5

COMPARISON OF BULK AND SUPPORTED V-P-O CATALYSTS

NEW INSIGHTS IN THE ACTIVE PHASE IN C4 OXIDATION TO MA

ABSTRACT
V-P-O catalysts supported on the surface of silica and titania particles were
studied and results are compared with those obtained from bulk V-P-O. The
catalytic performance was tested in the n-butane oxidation reaction to maleic
anhydride and the structure of the (equilibrated) catalysts after the catalytic
reaction was characterised with X-ray absorption spectroscopy (XAFS) and (lowtemperature) Electron Spin Resonance spectroscopy (ESR).
Our results show considerable differences in catalytic performance between
VPO/TiO2, on the one, and VPO/SiO2 and VPO/bulk, on the other hand. The yield
to maleic anhydride is comparable for VPO/bulk and VPO/SiO2.
The differences in catalytic behaviour are attributed to differences in the local
structure around vanadium (XAFS). Furthermore, different spin exchange
interactions between vanadium atoms in the three samples have been observed
(ESR). The combination of the characterisation methods employed suggests that
the structure of the supported V-P-O phase is possibly amorphous and differs
considerably from that of bulk crystalline vanadylpyrophosphate.
We therefore propose that the oxidation of n-butane to maleic anhydride takes
place over a surface V-P-O phase that does not equal vanadylpyrophosphate. This
finding has high relevance for our understanding of the catalytic activity of bulk
crystalline V-P-O catalysts as well.

57

Chapter 5
INTRODUCTION
Bulk catalysts based on vanadium-phosphorus-oxide (V-P-O) are industrially used
for the production of maleic anhydride (MA) from n-butane. The V-P-O system has
been under investigation for the last three decades, and although it is widely
accepted that vanadylpyrophosphate, (VO)2P2O7, is the main component in the
active catalyst, still little is known about the exact nature of the catalytic active
site [1,2,3,4].
At Utrecht University much effort has been devoted to the development of
supported V-P-O catalysts, as the supported V-P-O catalysts have superior
characteristics over bulk V-P-O, such as a cheap and reproducible preparation
procedure, a larger amount of active sites per unit surface area, a short activation
period, and a high mechanical strength [5,6,7]. Therefore, application of
supported V-P-O catalysts in a fluidised-bed process is very promising.
The newly developed supported catalysts show interesting catalytic properties.
For instance, titania-supported V-P-O is already active at moderate temperatures
(523 K vs. 673 K for the commercial bulk catalyst), although the selectivity is
rather low [5,7]. Silica-supported V-P-O catalysts, on the other hand, show
reasonable yields to maleic anhydride [6] and these systems have been further
optimised in our laboratory.
In the literature, only a few other examples of deposition of V-P-O on silica
[8,9,10], titania [11], alumina [11,12,13,14] and AlPO4 [15] have been described.
Generally, the lack of long-range order of the supported V-P-O moieties
compromises a proper characterisation of the supported V-P-O phase with
common techniques, such as X-ray diffraction and FTIR spectroscopy.
Nevertheless, most supported V-P-O catalysts are reported to consist of a phase
that strongly resembles V5+-phosphate, mostly -VOPO4 or -VOPO4 [16,17].
Generally, the catalytic activity of V-P-O catalysts has been related to the local
structure of surface vanadyl groups in vanadylpyrophosphate. However, with
most techniques, no explicit information about the catalytically active surface is
obtained. The structure of crystalline vanadylpyrophosphate consists of pairs of
edge-sharing pseudo octahedrally co-ordinated vanadium ions at a distance of
3.23 , isolated from other pairs by pyrophosphate groups [18]. This structural
unit is often used to model the active sites in V-P-O catalysts [2,19,20]. In
addition, various other models have been proposed, i.e. interfaces between
different VOPO4 phases and (VO)2P2O7 [16,17], V5+ sites on the surface of
(VO)2P2O7 [2], V5+ species in interaction with VO(PO3)2 [21], and amorphous V4+
and/or V5+ phases supported on crystalline (VO)2P2O7 [22,23].
In this chapter we describe our study on the structure-activity relationship of
supported V-P-O catalysts. To this end, we have applied X-ray absorption
spectroscopy (XANES/EXAFS) at the vanadium K-edge and ESR spectroscopy.
These techniques do not require crystallinity of the sample and provide information
on the local micro-structure around the vanadium site.

58

Comparison of bulk and supported V-P-O catalysts

EXPERIMENTAL
Catalyst preparation
The catalysts used in this study were bulk V-P-O, titania-supported V-P-O (8.2
wt% V), and silica-supported V-P-O (7.5 wt% V). All catalysts were prepared
with an atomic P/V ratio of 1.1. Bulk V-P-O was prepared in i-butanol according to
a well-known procedure [24] and exhibited a crystalline structure [25].
Supported V-P-O catalysts have been prepared using TiO2 (Degussa P25) and SiO2
(Engelhard C500-20) according to a procedure developed by Overbeek et al.
[5,6,26].
In brief, the method comprises an electrochemical reduction of V5+ to V3+ ions in
diluted hydrochloric acid solution, followed by homogeneous deposition
precipitation (HDP) of the V3+ species onto the supports in the presence of
NH4H2PO4. For the preparation of silica-supported catalysts, this method was
slightly adapted, because of the poor interaction of the V-P-O with silica [6,27].
First, an amount of V3+ precursor was precipitated in the absence of a phosphate
precursor. After drying in air at 393 K for 16 hours, the partially oxidised
supported vanadium oxide was subsequently impregnated with diluted phosphoric
acid to obtain silica-supported V-P-O. The supported catalysts will be referred to
as VPO/TiO2 and VPO/SiO2. The applied loadings compare to a calculated
monolayer coverage of 2.0 and 2.7 for VPO/SiO2 and VPO/TiO2, respectively.
Catalyst precursors were kept in N2 at 723 K for 16 hours prior to catalytic tests
and subsequent characterisation.
Catalyst performance
After calcination, bulk and supported V-P-O samples (1.5 ml) were tested in the
selective oxidation of n-butane using a 1.5% n-butane, 20.5% O2, 78.5% Ar flow
(50 ml/min, GHSV=2000 h-1) at atmospheric pressure. Formed gaseous products,
as well as unconverted reactants were analysed using an on-line Balzers QMA-420
mass spectrometer operating at 423 K to avoid condensation of products. Carbon
mass balances were in the range of 0.98 to 1.02. A detailed description of the
experimental conditions has been given in chapter 2 of this thesis [28].
Catalyst characterisation
All catalysts have been characterised with X-ray absorption spectroscopy
(XANES/EXAFS) and ESR spectroscopy.
XAS data were collected at Station 8.1 of the SRS facility in Daresbury (UK). The
energy of the electron beam was 2 GeV (average current ~100 mA). The Si (111)
double crystal monochromator was detuned to 50 % intensity to minimise the
presence of higher harmonics. The measurements were all carried out in
transmission mode using ionisation chambers as detectors. To minimise noise the
counting time per data point was taken to be 1000 ms and at least four scans
were recorded and averaged. All samples were recorded ex-situ in He at 77 K. The
energy calibration was performed by means of a V-foil (5m). The absolute value
of the vanadium-edge is 5465 eV.

59

Chapter 5
The catalyst samples were pressed into self-supporting wafers and mounted in an
EXAFS cell [29]. The thickness of the wafer was chosen in such a way as to give
an absorption (x) of 2.5 at the absorption edge for optimal signal-to-noise ratio.
To prevent self-absorption by the catalysts the amount of sample was chosen
such that a step in absorption of 1.0 in the edge region (x=1) was obtained. If
necessary, samples were diluted with boron nitride (BN). Standard procedures
were used to extract the EXAFS data from the measured absorption spectrum.
The background was subtracted using Victoreen routines [30].
The ESR experiments were performed on an X-band Bruker ESP300 spectrometer
equipped with an EN801 resonator (operating in TM110 cylindrical mode with
unloaded Q=1000). The microwave power was 1 mW, far below saturation levels
for the supported V-P-O samples. The magnetic field was modulated with a
frequency of 12.5 kHz and an amplitude of 1 Gauss. The sample temperature was
adjusted in the range of 3.7-300 K with an Oxford ESR900 helium flow cryostat
under control of an Oxford ITC503 temperature controller (temperature stability of
0.5 K).
RESULTS AND DISCUSSION
Catalyst performance
In Figure 1 the conversions as a function of the temperature over VPO/bulk,
VPO/TiO2 and VPO/SiO2 are represented. It is obvious that VPO/TiO2 is the most
active catalyst, already showing conversion at 473 K. In general, our titaniasupported catalysts are even more active at lower loadings [5]. Despite the lower
amount of V-P-O present in the silica-supported catalyst, this sample is more
active than the bulk catalyst. The higher specific surface areas of the active
component in the supported catalysts can explain this difference. The specific
surface area (BET method) of the samples was determined to be 9 m2/g for
VPO/bulk, 54 m2/g for VPO/SiO2 and 44 m2/g for VPO/TiO2, respectively.
Both VPO/bulk and VPO/SiO2 show about the same selectivity to maleic anhydride
(Figure 2). The optimum yield for VPO/bulk and VPO/SiO2 is 25-30% at a
conversion of 50%. This might indicate that the nature of the active sites of the
silica-supported and of the bulk catalyst is the same. However, with VPO/SiO2 the
selectivity decreases much more rapidly when the conversion is raised. The
selectivity of VPO/TiO2 at low conversions is substantially below that of the other
catalysts and only a yield of 6% is obtained at a conversion of 30%.
The differences in activity and selectivity between VPO/TiO2, on the one, and
VPO/SiO2 and VPO/bulk, on the other hand, have been ascribed to the stronger
interaction between V-P-O and titania [5,7]. Therefore, we will focus on the
structural characterisation of the various samples in the remaining part of this
chapter.

60

Comparison of bulk and supported V-P-O catalysts

100

n-Butane conversion (%)

80
60
40
20
0

400

500

600

700

800

T (K)
Figure 1

n-Butane conversion as a function of temperature for VPO/Bulk

(circles), VPO/SiO2 (squares) and VPO/TiO2 (triangles).

100

MA selectivity (%)

80
60
40
20
0

20

40

60

80

n-Butane conversion (%)

Figure 2

Selectivity to MA as a function of n-butane conversion for VPO/Bulk

(circles), VPO/SiO2 (squares) and VPO/TiO2 (triangles).

61

Chapter 5

Absorption (a.u.)

X-ray absorption spectroscopy

In Figure 3, the XANES spectra of VPO/bulk, VPO/TiO2 and VPO/SiO2 after being
used in the oxidation of n-butane are represented. The three spectra show clearly
different features in the positions and shape of the pre-edge and the main

VPO/bulk
VPO/TiO2
VPO/SiO2

5530

5550

absorption edge.
Figure 3
Vanadium K-edge XANES spectra of VPO/Bulk,
VPO/TiO2 employed in the oxidation of n-butane.

VPO/SiO2

5450

5470

5490

5510
Energy (eV)

and

Based on the work of Wong et al., these features can be attributed to different coordination geometries found in various vanadyl compounds, i.e. tetrahedral (VO4),
square pyramidal (VO5) or distorted octahedral (VO6) [31,32,33]. The XANES
spectrum of VPO/bulk agrees well with that reported for VO5 or distorted VO6
compounds, as expected from the single crystal XRD data [18,34]. However, the
XANES spectra of VPO/TiO2 and VPO/SiO2 are clearly different, indicating that
VO4 geometry is present in the titania-supported catalyst and distorted VO6
geometry in the silica-supported catalyst respectively.
However, the XANES features of the X-ray absorption data must first be
confirmed by the EXAFS spectra before conclusions can be drawn.
Figure 4 shows the k2 Fourier transforms of VPO/bulk and VPO/SiO2. It is obvious
that the imaginary parts of the two Fourier transforms differ significantly around
1.5 . This is the range where the contributions of the first-shell oxygen atoms are
located. Furthermore, an important additional contribution at about 2 is present
in the data of VPO/SiO2. For the higher co-ordination shells the spectra of bulk and
supported V-P-O differ considerably. A detailed description of the EXAFS data
62

Comparison of bulk and supported V-P-O catalysts

analysis of the bulk and of the supported V-P-O catalysts will be presented in
chapter 6 of this thesis.
Recently, Nguyen and co-workers have revealed the structure of crystalline
vanadylpyrophosphate [18]. Our EXAFS data of VPO/bulk could be fitted up to a
distance of 3.5 with the single crystal X-ray data of Nguyen et al. [18]. Above
this threshold, too many contributions must be included in the calculations, which
results in a low reliability of the final fit. The results of the EXAFS data analysis of

FT (a.u.)

VPO/bulk
VPO/SiO2

-1
0

R ()
VPO/bulk have been reported elsewhere [35,37].
Figure 4
Fourier transforms (k2, k= 3.5-10 -1) of the EXAFS data of
VPO/bulk (solid line) and VPO/SiO2 (dotted line).
Although an appreciable amount (about 30%) of our VPO/bulk catalyst consists of
an amorphous V-P-O phase [34,36], the EXAFS spectrum is dominated by the
structure of the ideally crystalline vanadylpyrophosphate phase [35]. This places
EXAFS, together with Raman spectroscopy and XRD, in the group of techniques
less appropriate or unsuitable to study possible amorphous (surface) contributions
in bulk V-P-O catalysts.
Figure 5 represents the k2 Fourier transforms of the EXAFS data of both supported
catalysts. These spectra are markedly different from each other both in the first
co-ordination shell and in the higher shells. It is important to note that the
spectrum of VPO/TiO2 also deviates from that of VPO/bulk. This means that the
structures of both supported V-P-O catalysts differ from that of bulk crystalline
(VO)2P2O7.

63

Chapter 5

FT (a.u.)

VPO/TiO2
VPO/SiO2

-1
0

R ()
Figure 5

Fourier transforms (k2, k= 3.5-10 -1) of the EXAFS data of

VPO/TiO2 (solid line) and VPO/SiO2 (dotted line).

As stated above, EXAFS is a technique that can probe the local structure around
the central vanadium atom, up to a distance of 4 . In principle, the EXAFS
spectrum represents the superposition of both bulk and surface contributions.
However, the EXAFS spectrum of VPO/SiO2 is not suffering from interfering
dominant bulk contributions and, hence, represents clearly the local structure of
the amorphous V-P-O phase at the surface.
Preliminary analyses of the EXAFS data indicate that VPO/TiO2 consists of small
particles, in which the isolated V atoms are tetrahedrally co-ordinated by oxygen
atoms, including oxygen atoms bridging V and Ti [35]. The VPO/SiO2 catalyst on
the other hand, consists of vanadyl groups in an octahedral environment [35,37].
This result raises the question concerning the distance between adjacent V atoms.
It is important to note that with both supported V-P-O catalysts no V-V
contributions are found in the EXAFS data up to a distance of 3.5 . To study the
V-V interaction in the supported catalysts we therefore have applied ESR
spectroscopy.
ESR spectroscopy
ESR spectroscopy is a powerful technique to probe paramagnetic centres. The
major part of V-P-O catalysts consists of V4+-phosphate (d1) with electronic spin
S=1/2. When the supported catalysts contain only isolated vanadyl groups, the
ESR spectrum will appear as a characteristic octet of the hyperfine coupling with
the 51V nucleus (I=7/2). In practice, the V hyperfine coupling is often unresolved
in solid state V-P-O samples [38].

64

Comparison of bulk and supported V-P-O catalysts

This absence is attributed to strong exchange interactions between adjacent
electron spins [39], which averages out the hyperfine interaction with the
vanadium nucleus.
Figure 6 represents the ESR spectra of VPO/TiO2 at various temperatures. At room
temperature the ESR spectra of the two supported catalysts correspond to the
spectrum of VPO/bulk [34,36]. However, when the temperature is decreased, the
spectrum of VPO/TiO2 starts revealing the hyperfine coupling with the V nucleus.
In contrast to this, with VPO/SiO2 no hyperfine coupling is visible, resulting in a
spectrum that resembles Figure 2 in chapter 3. This indicates that exchange
interactions between neighbouring vanadyl groups dominate the hyperfine
coupling down to very low temperatures with VPO/SiO2. The strength of this
interaction can be determined from the temperature dependence of the ESR
intensity [34,36]. For bulk V-P-O we have previously found that its ESR spectrum
contains three different contributions, i.e. strongly interacting vanadyl groups
(J/k=-65.7 K), weakly interacting vanadyl groups (J/k=-4.7 K), and

290 K

ESR signal (a.u.)

200 K
100 K
3.90 K

2850

3100

3350

3600

3850

B sweep (Gauss)
antiferromagnetic defects in a ratio of 10:7:2 [34,36].
Figure 6

X-band ESR spectra of VPO/TiO2 at different temperatures, i.e. 290 K,

200 K, 100 K and 3.90 K. The magnetic field is centred at 3400
Gauss, and swept over 1000 G. The spectrum has been taken at a
microwave frequency of 9.437 GHz.

65

Chapter 5
In the paramagnetic regime (T>TNeel), the ESR intensity, I, can be described by the
Curie-Weiss equation for antiferromagnets:

I(T)

(1)

ESR intensity (a.u.)

1/ESR intensity (a.u.)

The Curie-Weiss temperature, , can be extracted from a plot of the reciprocal of

the total integrated ESR intensity of VPO/SiO2 as a function of the temperature
(Figure 7).

20

40

60

80

T (K)
Figure 7

Total integrated ESR intensity (l) and the reciprocal values () of

these data of VPO/SiO2 as a function of temperature. The solid curve
represents a fit of the data assuming Curie-Weiss behaviour.

For antiferromagnetically coupled pairs the Curie-Weiss temperature is related to

the coupling parameter J via the relation k= -(3/2)J. The linear relation between
the reciprocal ESR intensity and T indicates that the compound is exhibiting CurieWeiss behaviour at low temperatures. The Curie-Weiss temperature is
extrapolated to =2 1 K. This value is significantly lower than that measured
with the VPO/bulk sample in which, in the amorphous part, an antiferromagnetic
coupling between vanadyl groups exists (J/k=-4.7 K, which corresponds to
=7.1 K). However, this interaction is still strong enough to average out the
hyperfine coupling in the ESR data down to the lowest temperatures accessible
with our equipment (3.7 K).

66

Comparison of bulk and supported V-P-O catalysts

The small value of the Curie-Weiss temperature in VPO/SiO2 suggests that the
average V-V distance in the amorphous V-P-O phase is considerably larger than
3.23 (the V-V distance in crystalline vanadylpyrophosphate), which points to a
distance probably too large to be observed with EXAFS.
CONCLUSIONS
Our results show considerable differences between VPO/TiO2 and VPO/SiO2 in the
catalytic performance, the local structure around vanadium (XANES/EXAFS) and
the spin exchange interactions between vanadium atoms (ESR). The combination
of characterisation methods used has revealed that the structure of the supported
V-P-O phase does not match that of crystalline vanadylpyrophosphate.
There are three strong indications that the active phase in VPO/SiO2 could
resemble the real active phase of bulk V-P-O catalysts: i) catalytic selectivity, ii)
catalytic yield, iii) XANES/EXAFS.
We are convinced that the local structure of our supported V-P-O phase can be
considered as a model for the amorphous surface of bulk V-P-O. Our results
support the view that the catalytic sites of bulk V-P-O catalysts are located in an
amorphous surface phase.
As it may serve as a model for the amorphous component in bulk V-P-O catalysts,
it will be interesting to study the structure of the VPO/SiO2 catalyst in more detail.
The analysis of the EXAFS data of this sample as well as application of other
characterisation techniques can provide more insight in the nature of the real
active site of the bulk V-P-O catalyst.
REFERENCES
1
2
3
4
5
6
7
8
9
10

B.K. Hodnett, Catal. Rev. -Sci. Eng. 27(3) (1985), 373

G. Centi, F. Trifir, J.R. Ebner, and V.M. Franchetti, Chem. Rev. 88
(1988), 55
G. Centi, Catal. Today 16 (1993), 5
M. Abon, J.C. Volta, Applied Catalysis: A General 157 (1997), 173
R.A. Overbeek, P.A. Warringa, M.J.D. Crombag, L.M. Visser, A.J. van
Dillen, J.W. Geus, Applied Catalysis A: General 135 (1996), 209
R.A. Overbeek, A.R.C.J. Pekelharing, A.J. van Dillen, J.W. Geus, Applied
Catalysis A: General 135 (1996), 231
M. Ruitenbeek, R.A. Overbeek, A.J. van Dillen, D.C Koningsberger, J.W.
Geus, Recl. Trav. Chim. Pays-Bas 115 (1996), 519
V.A. Zazhigalov, Yu. P. Zaitsev, V.M. Belousov, B. parlitz, W. Hanke, G.
Ohlman, React. Kinet. Catal. Lett. 32 (1986), 209
K.E. Birkeland, S.M. Babitz, G. K. Bethke, H.H. Kung, G.W. Coulston, S.R.
Bare, J. Phys. Chem. B 101 (1997), 6895
J.M.C. Bueno, G.K. Bethke, M.C. Kung, H.H. Kung, Catal. Today 43
(1998), 101

67

Chapter 5

11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37

68

M. Martinez-Lara, L. Moreno-Real, R. Pozas-Tormo, A. Jiminez-Lopez, S.

Bruque, P. Ruiz, G. Poncelet, Can. J. Chem 70 (1992), 5
M. Nakamura, K. Kawai, Y. Fujiwara, J. Catal. 34 (1974), 345
N.T. Do, M. Baerns, Appl. Catal. 45 (1988), 1
A. Ramstetter, M. Bearns, J. Catal 109 (1988), 303
P.S. Kuo, B.L. Yang, J. Catal. 117 (1989), 301
E. Bordes, Catal. Today 1 (1987), 499
G.J. Hutchings, A. Desmartin-Chomel, R. Olier, J.C. Volta, Nature 368
(1994), 41
P.T. Nguyen, R.D. Hoffman, and A.W. Sleight, Mater. Res. Bull. 30 (1995),
1055
P.A. Agaskar, L. DeCaul, R.K. Graselli, Catal. Lett. 23 (1994), 339
B. Schiott, K.A. Jorgensen, Catal. Today 16 (1993), 79
M.T. Sananes, A. Tuel, J.C. Volta, J. Catal. 145 (1994), 251
N. Harrouch Batis, H. Batis, A. Ghorbel, J.C. Vedrine, J.C. Volta, J. Catal.
128 (1991), 248
G. Bergeret, M. David, J.P. Broyer, J.C. Volta, G. Hecquet, Catal. Today 1
(1987), 37
K. Katsumoto, D.M. Marquis, US Patent 4,132,670 (1970)
R.A. Overbeek, M. Versluijs-Helder, P.A. Warringa, E.J. Bosma, J.W. Geus,
Stud. Surf. Sci. Catal. 82 (1994), 183
J.W. Geus, R.A. Overbeek, Eur. Patent 94201177.6 (1994)
R.A. Overbeek, E.J. Bosma, D.W.H. de Blauw, A.J. van Dillen, H.G. Bruil,
J.W. Geus, Applied Catalysis 163 (1997), 129
Chapter 2 of this thesis, and references therein
M. Vaarkamp, B.L. Mojet, J.T. Miller, D.C. Koningsberger, J. Phys. Chem.
99 (1995), 16067
J.B.A.D. van Zon, D.C. Koningsberger, H.F.J. van Blik, D.E.J. Sayers, J.
Chem. Phys. 82 (1985), 5742
J. Wong, F.W. Lytle, R.P. Mesmer, and D.H. Maylotte, Phys. Rev. B 30
(1984), 5596
J. Wong, R.P. Mesmer, D.H. Maylotte, and F.W. Lytle, , in: EXAFS and near
edge structure, A. Bianconi, L. Inoccia, and S. Stipcich, eds., Springer
Verlag, Berlin, 1983, 130
J. Wong, D.H. Maylotte, F.W. Lytle, R.B. Greegor, and R.L.St. Peters, in:
EXAFS and near edge structure, A. Bianconi, L. Inoccia, and S. Stipcich,
eds., Springer Verlag, Berlin, 1983, 206
Chapter 3 of this thesis, and references therein
M. Ruitenbeek, R.A. Overbeek, D.C. Koningsberger, J.W. Geus, in: Catalytic
Activation and Functionalisation of Light Alkanes, E.G. Derouane et al.
(eds.), Kluwer Academic Publishers, Dordrecht, 1998, 423
M. Ruitenbeek, A. Barbon, E.E. van Faassen, J.W. Geus, Catalysis Letters
54 (1998), 101
Chapter 6 of this thesis, and references therein

Comparison of bulk and supported V-P-O catalysts

38
39

Y. Zhang-Lin, M. Forissier, J.C. Vedrine, J.C. Volta., J. Catal. 145 (1994)

267
A. Bencini, D. Gatteschi, EPR of exchange coupled systems, SpringerVerlach, Berlin, 1990

69

Chapter 5

70