation Tool, CERN the basis of B(E2) measurements. Program Library Writeup W5013. [18] E. Caurier, Code ANTOINE, Strasbourg, 1989; E. Caurier, F. Nowacki, Acta Phys. Pol. 30 (1999) 705. References [19] W.A. Richter, et al., Nucl. Phys. A 523 (1991) 325. [20] A. Poves, E. Caurier, F. Nowacki, Nucl. Phys. A 694 (2001) [1] B.H. Wildenthal, B.A. Brown, I. Sick, Phys. Rev. C 32 (1985) 157. 2185. [21] M. Honma, T. Otsuka, B.A. Brown, T. Mizusaki, Phys. Rev. [2] B.A. Brown, R. Radhi, B.H. Wildenthal, Phys. Rep. 101 (1983) C 65 (2002) 061301(R). 313. [22] K.L. Yurkewicz, et al., Phys. Rev. C 70 (2004) 034301. [3] P.C. Sood, R.K. Sheline, B. Singh, Phys. Rev. C 51 (1995) [23] M. Fujishiro, Y. Satoh, K. Okamoto, T. Tsujimoto, Can. J. 2798. Phys. 58 (1980) 1712. [4] J. Fritze, et al., Phys. Rev. C 43 (1991) 2307. [24] M. Fujiwara, et al., Phys. Rev. C 35 (1987) 1257. [5] S.M. Lenzi, et al., Z. Phys. A 354 (1996) 117. [25] R. du Rietz, et al., Phys. Rev. Lett. 93 (2004) 222501. [6] S.M. Lenzi, et al., Phys. Rev. C 56 (1997) 1313. [26] A. Yokoyama, Phys. Rev. C 55 (1997) 1282. Physics Letters B 619 (2005) 95–104 www.elsevier.com/locate/physletb Charm production in antiproton–nucleus collisions at the J/ψ and the ψ thresholds L. Gerland a, L. Frankfurt b, M. Strikman c a SUBATECH, Laboratoire de Physique Subatomique et des Technologies Associées, University of Nantes, IN2P3/CNRS, Ecole des Mines de Nantes, 4 rue Alfred Kastler, F-44072 Nantes cedex 03, France b School of Physics and Astronomy, Raymond and Beverly Sackler Faculty of Exact Science, Tel Aviv University, Ramat Aviv 69978, Tel Aviv, Israel c Pennsylvania State University, University Park, PA 16802, USA Received 31 January 2005; received in revised form 3 May 2005; accepted 9 May 2005 Available online 16 May 2005 Editor: J.-P. Blaizot Abstract We discuss the production of charmonium states in antiproton–nucleus collisions at the ψ threshold. It is explained that measurements in pA¯ collisions will allow to get new information about the strengths of the inelastic J/ψN and ψ N interaction, ¯ on the production of Λc and D in charmonium–nucleon interactions and for the first time about the nondiagonal transitions ψ N → J/ψN. The inelastic J/ψ-nucleon cross section is extracted from the comparison of hadron–nucleus collisions with hadron–nucleon collisions. We extract the total J/ψ-nucleon cross section from photon–nucleon collisions by accounting for the color transparency phenomenon within the frame of the GVDM (generalized vector meson dominance model). We evaluate within the GVDM the inelastic ψ -nucleon cross section as well as the cross section for the nondiagonal transitions. Predictions for the ratio of J/ψ to ψ yields in antiproton–nucleus scatterings close to the threshold of ψ production for different nuclear targets are presented. 2005 Elsevier B.V. All rights reserved. 1. Introduction tion is played by the value of the total and the elastic cross sections for charmonium–nucleon interactions as well as the amplitude for the inelastic transition be- During the last two decades significant attention was given to the absorption of charmonium states pro- tween J/ψ and ψ states characterizing the role of duced in heavy ion collisions, see, e.g., Ref. [1] and color transparency phenomena. The aim of this Letter references therein. An important role in such evalua- is to extract these cross sections from photoproduction data following Ref. [2] and to make predictions for antiproton–nucleus collisions at the ψ threshold. We E-mail address: [email protected] (L. Gerland). demonstrate that in these collisions the cross section 0370-2693/$ – see front matter 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.physletb.2005.05.013 96 L. Gerland et al. / Physics Letters B 619 (2005) 95–104 for the nondiagonal transition ψ + N → J/ψ + N sible enhancement of the charmonium–nucleus cross can be measured. We account for the dependence section near threshold of charmonium–nucleus inter- of the cross sections on energy, and the dependence actions like that described in Ref. [8]. of the elastic cross section on the momentum trans- The amplitude of the J/ψ photoproduction close fer. to the threshold Eγ ∼ 9 GeV is dominated by the gen- The charmonium production at the ψ threshold eralized gluon density at large x1 − x2 ∼ 0.5. In such is well suited to measure the genuine charmonium– a kinematic, Fermi motion effects may lead to a sig- nucleon cross sections. At higher energies formation nificant enhancement. However, similar to the quark time effects makes the measurement of these cross sec- distribution functions one may expect a suppression tions more difficult [3]. These cross sections and the reflecting medium modifications of the nucleon struc- cross section for the nondiagonal transition ψ + N → ture functions (the analogue of the EMC effect). In a J/ψ + N are important for the analysis of charmo- genuine photoproduction experiment it would be very nium production data at SPS-energies [4,5]. At col- difficult to distinguish the EMC type effect from the lider energies, i.e., at RHIC and LHC, the forma- absorption due to the final state interaction. However, tion time effects will become dominant and charmo- the combination of a measurement at the GSI of the nium states will be produced only far outside of the p¯ + A → J/ψ + X reaction and of photoproduction nuclei [6]. However, measurements of the genuine at 12 GeV at the Jefferson-lab will make it possible to charmonium–nucleon cross sections as well as the measure the A dependence of the nuclear generalized cross section for the nondiagonal transition ψ + N → gluon distributions at large x. J/ψ + N are also important at collider energies for The ratio of the elastic to the total J/ψN cross sec- the evaluation of the interaction of charmonium states tion has been evaluated long ago in Ref. [10] in the with the produced secondary particles. vector meson dominance model (VDM), where the en- By using the appropriate incident energy in antipro- ergy dependence and the real part of the forward scat- ton–nucleus scattering, the case when the scattering tering amplitude were neglected and the t dependence occurs off the nucleons with small internal momenta of the elastic cross section was effectively adjusted to can be selected. Correspondingly, in this situation off the data on the soft QCD process of the ρ photopro- shell effects in the amplitude should be very small and, duction off the nucleon target. We will show in this hence, there will be no significant nuclear corrections Letter that all these effects and color transparency phe- due to possible modification of the nucleons in nuclei. nomenon should be taken into account. In addition, due to the large antiproton–nucleon cross In the beginning of this Letter we examine the en- section, those antiprotons which do not undergo ab- ergy dependence of the ratio of the elastic to the total sorption at the surface of the nuclear target will lose a J/ψN cross section as well as the influence of the real significant fraction of their energy. Therefore, at ener- part of the amplitude and treat the t dependence of gies close to the threshold the charmonium production the elastic cross section within a charmonium model. is almost impossible inside of the nuclear target. The final result is that the ratio of the elastic to the to- We discuss in this Letter the production of charm tal J/ψN cross section is still small (approximately at the pp¯ → J/ψ,ψ thresholds. To avoid difficulties 5–6.5%) but by a factor ≈ 2.5 larger than that given in with the specifics of low energy initial state interac- the previous evaluation. tion effects, which are actually included into the partial In Section 2 the amplitudes of the GVDM, the elas- width of the J/ψ → pp¯ decay, like those discussed tic form factor and the two-gluon-form factor are de- in Ref. [7], we will discuss ratios of cross sections, scribed. The amplitudes of the GVDM are described in in which the factor σ(pp¯ → J/ψ,ψ ) is canceled. more detail in Appendix A. In Section 3 the semiclas- We demonstrate that these ratios are well suited to sical Glauber model is described and the predictions measure the nondiagonal (ψ → ψ) cross section as for the future GSI experiment are shown. In Section 4 well as the inelastic J/ψ and ψ cross section. At the the results of this Letter are summarized. The phenom- same time the momentum of charmonium in the fi- ena considered in this Letter are complementary to the nal state is 5 GeV/c in the rest frame of the nuclear program of antiproton–nucleus scattering experiments target. Hence, one cannot probe in this reaction a pos- at the GSI range outlined in the recent review [11]. L. Gerland et al. / Physics Letters B 619 (2005) 95–104 97 Fig. 1. The four leading graphs that contribute to the elastic form factor of the J/ψ. 2. Model description and results the two-gluon-form factor [12], which is 1 In Ref. [10] the elastic and the total J/ψ-nucleon F 2 (t) = (6) 2g − t 4 (1 2 ) cross sections were evaluated within the vector me- m2g son dominance model (VDM). In this model, the J/ψ with m2 ≈ 1.1GeV2. photoproduction amplitude fγψ and the J/ψ-nucleon 2g And the two-gluon form factor of the J/ψ, calcu- elastic scattering amplitude fψψ are related as lated as the nonrelativistic limit of the diagrams shown e f = f . (1) in Fig. 1 is γψ f ψψ ψ − 2 dz Ψ(z,kt )∆(kt )Ψ (z, kt zqt )d kt z(1−z) Here, e is the charge of an electron and f is the Fψ (t) = . ψ Ψ(z,k )∆(k )Ψ (z, k )d2k dz J/ψ–γ coupling given by t t t t z(1−z) (7) e2 2 3 Γ(V → ee)¯ Here, Ψ is the wave function of the J/ψ, z is the = . (2) 4πfψ 4π mψ fraction of the longitudinal momentum of the char- monium state carried by the c-quark, while kt is the A similar relation like Eq. (1) can be written also for relative transverse momentum of the c-quark and the the ψ . From the optical theorem c¯-quark. ∆(kt ) is two-dimensional Laplace operator. Iψψ(t = 0) qt is the sum of the momenta of the two gluons. This σ (J/ψN) = , (3) tot 2p E form factor unambiguously follows from the analysis cm cm of Feynman diagrams for hard exclusive processes. By where Iψψ is the imaginary part of fψψ and the dif- definition it is equal to one at zero momentum trans- ferential elastic cross section fer Fψ (t = 0) = 1. To evaluate this form factor we use here the nonrelativistic wave functions of Ref. [13]. dσel 1 2 = |fψψ| (4) In the gluon exchange between the charmonium dt 64πp2 E2 cm cm and the target only one gluon polarization dominates. follows In QCD evolution only this contribution contains the 2 2 | | + 2 large logarithm ln(mc). Using the QCD Ward iden- σtot(J/ψN) teff (1 η ) σel = . (5) tity one can express the obtained formulae in terms of 16π the exchange by transversely polarized gluons like in |teff| comes from the integration of Eq. (4) over t. η is the derivation of the Weizsäcker–Williams approxima- the ratio of the real part to the imaginary part of the tion. In the nonrelativistic approximation the binding amplitude of J/ψN scattering. The t dependence of is dominated by a Coulomb potential. The Yang– the differential cross section is given by the square of Mills vertex between the Coulomb potential and trans- 98 L. Gerland et al. / Physics Letters B 619 (2005) 95–104 Fig. 2. The form factors squared of the diagonal (J/ψ → J/ψ Fig. 3. The form factors squared of the diagonal (J/ψ → J/ψ and and ψ → ψ ) and the nondiagonal transitions (J/ψ → ψ and ψ → ψ ) for two different nonrelativistic charmonium models. ψ → J/ψ). versely polarized gluons is zero. Therefore, only the glect the last term. Taking into account the elastic interaction between the two gluons and the two heavy form factor of the J/ψ reduces this value to |teff|= quarks of Fig. 1 have to be taken into account in this 0.3GeV2. calculation. Eq. (9) differs from the power law that arises in the The nonrelativistic approximation is justified at limit of large t, i.e., in large angle scattering where small momentum transfer because of the large mass of −t/s ∼ 1/2. In this regime, the selection of domi- the c-quark. Small momentum transfers are the most nant diagrams follows from the requirement to obtain important domain because the two-gluon form factor the lowest power of t. In the literature this is known decreases rather quickly with the momentum transfer. as power counting rules. However, in the processes The result for the elastic J/ψ form factor is shown in considered in this Letter, this integral is dominated − 2 ∼ Fig. 2. Additionally, Fig. 2 depicts the elastic ψ form by ( t)rN 1, where rN is the radius of a nucleon. factor as well as the nondiagonal transition from the This kinematical region does not overlap with high- J/ψ into the ψ .InFig. 3 the dependence of the form momentum transfers. factor on the charmonium model is shown. The elas- Two important phenomena are neglected in the tic form factors of the J/ψ and the ψ calculated in VDM model. One is the color transparency phenom- two different charmonium models are depicted. The enon due to production of cc¯ in configurations sub- charmonium models are from Refs. [13,14]. One can stantially smaller than the mean J/ψ size. As a result see that the dependence on the charmonium model is the effective cross section σtot(J/ψN) as extracted small in comparison to other uncertainties. The elastic from the J/ψ photoproduction off a nucleon is much cross section is then proportional to smaller than the genuine cross section of the J/ψN interaction. Another neglected effect is the hard con- | |= 2 2 teff dt F2g(t)Fψ (t). (8) tribution to σtot which rapidly increases with energy [9]. Therefore, we use the correspondence between the Integrating the two-gluon form factor of (J/ψ → GVDM and the QCD dipole model which leads to the J/ψ) over t yields parametrization of cross section see Ref. [2] s + m2 −3 2 1 2 2g dt F (t) = m 1 − 0.08 0.2 2g 3 2g m2 s s 2g σ (J/ψN) = 3.2mb + 0.3mb tot s s ≈ 0.4GeV2. (9) 0 0 (10) The approximation at the end of this equation is for 2 sufficiently high energies, where it is possible to ne- with s0 = 39.9GeV . L. Gerland et al. / Physics Letters B 619 (2005) 95–104 99 It is worth noting here that such a parametrization is cross section and the nondiagonal cross section (ψ + reasonable only for the energies where inelastic non- N → J/ψ + N). One can see that the nondiagonal diffractive channels (the lowest√ nondiffractive channel cross section (ψ + N → J/ψ + N) is comparable ¯ is Λc + D) are open, that is for s>4.15 GeV, in the with the elastic J/ψ-nucleon cross section. rest system of the nucleon this is ω>3.61 GeV. The amplitudes within the GVDM are related by e e fγψ = fψψ + fψ ψ , fψ fψ e e fγψ = fψψ + fψ ψ . (11) fψ fψ The amplitudes, fψψ and fψ ψ , that appear here ad- ditionally in comparison to the VDM in Eq. (1) are the amplitudes for the nondiagonal transitions J/ψ → ψ and ψ → J/ψ, respectively. The amplitudes follow- ing from Eqs. (10) and (11) are given in Appendix A. The results of Eqs. (5) and (10) (the total and the elastic cross section for J/ψN collisions) are shown in Fig. 4. Fig. 5 shows the same for ψ N collisions. The ratio of the elastic to the total cross section is de- picted in Fig. 6. The elastic cross section calculated with and with- out the real part of the amplitude is shown in Fig. 7. The real part contributes approximately 2% to the elas- Fig. 5. The elastic and the total ψ -nucleon cross section in depen- tic cross section in the discussed energy range. dence of the energy of the ψ in the rest frame of the nucleon. Fig. 8 shows the energy dependence of the elas- tic J/ψ-nucleon cross section, the elastic ψ -nucleon Fig. 6. The ratio of the elastic charmonium–nucleon cross section Fig. 4. The elastic and the total J/ψ-nucleon cross section in de- to the total charmonium–nucleon cross section in dependence of the pendence of the energy of the J/ψ in the rest frame of the nucleon. energy of the charmonium in the rest frame of the nucleon. 100 L. Gerland et al. / Physics Letters B 619 (2005) 95–104 tion of charmonium states in the antiproton–nucleus collisions at the ψ threshold. The direct production of J/ψ’s is suppressed here. However, a ψ is pro- duced and becomes an J/ψ in a further collision with a nucleon in the nuclear target. Since the pro- duced hidden charm state has a large momentum rel- ative to the nucleus target the semiclassical Glauber- approximation can be used. In our calculation we will neglect color transparency effects in the initial state for the production of J/ψ and ψ mesons [18], since the coherence length for the fluctuation of the incoming antiproton into a small configuration is very small at the relevant energies practically completely washing out the CT effect [19]. The production of a J/ψ at the threshold in a pA¯ collision and the subsequent production of a ψ in a rescattering of the J/ψ, is not well suited for the mea- Fig. 7. The elastic J/ψ-nucleon cross section in dependence of the surements of the nondiagonal√ cross sections. This is energy of the J/ψ in the rest frame of the nucleon is shown with because in a pA¯ collision at s = mψ = 3.1GeV and without the real part of the amplitude. the J/ψ is produced at rest in the center of mass system. This means the energy in the center of mass of the J/ψ and the nucleon is 4.5 GeV. The thresh- old for the production of a ψ in such a collision is mψ + mN = 4.626 GeV. This boundary is extended when nucleon Fermi motion within the nuclear target is taken into account (see Ref. [19] for the discussion of the role of Fermi motion effects in the production of charmonium states). However, this process is strongly suppressed by the phase space. At the same time, the process J/ψ+p → ¯ Λc + D is likely to dominate the inelastic cross sec- tion. Hence the measurement of the process p¯ + A → + ¯ + = 2 Λc D X in the vicinity of spp¯ mJ/ψ will allow a direct measurement of σinel(J/ψN). In the semiclassical Glauber-approximation the cross section to produce a ψ in an antiproton–nucleus collision is σ(p¯ + A → ψ ) Fig. 8. The elastic J/ψ-nucleon cross section, the elastic n ψ -nucleon cross section and the nondiagonal cross section = · p ¯ + → 2π db bdz1 ρ(b,z1)σ (p p ψ ) (ψ + N → J/ψ + N) in dependence of the energy of the char- A monium in the rest frame of the nucleon. z1 × exp − dzσpN¯ inelρ(b,z) ¯ 3. pA collisions at the ψ and the J/ψ threshold −∞ ∞ A program of studies of charmonium production × − dzσ ρ(b,z) . in a pA¯ collisions at a p¯ accumulator is planned exp ψ Ninel (12) [15]. Hence we discuss in this section the produc- z1 L. Gerland et al. / Physics Letters B 619 (2005) 95–104 101 In this formula, b is the impact parameter of the antiproton–nucleus collision, np is the number of pro- tons in the nuclear target, z1 is the coordinate of the production point of the ψ in beam direction, and ρ is the nuclear density. σ(p¯ + p → ψ ) is the cross sec- tion to produce a ψ in an antiproton–proton collision. σpN¯ inel is the inelastic antiproton–nucleus cross sec- tion. σψ Ninel is the inelastic ψ -nucleon cross section. All the factors in Eq. (12) have a rather direct inter- pretation z1 exp − dzσpN¯ inelρ(b,z) −∞ gives the probability to find an antiproton at the coor- Fig. 9. The ratio σ(p¯ + A → ψ + nuclear fragments)/σ (p¯ + A → ψ + nuclear fragments) is shown for 5 different sets of parameters dinates (b, z1), which accounts for its absorption, and np ρ(b,z )σ (p¯ + p → ψ ) is the probability to create (see text for further details). Shown are the nuclear targets O, S, Cu, A 1 W, and Pb. The lines are just to guide the eye. np a ψ at these coordinates. The factor A accounts for the fact that close to the threshold the antiproton can The factor produce a ψ only in an annihilation with a proton but z not with a neutron. The term 2 exp − dzσ ρ(b,z) ∞ ψ Ninel z1 exp − dzσψ inelρ(b,z) is the probability to find, at the coordinate (b, z2), z1 a ψ that was produced at (b, z1). σ(ψ + N → ψ + gives the probability that the produced ψ has no in- N)ρ(b,z2) is the probability that a ψ collides at the elastic collision in the nucleus, i.e., that it survives on coordinate (b, z2) with a nucleon and that a J/ψ is the way out of the nucleus. produced. Finally, Then, in the semiclassical Glauber-approximation ∞ J/ψ the cross section to subsequently produce a in an − antiproton–nucleus collision is exp dzσψinelρ(b,z) z2 ¯ + → + σ(p A J/ψ X) is the probability that the outgoing J/ψ has no in- np elastic interactions. θ(z2 − z1) = 0forz2 z1 Eqs. (12) and (13), the ratio does not depend on the cross section σ(p¯ + p → J/ψ+ X), which is not well × σ(ψ + N → ψ + N)ρ(b,z2) ∞ known at the threshold. In Fig. 9 we used five sets of parameters. “Normal” × − dzσ ρ(b,z) . exp ψNinel (13) means that the inelastic antiproton–nucleon cross sec- z 2 tion is σpN¯ inel = 50 mb, the inelastic cross section of 102 L. Gerland et al. / Physics Letters B 619 (2005) 95–104