Contrib. Plasma Phys.

41 (2001) 5, 494−503

Application of Emissive Probes for Plasma Potential

Measurements in Fusion Devices
R. Schrittwiesera) , C. Ionita
,
a)
, P.C. Balanb), J.A. Cabralc),
F.H. Figueiredoc), V. Pohoata ,
b)
, C. Varandasc)
a) Department of Ion Physics, Univertity of Innsbruck, Technikerstr. 25,
A-6020 Innsbruck, Austria
b) Faculty of Physics, University ”Al. I. Cuza”, B-dul Carol I 11, RO-6600 Ia,

Romania
c)
Centro de Fusão Nuclear, Association Euratom/IST, Insituto Superior
Técnico, Av. Rovisco Pais, P-1049-001 Lisbon, Portugal
e-mail: roman.schrittwieser@uibk.ac.at

Received 12 July 2000, in final form 23 March 2001

Abstract

In experimental fusion devices, up to now, only cold probes were used to determine the
plasma potential in the scrape-off layer (SOL), and their floating potential was assumed
to be proportional to the plasma potential. However, drifting electrons or beams shift the
current-voltage characteristic of a cold probe by a voltage, which corresponds to the mean
kinetic energy of the drifting electrons. This problem can be avoided by the use of electron
emissive probes, since an electron emission current is independent of electron drifts in the
surrounding plasma. In addition emissive probes are insensitive to electron temperature
fluctuations in the plasma. We have used an arrangement of three emissive probes in the
edge plasma region of ISTTOK (Instituto Superior Técnico tokamak) at Lisbon. The probes
have been mounted in such a way that the tips are positioned on the same poloidal meridian
but on different minor radii in the SOL. With this arrangement, the plasma potential has
been measured in the edge region of the ISTTOK, and first results are presented in this
contribution.

1 Introduction

A reliable and exact determination of the plasma potential and its fluctuations is
decisive in many plasmas, especially for a comparison of the experimental observations
with the theoretical models and numerical simulations. In plasma reactors a precise
knowledge of the potential drop across the space charge layer would be helpful for an
optimisation of plasma processes. The confinement and stability of magnetised fusion
plasmas and the radial transport across the scrape-off layer (SOL) is believed to be
determined by the radial potential profile of the SOL and by the turbulent fluctuations
present there, respectively. The question about the actual mechanism of the turbulent
transport through the SOL is still pending. There is a lot of evidence that the radial
potential profile is controlled by self-organised criticality (SOC), since the spectrum
of the fluctuations shows 1/f noise [1].
In this context, a radially varying Reynolds stress R [2, 3, 4, 5] seems to play a decisive
role. Also a measurement of the Reynolds stress requires an accurate determination
of the fluctuating electric fields (and thus of the plasma potentials). This is because
R is proportional to the averaged fluctuating components of the radial and poloidal

@ WILEY-VCH Verlag Berlin GmbH, 13086 Berlin, 2001 0863-1042/01/0509-0494 $ 17.50+.50/0

R. Schrittwieser et al., Plasma Potential Measurements in Fusion Devices 495

velocities ν̃r ν̃ϑ . These velocities are mainly determined by the E × B drifts. Since
fluctuations of B can be neglected, R ∝ Ẽr Ẽϑ , where Ẽr and Ẽϑ , are the radial and
poloidal components of the electric field fluctuations, respectively.
In fusion plasmas, as far as we know, hitherto only cold probes have been used for
a determination of the plasma potential and its fluctuations. Here we present for
the first time an investigation where instead of cold probes, electron-emissive probes
have been used in the edge plasma region of a small tokamak. In section 2 of this
paper it is shown that cold probes cannot deliver an accurate measure of the plasma
potential under all circumstances. Section 3 elucidates the design and the advantages
of electron-emissive probes. In section 4 the Innsbruck emissive probe design, which
is also used in ISTTOK, is laid forth. Section 5 shows the first results and section 6
presents our first conclusion of this investigation.

2 Drawbacks of cold (Langmuir) probes

Electric probes (or Langmuir probes) are very helpful for a quick localised determina-
tion of four important plasma parameters: the electron and ion densities ne,i , respec-
tively, the electron temperature Te and the plasma potential Φpl . The most accurate
measure of Φpl is obtained from the ”knee” of the current-voltage characteristic of the
probe.
Also density and potential fluctuations can in principle be registered with cold probes.
A comprehensive, exact theory of probes is very complicated. Therefore usually a
number of compromises have to be made for an easy and quick evaluation of the
probe signals. However, because of that, the above mentioned plasma parameters
can be subject to severe systematic errors. One of the gravest errors concerns the
determination of the plasma potential, since for the sake of simplicity it is usually
assumed that the floating potential Φf l is a measure for the plasma potential. Indeed,
in a plasma with Maxwellian velocity distribution functions of the ions and electrons
the two values are proportional to each other through the relation

Φf l = Φpl − µTe , (1)

where µ contains ln(mi /me ) (with mi,e being the ion and electron mass, respectively).
Since often not the absolute values but only the relative values of Φpl (or just of the
fluctuations Φ̃pl ) are of interest, it suffices to measure the Φ̃f l of the floating potential
of a cold probe. On the other hand, if for instance we want to register the temporal
evolution of Φpl , this method only works if we suppose that there are no temperature
fluctuations during the recording of Φf l.
Also electric fields (Ē and Ẽ, by which we mean the dc and ac components of the
electric field, respectively) have been determined by using two cold probes and taking
the difference between the floating potentials Φf l,1/2 of the two probes
Φf l,2 − Φf l,1
E = e21 , (2)
d
with e21 being the unit vector between the positions of the two probes and d the
distance. However, also this way of determining E requires the supposition that
there is no difference of the electron temperature between the positions of the two
probes. This is, however, often not the case, especially in the edge plasma region of
magnetically confined fusion plasmas.
496 Contrib. Plasma Phys. 41 (2001) 5

An additional, often neglected fact is the following: Any sufficiently strong electron
drift or an additional electron beam will distort the whole current-voltage characteristic
of a cold probe, in the simplest case by shifting it to the left. In such a case a
determination of the plasma potential from the ”knee” of the characteristic delivers
an erroneous result, and of course, also the floating potential is no longer related to
Φpl through Eq. 1.
In order to estimate the order of magnitude of an electron drift, which will distort
the characteristic, we recall that for thermal electrons (no drift or beam) in a one-
dimensional (sufficiently magnetised) plasma, the probe electron saturation current
density jes is given by [6]:


1 kB Te
jes = − ne eν̄e = −ne e , (3)
4 2πme
with e being the elementary charge and kB the Boltzmann constant.
However, for the case of a (for simplicity) cold one-dimensional electron beam, the
probe saturation current jes,b is:


2eVb
jes,b = −neb eνb = −neb e , (4)
me
where neb is the density of the beam electrons and νb their velocity. The term eVb is
the kinetic energy of the beam electrons. In such a case, the probe characteristic is
shifted to the left by Vb and the ”knee” will show an apparently negative value of Φpl .
If in a plasma simultaneously there are thermal electrons and drifting electrons or an
electron beam, respectively, the characteristic will be distorted if jes,b ≥ jes , i.e., for


 kB Te
neb 2eVb ≥ ne
2π
This reduces to
Te∗
n2eb Vb ≥ n2e , (5)
4π
with Te∗ being the kinetic electron temperature. So e.g., if the beam electron and
thermal electron density are equal, we see that an electron drift with a kinetic energy
1 ∼
corresponding to just 4π = 0.08 times the kinetic electron temperature suffices to make
such a measurement incorrect.
This fact is elucidated by Fig. 1a which shows the idealised characteristic of a cold
probe in a low-temperature magnetised plasma (Φpl = 0 V , Te = 0.2 eV), when the
entire electron population is drifting with a (minimum) kinetic energy of 2 eV. In this
case, due to their kinetic energy, the electrons are able to reach the probe already for
Vp = -2 V (Vp is the probe voltage), so that a determination of the plasma potential
from the ”knee” leads to a corresponding but false value of Φpl = -2 V. Of course also
the floating potential of the probe is shifted by 2 V to the left.

3 The advantages of electron emissive probes

All the problems, mentioned in section 2, can be circumvented when we use a probe,
which not only passively registers the flux of plasma electrons but actively emits an
R. Schrittwieser et al., Plasma Potential Measurements in Fusion Devices 497

Fig. 1: (a) Schematic characteristics of a Langmuir probe (not in scale) where the electrons
have a drift velocity equivalent to 2 eV to the positive direction; the short-dashed line shows
the electron current, the dotted line shows the ion current, the solid line shows the total char-
acteristic. The long-dashed line in the range -2 < Vp < 0 V shows the realistic characteristic.
The plasma potential Φpl is assumed to be zero. (b) Schematic characteristics of an electron
emissive probe (not in scale) under similar conditions as Fig. 1a; the dashed-dotted line
shows the ion current, the dotted line shows the electron emissive current. The long-dashed
line in the range -3 < Vp +1 V shows the realistic characteristic.
498 Contrib. Plasma Phys. 41 (2001) 5

electron current [6, 7, 8, 9, 10]. (The ion current can be entirely neglected in these
considerations.) An electron emission current will be able to flow from the probe to
the plasma as long as Vp is below the plasma potential Φpl , irrespective of the flow of
plasma electrons and of electron temperature fluctuations. For Vp ≥ Φpl , the emission
current drops and electron collection begins. According to the theory [7], the inflection
point of the characteristic is the most accurate measure of the true value of the plasma
potential, but usually simply the emissive probe floating potential is considered as the
plasma potential. There are, however, restrictions of this method when the electron
emission becomes too strong and a space charge of emitted electrons forms around the
probe [12]. An additional restriction lies in the fact that a high input impedance R is
needed to determine the floating potential of any probe. This, together with inevitable
cable and instrument capacities Ci , might lead to a rather high time constant RCi
which reduces the upper cut-off frequency of fluctuations, which can be picked up by
such a probe [9].
An emissive probe is usually realised by a small loop of tungsten wire, carried by
a double-bore ceramic tube, and heated by an external current so that the W-wire
becomes emissive. A necessary condition for sufficient electron emission is that the
magnitude of the thermionic electron emission saturation current density

jee = A∗Tw2 exp [−eWw /(kB Tw )] (6)

(with A* being the Richardson constant, and Tw and Ww the temperature and the
work function of the wire material, respectively) be at least twice the magnitude of
the electron collection current, i.e.:
   


 ∗ 2   2  
A Tw exp − eW w ∼  k B T e 
  = 2  ne + neb 2eVb . (7)
kB Tw  me 2π 

This is necessary for an overcompensation of the total collected electron saturation

current which guarantees that the floating potential of such an emissive probe is pulled
towards the true value of the plasma potential.
Fig. 1b shows the characteristic of the same probe and for the same plasma conditions
as Fig. 1a, but when the probe is emitting an electron current with a magnitude of
about twice that of the electron saturation current. So in addition to the ion current,
there is the electron emission current (which is in any case much larger than the ion
saturation current so that the latter can be neglected). From Fig. 1b we see that the
realistic characteristic crosses the voltage axis very close to the actual value of Φpl (=
0 V in this simplified case) so that the floating potential Φf l,e of the emissive probe
indeed yields a reasonably accurate measure for Φpl .

4 Emissive probe arrangement for the ISTTOK

To our knowledge, up to now, only cold probes have been used for measurements of
potential fluctuations in the edge region of fusion plasmas [13]. Although cold probes
in hot plasmas can become self-emissive due to the heating by the plasma [14], we
believe that it is more reliable to use an emissive probe, which is heated externally in
a controllable way from the start of the measurement.
ISTTOK has a major radius of 0.46 m and a minor radius of a=0.085 m, the latter
being determined by a metallic limiter, which can also be biased. The background
R. Schrittwieser et al., Plasma Potential Measurements in Fusion Devices 499

pressure is smaller than 10−7 mbar. Before each discharge, the chamber is filled with
H2 up to a pressure of around 10−4 mbar. Each shot has a limited duration, up to
40 ms. The strength of the toroidal magnetic field on the minor axis is usually 0.5 T,
the toroidal plasma current is typically 9 kA. The maximum attainable plasma density
is (5-10)·1018 m−3 and the electron temperature is in the range 80 - 220 eV. In the SOL
the density drops to values around (5-10)·1016 m−3 , and Te is on the order of 10 eV
[15, 16].
Under the assumption of ne = 1016 m−3 and Te =100 eV, and purely thermal electrons,
Eq. 3 yields an electron saturation current density of |jes | ∼
= 1.07·104 Am−2 . With an
emissive probe made of a tungsten wire (with a work function of Ww = 4.55 eV and
a Richardson constant of A∗ = 7.4 · 105 Am−2K−2 [17]), heated to Tw = 2500 K, from
Eq. 6 we obtain a thermionic electron emission current density of jee ∼ = 1.25·104 Am−2.
By stronger heating, the emission current can be increased further. E.g., at Tw =
3000 K, which is a realistic upper limit of the wire temperature before the metal softens,
we would obtain an emission current density of jee ∼ = 6.1·105 Am−2.
As it has turned out the meantime [18], this is sufficient even for the core plasma of
similar small fusion experiments, but might be too little for larger devices such as JET,
ASDEX or TEXTOR. On the other hand, recent experiments have indicated that a
full compensation of the electron saturation current might not be necessary to bring
the floating potential of an emissive probe sufficiently close to the plasma potential
[18].
Moreover, a further increase of the electron emission can be achieved by using a thori-
ated tungsten wire instead of the pure tungsten wire used up to now. W +0.6 % Th has
a work function of only 2.6 eV, while the thermal stability of the former is not much
impaired by the addition of thorium. So even though the Richardson constant for tho-
riated tungsten is only A∗ = 3 · 104 Am−2K−2 , for the two values of Tw , given above
(2500 K and 3000 K) we obtain electron emission current densities of 4.24 · 105 Am−2
and 1.23 · 107 Am−2, respectively. This latter value is almost equal to typical electron
saturation current densities in the SOLs of the above mentioned fusion experiments
with ne ∼ = 1019 m−3 .
In ISTTOK we have mounted an array of three emissive probes on one of the flanges.
Each probe consists of a ceramic tube (Al2O3 ) of 2.8 mm outer diameter and slightly
different lengths around 65 mm. Each of the Al2O3 tubes has four bores of 0.5 mm
diameter. Through two opposite bores, a 0.2 mm diameter tungsten wire is inserted
in such a way that on one side of the tube (at the ”hot end”) a W-wire loop of an
approximate total length of 5 mm is formed. In each of the bores, the W-wire extends
at least 5 cm towards the other end (the ”cold end” ”) of each ceramic tube. According
to the design of the IEPPG [9], before the insertion, each W-wires is spliced with about
12 copper threads with diameters of 0.05 mm. In this way, inside the bores the W-wires
are densely covered with a thin layer of Cu so that the conductivity of these parts is
increased.
The wrapping of the W-wires with Cu-threads is done in such a way that the electrical
and mechanical contact between the tungsten and the copper is very good. By careful
choice of the number of Cu-threads by which the W-wire is wrapped, the thickness of
the combined wire can be adjusted so that it tightly fits into the bores of each Al2O3
tube. This increases the electric and mechanical contact between the two materials
which can otherwise not be soldered or welded together. On the cold end of each tube,
only the spliced Cu-wires are protruding and can there be connected easily to any
further electrical lead.
500 Contrib. Plasma Phys. 41 (2001) 5

This treatment has the effect that only the exposed loop of the emissive probe is heated
when a current is passed through the probe wire. The total resistance of each of these
probes is about 0.11 Ω. The electric connection of each probe is made through two
vacuum-tight BNC connectors. The plane of each probe loop is directed so that it is
parallel to the magnetic field so that the effect of the Lorentz force j h × B on the loop
wire (with j h being the heating current density) is minimised.
The probes are mounted at a distance of 20 mm from each other, and the tips of the
loops of all three probes are situated on the same poloidal meridian, i.e., they have the
same toroidal coordinate ϕ, but different poloidal positions ϑ. Their lengths are such
that the probe tips are on different minor radii r1 = 86.1 mm, r2 = 87.3 mm and r3
= 88.1 mm. Thus all three probes are in the shadow of the limiter, hence not directly
in contact with the core plasma. With this arrangement, the plasma potential and
its fluctuations can be measured on three radial positions in the edge region of the
ISTTOK plasma, and thus an approximate radial potential profile can be determined
in the limiter shadow. Fig. 2 shows this arrangement schematically.

Fig. 2: Schematic of the three emissive probes in ISTTOK. The minor plasma radius is a
= 85 mm. The probe tips are situated on different minor radii r1 = 86.1 mm, r2 = 87.3 mm
and r3 = 88.1 mm.

5 Preliminary experimental results and discussion

Here we present preliminary results of measurements with the centre probe of the array,
the tip of which was situated at a distance of 2.3 mm from the edge of the plasma. The
probe (together with the entire heating circuit) was simply attached to a high input
impedance oscilloscope, and the temporal evolution of the probe floating potential
R. Schrittwieser et al., Plasma Potential Measurements in Fusion Devices 501

Φf l (t) was registered. By slowly increasing the heating of the probe from shot to shot,
the range of electron emission was approached. Thus the first measurements have still
been done with a cold probe, but with increasing heating, the probe became more and
more emissive. It has turned out that a current of at least 4.5 A is necessary to heat
the probes to electron emission. In this preliminary stadium, a regulated power supply
with ground-free outputs was used for the heating. In a later stage, a battery-powered
heating circuit will be used for each probe, in order to minimise 50 Hz noise.
Fig. 3a-b shows three temporal evolutions Φf l (t) with different heating currents as
indicated in the figure. The most essential features of the emissive probe behaviour
can be seen in the curves. Especially remarkable is the difference between the topmost
(Fig. 3a) and the lowermost curve (Fig. 3c). In the former case, we see just Φf l (t) of
a cold probe. In the latter case, we see a general rise of the entire curve by about
10 V, which is clear since Φpl is always more positive than the floating potential of a
cold probe. Thus we assume that in this case the probe was indeed emissive enough
to ensure that its floating potential was close to Φpl .
Three different periods can be distinguished in each curve: From t ∼ = -0.01 s to t = 0 s
the H2 gas is pre-ionised by a radio-frequency discharge. From t = 0 s to t ∼ = 0.008 s
a pre-discharge is breaking down the gas. Starting by t ∼ = 0.01 s until t ∼
= 0.02 s the
actual tokamak discharge takes place which should lead to a more or less constant
plasma current with a constant plasma potential. Here, however, we see two negative
peaks (at t ∼ = 0.014 s and ∼
= 0.019 s), which are present in all three figures. These are
evidences that the discharges have been somewhat unstable in these cases.
After about 50 ISTTOK shots the probe flange has been taken out and checked op-
tically under a microscope. All three probe wires appeared flawless and without any
trace of damage due to their exposure to the discharges or of exceeding evaporation
of tungsten from the wire. Moreover, during normal operation of the ISTTOK, the
presence of the probes in the SOL had no negative influence on the discharges.

6 Conclusion

In view of the fact that only electron emissive probes are able to deliver a correct
measure of the plasma potential – even in the presence of drifting electrons or an
electron beam, or when there are temperature fluctuations in the plasma – we have
installed a probe system consisting of three small electron emissive probes in the
SOL of ISTTOK. The objective of this installation was to check a novel method to
investigate the behaviour of the actual plasma potential in the SOL, i.e., to determine
an approximate radial (and azimuthal) profile of the dc values of the plasma potential
and of its fluctuations.
In a number of shots, the proof of principle of this method of plasma potential deter-
mination in a SOL of a small tokamak has been produced, without detrimental effect
on the quality of the discharges. The true value of the plasma potential has for the
first time been determined by electron emissive probes in the edge plasma region of
an experimental fusion device.
Based on these data, more evidence of typical edge plasma phenomena in a magnetised
fusion device, like ELMs, possible flicker noise, radial transport in connection with
Reynolds stress etc. shall be gained in future.
We hope that our method of determining the plasma potential with emissive probe will
be applicable also for other fusion experiments, as it has actually been done recently
with very good success even in the core plasma of the CASTOR tokamak in Prague
502 Contrib. Plasma Phys. 41 (2001) 5

Fig. 3: Temporal evolution of the floating potential of the centre probe during three shots of
the ISTTOK. (a) For a cold probe (almost no heating, Iph = 2 A), (b) for a slightly emissive
probe (Iph = 4 A), (c) for a strongly heated, fully emissive probe (Iph = 4.75 A). The topmost
curve contains a peak which seems to be a pure temperature fluctuation but not a fluctuation
of the plasma potential.

[18]. With modifications, such as the use of thoriated tungsten wires instead of pure
tungsten wires (thereby strongly increasing the electron emission), probes could be
used even in the SOL of larger fusion experiments.

Acknowledgements

Four of the authors (R.S. C.I, P.C. B. and V. P.) would like to thank for the hospitality
extended to them by the Centro de Fusão Nuclear, Instituto Superior Técnico, Lisbon,
Portugal, at the occasion of their various visits at Lisbon. The authors would like to
thank Carlos Hidalgo and Jens Juul Rasmussen for inspiring discussions. This work
has been supported by the Fonds zur Förderung der wissenschaftlichen Forschung
(Austria) under grant No. P-12145 and by the University of Innsbruck, and has been
part of the Association EURATOM-ÖAW under contract No. ERB 5004 CT 96 0020.
R. Schrittwieser et al., Plasma Potential Measurements in Fusion Devices 503

References

[1] P. Bak, C. Tang, K. Wiesenfeld, Phys. Rev. Lett. 59 (1987), 381; Phys. Rev. A 38
(1988), 364.
[2] P.H. Diamond, Y.B. Kim, Phys. Fluids B3 (1991), 1626.
[3] B.A. Carreras, D. Newman, P.H. Diamond, Y.-M. Liang, Phys. Plasmas 1
(1994), 4014.
[4] L. García et al., Proc. Int. Conf. Plasma Phys. Contr. Nucl. Fus. Res. (Würzburg,
1992), IAEA (Vienna, 1993), Vol. 2, 225.
[5] C. Hidalgo, C. Silva, M.A. Pedrosa, E. Sánchez, H. Fernandes, C.A.F.
Varandas, Phys. Rev. Lett. 83 (1999), 2203.
[6] P.C. Stangeby, Phys. Fluids 27 (1984), 682
[7] R.F. Kemp, J.M. Sellen Jr., Rev. Sci. Instrum. 37 (1966), 455; J.R. Smith, N.
Hershkowitz, P. Coakley, Rev. Sci. Instrum. 50 (1979), 210.
[8] R.W. Motley, J. Appl. Phys. 43 (1972), 3711; H. Fujita, S. Yagura, Jpn. J. Appl.
Phys. 22 (1983), 148.
[9] S. Iizuka, P. Michelsen, J.J. Rasmussen, R. Schrittwieser, R. Hatakeyama,
K. Saeki, N. Sato, J. Phys. E: Sci. Instrum. 14 (1981), 1291.
[10] A. Siebenförcher, R. Schrittwieser, Rev. Sci. Instrum. 67 (1996), 849.
[11] M.A. Makowski, G.A. Emmert, Rev. Sci. Instrum. 54 (1983), 830.
[12] K. Reinmüller, Contrib. Plasma Phys. 38 (1998), 7.
[13] e.g.: H. Niedermeyer, M. Endler, L. Giannone, A. Rudyj, G. Theimer, AS-
DEX Team and W7-AS Team, Contrib. Plasma Phys. 36 (1995), 131; C. Hidalgo,
R. Balbín, M.A. Pedrosa, I. García-Cortés, E. Anabitarte, J.M. Sentíes,
M.A.G. San José, E.G. Bustamante, L. Giannone, H. Niedermeyer, ibid. 36
(1995), 139.
[14] e.g.: N. Hershkowitz, B. Nelson, J. Pew, D. Gates, Rev. Sci. Instrum. 54 (1983),
29.
[15] J.A.C. Cabral, C.A.F. Varandas, A. Malaquias, A. Praxedes, M.P. Alonso,
P. Belo, R. Canário, H. Fernandes, J. Ferreira, C.J. Freitas, R. Gomes, J.
Pires, C. Silva, A. Soares, J. Sousa, P.H.M. Vaessen, Plasma Phys. and Contr.
Fusion 38, (1996), 51.
[16] J.A.C. Cabral, C.A.F. Varandas, M.P. Alonso, P. Belo, R. Canário, H.
Fernandes, R. Gomes, A. Malaquias, P. Malinov, F. Serra, F. Silva, A.
Soares, Plasma Phys. and Contr. Fusion 40, (1998), 1001.
[17] W.B. Nottingham, in: Encylcopedia on Physics, Vol. XXI, ed. by S. Flügge
(Springer, Berlin, 1956), p. 1; C. Kittel, in: Introduction to Solid State Physics,
2nd ed. (Wiley, New York, 1965), p. 267.
[18] P. Balan, M. Hron, C. Ionita, E. Martines, V. Pohoata, R. Schrittwieser,
J. Stöckel, M. Tichy, G. Van Oost, to be published.