The Astrophysical Journal Supplement Series, 189:240–252, 2010 July doi:10.1088/0067-0049/189/1/240
C
2010. The American Astronomical Society. All rights reserved. Printed in the U.S.A.
THE JINA REACLIB DATABASE: ITS RECENT UPDATES AND IMPACT ON TYPE-I X-RAY BURSTS
Richard H. Cyburt
1,2,9
, A. Matthew Amthor
1,2,3
, Ryan Ferguson
1,2,3
, Zach Meisel
1,2,3
, Karl Smith
1,2,3
,
Scott Warren
1,2,4
, Alexander Heger
1,5
, R. D. Hoffman
6
, Thomas Rauscher
7
, Alexander Sakharuk
1,2
,
Hendrik Schatz
1,2,3
, F. K. Thielemann
7
, and Michael Wiescher
1,8
1
Joint Institute for Nuclear Astrophysics (JINA), http://www.jinaweb.org
2
National Superconducting Cyclotron Laboratory, Michigan State University, East Lansing, MI 48824, USA
3
Department of Physics and Astronomy, Michigan State University, East Lansing, MI 48824, USA
4
Department of Electrical and Computer Engineering, Michigan State University, East Lansing, MI 48824, USA
5
School of Physics & Astronomy, University of Minnesota, Twin Cities, Minneapolis, MN 55455, USA
6
Lawrence Livermore National Laboratory, P.O. Box 808, L-414 Livermore, CA 94550, USA
7
Department of Physics, University of Basel, 4056 Basel, Switzerland
8
Department of Physics, University of Notre Dame, Notre Dame, IN 46556, USA
Received 2009 November 23; accepted 2010 May 12; published 2010 June 30
ABSTRACT
We present results from the JINA REACLIB project, an ongoing effort to maintain a current and accurate library
of thermonuclear reaction rates for astrophysical applications. Ongoing updates are transparently documented
and version tracked, and any set of rates is publicly available and can be downloaded via a Web interface at
http://groups.nscl.msu.edu/jina/reaclib/db/. We discuss here our library V1.0, a snapshot of recommended rates for
stable and explosive hydrogen and helium burning. We show that the updated reaction rates lead to modest but sig-
nificant changes in full network, one-dimensional X-ray burst model calculations, compared with calculations with
previously used reaction rate sets. The late time behavior of X-ray burst light curves shows significant changes,
suggesting that the previously found small discrepancies between model calculations and observations may be
solved with a better understanding of the nuclear input. Our X-ray burst model calculations are intended to serve
as a benchmark for future model comparisons and sensitivity studies, as the complete underlying nuclear physics
is fully documented and publicly available.
Key words: nuclear reactions, nucleosynthesis, abundances – X-rays: bursts
Online-only material: color figures
1. INTRODUCTION
Nuclear astrophysics addresses questions related to the ori-
gin and evolution of the chemical elements, as well as astro-
physical events powered by nuclear processes. Pioneering ef-
forts identified and disentangled many of the nuclear processes
needed to explain observations (Burbidge et al. 1957; Cameron
1957; Wagoner 1969;Howardetal.1971; Audouze et al. 1973)
and many more have been discovered since (see Clayton 1968;
Wallace & Woosley 1981; Rolfs & Rodney 1988; Woosley et al.
1990; Woosley & Hoffman 1992; Pagel 1997; Wallerstein et al.
1997; Schatz et al. 1998; Iliadis 2007;Boyd2008;Fr
¨
ohlich
et al. 2006; Pruet et al. 2006 for recent reviews). Key to this
exploration is reliable and up-to-date nuclear physics input. Of
foremost importance in most astrophysical scenarios are ther-
monuclear reaction rates. These needs drove many efforts into
the systematization and formulation of reaction rate compila-
tions which have played a central role in the field early on
(Fowler et al. 1967, 1975; Harris et al. 1983; Caughlan et al.
1985; Caughlan & Fowler 1988). Here we focus on charged
particle reactions of relevance in hydrogen and helium burning
scenarios in stars, core collapse supernovae, novae, and X-ray
bursts. Caughlan & Fowler (1988) were the last of a series of
widely used compilations summarizing mostly charged particle
reaction rates on stable targets taking into account experimental
and theoretical nuclear physics information. An updated com-
pilation of similar scope was presented by the NACRE collab-
9
Corresponding author: cyburt@nscl.msu.edu
oration (Angulo et al. 1999) focusing exclusively on charged
particle-induced reaction rates on mainly stable targets in the
A = 1–28 mass range. Triggered by the advances of radioactive
beam experiments over the last few decades, efforts to make
reaction libraries complete for explosive hydrogen burning sce-
narios resulted in compilations relevant for novae (Wiescher
et al. 1986) and the rp-process in X-ray bursts (van Wormer
et al. 1994; Schatz et al. 1998). Iliadis et al. (2001) more re-
cently presented a compilation of proton capture reaction rates
in the A = 20–40 mass range that included all relevant reactions
on neutron deficient radioactive targets, including theoretical
reaction rates. Neutron capture reactions have also been exten-
sively compiled (Allen et al. 1971;Bao&K
¨
appeler 1987; Beer
et al. 1992; Bao et al. 2000). The KADoNiS project (Dillmann
et al. 2006) has combined these neutron capture rate evaluations,
supplemented with more recent experiments and provided easy
Web-access to their database
10
. These compilations are comple-
mented by large data sets of theoretical rate calculations based
on shell model (Herndl & Brown 1997; Fisker et al. 2001)or
Hauser–Feshbach models (Hauser & Feshbach 1952;Holmes
et al. 1976; Woosley et al. 1978; Woosley & Hoffman 1992;
Rauscher & Thielemann 1998, 2000; Rauscher 2008a; Goriely
1998; Arnould & Goriely 2003).
However, compilations only cover a small subset of the types
of rates and mass ranges needed in modern nuclear reaction
network calculations. This has led astrophysical modelers to
compile their own complete set of rates making it difficult to
10
http://www.kadonis.org
240
No. 1, 2010 JINA REACLIB DATABASE, ITS RECENT UPDATES AND IMPACT 241
compare model calculations by different groups and to identify
reaction rates that have been used in specific calculations.
Another problem is that compilations are typically frozen at
some cutoff date prior to the time of publication, often as
a one-time project or with very long publication intervals.
Because of this, new experimental or theoretical results are
often not taken into account in astrophysical models. To address
these problems, we present here a new public database for
thermonuclear reaction rates maintained by the Joint Institute for
Nuclear Astrophysics, the JINA REACLIB database. It is based
on an updated version of Thielemann’s REACLIB reaction rate
library that has been used by various groups over the last decades
(Thielemann et al. 1987; Wiescher et al. 1986; van Wormer
et al. 1994; Schatz et al. 1998). It represents a reaction rate
compilation that is continuously updated yet provides well-
defined snapshots at regular intervals to allow comparison of
model calculations by different groups. The main criterion for
updates is to provide the best possible choice of reaction rates
based on what is available in the literature at any given time.
Data on reaction rates that require a thorough evaluation of
previous work are only included once such an evaluation has
been published. Reaction rates are presented in an analytic
form, and fit within 5% of literature values unless otherwise
noted. Version tracking allows users to document and reference
a specific reaction rate set used in a calculation, which can then
be looked up in the database. This is done through a Web-
interface system, where users can access the database
11
.
Our database is complementary to the BRUSLIB database and
NETGEN reaction network generator Web interface
12
, which is
also an effort to maintain complete reaction libraries (Aikawa
et al. 2005). BRUSLIB contains experimental-based rates from
the NACRE collaboration (Angulo et al. 1999), Iliadis et al.
(2001) and Bao et al. (2000), and has been further supplemented
with other experimental- and theory-based thermonuclear and
weak reaction rates last updated in 2005 November (See Aikawa
et al. 2005 for details). Instead of the parameterized REACLIB
format, BRUSLIB presents data in tabular form and includes
estimates for rate uncertainties from the NACRE collaboration.
The REACLIB release presented here focuses on reaction
rates needed to model hydrogen and helium burning environ-
ments. A particular goal of this release was to improve models
of Type-I X-ray bursts (Schatz & Rehm 2006; Strohmayer &
Bildsten 2006). After their discovery (Grindlay 1976; Evans
et al. 1976), X-ray bursts were soon explained as resulting from
unstable hydrogen and helium burning in material on the sur-
face layers of neutron stars accreted from a companion star
(Hansen & van Horn 1975; Woosley & Taam 1976;Joss1977).
X-ray bursts release about 10
39
–10
40
erg in 10–100 s and ex-
hibit recurrence times of hours to days. With over 70 known
sources they are the most frequent thermonuclear explosions
observed in the Galaxy. Current X-ray observatories have ac-
cumulated a vast body of detailed observational data. These
have revealed and reinforced many puzzles and open ques-
tions such as bursts with multiple peaks (Hoffman et al. 1980;
Sztajno et al. 1985; van Paradijs et al. 1986; Watts & Maurer
2007), the unexplained burst behavior at high accretion rates
(Kuulkers et al. 2002; Cornelisse et al. 2003), or the origin of
12
C in the burst ashes thought to be required to power the rarely
observed superbursts (Cumming & Bildsten 2001; Strohmayer
&Brown2002). Improved nuclear data are needed to clarify
11
http://groups.nscl.msu.edu/jina/reaclib/db/
12
http://www-astro.ulb.ac.be/Netgen
whether these issues reflect problems in the nuclear physics in-
put or require advances in astrophysical modeling. Improved
nuclear physics is also needed to extract system parameters
such as accreted hydrogen content, accretion rate, or neutron
star properties from detailed comparisons of model bursts with
observations, and to predict possible spectral signatures from
ejected ashes that could be targets of current and future X-ray
observatories (Weinberg et al. 2006). Despite the problems of
current burst models to explain certain observational features,
in some cases overall good agreement between burst calcu-
lations and observations has been found (Heger et al. 2007),
though some discrepancies remain. However, sensitivity studies
have demonstrated that burst light curves do vary significantly
within nuclear physics uncertainties (Woosley et al. 2004;Parikh
et al. 2008), leaving the possibility that such agreement is for-
tuitous with deficiencies in the nuclear physics compensating
for deficiencies in the astrophysical models or the chosen model
parameters.
The principal nuclear reaction sequences in X-ray bursts have
recently been delineated in detail by Fisker et al. (2008). These
are characterized by ignition driven by the 3α reaction and rapid
breakout from the CNO cycles, followed by helium burning via
the αp-process and hydrogen burning via the rp-process ending
under the most favorable conditions (high hydrogen contents in
the accreted matter, low metallicity, and high accretion rate) in
a SnSbTe cycle (Schatz et al. 2001).
Our work presented here includes an update of the relevant
reaction rates of 3α,(α,p), (α,γ ), and (p,γ ) reactions from H to
Te using newly published reaction rates based on experimental
results. We also present new rules for fitting reaction rates that
avoid problems that were present in REACLIB in the past,
such as charged particle reaction rates that become non-physical
at low temperatures. Weak interaction decay rates that do not
depend on density are also updated and included. In addition, we
present a new set of theoretical reaction rates calculated with
the code NON-SMOKER
WEB
v5.0w and use updated nuclear
masses that take into account new experimental information
from mass measurements and nuclear lifetime constraints. In
Section 2, we discuss how reaction rates are updated and verified
in the REACLIB database. In Section 3, we discuss the new
content of the database. We then use in Section 4 astateof
the art multi-zone X-ray burst model (Woosley et al. 2004)to
calculate a sequence of X-ray bursts with the updated reaction
library. The importance of these calculations is two fold. First,
using updated reaction rates leads to more reliable calculations
that either validate or falsify conclusions based on earlier model
calculations. Second, to our knowledge the calculation presented
here is the first full one-dimensional X-ray burst simulation with
fully documented and publicly available nuclear physics input.
It is intended to serve as a benchmark and starting point to
compare different burst models from various groups, and to
determine the impact of future improvements in the nuclear
physics. We conclude our results and discuss future prospects
for this research in Section 5.
2. THE JINA REACLIB DATABASE
The JINA REACLIB database is completely public and acces-
sible to the community via the World Wide Web. The interface
11
is PHP-driven
13
and connected to a MySQL database
14
. The cur-
rent version of the database stores reaction rates as a function
of temperature in the seven-parameter rate parameterization of
13
http://www.php.net
14
http://www.mysql.com
242 CYBURT ET AL. Vol. 189
Thielemann et al. (1987) and F.-K. Thielemann (1995, private
communication):
λ = exp
a
0
+
5
i=1
a
i
T
2i−5
3
9
+ a
6
ln T
9
. (1)
These rates go into a set of stiff coupled differential equations,
and are then evolved to solve the abundance changes of the
nuclides in the network. For a single reaction channel (A +B →
C + D), the equations take the form
−
1
N
A
∂
t
Y
A
=−
1
N
B
∂
t
Y
B
=
1
N
C
∂
t
Y
C
=
1
N
D
∂
t
Y
D
=
Y
N
A
A
N
A
!
Y
N
B
B
N
B
!
ρ
ν
baryon
λ, (2)
where Y
i
are the molar abundances per gram and N
i
is the
number of nuclides of type i produced or destroyed in the
reaction and ν =
N
A
+ N
B
− 1. By definition, the reaction
rate or “rate of reaction” is given by the entire right-hand side of
Equation (2), but the term reaction rate is used synonymously
for the “reduced” reaction rate or reactivity, λ, throughout this
paper and in the REACLIB database. For a network of reactions,
each ∂
t
Y
i
|
A+B→C+D
is summed over all participating reactions,
including their reverse rates. For unary rates, λ = 1/τ has
units of s
−1
, inversely proportional to the mean lifetime. For
binary rates, λ = N
A
<σ v> has units of cm
3
s
−1
mol
−1
, while
for trinary rates, λ has units of cm
6
s
−1
mol
−2
. Multiple sets
of parameters can be added to fit more complex temperature
dependencies.
Reaction rates are continuously updated to ensure that the
latest progress in nuclear physics is available to address as-
trophysical problems. Rather than delete old rates as they are
supplanted by newer evaluations, we keep them in the database
under different version numbers. While only one version is rec-
ommended, this gives users a choice. A rate detail page allows
detailed comparison between different reaction rate versions in
tabular and graphical forms.
In some cases it might be desirable to carry out an astro-
physical model calculation with a well-known set of rates that,
for example, is being used by other groups. This allows one to
compare results from different models in a meaningful way. To
address this need, we release on a regular basis snapshots of
the currently recommended rates. The reaction rates discussed
in this paper are such a snapshot called REACLIB V1.0. Users
can also create their own snapshots and store them in the sys-
tem. Snapshots can be referenced in publications, and can be
accessed through the Web interface so that readers can look up
reaction rates used in a particular study, or can download the
same set of reaction rates for their own calculations.
Reaction rate updates are considered as rates appear in the
literature or are suggested by users. This process is documented
on the database Web site, and a complete history of updates
is available
15
. In addition, documents created in the process of
evaluating a reaction rate are stored in the database as well. Pos-
sible new rate entries are found from several sources. (1) Rates
can be recommended by the community as new experimental/
theoretical work is completed (published). This can be done via
our Web interface or by direct communication. (2) Papers with
relevant reaction rate information in the JINA Virtual Journal
15
http://groups.nscl.msu.edu/jina/reaclib/db/status.php
of Nuclear Astrophysics
16
(VJ), a weekly compilation of new
publications in nuclear astrophysics, are flagged by the editor,
and information is transferred through Web-based tools into the
update process for the REACLIB database. (3) Rates can also
be submitted and evaluated at the http://www.nucastrodata.org
repository, in development at ORNL (Smith et al. 2008).
The main motivation for updating our database is to provide
the best reaction rates available in the literature at any given
time. Relevant reaction rate information is collected on a
regular basis for each reaction rate, compared with previously
published information, and then subjected to an initial screening
process. The possible outcomes of this screening process are a
recommendation to either (1) enter the published reaction rate
directly into the database (“Implement As Published”), (2) store
the information for a future detailed evaluation (“Evaluation
Needed”), (3) ignore the information (“No Action Needed”).
Immediate implementation is typically recommended when the
published work is a thorough evaluation taking into account
all previous work, or if it represents an obvious improvement
compared with the previously available reaction rate. Examples
for an obvious improvement include the replacement of theory
with credible experimental data or a dramatically improved
experiment. In most cases, these are reaction rates where very
little previous experimental data are available. Storing the
information for future evaluation is typically recommended for
reaction rates where a lot of information is available that has
to be taken into account in a consistent way and where the
published work does not provide such a complete evaluation.
Other cases include a conflict with previous work without
a clear explanation or obvious improvement, or incomplete
information that requires a major effort to extract a reaction rate.
Once a published complete evaluation becomes available, the
result is again considered for implementation into the database.
Leaving a reaction rate out is recommended for cases where the
information turns out to be not relevant for the astrophysical
reaction rate, or if it results in no significant difference to
previous work. The decisions are documented on the database
Web site and can be discussed by the community through
discussion threads.
New rates are fit to the standard seven-parameter REACLIB
form given in Equation (1). This format is capable of handling
all reaction types. Multiple sets of Equation (1) with differing
parameters, a
0
through a
6
, can be summed in order to properly
fit rates with numerous resonant and non-resonant contributions.
Non-physical behavior of the reaction rates outside of the fitted
temperature range is avoided by enforcing physical constraints
on the parameters (Wagoner 1969; Woosley et al. 1978). The
enforcement of physical constraints on the fit parameters is
an improvement brought to the REACLIB database with the
V1.0 update. Unique rules exist for assigning values to the
rate parameters a
0
–a
6
in the cases of charge-induced non-
resonant, neutron-induced non-resonant, and narrow resonant
rate contributions. In practice, these rules should serve more
as guidelines so that actual parameter values used are within
proximity to those theoretically assigned. For positive Q-value
reactions these are summarized in Table 1.
In Table 1 N
A
is Avogadro’s number, σ is the cross section
in cm
2
, v is the center of mass relative speed in cm s
−1
, E
is the center of mass relative energy in MeV, Z
1
and Z
2
are
the target and reactant charges, A is the reduced mass of the
reactants in atomic mass units, S(0) is the astrophysical S-factor
16
http://groups.nscl.msu.edu/jina/journals/jinavj/
No. 1, 2010 JINA REACLIB DATABASE, ITS RECENT UPDATES AND IMPACT 243
Tab le 1
Shown are the Fitting Rules for Various Types of Reaction Rates, Including Non-resonant (NR) Charge-induced (q-induced) and n-induced Reactions, as well as
Narrow Resonant Reactions
q-induced NR n-induced NR Narrow Resonant
a
0
= ln[B(
Z
1
Z
2
A
)
1
3
S(0)] a
0
= ln(N
A
C
Γ(+3/2)
Γ(3/2)
(
σv
E
)
E=0
) a
0
= ln(DA
−3/2
ωγ )
a
1
= 0 a
1
= 0 a
1
=−11.6045E
r
a
2
=−4.2486(Z
2
1
Z
2
2
A)
1
3
a
2
= 0 a
2
= 0
a
3
= float a
3
= float a
3
= 0
a
4
= float a
4
= float a
4
= 0
a
5
= float a
5
= float a
5
= 0
a
6
=−2/3 a
6
= a
6
=−3/2
Notes. These rules enforce the proper low-temperature analytics. All parameters with numerical values can be fixed, and
those specified as “float” can be varied to accommodate the rate changes at higher temperature.
in MeV-barn at zero energy, is the minimum orbital angular
momentum value in units of ¯h, Γ(z) is the Gamma function,
ωγ is the narrow resonance strength in MeV, E
r
is the narrow
resonance energy in MeV, B = 7.8318 × 10
9
cm
3
s
−1
mole
−1
MeV
−1
barn
−1
, C = 0.08617 MeV, and D = 1.5394 × 10
11
cm
3
s
−1
mole
−1
MeV
−1
. Reaction rates from shell model or
statistical model calculations with high level densities should
follow the non-resonant prescriptions. If a rate is comprised
of both non-resonant and resonant pieces, multiple sets of the
seven-parameter fits may be used to describe the rate. It may also
be necessary to fit multiple sets to obtain the requisite precision.
Proper use of REACLIB form and fitting procedure should yield
a reaction rate fit that is within 5% of the data. 5% fit accuracy
is adhered to in the database and is considered acceptable by the
authors who note that experimental error is rarely better than
10%–15% and theoretical errors are often more than 30%. It
may occur that a fit to 5% precision is not possible. Such cases
are listed on an automatically generated deviations list and an
effort is made to constantly improve such cases. Future updates
may include adding two more terms in each exponential set
(i.e., T
7
3
9
and T
3
9
terms) to improve fitting performance and fit
precision.
Once a rate is fitted it will be entered into the database as a
“Future” rate, which is not visible as part of the database. The
rate is then independently verified to ensure quality control, and
actual rate values are entered into a verification database that is
run automatically on a regular basis displaying any unacceptable
discrepancies (more than 5%) on the Web site. This documents
cases of inaccurate fits and ensures the integrity of the database
over time. Once this process is completed, the rate will become
available in the database. In some cases the new rate might
not become the recommended rate version, but will be made
available as a choice.
3. NEW REACLIB CONTENT
Our library has been updated using available information
from experiments (e.g., cross sections, resonance strengths,
etc.) as well as theoretical rate predictions. The bulk of the
rates are from new statistical model calculations with a recently
updated version of NON-SMOKER
WEB
(see Section 3.3)using
updated nuclear masses (see Section 3.2). In the sd- and fp-
shell, shell model based reaction rates are used where available
(see Section 3.3). These theoretical rates are replaced with
experimentally based reaction rates for the relatively small
number of cases where these are available (see Section 3.1). In
addition, several reaction rates that were taken over from older
REACLIB versions were refitted to avoid unphysical behavior.
3.1. Experiment-based Rates
Many of the experiment-based reaction rates were taken from
the compilations of the NACRE collaboration (Angulo et al.
1999) and from Iliadis et al. (2001) for stable and unstable nu-
clei respectively. In some cases other reaction rates have been
chosen, mostly because more recent experimental information
became available. These cases are discussed in the following.
Stellar enhancement factors that take into account the population
of excited states in the target nucleus when immersed in the as-
trophysical plasma, are taken from the relevant NON-SMOKER
statistical model calculations (Rauscher & Thielemann 1998,
2000; Rauscher 2008a) except for
32
Cl(p,γ ) where a stellar en-
hancement factor is given in the most recent evaluation of the
rate by Schatz et al. (2005).
4
He(αα, γ )
12
C is a key reaction in several sites of nucleosyn-
thesis. It is the reaction that triggers the thermonuclear runaway
in X-ray bursts. In addition to the dominant contribution of the
Hoyle state in
12
C, the NACRE compilation (Angulo et al. 1999)
includes an extra contribution from a theoretically predicted 2
+
resonance at 9.1 MeV. New experimental data of the inverse
process (
12
C
∗
→ 3α) have been obtained by Fynbo et al. (2005)
providing new information on additional resonances beyond the
Hoyle state. They find a number of interfering broad resonances,
which they include in their compiled reaction rate, but conclude
that the presence of a state at 9.1 MeV is unlikely based on their
data. We therefore recommend the reaction rate by Fynbo et al.
(2005). The differences between the Fynbo et al. (2005) and the
NACRE rate (Angulo et al. 1999) are negligible for X-ray burst
temperatures. Recently, after the cutoff date for this compila-
tion, some evidence for a 2+ state at 9.6 MeV was found (Freer
et al. 2009; Diget et al. 2009). If correct it would affect the rate
at very high temperatures, though its impact will be lessened
compared with the prediction of NACRE because of the higher
energy of the state.
12
C(α, γ )
16
O is a difficult reaction to measure at energies rel-
evant for astrophysics. The reaction is important in hydrostatic
and explosive helium burning regimes. Buchmann (1996) and
Kunz et al. (2002)usedanR-matrix formalism that combines
information about
16
O structure,
12
C-α scattering, as well as
direct capture measurements available at the time to derive the
low energy behavior of this cross section. Since then, signifi-
cant experimental progress has been made (Kunz et al. 2001;
Sch
¨
urmann et al. 2005a, 2005b; Tang et al. 2007, 2008). Kunz
et al. (2001) determine the angular distributions of γ rays from
the direct reaction at 20 energies from 0.95 to 2.8 MeV. These
data can be used to determine the E1 and E2 components of
the S-factor separately. Tang et al. (2007, 2008) measure the
244 CYBURT ET AL. Vol. 189
β-delayed α decay of
16
N, extracting constraints on the E1 com-
ponent of the S-factor at 300 keV. New data from Sch
¨
urmann
et al. (2005a, 2005b) measure the total
12
C(α, γ )
16
O cross sec-
tion at energies between 1.9 and 4.9 MeV in inverse kinematics
via use of the ERNA recoil separator. However, as discussed
for example in Buchmann (2008), the total S-factor obtained
in more recent evaluations agrees well with the value obtained
by Buchmann (1996). We therefore continue to recommend the
Buchmann (1996) rate.
13
N(p, γ )
14
O is an important reaction in the hot CNO cycle.
It has undergone several experimental updates since the NACRE
collaboration’s recommendation (Angulo et al. 1999). Tang et al.
(2004) use the peripheral reaction
14
N(
13
N,
14
O)
13
C to extract
an asymptotic normalization constant (ANC) for
14
O →
13
N+p.
This ANC is then used to calculate the direct capture component
of this cross section. The rate is dominated by a low energy
resonance at E
R
= 528 keV. Tang et al. (2004) infer that this
state interferes with the direct capture component enhancing the
low energy cross section. More recently Li et al. (2006)have
re-examined this reaction by similarly measuring the ANC, but
through the reaction
13
N(p, γ )
14
O. They confirm the results
from Tang et al. (2004). We therefore adopt the most recent Li
et al. (2006) results as our recommended value.
The
14
N(p, γ )
15
O reaction is the slowest reaction in the low
temperature CNO cycle. Recently, a number of new direct
measurements have been performed (Imbriani et al. 2005;
Runkle et al. 2005) that have extended the measured cross
sections to significantly lower energies. Both groups evaluate
the available data and obtain comparable S-factors, so we chose
to implement Imbriani et al. (2005). The new reaction rate
is significantly lower at low temperatures compared with the
rate given in NACRE (Angulo et al. 1999) and Caughlan &
Fowler (1988), but agrees well with previous compilations
for 0.2 <T
9
< 2. More recently, the LUNA Collaboration
et al. (2006) measured the total cross section down to 70 keV.
In addition, Marta et al. (2008) have explored the critical
issue of the interference between the 259 keV resonance and
direct capture component and provide a stringent limit for
the extrapolation to lower energies. However, a comprehensive
evaluation that would allow us to include the new data in our
database was not available at the cutoff date for this work.
14
N(α, γ )
18
F is important in early phases of He burning,
taking place before the triple-α reaction. It is also the main
source of
22
Ne, via another α capture, which is a neutron source
for the s-process (
22
Ne(α, n)
25
Mg). G
¨
orres et al. (2000)have
recently measured the lowest lying resonance properties and
the direct capture into the ground state. This yields changes in the
adopted rate of factors of 2–5 compared with Caughlan & Fowler
(1988) and NACRE (Angulo et al. 1999) in the astrophysical
temperature range of interest 0.1 <T
9
< 0.5.
15
N(α, γ )
19
F is a possible source of
19
F in Asymptotic Giant
Branch stars. Recent experimental efforts by Wilmes et al.
(2002) have measured the resonance properties of several states
in
19
F. Besides observing two levels for the first time, they were
also able to identify two levels as α-cluster states. The resulting
thermonuclear reaction rate is identical to NACRE (Angulo et al.
1999), though with reduced uncertainties. Wilmes et al. (2002)
is recommended in the REACLIB database.
The reaction
14
O(α, p)
17
F is important in the hot CNO cycle.
It bypasses
14
O β-decay for T
9
> 0.35. Hahn et al. (1996)
evaluated this reaction rate, tabulating two rates differing only
in the sign of the assumed interference between the non-
resonant and E = 6.25 MeV resonance. We have adopted
the constructive interference rate as our recommended value.
The two rates agree at temperatures beyond T
9
∼ 0.5, but are
an order of magnitude different at lower temperatures. Recent
measurements suggest that the rate adopting the constructive
interference is accurate to within 50% (J. C. Blackmon 2009,
private communication).
15
O(α, γ )
19
Ne is an important hot-CNO breakout reaction for
novae and X-ray bursts, competing with
18
Ne(α, p)
21
Na. The
dominant uncertainty stems from the α-width of a resonance at
E
X
= 4033 keV. The α branching ratio B
α
is strongly Coulomb
suppressed, since it is only 500 keV above threshold. Previous
analyses have estimated its strength using iso-spin symmetry
(Mao et al. 1995). An experimental upper limit was placed by
Davids et al. (2003), B
α
< 4.3 × 10
−4
at 90% confidence.
Tan et al. (2007) find B
α
= (2.9 ± 2.1) × 10
−4
, which would
yield a 90% confidence upper limit of B
α
< 5.6 × 10
−4
if
errors were normally distributed. We adopt the thermonuclear
reaction rate calculated in Mao et al. (1995) and tabulated in
Hahn et al. (1996), which finds B
α
≈ 1.2 × 10
−4
. Rates from
Caughlan & Fowler (1988), Hahn et al. (1996), and Fisker et al.
(2007) are within ∼50% of each other at breakout temperatures
(T
9
∼ 0.4–0.6).
17
O(p, γ )
18
F and
17
O(p, α)
14
N are important in hydrogen
burning nucleosynthesis, and compete against each other in
the CNO cycle. Recently, both Fox et al. (2005) and Chafa
et al. (2007) measured resonance properties in the
18
Fsystem.
They observed previously unobserved states important for the
low energy nuclear cross section. Fox et al. (2005) estimated
the direct capture components using a potential model, opting
to ignore the low energy data, due to issues with resonance
subtraction. Chafa et al. (2007) perform this subtraction and
find agreement with the shape of the direct capture, though with
a higher value, to match the data. These differences are washed
out by the large uncertainties assigned to this component. For
the resonant contribution to the reaction rate, a new resonance
at E
R
= 183.3 keV plays an important role. For (p,α), the
resonance strengths given by Fox et al. (2005) and Chafa et al.
(2007) agree and have been confirmed in a new experiment by
Moazen et al. (2007). The resulting (p, α) reaction rate agrees
with NACRE (Angulo et al. 1999) within a factor of 2 from
0.5 <T
9
< 2. The largest deviation of a factor of 20 is at
T
9
= 0.2 because of the new resonance. For (p,γ ), the resonance
strengths obtained by Fox et al. (2005) and Chafa et al. (2007)
disagree however by almost a factor of 2 (more than one standard
deviation). The resulting (p, γ ) rates are within 40% of each
other for T
9
< 0.5 and within 10% from 0.5 <T
9
< 5. As no
detailed evaluation of this situation is available in the literature,
we for now adopt the (p,γ ) and (p,α) rates by Chafa et al. (2007).
17
F(p,γ )
18
Ne is an important reaction in the hot CNO cycle,
dominating over
17
F β-decay when T
9
> 0.093. The recom-
mended rate is that by Bardayan et al. (2000). The new theory
calculation by Dufour & Descouvemont (2004) agrees quite
well with Bardayan et al. (2000).
18
Ne(α, p)
21
Na is a hot CNO cycle breakout reaction, com-
peting with
15
O(α, γ )
19
Ne. We adopt a rate based on two
compilations: at low temperature we use G
¨
orres et al. (1995),
while at high temperature we use Bradfield-Smith et al. (1999).
New constraints come from experiments by Chen et al. (2001)
and Chae et al. (2009), populating states in
22
Mg via the
12
C(
16
O,
6
He)
22
Mg and
24
Mg(p,t)
22
Mg reactions, respectively.
The rates determined from Chen et al. (2001) and Chae et al.
(2009) agree quite well with each other and within a factor of
4 of the combined G
¨
orres et al. (1995) and Bradfield-Smith
