Professional Documents
Culture Documents
I. INTRODUCTION
The description of energy production in stars and of nucleosynthesis of the elements requires the calculation of thousands of thermonuclear reaction rates. To accompany the evolution of a star these rates must be calculated as a function of the stellar temperature and for each temperature the reactions of different species will be important as determined by folding the nuclear cross sections with the MaxwellBoltzmann distribution. Normally a combination of numerical integration and analytical techniques appropriate to the energy dependence of the cross section for each reaction is used [13]. While the importance of analytical techniques in numerical evaluations is diminished by increased computing power they will always be useful for understanding and checking numerical results (Ref. [4] reviews analytic techniques). Reference [1] calculated the contribution of narrow resonances in the simplest possible manner, which is to approximate the Maxwell-Boltzmann distribution by its value at the resonance energy. This approximation can be expected to be good for resonances which do not overlap signicantly with the Gamow peak. Here we derive an improved formula for the inclusion of resonances in charged particle reactions which removes this restriction. In Ref. [5] we developed an asymptotic expansion for the thermonuclear reaction rate in terms of the effective astrophysical S factor Seff using the method of Dingle [6]. It may used in case cases where SE can be reliably expanded as a Taylor series. Two alternative expressions for Seff were obtained by expanding SE about E = 0 and about E = E0, E0 being the effective energy of the Gamow window. From these expressions, all the approximate formulas for Seff com-
monly used [7] when SE is slowly varying may be obtained as special cases. The validity of the expressions derived in Ref. [5] is limited by the radii of convergence of the Taylor series expansions of SE, that is, by the location of the poles of SE. The poles may be due bound states of the composite nucleus. Illustrative is the case of 7Be+ p 8B + [8] whose SE has a pole at E = EB, EB = 137.5 keV being the binding energy of 8 B. (Ref. [9] discusses effects other than the sub-threshold pole which can produce the low energy rise seen in the S-factors of several capture reactions.) There may also be poles in SE due to resonances of the reaction pair in which case the cross section may be parametrized in the region of the resonance by the Breit-Wigner form (see Eq. (7) below) which has two poles, or the Lorentzian form [10] which has four poles. In this paper we modify the method of Ref. [5] to obtain an asymptotic expansion of the thermonuclear reaction rate for charged particles for the case that the cross section contains a single narrow or wide resonance described by a Breit-Wigner shape. Our result is uniformly valid as the resonance and Gamow peak coalesce or separate and as the resonance width is varied from narrow to broad. Basic denitions follow in Sec. II, the limitations of conventional approximations to the reaction rate are discussed in Sec. III and the uniform approximation is derived and discussed in Sec. IV. In Sec. V we apply our formula to the reaction 12C p , 13N which is important in the CNO cycle and some conclusions are drawn in Sec. VI.
II. BASIC DEFINITIONS
For a given reaction cross section E, the Maxwellian averaged reaction rate per particle pair is given by [11] v =
Electronic address: adams@fma.if.usp.br Electronic address: mpato@if.usp.br Electronic address: hussein@fma.if.usp.br 0556-2813/2004/70(2)/025802(6)/$22.50
8
1/2
1 k BT
3/2
dEEEeE/kBT ,
where = M 1M 2 / M 1 + M 2 is the reduced mass of the two collision partners whose masses are M 1 and M 2, kB is the
70 025802-1 2004 The American Physical Society
Boltzmann constant, and T the environmental temperature. Expressing the cross-section, E, in terms of the astrophysical S-factor, SE, SE E /E e G , E = E the reaction rate becomes v = where FE = E/kBT + EG/E , where E G = b 2 c Z 1Z 2 ,
2 2 2
v = where
8
1/2
1 k BT
5/2
Sr
2 r I x 0, x p , 4
13
I x 0, x p =
dx
e f x 2 x x r 2 + r /4
14
8
1/2
1 k BT
3/2
dESEeFE ,
2 J J x 0, x p , r
15
J x 0, x p =
dx
e f x , x xp
16 17 18 19
being the ne structure constant. The function FE has a single minimum in the interval 0 , from which it increases monotonically to innity as E approaches zero and innity. The position of the minimum is given by
1/2 kBT/22/3 , E0 = EG
and is the effective energy of the Gamow window for thermonuclear reactions at temperature T. We assume that for low enough energy, SE for the radiative capture of a charged particle through a single isolated resonance may be approximated by SE = Sr E
2 r /4 2 E r + 2 , r /4
We have given I and J the arguments x0 and x p, the minimum of f x and the position of the pole due to the resonance, respectively. These are the critical points [6,12] of the integrand. The principal contributions to an integral such as we are considering come from the neighborhood of the critical points.
III. CONVENTIONAL APPROXIMATIONS
where Er is the energy of the resonance, r = Er is its width at Er and Sr = SEr. In terms of the resonance strength r = ErrEr / 22, we have 2 2 r E / E Sr = e G r. The reaction rate becomes v = 8
When the resonance and the Gamow window are well separated the contribution from each to the reaction rate can be estimated separately and the two contributions summed. Let us write I Ir + IG , 20
8
1/2
1 k BT
3/2 2 S r r /4
where Ir is the contribution from the resonance and IG the contribution from the Gamow window. The resonance contribution may be estimated [11] by approximating the Gamow window exponential in Eq. (14) by its value at the resonance energy, xr: I r = e f xr
dE
e FE 2 . E E r 2 + r /4 9
dx e xr xG/xr = /2 + r , 2 r/2 x x r 2 + r /4
21 with
r = tan12xr/r .
10
22
Similarly, the Gamow window contribution is estimated by approximating the Breit-Wigner factor by its value at the location of the Gamow peak, x0: IG = e f x0 2 x 0 x r 2 + r /4
f x0
11 12
e 2 x 0 x r 2 + r /4
dxe f x f x0 dxex x0
2/ /22 x
23
FIG. 1. A comparison of exact reaction rate with conventional approximations. The dashed lines show log10 I [Eq. (14)], the long dashed lines log10 Ir [Eq. (21)], the dotted lines log10 IG [Eq. (24)] vs resonance energy (shifted by the effective energy). See also Table I.
2 e 2 . 3 x 0 x r 2 + r /4
24
tx = 0 = , tx = = , tx = x p = t p, tx = x0 = 0. The position of the pole in t-space is then given by t p = f x p f x01/2 Er ir/2 E0 + = k BT Equation (15) becomes J x 0, x p = 29
In Eq. (24) a Gaussian approximation to the Gamow window exponential has been employed [11] and we have introduced x = 8/Fx0 = 4x0/3, 25
dt
EG Er ir/2
EG E0
1/2
. 30
= f x 0 = 3 x 0 .
26
Equation (24) is the zeroth-order term of an asymptotic expansion of IG in powers of 1 / [5]. For the reaction rate one thus obtains v 2/1/2 1 1/2SEreEr/kBT EG/Er kBT3/2 27
dx t t p et dt x x p t t p
2
31
+ S E 0 e ,
=e where
e t dtt , t tp
32
where = 4E0kBT / 31/2 is the width of the Gamow window. In Fig. 1 we plot Ir, IG and Ir + IG together with the exact I for comparison over a range of parameters. The left- and right-hand columns are representative of narrow resonances, r / 2 = 0.1, and wide resonances, r / 2 = 5.0, respectively. The three rows show the evolution with respect to a, which determines both the effective energy x0 and effective width of the Gamow window x. It is seen from Fig. 1 that when xr x0 the method of approximating I by Ir + IG fails for both narrow and wide resonances: for narrow resonances, IG alone overestimates I whilst for wide resonances Ir alone overestimates I.
IV. UNIFORM APPROXIMATION
t =
dx t t p . dt x x p
33
In order to obtain an expansion which is uniform as x p is made to approach or coincide with x0, t is expanded as follows [1214]: t =
n=0
n + t tpntt tpn .
34
The expansion given by Eq. (34) has the property that it gives the value of the function t and all its derivatives exactly at both t = 0 and t = t p. Let us approximate t by the truncation 0 t = 0 + t t p 0 , Then Jx0 , x p is approximated by 35
025802-3
FIG. 2. The dashed lines show I [Eq. (14)] and the solid lines the uniform approximation I0 [Eq. (40)] vs resonance energy (shifted by the effective energy). See also Table I.
J 0 x 0, x p = e
e t dt0t t tp
q I =
0
r/2
e J
2 , 3 x 0 x p 44
= 0e
e t dt + 0e t tp
dtet
36
= IG ,
= 0e
where IG is given by Eq. (24). Substituting Eqs. (40)(44) into Eq. (13) we nally obtain v = 2/1/2
2 1 e SErJ ietp erfc it p + kBT3/2 tp
e t dt + 0e . t tp 2
37
+ S E 0 , 38
0 =
1 . + t p 3 x 0 x p
45
where
dt
e t 2 = iez erfc iz , tz
39
we obtain an approximation to Ix0 , x p which corresponds to the truncation of the Taylor series we made in Eq. (35): I 0 = I 0 + I 0 , 40
where we have expressed our result using the astrophysical S-factor, SE = EEeEG/E and the width of the Gamow window = 4E0kBT / 31/2. In Fig. 2 we compare I with the uniform approximation I0. We see that I0 gives a good account of I for the whole of the parameter range investigated (Table I). In order to understand the contribution of the three terms on the r.h.s. of Eq. (45) let us consider Eq. (30) for t p. Writing x p = x0 + x where x = x p x0 we obtain t p = x p + xG/x p 3x01/2 = x0 + x + xG/x01 + x/x01/2 3x01/2 . Noting that xG / x0 = 2x0 and making the binomial expansion
TABLE I. Parameter values used in Figs. 13.
41
42
x0
where
p = I
0
1 e J , tp r/2
43
025802-4
a xG x
FIG. 3. Comparison of the three terms of the uniform approximation with the exact I (dashed line) [Eq. (14)] as a function of r = r / kBT. The long dashed lines show log10 I0 [Eq. (41)], the p solid lines log10I [Eq. (43)] and the dotted lines log10 IG [Eq. 0 (24)].
FIG. 4. The reaction rate for 12C p , 13N as function of T9. The circles show the rates from Ref. [1], the squares Ref. [17], the solid lines the uniform approximation, Eq. (45), and the dotted lines the resonance-plus-tail approximation, Eq. (27). We took the resonance parameters from Ref. [16] (Er = 457 KeV and r pEr = 39 KeV [both being laboratory frame values] and Er = 130 b).
1 + x/x01/2 =
j=0
1/2 j
j x j/x0
ez2 erfcz = 1
z ietp erfc it p +
2
1 x 3 x2 5 x3 =1 + + , 46 2 x0 8 x2 16 x3 0 0 we obtain t1 p = = 1
2x0 2x0/3
x
3 x2 5 x3 + 8 x2 16 x3 0 0
1/2 8 j x j/x0 3 j=0 j + 2 tp 2x0/3 x p x0 Er E0 ir/2 . /2
1/2
is O1 / x3. Thus I IG as x since IG O1 / x. This behavior is also conrmed by Fig. 3 for xr = x0. When xr x0 0 the conventional approximation [Eq. (20)] may be used. . 47
V. REACTION RATE FOR
12
1/2
Cp , 13N
49
We have calculated the reaction rate for 12C p , 13N which is important in the CNO cycle using Eqs. (45) and (27) including only the lowest resonance at 457 keV [16] and the results are displayed in Fig. 4. The exact integral is not shown as it is indistinguishable from the uniform approximation for this case. It is apparent that the resonanceplus-tail approximation cannot be used at temperatures for which Er E0. For comparison we have also shown the rates of Ref. [17] which used a different analytic approach and those of Ref. [1] which also included the contribution of the second resonance at 1698 keV.
VI. CONCLUSIONS
Substituting this expression for t p into Eq. (43) we see that q p cancels I so that I I0. This behavior is conrmed I 0 0 numerically by Fig. 3 for xr = x0. The result I Ir [Eq. (21)] may be recovered by making the crude approximation that erfcit p 1 for t p 0 and assuming that r Er. To analyze the limit x we write it p = z and employ the asymptotic series [15]
In this paper we have derived a new approximate analytic expression for the thermonuclear reaction rate when the cross section has a single isolated resonance. The new expression has a much wider range of validity than simpler approximations which are commonly used. In particular it is uniformly
025802-5
valid as the resonance energy coalesces with the effective energy of the Gamow window and as the width of the resonance is varied from narrow to broad. An interesting problem which still has to be addressed is the generalization to several resonances.
ACKNOWLEDGMENTS
0 =
0 t p 0 0 = tp
1 dx + t p dt
1 . t=0 x0 x p
A2
+ f x
d 2x = 2. dt2
The coefcients 0 and 0 are obtained from Eq. (35) together with Eq. (29). One nds
0 = 0 t p
= dx dt dx dt
t=t p
t tp x xp dt dx
t=t p
we nd
t=t p
t=t p
=1 and
A1
dx dt
= 2/ f x0 =
t=0
2 , 3
A3
[1] C. Angulo, M. Arnould, M. Rayet, P. Descouvemont, D. Baye, C. Leclercq-Willain, A. Coc, S. Barhoumi, P. Aguer, C. Rolfs et al., Nucl. Phys. A656, 3 (1999). [2] C. Iliadis, J. M. DAuria, S. Starreld, W. J. Thompson, and M. Wiescher, Astrophys. J., Suppl. Ser. 134, 151 (2001). [3] H. Herndl, R. Honger, J. Jank, H. Oberhummer, J. Grres, M. Wiescher, F.-K. Thielemann, and B. A. Brown, Phys. Rev. C 60, 064614 (1999). [4] A. M. Mathai and H. J. Haubold, Astrophys. Space Sci. 282, 265 (2002). [5] M. Ueda, A. J. Sargeant, M. P. Pato, and M. S. Hussein, Phys. Rev. C 61, 045801 (2000). [6] R. B. Dingle, Asymptotic Expansions: Their Derivation and Interpretation (Academic, New York, 1973). [7] E. G. Adelberger, S. M. Austin, J. N. Bahcall, A. B. Balantekin, G. Bogaert, L. S. Brown, L. Buchmann, F. E. Cecil, A. E. Champagne, L. de Braeckeleer et al., Rev. Mod. Phys. 70, 1265 (1998).
[8] B. K. Jennings, S. Karataglidis, and T. D. Shoppa, Phys. Rev. C 58, 3711 (1998). [9] A. M. Mukhamedzhanov and F. M. Nunes, Nucl. Phys. A708, 437 (2002). [10] S. Goriely, Phys. Lett. B 436, 10 (1998). [11] C. E. Rolfs and W. S. Rodney, Cauldrons in the Cosmos (Chicago Univ. Press, Chicago, 1988). [12] N. Bleistein and R. A. Handelsman, Asymptotic Expansions of Integrals (Holt, Rinehart and Winston, New York, 1986) (reprinted in 1986 by Dover, New York). [13] C. Chester, B. Friedman, and F. Ursell, Proc. Cambridge Philos. Soc. 53, 599 (1957). [14] M. V. Berry, Proc. Phys. Soc. London 89, 479 (1966). [15] Handbook of Mathematical Functions, edited by M. Abramowitz and I. A. Stegun (Dover, New York, 1970). [16] C. Rolfs and R. E. Azuma, Nucl. Phys. A227, 291 (1974). [17] G. R. Caughlan and W. A. Fowler, At. Data Nucl. Data Tables 40, 283 (1988).
025802-6