DIELECTRIC POLARIZATION AND ELECTRIC FIELD DISTORTION

DUE TO HEAVY IONS IMPINGING ON SILICON DETECTORS

M. Pârlog,1, 2, H. Hamrita,2 B. Borderie,2 L. Lavergne,2 M. F. Rivet,2 and J. P. Wieleczko1
1 GANIL, CEA et IN2P3-CNRS, B.P. 55027, F-14076 Caen Cedex5, France.
2 Institut de Physique Nucléaire, IN2P3-CNRS, F-91406 Orsay Cedex, France.
Abstract
The polarization of the electron-hole pairs induced by 80 MeV 12 C in a silicon detector was considered
and connected to the relative dielectric permittivity, locally increased. The exact coordinate dependence
of the modified electric field - inside and outside the ion range - was found as the solution of the one
dimension Poisson’s equation for the electric potential in this inhomogeneous medium. The improvement
of the signal simulation is encouraging, as compared to an undisturbed electric field case.

1. INTRODUCTION thickness; in the expression of the depletion bias

We attempt to exploit the sensitivity of the en-
tire current pulse shape in solid state detectors
to the carrier density along the ionization track,
and thus to the charge (Z ), mass (A) and en-
ergy (E ) of the detected ion. This fact becomes
possible with the advent of modern current sen-
sitive preamplifiers, giving access to the nearly
undisturbed current flowing inside the detector
[1]. The perfect simulation of the experimen-
tal signal produced by a given ion (Z A E ) ap-
pears to be of major interest in relation with Pulse
Shape Discrimination technique too. The first
step - approached in the following - is to consider
the electric field distortion, if any, due to the pres-
ence of the ionized track.
The silicon detectors used in this work are
overdepleted abrupt p-n junctions with a 300
m
n-type bulk compared to a negligible-thin p-type
voltage Vd = eND d2 =2r 0 (Vd < V ) enter also
the electron charge magnitude e, the volume con-
centration of donors ND and the silicon dielectric
permittivity  = r 0 written as the product of its
relative dielectric permittivity r (for silicon r =
11.7) and the vacuum permittivity 0 . An ioniz-
ing light charged particle (lcp) induces low con-
centration carriers. Driven by the electric field,
they drift towards the electrodes. Every moving
charge - electron (e) or hole (h) - contributes to
the electric current [2]. The pulse of, e.g., 5 MeV
protons is very well reproduced [1] by this ”stan-
dard model”. Conversely, the same model com-
pletely fails in case of 80 MeV 12 C - Fig. 1.
0 10 20
t (ns)
30 40 50
0
−10
−20
U (mV)

−30 12

zone; p, d and 12 C of known energies were im- −40

C
80 MeV
pinging on the n-type rear side, in order to get −50
a higher sensitivity to the particle nature. En- −60
ergy measurements using charge preamplifiers −70
showed that charge collection was complete in
all cases, i.e., no pulse height defect has been ob- Fig. 1: The experimental current signal induced by a 80
served. MeV 12 C ion impinging on the rear side of a silicon detec-

For x considered from the rear side, the electric tor [1] - full line; the simulation using a linear electric field
field reads: F (x) = 2Vd x=d2 ; (Vd ; V )=d. V - dashed line - completely fails.
is the applied bias voltage and d is the detector What makes lcp and carbon ion so differ-
ent? The answer is: the mean stopping power
e-mail:parlog@ganil.fr; Permanent address: Nat. (E=Range), which greatly varies as shown in
Inst. for Physics and Nuclear Engineering, Bucharest- column 3 of Table I. As a consequence, the mean
Măgurele, Romania. electron-hole (e-h) pair concentration - column 4
- is at least one order of magnitude higher in case
of carbon:
1017, but still far below the free
electron concentration in a conductor, of the or-
der of Avogadro’s number. A cylinder radius rc =
750 nm was considered for our estimation [3, 4].
Table I: Mean values of stopping power and carrier pair
concentration for 5 MeV protons and 80 MeV 12 C (see text
for explanations).
ion E E=Range concentration
[MeV/(mg/cm2 )] [cm;3 ]
[MeV]
3:6 1015
p 5 0.10
12C 9:6 1016
80 2.62
Historically, this difference, associated to a
slower carrier collection for heavy ions (HI), was
quantified as a plasma delay HI pd . This would
be one of the main components (besides the
preamplifier rise-time) of the current signal rise-
time,  r , experimentally measured: HIr > r
lcp
[3, 5–7]. Should be considered a linear [8] or a
quadratic [9] composition of these components?
The question becomes obsolete when the whole
signal is examined.
2. DIELECTRIC POLARIZATION
Traditionally, the plasma delay scenario is
based on the analogy with a conductor placed in
an external electric field, or eventually with an
electronic tube. The plasma of separated nega-
tive and positive charges, polarized by the exter-
nal electric field, acts as an electric shield: in-
side, the electric field cancels out (F = 0), which
is equivalent to an infinite dielectric permittivity
( ! 1). Then, the plasma undergoes an erosion
due to the diffusion of the carriers. The electric
field is finally restored and the carriers may start
their drift towards the electrodes. But, the delay
time involved by the diffusion process is two or
three orders of magnitude longer [10] than the ex-
perimentally measured rise times. To overcome
these short-comings, one had to resort to artifices
as, for example, to allow the penetration of the
external electric field into a thin sheet of plasma,
in order to speed up its erosion [8].
The above image has a basic difficulty: electric
shielding and hence infinite permittivity are com-
patible with a conductor while the silicon is, in
the framework of the electrodynamics, a dielec-
tric.
Actually, the alternative to this ”conductor”
scenario is the dielectric one, in which the e-h
pairs - induced by ionization - are considered as
dipoles ”oriented” by the initial electric field in
such a way that the barycentres of the negative
and positive charges inside the ion range l are
slightly shifted: the dielectric polarization. Con-
sequently, as we shall show in the next section,
the electric field strength at any position x inside
the plasma column Fin(x) will be lower than the
standard value F (x) in the same point, but not
completely cancelled out: 0 < Fin (x) < F (x).
Albeit diminished, Fin (x) may immediately drive
the carriers, or at least part of them, towards the
two electrodes of the detector. The diffusion,
even if present and playing a role in the disso-
ciation of the closest e-h pairs, is no more seen as
a necessary stage to set in the charge collection.
As a consequence of the dielectric polarization,
the local dielectric permittivity (x) inside the
ionization track will increase in comparison with
the usual silicon permittivity : (x) >  for
x  l. This enhancement is quantified in the fol-
lowing by connecting it, in the framework of an
exact, infinitesimal treatment, to the local energy
loss -(dE /dx) of the ionizing particle. In the an-
cient ”conductor + diffusion” vision, a change of
the dielectric constant was crudely considered in
the early nineties [11].
The electric field induced by a free charge den-
sity free in a dielectric is given by the third equa-
tion of Maxwell:
~
!
r F~ + P = free () rF~ = free + pol :
0 0 0
Here P~ is the dielectric polarization vector and
the term pol = ;rP~ is the polarization charge
density given by the product of the dipole electric
charge e and the dipole volume concentration N :
jpolj = Ne. P = erN has the physical meaning
of dipolar moment of the volume unit, r being the
mean relative distance between individual dipole
charges.
In the first order approximation: P~ = 0F~ ,
being the dielectric susceptibility, r = 1 + and
the above equation becomes: is: n0 = 1=( rc2 w )jdE=dxj, with the energy per
h i free   free e-h pair at 300 K w = 3.62 keV, rc being the fidu-
r (1 + ~
)F = () r  F~ = : r cial cylinder radius.
0 0 For the rear side entrance, the charge distribu-
In an homogeneous medium (r = constant): tion (x) = free in the overdepleted reverse bi-
ased junction reads:
rF~ = free
 () r ~ = free :
F  (1) 
0 < x  xn
(x) = ;eN
r 0 D
eNA xn < x  xn + xp
By introducing the electric potential ': F~ =
;r', eq. (1) becomes the Poisson’s equation: where NA is the acceptor concentration in the
r2' = ; free : (2)
thin p-type part of thickness xp and xn is the
thickness of the n-type zone. We deal with a
cylindrical, longitudinal geometry and the Pois-
When the dipole concentration varies from N to son’s equation (2) may be considered in one di-
N 0 = N + n, the polarization varies from P to mension, with the boundary conditions: '(0) =
P 0 = erN + er0n, where we supposed different V and '(xn + xp ) = 0. One get its solu-
mean relative distances r0 6= r - both of them pro- tion by making use of the mathematical induc-
portional to the electric field - for the supplemen- tion method and also of the continuity condition
tary n and the previous N dipoles, respectively. of the normal component F of the electric in-
Some algebra leads to the relation between the duction D ~ = 0 F~ + P~ at each separation surface
relative permittivities in the two cases: between two homogeneous infinitesimal zones.
0r = r + a1n (3)
3. RESULTS
which will be used to connect the new dielectric From the charge conservation condition, imply-
constant in the ionized column to the carrier con- ing: ND xn = NA xp and because in our detectors
centration n via the constant a1 - our model pa- NA  ND , one get xp xn and xn
d =
rameter. Just after the ionization, the local con- xn + xp. Thus, the electric field strength inside
centration in an infinitesimal cylindrical slice dx and outside the ion range may be written:

Rl
0 (x) dx + 2 (d ; l ) ; V
x eND 2 2
Fin = eN D x ;  eND <xl
(x) (x) R 0
 0l dx(x) + d ; l
Rl x dx ; eND (d2 ; l2 ) ; V
(4)
eND 0 (x)
Fout = eN D x ; R l dx
2
l<xd
 0 (x) + d ; l

Both expressions depend on the same parameter x. The discontinuity appears at x = l. The mod-
a1 (including rc ) from the permittivity eq. (3). ified field is lower than the standard one inside
the ion range but higher outside it. In both cases
The corresponding curve for 80 MeV 12 C ions the integral along the detector thickness d is the
impinging on the rear side of the detector of 2500 same: V . The distorted field values correspond to
 cm resistivity (ND
1:9 1012 atoms/cm3 ) is
the initial e-h pairs density, calculated by means
plotted in Fig. 2 a) as a function of the coordinate
 1500
1000
F (kV/m)
F (kV/m)
500
(a)
0
0 100
X (µm)
200 300
Fig. 2: a) The distorted electric field strength inside the
silicon detector, due to the ionized column induced by 80
MeV 12 C - dotted curve. b) The same for protons of 5 MeV
- dotted curve. For comparison, the standard electric field
is plotted too - dashed straight line.
of the stopping power tables in ref. [12]. The
parameter a1 was adjusted to reproduce - for this
field curve - the amplitude of the experimental
signal - Fig. 3.
0 20
t (ns)
0
−5
−10
U (mV)
−15
−20
−25
−30
Fig. 3: The experimental current signal induced by a 80
MeV 12 C ion impinging on the rear side of a silicon detec-
tor [1] - full line; the simulation using a distorted electric
field - dotted line.
For protons of 5 MeV, the electric field in a de-
tector of 12700  cm resistivity (ND
3:7 1011
atoms/cm3 ) is slightly distorted and only in a thin
zone corresponding to the peak of Bragg - Fig. 2
b). This fact explains why the standard descrip-
tion of the signal shape is successful in this lcp
case.
4. PROSPECTS AND CONCLUSIONS
In the framework of the simulation shown in
Fig. 3, the drift of all the carriers was instan-
taneously allowed in the modified electric field
plotted in Fig. 2a). In fact, this field was kept
frozen during the whole transit time. However,
once the dipole charges are separated (n = 0), the
electric field should be restored. The alternative
is the following: the e-h pairs are progessively
carried out of each infinitesimal slice concomi-
tant with a progressive restoration of the electric
500
field. This procedure implies, of course, a second
400
parameter for the rate of the carrier extraction and
it will be addressed elsewhere.
300
200
(b)
In conclusion, we have considered the polariza-
100
tion - in the electric field existing in a reverse bi-
0
0 100
X (µm)
200 300 ased silicon junction - of the e-h pairs induced by
an ionizing particle in a cylinder along its path,
parallel to the field. The carrier concentration
was connected to the local stopping power in-
side the particle range. We gave an exact solution
to the Poisson’s equation for the electric poten-
tial in this inhomogeneous medium. In this in-
finitesimal treatment, we established the expres-
sions of the modified electric field strength, in-
side and outside the ion range. The influence of
this field distortion on the shape of the pulse in-
40 60
duced by 80 MeV 12 C was found to be dominant.
The comparison with the experimental signal has
shown an important improvement of the simu-
12
lation result compared to the undisturbed, linear
C electric field strength case. A time dependent ap-
80 MeV
proach, towards a complete description, is under
progress.
5. REFERENCES
[1] H. Hamrita et al., in Proc. Int. Workshop on Multi-
fragmentation 2003, Caen, France, 2003 .
[2] E.Gatti and P. F. Manfredi, Revista del Nuovo Ci-
mento Vol. 9, N. 1 (1986) 1.
[3] W. Seibt, Nucl. Instr. and Meth. 113 (1973) 317.
[4] M. Parlog et al. (INDRA collaboration), Nucl. Instr.
and Meth. in Phys. Res. A 482 (2000) 674.
[5] H. O. Neidel and H. Henschel, Nucl. Instr. and Meth.
178 (1980) 137.
[6] W. Bohne et al., Nucl. Instr. and Meth. in Phys. Res.
A 240 (1985) 145.
[7] J. B. England, G. M. Field, T. R. Ophel, Nucl. Instr.
and Meth. in Phys. Res. A 280 (1989) 291.
[8] P. A. Tove, W. Seibt and. W. Leitz, Nucl. Instr. and
Meth. 51 (1967) 304.
[9] A. Alberigi Quaranta, A. Taroni, G. Zanarini, IEEE
Trans. Nucl. Sci. NS-15 (1968) 373.
[10] G. Dearnaley and. D. C. Northrop, Semiconductor
counters for nuclear radiation, 2-nd ed. 1966 .
[11] I. Kanno, Nucl. Instr. and Meth. in Phys. Res. A 353
(1994) 93, and references therein.
[12] J. F. Ziegler et al., The Stopping and Range of Ions in
Matter (SRIM), Pergamon Press, New York 1985.