GEAMAG Results, known limitations and bugs
The figure below shows nine sample trajectories
in the horizontal plane
traced through the spectrometer with the interactive
version of GEAMAG.
Plan view of the MAGNEX layout with horizontal trajectories
traced through the spectrometer into the focal plane detector.
The three groups of rays in the focal plane correspond
to a central momentum p0 and p0 ± 5%.
The initial angles
θi are 0o and ± 4o.
The figure shows that to a first order the program is correctly
bringing particles with different angles to a proper focus in the
detector and that particles with different momenta are dispersed as
expected. A more detailed analysis requires the batch version of
the program, as described in (MAG99).
The next figure shows sample trajectories
in the vertical plane
traced with the interactive
version of GEAMAG.
Side view of the MAGNEX layout with vertical trajectories
traced through the spectrometer into the focal plane detector.
The rays in the correspond
to initial angles of
φ = 0o, ± 3.5o and ± 7o.
The vertical scale has been magnified to four times the horizontal.
The vertical trajectories are properly focussed by the quadrupole
through the 20 cm dipole gap into the focal plance detector. Note that
unlike the figure from the ray-tracing program in the MAGNEX report (MAG97),
which is plotted ``unfolded'' along the central trajectory,
this figure is a direct projection of the three-dimensional
view of the system.
In fact, trial simulations have shown that
the location of focal plane is about 2 cm further back from
the dipole exit field boundary
than ZGOUBI calculated (nominal distance 130.0 cm).
The value of the dipole field required to bend nominally
40Ca20+
particles to the center of the focal plane is in disagreement with
the calculated Bρ.
In addition, the momentum dispersion
in the focal plane for 40Ca is anomalous:
we obtain a value of 2.86 cm/% instead of the
expected value of about 4.52 cm/%, which is
indeed obtained
the lighter ions such as 6Li.
This turns out to be because of the average ionic charge state
assumed by GEANT for heavy-ions. This approximation is used
to simulate the continually-changing charge state of ions as they
pass through materials, losing and picking up electrons.
Erroneously, this average
(non-integer) charge-state is also used
by the GTHION routine for heavy ions in vacuum.
Since, the average charge state is a function of energy, it results
in an anomalous dispersion across the focal plane.
To work around this problem, we have written our own version of
the GUSWIM routine that in homogeneous
magnetic fields (field type 1
in geageom.f)
always uses an integer charge state. This charge state is either
Q set in the
user's GUKINE routine and
passed to GUSWIM through a COMMON block
or (default) is the rounded integer of the average
charge state used for energy-loss
calculations by GTHION. In some GUKINE examples, we have simply fixed
Q to a constant (note that this is independent of
the one used by GTHION for straggling and energy-loss).
In more complicated cases, we have calculated in GUKINE the charge state
probability distribution based on the formulae in Ref.
[2]
and used this to select a charge state for a given event.
GEANT 3.21 appears to "miss" the first 8 μm of certain volumes. This has caused
problems for the apparent target and PSD foil thicknesses, which in reality
are much thinner than 8 μm. The effect is seen both
in the observed tracking step length and the energy loss. As a consequence, the
target foil and PSD foil need to be "made" of 1/10th
desity material, made
10x thicker with an additional 8 μm added.
Bibliography
[1] F. Meot and S. Valero, ZGOUBI Users' Guide,
Version 3, SATURNE, LNS/GT/93-12 (1996).
[2] Baudinet and Robinet, Nucl. Instrum. Methods 190 (1981) 197.
Back to Melbeck's Home page |
Back to GEAMAG index |
Last modified: 11 March 2002