Physics Letters B 566 (2003) 8–18

www.elsevier.com/locate/npe

Effective number of neutrinos and baryon asymmetry

from BBN and WMAP
V. Barger a,f , James P. Kneller b , Hye-Sung Lee a ,
Danny Marfatia c,f , Gary Steigman d,e,f
a Department of Physics, University of Wisconsin, Madison, WI 53706, USA
b Department of Physics, North Carolina State University, Raleigh, NC 27695, USA
c Department of Physics, Boston University, Boston, MA 02215, USA
d Department of Physics, The Ohio State University, Columbus, OH 43210, USA
e Department of Astronomy, The Ohio State University, Columbus, OH 43210, USA
f Kavli Institute for Theoretical Physics, University of California, Santa Barbara, CA 93106, USA

Received 12 May 2003; accepted 23 May 2003

Editor: M. Cvetič

Abstract
We place constraints on the number of relativistic degrees of freedom and on the baryon asymmetry at the epoch of Big Bang
Nucleosynthesis (BBN) and at recombination, using cosmic background radiation (CBR) data from the Wilkinson Microwave
Anisotropy Probe (WMAP), complemented by the Hubble Space Telescope (HST) Key Project measurement of the Hubble
constant, along with the latest compilation of deuterium abundances and H II region measurements of the primordial helium
abundance. The agreement between the derived values of these key cosmological and particle physics parameters at these
widely separated (in time or redshift) epochs is remarkable. From the combination of CBR and BBN data, we find the 2σ
ranges for the effective number of neutrinos Nν and for the baryon asymmetry (baryon to photon number ratio η) to be 1.7–3.0
and 5.53–6.76 × 10−10 , respectively.
 2003 Published by Elsevier B.V. Open access under CC BY license.

1. Introduction WMAP precision measurements of the CBR tempera-

ture and polarization anisotropy spectra [1]. However,
despite the impressive successes of the standard model
The concordance model of cosmology, with dark
in describing a wide range of cosmological data, the
energy, dark matter, baryons, and three flavors of light
possibility remains that there could be non-standard
neutrinos, provides a consistent description of BBN
model contributions to the total energy density in the
(∼20 min), the CBR (∼380 Kyr), and the galaxy for-
radiation era from additional relativistic particles.
mation epochs of the universe (
1 Gyr). The stan-
In this Letter new constraints are placed on any
dard model has received recent confirmation from the
physics beyond the standard model that contributes to
the energy density like radiation (i.e., decreases with
E-mail address: [email protected] (D. Marfatia).
the expansion of the universe as the fourth power of

0370-2693  2003 Published by Elsevier B.V. Open access under CC BY license.

doi:10.1016/S0370-2693(03)00800-1
 V. Barger et al. / Physics Letters B 566 (2003) 8–18 9

the scale factor, independent of the sign of that con- standard model,
tribution). While such new physics may or may not
be due to extra relativistic degrees of freedom, it is Pre-e± annihilation: tTγ2 = 0.738 MeV2 s, (3)
±
assumed that the non-standard contribution to the en- Post-e annihilation: tTγ2 = 1.32 MeV s.
2
(4)
ergy density may be parameterized as such. Simulta-
The most straightforward variation of the standard
neously, constraints are placed on the baryon density
cosmology is “extra” energy contributed by new,
at widely differing epochs in the evolution of the uni-
light (relativistic at BBN) particles “X”. These might,
verse. The keys to these constraints are the recently
but need not be sterile neutrinos. When the X are
released measurements of the CBR anisotropy spectra
decoupled, in the sense that they do not share in the
by the WMAP Collaboration, the most recent compila-
energy released by e± annihilation, it is convenient
tion of high redshift, low metallicity deuterium abun-
to account for the extra contribution to the standard
dances [2] and 4 He abundances relevant to BBN.
model energy density by normalizing it to that of an
“equivalent” neutrino [3],
2. Modified relativistic energy density 7
ρX ≡ Nν ρν = Nν ργ . (5)
8
The cosmology of interest here begins when the For SBBN Nν = 0, where Nν ≡ 3 + Nν . For
universe is already a few tenths of a second old each additional “neutrino-like” particle (i.e., any two-
and the temperature is a few MeV. At such early component fermion), if TX = Tν , then Nν = 1; if X
epochs the total energy density receives its dominant is a scalar, Nν = 4/7. However, it may well be that
contribution from all the relativistic particles present the X have decoupled even earlier in the evolution of
(the evolution of the universe is said to be “radiation- the universe and have failed to profit from the heat-
dominated” (RD)). In the standard cosmology, prior ing when various other particle–antiparticle pairs an-
to e± annihilation, these relativistic particles are: nihilated (or unstable particles decayed). In this case,
photons, e± pairs and three flavors of left-handed (i.e., the contribution to Nν from each such particle will
one helicity state) neutrinos (and their right-handed be < 1 (< 4/7). We emphasize that, in principle, we
antineutrinos). Then, the energy density is are considering any term in the energy density which
43 scales like a −4 , where a is the scale factor. In this
ρTOT = ρR = ργ + ρe + 3ρν = ργ , (1) sense, the modification to the usual Friedman equation
8
due to higher dimensional effects, as in the Randall–
where ργ is the energy density in photons (which by Sundrum model [4] (see also, [5–12]), can be included
today have redshifted to become the CBR photons at a as well. An important interest in this latter case is that
temperature of about 2.7 K). it permits the possibility of a negative contribution to
In “standard” BBN (SBBN) it is assumed that the the radiation density (Nν < 0; Nν < 3).
neutrinos are fully decoupled prior to e± annihilation In the presence of such a modification to the
and do not share in the energy transferred from the relativistic energy density, the pre-e± annihilation
annihilating e± pairs to the CBR photons. In this energy density in Eq. (1) is changed to,
approximation, the photons in the post-e± annihilation

universe are hotter than the neutrinos by a factor 43 7Nν
(ρR )pre = 1+ ργ . (6)
Tγ /Tν = (11/4)1/3, and the relativistic energy density 8 43
is Any extra energy density (Nν > 0) speeds up the
expansion of the universe so that the right-hand side of
ρR = ργ + 3ρν = 1.6813 ργ . (2)
the time-temperature relation in Eq. (3) is smaller by
During the RD epoch the age and the energy the square root of the factor in parentheses in Eq. (6),
density are related by 43 ρR t 2 = 1 (we have chosen

 7Nν 1/2
units in which 8πG = 1), so that once the particle Spre ≡ (t/t )pre = 1 +
content (ρR ) is specified, the age of the universe is 43
known as a function of the CBR temperature. In the = (1 + 0.1628Nν )1/2 , (7)
10 V. Barger et al. / Physics Letters B 566 (2003) 8–18

where t  is the age of the universe with the extra energy

density. In the post-e± annihilation universe the extra
energy density is diluted by the heating of the photons,
so that,

(ρR )post = 1.6813(1 + 0.1351Nν )ργ , (8)

and

Spost ≡ (t/t  )post = (1 + 0.1351Nν )1/2 . (9)

This latter expression (Eq. (9)) is also relevant for the
modification to the spectrum of temperature fluctua-
tions in the CBR (when compared with the standard
Nν = 3 case). Fig. 1. The CBR degeneracy between ωM and Nν is evident from
the 1σ and 2σ contours from the WMAP data.

3. Constraints on Nν from the CBR • In the matter dominated (MD) era before re-
combination, the density contrast (δρ/ρ) of the pres-
The competition between gravitational potential sureless cold dark matter (CDM) grows unimpeded
and pressure gradients is responsible for the peaks and (as t 2/3 ) while the density contrast of the baryons is
troughs in the CBR power spectrum. The redshift of either oscillating or decaying. The longer this pre-
matter-radiation equality, recombination MD era lasts, the more suppressed are
ωM the amplitudes of the peaks.
zeq = 2.4 × 104 2
, (10)
Spost Conversely, if matter-radiation equality is delayed, the
affects the time (redshift) duration over which this gravitational potential is dominated by the photon–
competition occurs. Here, ωM ≡ ΩM h2 is the total baryon fluid closer to recombination resulting in a
matter density (comprised, for nearly massless neutri- more pronounced peak structure.
nos, of baryons and cold dark matter) and h (H0 ≡ An increase in the relativistic content causes the
100h km s−1 Mpc−1 ) is the normalized Hubble con- universe to be younger at recombination with a corre-
stant. The direct correlation between ωM and Nν spondingly smaller sound horizon s∗ . Since the loca-
[13] is evident in Fig. 1 which results from our analy- tion of the nth peak scales roughly as nπD∗ /s∗ (where
sis described below. The primary effects of relativistic D∗ is the comoving angular diameter distance to re-
degrees of freedom (other than photons) on the CBR combination), the peaks shift to smaller angular scales
power spectrum result essentially from changing the (larger l) and with greater separation. These features
redshift of matter-radiation equality. If the radiation are clearly visible in Fig. 2.
content is increased, matter-radiation equality is de- The heights and locations of the peaks also depend
layed, and occurs closer (in time and/or redshift) to on the history of the universe after recombination.
the epoch of recombination. At the end of matter domination and the onset of
The redshift of matter-radiation equality is impor- dark energy domination, further and much slower
tant for two reasons [14]: (compared to that in the radiation epoch) potential
decay occurs. The more gradual potential decay causes
• Radiation causes potential decay which blueshifts the induced anisotropy to be suppressed by a factor
the photons because they do not have to climb out of of l. The amplification of the power in the lowest
such deep wells. Moreover, the concurrent decay in l’s from this late decay serves as a probe of dark
the spatial curvature doubles the blueshift effect by energy (or another probe of the matter content in a
contracting the wavelength of the photons relative to flat universe). In principle, the degeneracy between
the pure cosmological expansion. Nν and ωM is broken by this effect and by the
 V. Barger et al. / Physics Letters B 566 (2003) 8–18 11

• 0.64  h  0.8 in steps of size 0.02.

• 0.018  ωB  0.028 in steps of size 0.001.
• 0.11  ωM  0.27 in steps of size 0.01 and ωM =
0.07, 0.3.
• 1  Nν  3.5 in steps of size 0.25, 4  Nν  9 in
steps of size 0.5 and Nν = 0, 0.5.
• 0  τ  0.3 in steps of size 0.025.
• 0.90  ns  1.02 in steps of size 0.01 and ns =
0.80, 0.84, 0.88, 1.04, 1.08, 1.12, 1.16, 1.20.
• The normalization of the spectrum is a continuous
parameter.

The first year WMAP data are in the form of

899 measurements of the TT power spectrum from
Fig. 2. The power spectrum for the best-fit (Nν = 2.75) to the
WMAP data is the solid line. With all other parameters and the
l = 2 to l = 900 [20] and 449 data points of the
overall normalization of the primordial spectrum fixed, the spectra TE power spectrum [21]. Although the effect of
for Nν = 1, Nν = 5 and Nν = 7 are the dotted, dot-dashed and relativistic degrees of freedom on the TE spectrum
dashed lines, respectively. The data points represent the binned TT is insignificant, it is included in our analysis for
power spectrum from WMAP. completeness. The likelihood of each model of our
grid is computed using Version 1.1 of the code
provided by the WMAP Collaboration [22]. The code
accompanying change in the redshift at which the computes the covariance matrix under the assumption
matter dominated epoch ends. However, note that that the off-diagonal terms are subdominant. This
the lowest multipoles also have the largest cosmic approximation breaks down for unrealistically small
variance. amplitudes. When the height of the first peak is
The TT and TE power spectra are computed us- below 5000 µK2 (which is many standard deviations
ing the Code for Anisotropies in the Microwave Back- away from the data), only the diagonal terms of the
ground or CAMB [15] which is a parallelized version covariance matrix are used to compute the likelihood.
of CMBFAST [16]. The Universe is assumed to be flat, The best-fit parameters are h = 0.68, ωB = 0.023
in accord with the predictions of inflation [17], and the (η10 = 6.3), ωM = 0.14, Nν = 2.75, τ = 0.13, and
dark energy is assumed to behave as a cosmological ns = 0.97 with a χ 2 = 1429.13 for 1341 degrees of
constant Λ. The restriction of a flat geometry allows freedom. The allowed parameter space in the η10 –
us to relate the dark energy and matter densities at the Nν plane is shown in Fig. 3. The solid (dotted) lines
present time: ΩΛ = 1 − ΩM . The angular power spec- correspond to the 1σ - and 2σ -regions1 for t0 > 11
trum is calculated on a grid defined by h, the baryon (12) Gyr. The cross identifies the best-fit point. Note
density ωB ≡ ΩB h2 (or η10 ≡ 1010 nB /nγ = 274ωB ), that while this best fit point lies at Nν < 3, the Nν
ωM , the number of equivalent neutrinos Nν , the reion- distribution is very broad. After marginalizing over η,
ization optical depth τ , and the spectral index ns of the the 2σ range in Nν extends from 0.9 (Nν = −2.1)
primordial power spectrum. Two priors are imposed to to 8.3 (Nν = 5.3). Although similar CBR analyses
largely break the degeneracy between ωM and Nν . (see Ref. [23]) have included different, additional data
For h a top-hat distribution is chosen corresponding to that from WMAP alone, making direct comparisons
to the HST measurement, h = 0.72 ± 0.08 [18], and difficult, our results are in good agreement with them.
we require that the universe be older than the globular The prior on t0 has a significant effect on the allowed
clusters (which, at 2σ , are older than 11 Gyr [19]). For values of Nν [24] for a simple reason. Since flatness
comparison, we also consider the case when the age of
the universe t0 exceeds 12 Gyr.
Our top-hat grid, consisting of over 10 million 1 For 2-dimensional constraints, the 1σ -, 2σ - and 3σ -regions are
points, is: defined by χ 2 = 2.3, 6.17 and 11.83, respectively.
12 V. Barger et al. / Physics Letters B 566 (2003) 8–18

nos in thermal equilibrium with the electron–positron–

photon plasma become slow compared to the universal
expansion rate and the neutrinos decouple, although
they do continue to interact with the neutrons and pro-
tons via the charged-current weak interactions. Prior
to e± annihilation, when the temperature drops below
∼0.8 MeV and the universe is ≈1 second old, these
interactions, interconverting neutrons and protons, be-
come too slow (compared to the universal expansion
rate) to maintain n–p equilibrium and the neutron-to-
proton ratio begins to deviate from (exceeds) its equi-
librium value ((n/p)eq = exp(−m/T )), where m
is the neutron–proton mass difference. Beyond this
point, often described as neutron–proton “freeze-out”,
the n/p ratio continues to decrease, albeit more slowly
Fig. 3. The 1σ and 2σ contours in the η10 –Nν plane from WMAP than would have been the case in equilibrium. Since
data. The solid (dotted) lines correspond to t0 > 11 (12) Gyr. The there are several billion CBR photons for every nu-
cross marks the best-fit at ωB = 0.023 and Nν = −0.25. cleon (baryon), the abundances of any complex nuclei
are entirely negligible at these early times.
is assumed, t0 depends only on the matter content and We note here that if there is an asymmetry be-
the Hubble parameter via tween the numbers of νe and ν̄e (“neutrino degen-

√  eracy”), described by a chemical potential µe , then
2 1 1 + 1 − ΩM
H0 t0 √ ln √ . (11) the equilibrium neutron-to-proton ratio is modified to
3 1 − ΩM ΩM (n/p)eq = exp(−m/T − µe /T ). In place of the neu-
The combination of the HST prior on h and the t0 prior trino chemical potential, it is convenient to introduce
restricts ωM and help to break the degeneracy between the dimensionless degeneracy parameter ξe ≡ µe /T .
Nν and ωM . A positive chemical potential (ξe > 0; more νe than
The best fit WMAP-determined baryon density is ν̄e ) leads to fewer neutrons and less 4 He will be syn-
η10 = 6.30 (ωB = 0.0230), in excellent agreement thesized in BBN.
with Spergel et al. [25] and other similar analyses [23]. BBN begins in earnest after e± annihilation, at
The CBR 2σ range extends from η10 = 5.58 (ωB = T ≈ 0.08 MeV (t ≈ 3 min), when the number density
0.0204) to η10 = 7.26 (ωB = 0.0265). of those CBR photons with sufficient energy to pho-
These CBR constraints on Nν and ωB apply to todissociate deuterium (those in the tail of the black
epochs in the evolution of the universe
380 Kyr. body distribution) is comparable to the baryon den-
An important test of the standard models of cosmol- sity. By this time the n/p ratio has further decreased
ogy and particle physics is to compare them with cor- (the two-body reactions interconverting neutrons and
responding constraints from the much earlier epoch protons having been somewhat augmented by ordi-
probed by BBN. nary beta decay; τn = 885.7 s), limiting (mainly) the
amount of helium-4 which can be synthesized. As a
result, the predictions of the primordial abundance of
4 He depend sensitively on the early expansion rate and
4. The roles of Nν and η10 in BBN
on the amount—if any—of a νe –ν̄e asymmetry.
At T ∼ few MeV, the neutrinos are beginning to In contrast to 4 He, the BBN-predicted abundances
decouple from the γ − e± plasma and the neutron of deuterium, helium-3 and lithium-7 (the most abun-
to proton ratio, crucial for the production of primor- dant of the nuclides synthesized during BBN) are de-
dial 4 He, is decreasing. As the temperature drops be- termined by the competition between the various two-
low ∼2 MeV, the two-body collisions between neu- body production/destruction rates and the universal
trinos and e± pairs, responsible for keeping the neutri- expansion rate. As a result, the D, 3 He, and 7 Li abun-
 V. Barger et al. / Physics Letters B 566 (2003) 8–18 13

dances are sensitive to the post-e± annihilation expan- Li/H, omitted for clarity, are similar in behavior to
sion rate, while that of 4 He depends on both the pre- that of D/H). The trends illustrated in Fig. 4 are easy
and post-e± annihilation expansion rates; the former to understand in the context of the discussion above.
determines the “freeze-in” and the latter modulates the The higher the baryon density (η10 ), the faster pri-
importance of beta decay (see, e.g., Kneller, Steigman mordial D is destroyed, so the relic abundance of D
[26]). Also, the primordial abundances of D, 3 He, and is anticorrelated with η10 . But, the faster the uni-
Li, while not entirely insensitive to neutrino degen- verse expands (Nν > 0), the less time is available
eracy, are much less effected by a non-zero ξe (e.g., for D-destruction, so D/H is positively, albeit weakly,
[27]). correlated with Nν . In contrast to D (and to 3 He and
Of course, the BBN abundances do depend on the Li), since the incorporation of all available neutrons
baryon density which fixes the nuclear reactions rates into 4 He is not limited by the very rapid nuclear re-
and also, through the ratio of baryons to photons, reg- action rates, the 4 He mass fraction is relatively insen-
ulates the time/temperature at which BBN begins. As sitive to the baryon density, but it is very sensitive to
a result, the abundances of at least two different relic both the pre- and post-e± annihilation expansion rates
nuclei are needed to break the degeneracy between the (which control the neutron-to-proton ratio). The faster
baryon density and a possible non-standard expansion the universe expands, the more neutrons are available
rate resulting from new physics or cosmology, and/or for 4 He. The very slow increase of Y with η10 is a
a neutrino asymmetry. In this Letter only the former reflection of the fact that for higher baryon density,
possibility is considered; in another publication sev- BBN begins earlier, when there are more neutrons.
eral of us (along with Langacker) have explored the As a result of these complementary correlations, the
consequences of neutrino degeneracy and we studied pair of primordial abundances yD ≡ 105 (D/H) and the
the modifications to the constraints on Nν when both 4 He mass fraction Y, provide observational constraints

of these non-standard effects are simultaneously in- on both the baryon density and the universal expan-
cluded. sion rate when the universe was some 20 minutes old.
While the abundances of D, 3 He, and Li are most Comparing these to constraints when the universe was
sensitive to the baryon density (η), the 4 He mass frac- some 380 Kyr old, from the WMAP observations of
tion (Y) provides the best probe of the expansion the CBR spectra, provides a test of the consistency
rate. This is illustrated in Fig. 4 where, in the Nν – of the standard models of cosmology and of particle
η10 plane, are shown isoabundance contours for D/H physics and further constrains the allowed range of the
and Y (the isoabundance curves for 3 He/H and for present baryon density of the universe.

5. Primordial abundances

It is clear from Fig. 4 that while D (and/or 3 He

and/or 7 Li) largely constrains the baryon density and
4 He plays the same role for N , there is an interplay
ν
between η10 and Nν which is quite sensitive to the
adopted abundances. For example, a lower primordial
D/H increases the BBN-inferred value of η10 , leading
to a higher predicted primordial 4 He mass fraction. If
the primordial 4 He mass fraction derived from the data
is “low”, then a low upper bound on Nν will be in-
Fig. 4. Isoabundance curves for D and 4 He in the η10 –Nν plane. ferred. Therefore, it is crucial to make every effort to
The nearly horizontal curves are for 4 He (from top to bottom:
avoid biasing any conclusions by underestimating the
Y = 0.25, 0.24, 0.23). The nearly vertical curves are for D (from
left to right: 105 (D/H) = 3.0, 2.5, 2.0. The data point with error present uncertainties in the primordial abundances de-
bars corresponds to yD = 2.6 ± 0.4 and Y = 0.238 ± 0.005; see the rived from the observational data. For this reason deu-
text for discussion of these abundance values. terium is adopted as the baryometer of choice. Primar-
14 V. Barger et al. / Physics Letters B 566 (2003) 8–18

ily, this is because its observed abundance should have ξe = 0), at ±1σ this corresponds to a baryon density
only decreased since BBN [28], but also because the η10 = 6.1+0.7
−0.5 (ωB = 0.022 ± 0.002).
3

deuterium observed in the high redshift, low metallic- A similar, less than clear situation exists for de-
ity QSO absorption line systems (QSOALS) should be terminations of the primordial abundance of 4 He.
very nearly primordial. In contrast, the post-BBN evo- At present there are two, largely independent, esti-
lution of 3 He and of 7 Li are considerably more compli- mates based on analyses of large data sets of low-
cated, involving competition between production, de- metallicity, extragalactic H II regions. The “IT” [39,
struction, and survival. As a result, at least so far, the 40] estimate of Y(IT) = 0.244 ± 0.002, and the “OS”
current, locally observed (in the Galaxy) abundances determination [41–43] of Y(OS) = 0.234 ± 0.003
of these nuclides have been of less value in constrain- which differ by nearly 3σ . The recent analysis of
ing the baryon density than has deuterium. Nonethe- high quality observations of a relatively metal-rich
less, inferring the primordial D abundance from the (hence, chemically evolved and post-primordial) H II
QSOALS has not been without its difficulties, with region in the Small Magellanic Cloud (SMC) by Pe-
some abundance claims having been withdrawn or re- imbert, Peimbert, and Ruiz (PPR) [44] yields an abun-
vised. Presently there are 5–6 QSOALS with reason- dance YSMC = 0.2405 ± 0.0018. When PPR extrap-
ably firm deuterium detections [2,29–33]. However, olated this abundance to zero metallicity, they found
there is significant dispersion among the abundances Y(PPR) = 0.2345 ± 0.0026, lending support to the
and the data fail to reveal the anticipated “deuterium OS value. These comparisons of different observations
plateau” at low metallicity or at high redshift [34]. Fur- and analyses suggest that unaccounted systematic er-
thermore, subsequent observations of the D’Odorico rors may dominate the statistical uncertainties. Indeed,
et al. [33] QSOALS by Levshakov et al. [35] re- Gruenwald, Steigman, and Viegas [45] argue that if
vealed a more complex velocity structure and led to a unseen neutral hydrogen in the ionized helium region
revised—and more uncertain—deuterium abundance. of the observed H II regions is accounted for, the IT es-
This sensitivity to the often poorly constrained veloc- timate of the primordial abundance should be reduced
ity structure in the absorbers is also exposed by the to Y(GSV) = 0.238 ± 0.003 (see also [46,47]). Here,
analyses of published QSOALS data by Levshakov we adopt this latter estimate for the central value but,
and collaborators [36–38], which lead to consistent, as we did with deuterium, the uncertainty is increased
but somewhat higher deuterium abundances than those in an attempt to account for likely systematic errors:
inferred from “standard” data reduction analyses. In Y = 0.238 ± 0.005, leading to a 2σ range, 0.228 
the absence of a better motivated choice, here we Y  0.248; this range is in accord with the estimate
adopt the five abundance determinations collected in adopted by Olive, Steigman, and Walker (OSW) [48]
the recent Letter of Kirkman et al. [2]. The weighted in their review of SBBN. Although we will comment
mean value of yD is 2.6.2 But, the dispersion among on the modification to any conclusions if Y(IT) is sub-
these five data points is very large. For this data set stituted for Y(OSW), Figs. 4–8 are shown for yD =
χ 2 = 15.3 for four degrees of freedom, suggesting that 2.6 ± 0.4 and Y(OSW) = 0.238 ± 0.005.
one or more of these abundance determinations may
be in error, perhaps affected by unidentified and un-
accounted for systematic errors. For this reason, we 6. Standard BBN
follow the approach advocated by [31] and [2] and
adopt for the uncertainty in yD the dispersion divided
Before proceeding to our main goal of constrain-
by the square root of the number of data points. Thus,
ing new physics using BBN, it is worthwhile to set the
the primordial abundance of deuterium to be used here
scene by considering the standard model case (Nν = 3,
is chosen to be: yD = 2.6 ± 0.4. For SBBN (Nν = 3,
ξe = 0) first. The result of this comparison is well

2 This differs from the result quoted in Kirkman et al. because 3 We have purposely avoided quoting the baryon density to
they have taken the mean of log(yD ) and then used it to infer yD more significant figures than is justified by the accuracy of the
(yD ≡ 10log(yD ) ). D-abundance determination.
 V. Barger et al. / Physics Letters B 566 (2003) 8–18 15

baryon density [49]. At the same time we strongly em-

phasize that this “discrepancy” is only at the ∼ 2σ
level and we should celebrate the excellent agreement
between the baryon density determined when the uni-
verse was only 20 minutes old and when the universe
was some 380 Kyr old (CBR).

7. Non-standard BBN: Nν
= 0

As noted above, for SBBN (Nν = 3) the observa-

tionally inferred primordial abundance of 4 He is too
small (by ∼2σ ) for the baryon density inferred either
from the D abundance or from the non-BBN analysis
of Spergel et al. [25]. This suggests that the early uni-
verse expansion rate may have been too fast, leaving
too many neutrons available for the synthesis of 4 He.
If this tension between 4 He and D (or, between 4 He
Fig. 5. The band is the SBBN predicted relation between the and ωB ) should persist, it could be a signal of non-
primordial abundances of D and 4 He, including the errors (±1σ )
standard physics corresponding to S < 1 (Nν < 0).
in those predictions from the uncertainties in the nuclear and weak
interaction rates. The point with error bars is for the relic abundances Indeed, in Fig. 4 it can be seen that for the adopted
of D and 4 He adopted here (see the text). primordial abundances of D and 4 He, there is a “per-
fect” fit (χ 2 = 0) for Nν ≈ −0.7 (Nν ≈ 2.3) and
η10 ≈ 5.7. Although Nν = 3 is only disfavored by
known: there is a “tension” between the primordial ∼ 2σ , any increase in the early universe expansion
abundances of D and 4 He inferred from the observa- rate (S > 1, Nν > 3) is strongly disfavored. This is
tional data and those predicted by SBBN. For example, illustrated in Fig. 6 which shows the 1σ -, 2σ - and
if yD is used to fix the baryon density (yD = 2.6 ± 0.4) 3σ -contours in the η10 –Nν plane for the adopted D
SBBN = 6.1+0.7 (ω = 0.022 ± 0.002),
then, at ±1σ , η10 −0.5 B and 4 He (OSW) abundances. The shape of these con-
the corresponding predicted 4 He abundance is Y = tours reflects our much discussed complementarity be-
0.248 ± 0.001 (1σ ), which is some 2σ higher than tween D and 4 He: D provides the best constraint on the
either the OSW or the IT estimates. This is illus- baryon density while 4 He is most sensitive to the early
trated in Fig. 5 which shows the SBBN-predicted re- universe expansion rate, the latter providing an excel-
lation between the relic abundances of D and 4 He lent probe of possible new physics.
along with the Y(OSW) abundance estimate adopted With reference to that Fig. 6 we note that even
here. The D-inferred baryon density is in excellent one extra, fully thermalized neutrino (Nν = 1) is
agreement with the baryon density determined inde- strongly disfavored. This seemingly eliminates the
pendently (non-BBN; Nν = 3) by Spergel et al. [25] sterile neutrino suggested by the LSND experiment
from a combination of CBR and Large Scale Struc- [50]. For the LSND parameters, in the absence of
ture data (2dF + Lyman α): η10 non-BBN
= 6.14 ± 0.25 significant neutrino asymmetry, this “sterile” neutrino
(ωB = 0.0224 ± 0.0009). Thus, it appears that 4 He
4
would have been mixed with the active neutrinos
is the problem: the primordial abundance of 4 He is and thermalized prior to neutrino decoupling (prior
smaller than predicted for SBBN given either the ob- to BBN) [51]. If such a neutrino were to exist, the
served deuterium abundance or the non-BBN inferred “new” standard model would correspond to Nν = 4.
This would be a disaster since for Nν = 4 and the
4 It should be noted that from the CBR alone Spergel et al. find OSW 4 He abundance the minimum χ 2 (which now
CBR = 6.6 ± 0.3 (ω = 0.024 ± 0.001). It is this value which is
η10 B
occurs at η10 ≈ 6.1) is greater than 20. The situation is
most directly comparable to our CBR and joint BBN/CBR results. even worse for the IT 4 He abundance since the smaller
16 V. Barger et al. / Physics Letters B 566 (2003) 8–18

Fig. 6. The 1σ -, 2σ - and 3σ -contours in the η10 –Nν plane for Fig. 7. The 1σ -, 2σ - and 3σ -contours (solid lines) in the η10 –Nν
the adopted D and 4 He (OSW) abundances (solid lines). The cross plane for Nν  3 and the adopted D and 4 He (OSW) abundances.
marks the best fit BBN point. The 1σ - and 2σ -contours from The corresponding 1σ - and 2σ -contours from WMAP (dashed
WMAP (dashed lines) are shown for comparison. lines) are shown for comparison.

uncertainty in Y forces a much smaller baryon density

(η10 ≈ 4.7), with χmin
2 > 60!

7.1. Requiring Nν  3

Since, as is well known from LEP [52], there are

three flavors of active, left-handed neutrinos (and their
right-handed antiparticles), any extra contributions
to the relativistic energy density at BBN should
result in Nν > 3. Of course, “new physics” in the
form of non-minimally coupled fields (e.g., [26,53]
and references therein) or from higher-dimensional
phenomena (e.g., [4–12]) may result in an effective
Nν < 3. If, however, the class of non-standard physics
of interest is restricted to Nν  0, then the BBN
constraints presented above (and those from the CBR) Fig. 8. The 1σ -, 2σ - and 3σ -contours in the η10 –Nν plane from
will change. With a prior of Nν  3, the best fit a combination of WMAP data and the adopted D and 4 He (OSW)
BBN-determined values of the baryon-to-photon ratio abundances.
and Nν (for Y(OSW)) shift from η10 ≈ 5.7 and
Nν ≈ −0.7, to η10 ≈ 5.9 and Nν = 0. The value WMAP and from BBN separately is excellent. Guided
2 changes to 4.2. The corresponding confidence
of χmin by this, the BBN and CBR results are combined to
contours in the η10 –Nν plane are shown in Fig. 7. obtain the joint fit in the η–Nν plane shown in
Fig. 8. To a good first approximation, BBN (and
primordial 4 He) determines Nν while WMAP fixes
8. Joint constraints and summary η10 (with some help from BBN and primordial D). The
corresponding figure for Nν  3 for the joint fit has not
As may be seen from Figs. 6 and 7, the agreement been shown because again the Nν range is almost
between the values obtained for Nν and η10 from identical to that from BBN alone (Fig. 7).
 V. Barger et al. / Physics Letters B 566 (2003) 8–18 17

Table 1
The 2σ ranges (for 1 degree of freedom) of Nν and η10 from
analyses of WMAP data, deuterium and helium abundances and
their combinations. The WMAP analysis involves the assumption
of a flat universe, along with the strong HST prior on h and the age
constraint t0 > 11 Gyr. For BBN the adopted primordial abundances
are: yD ≡ 105 (D/H) = 2.6 ± 0.4, Y(OSW) = 0.238 ± 0.005, and
Y(IT) = 0.244 ± 0.002
Nν (2σ range) η10 (2σ range)
WMAP 0.9–8.3 5.58–7.26
yD + Y(OSW) 1.7–3.0 4.84–7.11
yD + Y(IT) 2.4–3.0 5.06–7.33
WMAP + yD + Y(OSW) 1.7–3.0 5.53–6.76
WMAP + yD + Y(IT) 2.4–3.0 5.58–6.71

Table 2 Fig. 10. The marginalized likelihood distributions for η10 from the
The same as Table 1, except that the constraint Nν  3 is imposed joint WMAP and BBN analysis for two choices of the primordial
Nν (2σ bound) η10 (2σ range) abundance of 4 He (solid: OSW, dashed: IT).

WMAP 8.3 5.64–7.30

yD + Y(OSW) 3.3 5.04–7.18 for either the OSW or the IT 4 He abundances, each
yD + Y(IT) 3.1 4.89–6.56 are consistent with Nν = 3 at ∼2σ . These results are
WMAP + yD + Y(OSW) 3.3 5.66–6.80 in excellent agreement with those of Hannestad [23],
WMAP + yD + Y(IT) 3.1 5.54–6.60
with which they are most directly related.
Clearly, BBN constrains Nν much more stringently
than WMAP, while the measurement of η10 by WMAP
is at a precision superior (by a factor ∼2) to that
from BBN. In this sense, the CBR and BBN are quite
complimentary. Indeed, while the constraint on Nν
barely changes with the inclusion of the WMAP data
in a joint analysis with BBN, it is sensitive to the
adopted 4 He abundance; see Fig. 9. On the other hand,
the joint constraint on η is extremely insensitive to
the choice of the 4 He abundance, being dominated by
the WMAP data (and the primordial D abundance);
see Fig. 10. However, BBN and WMAP do provide
a very important consistency check of the standard
model of cosmology at widely separated epochs,
using significantly different physics. The excellent
Fig. 9. The marginalized likelihood distributions for Nν from the agreement between Nν and η10 when the universe was
joint WMAP and BBN analysis for two choices of the primordial 20 minutes and 380 000 years old is a major triumph
abundance of 4 He (solid: OSW, dashed: IT). for the (new) standard model of cosmology.

Our results are summarized in Tables 1 and 2 and

Figs. 9 and 10. BBN and the primordial D abundance Acknowledgements
combine to provide a quite accurate determination
of the baryon density (0.020  ωB  0.025 at 2σ ). Extensive computations were carried out on the
The currently large uncertainty in the primordial CONDOR system at the University of Wisconsin,
abundance of 4 He is responsible for the larger allowed Madison with parallel processing on up to 200 CPUs.
range of Nν . While the best fit value for Nν is < 3 We thank S. Dasu, W. Smith, D. Bradley and S. Rader
18 V. Barger et al. / Physics Letters B 566 (2003) 8–18

for providing access to and assistance with CON- E. Pierpaoli, astro-ph/0302465;

DOR. This research was supported by the US DOE S. Hannestad, astro-ph/0303076.
under Grants Nos. DE-FG02-95ER40896, DE-FG02- [24] See, e.g., J.P. Kneller, R.J. Scherrer, G. Steigman, T.P. Walker,
Phys. Rev. D 64 (2001) 123506;
91ER40676, DE-FG02-02ER41216, and DE-FG02- S.H. Hansen, G. Mangano, A. Melchiorri, G. Miele, O. Pisanti,
91ER40690, by the NSF under Grant No. PHY99- Phys. Rev. D 65 (2002) 023511.
07949 and by the Wisconsin Alumni Research Foun- [25] D.N. Spergel, et al., astro-ph/0302209.
dation. V.B., D.M. and G.S. thank the Kavli Institute [26] J.P. Kneller, G. Steigman, Phys. Rev. D 67 (2003) 063501.
for Theoretical Physics at the University of California, [27] H.S. Kang, G. Steigman, Nucl. Phys. B 372 (1992) 494.
[28] R. Epstein, J. Lattimer, D.N. Schramm, Nature 263 (1976) 198.
Santa Barbara for its support and hospitality. [29] S. Burles, D. Tytler, Astrophys. J. 499 (1998) 699.
[30] S. Burles, D. Tytler, Astrophys. J. 507 (1998) 732.
[31] J.M. O’Meara, et al., Astrophys. J. 552 (2001) 718.
References [32] M. Pettini, D.V. Bowen, Astrophys. J. 560 (2001) 41.
[33] S. D’Odorico, M. Dessauges-Zavadsky, P. Molaro, Astron.
Astrophys. 338 (2001) L1.
[1] C.L. Bennett, et al., astro-ph/0302207. [34] G. Steigman, To appear in the Proceedings of the STScI
[2] D. Kirkman, D. Tytler, N. Suzuki, J. O’Meara, D. Lubin, astro- Symposium, in: M. Livio (Ed.), The Dark Universe: Matter,
ph/0302006. Energy, and Gravity, 2–5 April 2001, astro-ph/0107222.
[3] G. Steigman, D.N. Schramm, J.E. Gunn, Phys. Lett. 66B [35] S.A. Levshakov, M. Dessauges-Zavadsky, S. D’Odorico,
(1977) 202. P. Molaro, Astrophys. J. 565 (2002) 696.
[4] L. Randall, R. Sundrum, Phys. Rev. Lett. 83 (1999) 3370; [36] S.A. Levshakov, W.H. Kegel, F. Takahara, Astrophys. J. 499
L. Randall, R. Sundrum, Phys. Rev. Lett. 83 (1999) 4690. (1998) L1.
[5] J.M. Cline, C. Grojean, G. Servant, Phys. Rev. Lett. 83 (1999) [37] S.A. Levshakov, W.H. Kegel, F. Takahara, Astron. Astro-
4245. phys. 336 (1998) 29L.
[6] P. Binetruy, C. Deffayet, U. Ellwanger, D. Langlois, Phys. Lett. [38] S.A. Levshakov, W.H. Kegel, F. Takahara, Mon. Not. R.
B 477 (2000) 285. Astron. Soc. 302 (1999) 707.
[7] R. Maartens, D. Wands, B. Bassett, I. Heard, Phys. Rev. D 62 [39] Y.I. Izotov, T.X. Thuan, V.A. Lipovetsky, Astrophys. J. Ser. 108
(2000) 041301. (1997) 1.
[8] D. Langlois, R. Maartens, M. Sasaki, D. Wands, Phys. Rev. [40] Y.I. Izotov, T.X. Thuan, Astrophys. J. 500 (1998) 188.
D 63 (2001) 084009. [41] K.A. Olive, G. Steigman, Astrophys. J. Ser. 97 (1995) 49.
[9] S. Mizuno, K. Maeda, Phys. Rev. D 64 (2001) 123521. [42] K.A. Olive, E. Skillman, G. Steigman, Astrophys. J. 483
[10] J.D. Barrow, R. Maartens, Phys. Lett. B 532 (2002) 155. (1997) 788.
[11] K. Ichiki, M. Yahiro, T. Kajino, M. Orito, G.J. Mathews, Phys. [43] B.D. Fields, K.A. Olive, Astrophys. J. 506 (1998) 177.
Rev. D 66 (2002) 043521. [44] M. Peimbert, A. Peimbert, M.T. Ruiz, Astrophys. J. 541 (2000)
[12] J.D. Bratt, A.C. Gault, R.J. Scherrer, T.P. Walker, Phys. Lett. 688.
B 546 (2002) 19. [45] R. Gruenwald, G. Steigman, S.M. Viegas, Astrophys. J. 567
[13] R. Bowen, S.H. Hansen, A. Melchiorri, J. Silk, R. Trotta, Mon. (2002) 931.
Not. R. Astron. Soc. 334 (2002) 760. [46] S.M. Viegas, R. Gruenwald, G. Steigman, Astrophys. J. 531
[14] For a recent review see, W. Hu, S. Dodelson, astro-ph/0110414. (2000) 813.
[15] A. Lewis, A. Challinor, A. Lasenby, Astrophys. J. 538 (2000) [47] D. Sauer, K. Jedamzik, Astron. Astrophys. 381 (2002) 361.
473, astro-ph/9911177, http://camb.info/. [48] K.A. Olive, G. Steigman, T.P. Walker, Phys. Rep. 333 (2000)
[16] U. Seljak, M. Zaldarriaga, Astrophys. J. 469 (1996) 437, astro- 389.
ph/9603033. [49] R.H. Cyburt, B.D. Fields, K.A. Olive, astro-ph/0302431.
[17] A.H. Guth, Phys. Rev. D 23 (1981) 347. [50] A. Aguilar, et al., LSND Collaboration, Phys. Rev. D 64 (2001)
[18] W.L. Freedman, et al., Astrophys. J. 553 (2001) 47, astro- 112007, hep-ex/0104049.
ph/0012376. [51] P. Langacker, UPR-0401T.
[19] L.M. Krauss, B. Chaboyer, Science 299 (2003) 65. [52] K. Hagiwara, et al., Particle Data Group Collaboration, Phys.
[20] G. Hinshaw, et al., astro-ph/0302217. Rev. D 66 (2002) 010001.
[21] A. Kogut, et al., astro-ph/0302213. [53] X. Chen, R.J. Scherrer, G. Steigman, Phys. Rev. D 63 (2001)
[22] L. Verde, et al., astro-ph/0302218. 123504.
[23] P. Crotty, J. Lesgourgues, S. Pastor, astro-ph/0302337;