Elastic Electron Scattering off Exotic Nuclei

(a)

arXiv:nucl-th/0607024v1 13 Jul 2006

I.

C.A. Bertulani(a)
Department of Physics, University of Arizona, Tucson, AZ 85721

INTRODUCTION

The use of radioactive nuclear beams produced by fragmentation in high-energy heavyion reactions lead to the discovery of halo nuclei, such as 11 Li, about 20 years ago [1].
Nowadays a huge number of -unstable nuclei far from stability are being studied thanks
to further technical improvements. Unstable nuclei far from stability are known to play an
important role in nucleosynthesis. Detailed studies of the structure and their reactions will
have unprecedented impact on astrophysics [2].
The first experiments with unstable nuclear beams aimed at determining nuclear sizes
by measuring the interaction cross section in high energy collisions [1]. Successive use of
this technique has yielded nuclear size data over a wide range of isotopes. Other techniques,
e.g. isotope-shift measurements, have allowed to extract the charge size. The growth of a
neutron skin with the neutron number in several isotopes have been deduced from nuclearand charge-size data [3].
Undoubtedly, electron scattering off nuclei would provide the most direct information
about charge distribution, which is closely related to the spatial distribution of protons
[4]. A technical proposal for an electron-heavy-ion collider has been incorporated in the
GSI/Germany physics program [5]. A similar program exists for the RIKEN/Japan facility
[6]. In both cases the main purpose is to study the structure of nuclei far from the stability
line. The advantages of using electrons in the investigation of the nuclear structure are
mainly related to the fact that the electron-nucleus interaction is relatively weak. For
this reason multiple scattering effects are usually neglected and the scattering process is
described in terms of perturbation theory. Since the reaction mechanism in perturbation
theory is well under control the connection between the cross section and quantities such as
charge distributions, transition densities, response functions etc., is well understood [7].
Under the impulse approximation, or plane wave Born approximation, the charge form
factor can be determined from the differential cross section of elastic electron scattering.
Since the charge distribution, ch (r), is obtained from the charge form factor by a Fourier
transformation, one can experimentally determine ch (r) by differential cross-section measurements covering a wide range of momentum transfer q. One can obtain information on
the size and diffuseness when the charge form factor is measured at least up to the first
maximum. To do this within a reasonable measuring time of one week, a luminosity larger
than 1026 cm2 s1 is required, for example for the 132 Sn isotope [5].
On the theoretical side the difference between the proton and neutron distributions can
be obtained in the framework of Hartree-Fock (HF) method (see for example [8]) or HartreeFock-Bogoliubov (HFB) method (see for example [9, 10]). As a rule of thumb, a theoretical
calculation of the nuclear density is considered good when it reproduces the data on elastic
electron scattering. But some details of the theoretical densities might not be accessible
in the experiments, due to poor resolution or limited experimental reach of the momentum
transfer q.

II.

ELASTIC ELECTRON SCATTERING

In the plane wave Born approximation (PWBA) the relation between the charge density
and the cross section is given by
 
d
M
|Fch (q) |2 ,
(1)
=
d PWBA 1 + (2E/MA ) sin2 (/2)
where M = (Z 2 2 /4E 2 ) cos2 (/2) sin4 (/2) is the Mott cross section, the term in the
denominator is a recoil correction, E is the electron total energy, MA is the mass of the
nucleus and is the scattering angle.
The charge form factor Fch (q) for a spherical mass distribution is given by
Z
Fch (q) = 4
dr r 2 j0 (qr) ch (r) ,
(2)
0

wherepq = 2k sin (/2) is the momentum transfer, ~k is the electron momentum, and
E = ~2 k 2 c2 + m2e c4 . The low momentum expansion of eq. 2 yields the leading terms
2
Fch (q) /Z = 1 q 2 hrch
i /6 + . Thus, a measurement at low momentum transfer yields a
2 1/2
direct assessment of the mean square radius of the charge distribution, hrch
i . However, as
more details of the charge distribution is probed more terms of this series are needed and,
for a precise description of it, the form factor dependence for large momenta q is needed.
A theoretical calculation of the charge density entering eq. 2 can be obtained in many
ways. Let p (r) and n (r) denote the point distributions of the protons and the neutrons,
respectively, as calculated, e.g. from single-particle wavefunctions obtained from an average
one-body potential well, the latter in general being different for protons and neutrons. If
fEp (r) and fEn (r) are the spatial charge distributions of the proton and the neutron in the
non-relativistic limit, then the charge distribution of the nucleus is given by
Z
Z

3
ch (r) = p (r ) fEp (r r ) d r + n (r ) fEn (r r ) d3 r .
(3)
The second term on the right-hand side of eq. 3 plays an important role in the interpretation of the charge distribution of some nuclear isotopes. For example, the half-density
charge radius increases 2% from 40 Ca to 48 Ca, whereas the surface thickness decreases by
10% with the result that there is more charge in the surface region of 40 Ca than of 48 Ca
[11]. This also results that the rms charge radius of 48 Ca is slightly smaller than that of
40
Ca. The reason for this anomaly is that the added f7/2 neutrons contribute negatively to
the charge distribution in the surface and more than compensate for the increase in the rms
radius of the proton distribution.
For the proton the charge density fEp (r) in eq. 3 is taken as an exponential function,
what corresponds to a form factor GEp (q) feEp (q) = (1 + q 2 /2 )1 . For the neutron a
good parametrization is GEn (q) = n GEp (q)/(1 + p ), where n is the neutron magnetic
dipole moment and = q 2 /4mN . One can use 2 = 0.71 fm2 (corresponding to a proton
rms radius of 0.87 fm) and p = 5.6, which Galster et al. [12] have shown to reproduce
electron-nucleon scattering data.
Eqs. 1-3 are based on the first Born approximation. They give good results for light nuclei
(e.g. 12 C) and high-energy electrons. For large-Z nuclei the agreement with experiments
is only of a qualitative nature. The effects of distortion of the electron waves have been
2

studied by many authors (see, e.g. ref. [13, 14, 15]). More important than the change in the
normalization of the cross section is the displacement of the minima. It is well known that a
very simple modification can be done in the PWBA equation that reproduces the shift of the
minima to lower qs. One replaces the momentum transfer q in the form factor of eq. 1 with
the effective momentum transfer qef f = q (1 + 3Ze2 /2Rch E), where E is the electron energy
and R 1.2 A1/3 fm. This is because a measurement at momentum transfer q probes in
fact |F (q)|2 at q = qef f due to the attraction the electrons feel by the positive charge of
the nucleus. This expression for qef f assumes a homogeneous distribution of charge within
a sphere of radius R.
A realistic description of the elastic electron scattering cross section requires the full
solution of the Dirac equation. The Dirac equation for the elastic scattering from a charge
distribution can be found in standard textbooks, e.g. [16]. Numerous DWBA codes based
on Dirac distorted waves have been developed and are public.
III.
A.

SKINS AND HALOS

Neutron Skins

Appreciable differences between neutron and proton radii are expected [17] to characterize
the nuclei at the border of the stability line. The liquid drop formula expresses the binding
energy of a nucleus with N neutrons and Z protons as a sum of bulk, surface, symmetry and
Coulomb energies E/A = aV A + aS A2/3 + S(N Z)2 /A + aC Z 2 /A1/3 ap A1/2 ,where aV ,
aS , ap , S and aC are parameters fitted to the experimental data of binding energy of nuclei.
This equation does not distinguish between surface (S) and volume (V ) symmetry energies.
As shown in ref. [18], this can be achieved by partitioning the particle asymmetry as N Z =
NS ZS + NV ZV . The total symmetry energy S then takes on the form S = SV (NV
ZV )2 /A+SS (NS ZS )2 /A2/3 . Minimizing under fixed N Z leads to an improved liquid drop
formula [18] with the term SV (N Z)2 /A replaced by SV (N Z)2 /A[1 + (SV /SS ) A1/3 ].
The same approach also yields a relation between the neutron skin Rnp = Rn Rp , and
SS , SV , namely [18]Rn Rp /R = (A/6NZ)[N Z (aC /12SV ) ZA2/3 ]/[1 + (SS /SV ) A1/3 ],
where R = (Rn + Rp )/2.
Here the Coulomb contribution is essential; e.g. for N = Z the neutron skin Rnp is
negative due to the Coulomb repulsion of the protons. A wide variation of values of SV and
SS can be found in the literature. These values have been obtained by comparing the above
predictions for energy and neutron skin to theoretical calculations of nuclear densities and
experimental data on other observables [18, 19]. For heavy nuclei, with A 1, NZ A2 /2,
and using aC = 0.69 MeV, SV = 28 MeV and R = 1.2A1/3 fm, one gets
 


SV
Rnp = Rn Rp 0.4
2.05 103 ZA1/3 fm,
(4)
SS
where = (N Z) /A is the asymmetry parameter. If one assumes that the central densities
for neutrons and protons are roughly the same and that they are both described by a uniform
distribution with sharp-cutoff radii, Rn and Rp , one finds Rnp 0.8A1/3 fm. Of course, the
sharp sphere model is too simple.
On the experimental front, a study of antiprotonic atoms published in reference [20]
obtained the following fitted formula for the neutron skin of stable nuclei in terms of the
3

root mean square (rms) radii of protons and neutrons

1/2
2 1/2
rnp = rn2
rp
= (0.04 0.03) + (1.01 0.15) fm.

(5)

The relation of the mean square radii with the half-density radii is given by hrn2 i = 3Rn2 /5 +
7 2 a2n /5, where an is the diffuseness parameter. For heavy nuclei, assuming an = ap
Rn , Rp , one gets the same linear dependence on the asymmetry parameter as in eq. 4.
Eq. 5 can be used as the starting point for accessing the dependence of electron scattering
on the neutron skin of heavy nuclei. Applying it to calcium isotopes as an example, one
obtains that the neutron skin varies from 0.15 fm for 35 Ca (proton-rich with negative
neutron skin) to 0.25 fm for 53 Ca. A negative neutron skin obviously means an excess of
protons at the surface.
For heavy nuclei the charge and neutron distributions can be described by a Fermi distribution. The diffuseness is usually much smaller thanpthe half-density radius, ap,n Rp,n .
The neutron skin is then given by Rnp = Rn Rp 5/3 rnp . One can assume further
that the nuclear charge radius is given by Rp = 1.2A1/3 fm Rnp /2 with rnp given by eq.
5. The first and second minima of the form factors occur at q1 = 4.49/Rp and q2 = 7.73/Rp ,
respectively, corresponding to the zeroes of the transcendental equation tan (qRp ) = qRp .
The linear dependence of Rp with the neutron skin (and with the asymmetry parameter
), also imply a linear dependence of the position of the minima,

1
3.74
(N Z)
q1 1/3 1 0.535
fm1 ,
and
q2 = 1.72 q1 .
(6)
A
A4/3
For 100 Sn the first minimum is expected to occur at q1 = 0.806 fm1 = 159 MeV/c, while
for 132 Sn it occurs at q1 = 0.754 fm1 = 149 MeV/c.
The variation of q1 with the neutron skin of neighboring isotopes, q1 2/A8/3 fm1 , is
too small to be measured accurately. The first minimum, q1 , changes from 220 MeV/c for
35
Ca to 204 MeV/c for 53 Ca, an approximate 7%, which is certainly within the experimental
resolution. Of course, sudden changes of the neutron skin with might happen due to shell
closures, pairing, and other microscopic effects.
To be more specific, let us assume that a reasonable goal is to obtain accurate results for

1/2

1/2
the charge radius rp2
, so that rp2
< 0.05 fm. This implies that the measurement of
 1 
q1 has to be such that (q1 /q1 ) < q1 fm
%, with q1 in units of fm1 and the right-hand
side of the inequality yielding the percent value. For 53 Ca, q1 = 1.11 fm1 meaning that the

1/2
experimental resolution on the value of q1 has to be within 1% if rp2
< 0.05 fm is a
required precision. Of course, the ultimate test of a given theoretical model will be a good
reproduction of the measured data, below and beyond the first minimum.
B.

Neutron halos

Elastic electron scattering will be very important to determine charge distributions in

proton-rich nuclei. This will complement the basic information on the charge distribution
in, e.g. 8 B, obtained in nucleon knockout reactions [21]. For light nuclei composed by a
core nucleus and an extended distribution of halo nucleons, the nuclear matter form factor
can be fitted with the simple expression F (q) = (1 g) exp (q 2 a21 /4) + g/(1 + a22 q 2 ), with
the density normalized to one. g is the fraction of nucleons in the halo. In this expression
4

the first term follows from the assumption that the core is described by a Gaussian and
the halo nucleons by an Yukawa distribution. Taking 11 Li as an example, the following
set of parameters can be used g = 0.18, a1 = 2.0 fm and a2 = 6.5 fm. Although only few
nucleons are in the halo they change dramatically the appearance of the squared form factor.
Even when the individual contribution of the halo nucleons is small and barely visible in a
linear plot of the matter distribution, it is very important for the form factor of the total
matter distribution. It is responsible for the narrow peak which develops at low momentum
transfers. This signature of the halo was indeed the motivation for the early experiments
with radioactive beams. The narrow peak was observed in momentum distributions following
knockout reactions [1].
Elastic electron scattering will not be sensitive to the narrow peak of |F (q)|2 at small
momentum q, but to the form factor of the charge distribution, |Fch (q)|2 . The determination
of this form factor will tell us if the core has been appreciably modified due to the presence
of the halo nucleons.
In order to explain the spin, parities, separation energies and size of exotic nuclei consistently a microscopic calculation is needed. One possibility is to resort to a Hartree-Fock
(HF) calculation. Unfortunately, the HF theory cannot provide the predictions for the separation energies within the required accuracy of hundred keV. I have used a simple and
tractable HF method [22] to generate synthetic data for the charge-distribution of 11 Li.
Details of this method is described in ref. [23]. Assuming spherical symmetry, the equation
for the Skyrme interaction can be written as


~2
+ V (r) i (r) = i i (r)
(7)
2m (r)
where m (r) is the effective mass. The potential V (r) has a central, a spin-orbit and a
Coulomb term. The central potential is multiplied by a constant factor f only for the last
neutron configuration:

f 6= 1 for last neutron configuration
Vcentral (r) = f VHF (r),
(8)
f = 1 otherwise.
As the effective interaction, a parameter set of the density dependent Skyrme force, so
called BKN interaction [24], is adopted. The parameter set of BKN interaction has the
effective mass m /m =1 and gives realistic single particle energies near the Fermi surface
in light nuclei. The original BKN force has no spin-orbit interaction. In the calculations,
the spin-orbit term was introduced in the interaction so that the single-particle energy of
the last neutron orbit becomes close to the experimental separation energy. In this way, the
asymptotic form of the loosely-bound wave function becomes realistic in the neutron-rich
nucleus. In order to obtain the nuclear sizes, the rms radii of the occupied nucleon orbits
are multiplied by the shell model occupation probabilities, which are also obtained in the
calculations.
The elastic form factor for the matter distribution obtained in the HF calculations are
very close to the ones calculated by the empirical formula given at the beginning of this
section. The lack of minima, and of secondary peaks (as in the empirical formula), make
it difficult to extract from |Fch (q)|2 more detailed information on the charge-density profile.
For example, in the case of 6 Li a good fit to experimental data was obtained with [25]
|Fch (q)|2 exp(a2 q 2 ) Cq 2 exp(b2 q 2 ), with a = 0.933 fm, b = 1.3 fm, and C = 0.205.
However, the data cannot be fitted by using a model in which the nucleons move in a
single-particle potential.
5

C.

Proton Halos

I will consider 8 B as a prototype of proton halo nucleus. This nucleus is perhaps the most
likely candidate for having a proton halo structure, as its last proton has a binding energy of
only 137 keV. The charge density for this nucleus can be calculated in the framework of the
Skyrme HF model. I use here the results obtained in ref. [26], where axially symmetric HF
equations were used with SLy4 [27] Skyrme interaction which has been constructed by fitting
the experimental data on radii and binding energies of symmetric and neutron-rich nuclei.
Pairing correlations among nucleons have been treated within the BCS pairing method. The
form factor squared for the charge density in 8 B is calculated.
The width of the charge form factor squared is ch = 0.505 fm1 = 99.6 MeV/c. The
corresponding values for the neutron and the total matter distributions are, respectively,
n = 0.512 fm1 = 101 MeV/c and tot = 0.545 fm1 = 108 MeV/c. This amounts to
approximately 10% differences of matter and charge form factors in 8 B.
The proton halo in 8 B is mainly due to the unpaired proton in the p3/2 orbit. It is clear
that for such a narrow halo the size of the nucleon also matters. A slice of the nucleon is
included in a thin spherical shell of radius r and thickness dr from the center of the nucleus.
If the position of the nucleon is given by R, the part of the proton charge included in the
spherical shell is given by
Z
2
dch = 2r dr
d p (x) sin ,
(9)
0

where p (x) is the charge distribution inside a proton at a distance x from its center.
The coordinates are related by x2 = r 2 + R2 2rR cos . The contribution to the nuclear
charge distribution from a single-proton in this spherical shell is thus given by Np (R, r) =
dch /4r 2 dr.
Assuming that the charge distribution of the proton is described either by a Gaussian or
a Yukawa form, the integral in eq. 9 can be performed analytically, yielding
n
h
i
h
io
(Rr)2
(R+r)2
1

exp a2
exp a2
, for a Gaussian dist.
1

n
h
i
h
io
Np(G) (R, r) = 1/2
4 arR 1 exp |Rr| exp |R+r| , for a Yukawa dist.,
2
a
a
(10)
where a is the proton radius parameter.
The charge distribution at the surface of a heavy proton-rich nucleus, ch (r), may be
described as a pile-up of protons forming a skin. Let ni be the number of protons in the
skin and Ri their distance to the center of the nucleus. One gets
X
ch (r) =
ni Np (r, Ri ) .
(11)
i

Assuming Ri to be constant, equal to the nuclear charge radius R, and using eq. 10 it
is evident that while the density at the surface increases, its size R and width a, remain
unaltered. The form factor associated with this charge distribution is given by
Z

X sin (qRi )
4 X
ni
Fch (q) =
dr r Np (r, Ri ) sin (qr) = exp qa2
ni
,
(12)
q i
qR
i
0
i
6

where the last result is for the Gaussian distribution. An analytical expression can also be
obtained for the Yukawa distribution. For Ri = R, expression 12 shows that the increase of
density in the skin does not change the shape of the form factor, or of the cross section, but
just its normalization. The fall down of the cross section is determined by a alone, and not
by ni . If the charge of additional protons is distributed homogeneously across the nucleus
including the skin, the normalized form factor will not change, except for a small change in
R.
P

i ni
RFor a 2 proton halo nucleus it is more appropriate to replace
4 dR R Np (r, R) ch (R), where ch (R) is the density change created by the extended
wavefunction of the halo protons. One then recast eq. 12 in the form
Z


4
2
Fch (q) =
exp qa
dR R ch (R) sin (qR) .
(13)
q
0
The shape of the form factor is here dependent not only on the proton size but also on the
details of the halo density distribution. For 8 B, the halo size is determined by the valence
proton in a p3/2 orbit. The density ch (R) due this proton can be calculated with a WoodsSaxon model. Using the same potential parameters as in ref. [28] I compare the form factor
|Fch (q)|2 to the charge form factor |Fch (q)|2 . The halo is found to contribute to a narrow
form factor. However, in contrast to the neutron halo case, the charge form factor of 8 B
does not show a pronounced influence of the halo charge distribution.
2
The rms radius of the charge distribution can be calculated from hrch
i =
2
8
2 1/2
6 [dFch /d (q )]q2 =0 . Applying this relationship to the B charge form factor we get hrch i =
1/2

2
2.82 fm which is close to the experimental value hrch
iexp = 2.76 0.08 fm. The shape of the
charge form factor can also be described by a Gaussian distribution with radius parameter
a = 2.30 fm. In contrast to the case of 11 Li, the proton halo in 8 B does not seem to build
up a two-Gaussian shaped form factor. This observation also seems to be compatible with
the momentum distributions of 7 Be fragments in knockout reactions using 8 B projectiles in
high energy collisions [21]. Electron scattering experiments will help to further elucidate
this property of proton halos.

IV.

INVERSE SCATTERING PROBLEM

In PWBA the inverse scattering problem can be easily solved. It is possible to extract
the form factor from the cross section and then, with an inversion of the Fourier transform,
to get the charge density distribution
Z
1
ch (r) = 2
dq q 2 j0 (qr) Fch (q) .
(14)
2 0
The PWBA approximation can be justified only for light nuclei (e.g. 12 C) in the region far
from the diffraction zeros. For higher Z values the agreement with experiment is only of a
qualitative nature.
It is very common in the literature to use a theoretical model for ch (r), e.g. the HF
calculations discussed in the previous sections and compare the calculated F (q) with experimental data. When the fit is reasonable (usually guided by the eye) the model is
considered a good one. However, whereas the theoretical ch (r) can contain useful information about the central part of the density (e.g. bubble-like nuclei, with a depressed central
7

density), an excellent fit to the available experimental data does not necessarily mean that
the data is sensitive to those details. The obvious reason is that short distances are probed
by larger values of q. Experimental data from electron-ion colliders will suffer from limited
accuracy at large values of q, possibly beyond q = 1 fm1 .
In order to obtain an unbiased
experimental ch (r) one usually assumes that the density
P
a
f
is expanded as ch (r) =
n=1 n n (r) , where the basis functions fn (r) are drawn from any
convenient complete set and the expansion coefficients an are adjusted to reproduce the
differential elastic cross section. The corresponding Fourier transform then takes the form
e(q) Fch (q) =

X
n=1

an fen (q) ,

fen (q) = 4

dr r 2 j0 (qr) fn (r) .

(15)

Evidently the sum in n has to be truncated and this produces an error in the determination
of the charge density distribution. Another problem is that, as shown by eq. 14, the solution
of the inverse scattering problem requires an accurate determination of the cross section
up to large momentum transfers. Electron scattering experiments in electron-ion colliders
will be performed within a limited range of q and this will produce an uncertainty in the
determination of the charge density distribution.
Two bases have been found useful [29] in the analysis of electron or proton scattering
data. The present discussion is limited to spherical nuclei, but generalizations to deformed
nuclei can be done. The Fourier-Bessel (FB) expansion (i.e. with fn taken as spherical
Bessel functions) is useful because of the orthogonality relation between spherical Bessel
functions
Z Rmax
1
dr r 2 jl (qn r) jl (qm r) = Rc3 [jl+1 (qn Rmax )]2 nm ,
(16)
2
0
where the qn are defined such as jl (qn Rmax ) = 0. The FB basis implies that the charge
density ch (r) should be zero for values of r larger than Rmax . For example, the basis can
be defined as follows
4 (1)n Rmax
j0 (qRmax ) ,
fen (q) =
q 2 qn2

fn (r) = j0 (qn r) (Rmax r) ,

(17)

where is the step function, Rmax is the expansion radius and qn = n/Rmax .
In principle it is possible to obtain the an coefficients measuring directly the cross section
at the qn momentum transfer. If the form factor (2) is known at qn , the coefficients an can
be obtained inserting (16) and (17) in the definition (2) of the form factor, leading to
an =

Fch (qn )
.
3
2Rmax
[j1 (qn Rmax )]2

(18)

In general the cross sections are measured at q values different from qn . Using the
expansion (17) of the charge density one finds for the form factor the relation
4 X
(1)n
Fch (q) =
an 2
sin (qRmax ) .
q n
q qn2

(19)

By fitting the experimental Fch (q) one obtains the an parameters and reconstruct the nuclear
charge density. Not all an s are needed. Since the integral of the density, or F (q = 0), is
8

fixed to the charge number there is one less degree of freedom. Also, densities tend to zero
at large r. Thus another condition can be used, e.g. that the derivative of the density is
zero at Rmax . Thus, when one talks about n expansion coefficients one means in fact that
only n 2 coefficients need to be used in eq. 19. For experiments performed up to qmax the
number of expansion coefficients needed for the fit is determined by nmax qmax Rmax /.
A disadvantage of the FB expansion is that a relatively large number of terms is often
needed to accurately represent a typical confined density, e.g. for light nuclei. One can use
other expansion functions which are invoke less number of expansion parameters, e.g. the
Laguerre-Gauss (LG) expansion,


2
2
fn (r) = e L1/2
2
,
n

and

2
2
fen (q) = 4 3/2 3 (1)n e L1/2
2
,
n

where = r/, = q/2, and Lpn is the generalized Laguerre polynomial. Another possibility is to use an expansion on Hermite (H) polynomials. In both cases, the number of
terms needed to provide a reasonable approximation to the density can be minimized by
choosing in accordance with the natural radial scale. For light nuclei = 1 2 fm can
be chosen, consistent with the parametrization of their densities. Then the magnitude of an
decreases rapidly with n, but the quality of the fit and the shape of the density are actually
independent of over a wide range.
For real data, the expansion coefficients an are obtained by minimizing
X  yi y(qi, an ) 2
2
=
,

i
i
where y(qi , an ) is the fitted value of the cross section (form factor) with a set of coefficients
an and yi are the experimental data at momentum qi with uncertainty i .
Increasing the number of coefficients does not improve the quality of the fit. It only
produces more oscillations of the density. The reason is that terms with larger ns are only
needed to reproduce the data at larger values of momentum transfer. The fit to the data
for q < qmax is not affected but the presence of these new terms introduces oscillations in
the charge distribution. A possible fix to this problem is to include pseudodata in addition
to experimental data. This method is well known in the literature [29]. The pseudodata
are used to enforce constraints and to estimate the incompleteness error associated with the
limitation of experimental data to a finite range of momentum transfer.
The author is grateful to Haik Simon for beneficial discussions. This work was supported
by the U. S. Department of Energy under grant No. DE-FG02-04ER41338.

[1] I. Tanihata et al., Phys. Rev. Lett. 55 (1985) 2676.

[2] C.A. Bertulani, M.S. Hussein, and G. M
unzenberg, Physics of Radioactive Beams (Nova
Science Publishers, Hauppage, NY, 2002).
[3] A. Ozawa, T. Suzuki and I. Tanihata, Nucl. Phys. A693 (2001) 32.
[4] R. Hofstadter, Nobel Lectures, available at http://nobelprize.org/physics/laureates/1961/hofstadter-lecture.pd
[5] Technical Proposal for the Design, Construction, Commissioning, and Operation of the ELISe
setup, spokesperson Haik Simon, GSI Internal Report, Dec. 2005.

[6] T. Suda, K. Maruayama, Proposal for the RIKEN beam factory, RIKEN, 2001; M. Wakasugia,
T. Suda, Y. Yano, Nucl. Inst. Meth. Phys. A 532, 216 (2004)
[7] R. Hofstadter, Rev. Mod. Phys. 28, 214 (1956).
[8] F. Hofmann and H. Lenske, Phys. Rev. C57, 2281 (1998).
[9] S. Mizutori et al., Phys. Rev. C61, 0443261 (2000).
[10] A.N. Antonov et al., Phys. Rev. C72, 044307 (2005).
[11] R.F. Frosch et al., Phys. Rev. 174, 1380 (1968).
[12] S. Galster et al., Nucl. Phys. B32(1971) 221 .
[13] W.A. McKinley and H. Feshbach, Phys. Rev. 74, 1759 (1948).
[14] D.R. Yennie, D.G. Ravenhall, and R.R. Wilson, Phys. Rev. 92, 1325 (1953); Phys. Rev. 95,
500 (1954).
[15] L.S. Cutler, Phys. Rev. 157, 885 (1967).
[16] J.M. Eisenberg and W. Greiner, Excitation Mechanisms of the Nucleus, (North-Holland,
Amsterdam, 1988).
[17] J. Dobaczewski,W. Nazarewicz, and T. R. Werner, Z. Phys. A 354, 27 (1996).
[18] P. Danielewicz, Nucl. Phys. A 727 (2003) 233.
[19] A.W. Steiner et al., Phys. Rep. 411 (2005) 325.
[20] A. Trzcinska et al., Phys. Rev. Lett. 87, 082501 (2001).
[21] W. Schwab et al., Z. Phys. A350, 283 (1995).
[22] G.F. Bertsch, A.B. Brown and H. Sagawa, Phys. Rev. C39, 1154 (1989).
[23] H. Sagawa and C.A. Bertulani, Prog. Theo. Phys. Suppl. 124, 143 (1996).
[24] P. Bonche, S. Koonin and J. W. Negele, Phys. Rev. C13 (1976) 1226.
[25] L.R. Sueltzle, M.R. Yearian, and H. Crannel, Phys. Rev. 162, 992 (1967).
[26] S.S. Chandel, S.K. Dhiman and R. Shyam, Phys. Rev. C68, 054320 (2003).
[27] E. Chabanat, P. Bonche, P. Haensel, J. Meyey and R. Schaeffer, Nucl. Phys. A627 (1997)
710.
[28] P. Navratil, C.A. Bertulani and E. Caurier, Phys. Lett. B634 (2006) 191.
[29] J. L. Friar and J. W. Negele, Nucl. Phys. A212 (1973) 93.
[30] K.T.R. Davies and J.R. Nix, Phys. Rev. C14, 1977 (1976).
[31] M.E. Grypeos, G.A. Lalazissis, S.E. Massen and C.E Panos, J. Phys. G 17 (1991) 1093.
[32] R.E. Kozak, Am. J. Phys. 59, 74 (1991).

10