Catalysis Today 69 (2001) 247–252
Development of a monolith-based process for H2 O2 production:
from idea to large-scale implementation
R. Edvinsson Albers∗ , M. Nyström, M. Siverström, A. Sellin, A.-C. Dellve,
U. Andersson, W. Herrmann, Th. Berglin
Akzo Nobel, Eka Chemicals, Bleaching Chemicals Europe, 44580 Bohus, Sweden
Abstract
Akzo Nobel, Eka Chemicals, produces hydrogen peroxide in a large scale using the anthraquinone (AQ) autooxidation
process. The key step is the highly selective liquid-phase hydrogenation of the AQs to their corresponding hydroquinones.
For this step, a unique hydrogenation technology employing a monolithic catalyst has been developed and implemented.
The present contribution outlines the development of this technology from the initial idea to implementation in industrial
scale. Examples taken from Ekas patents in this area are used for illustrative purposes. © 2001 Elsevier Science B.V. All
rights reserved.
Keywords: Monolithic catalyst; Liquid-phase hydrogenation; Hydrogen peroxide; Scale-up
1. Introduction A variety of AQs can be used and among those
commonly used, we find 2-alkyl-substituted ones, e.g.
Although H2 O2 can be produced in many ways, 2-ethyl AQ. Further details of this process can be found
the by far dominating process is the anthraquinone in [1,2]. It is worth noticing that a very high, well
(AQ) autooxidation process (see Fig. 1). A mixture above 99%, overall selectivity is required.
of AQ derivatives is dissolved in organic solvents to The hydrogenation step requires a catalyst. Initially,
form the working solution, which is cycled between Raney nickel was used, but today most producers use
three main steps: palladium-based catalysts. Hydrogenation technolo-
gies used for this process include fixed-bed reactors
1. a hydrogenator where AQ is hydrogenated catalyt- using a supported Pd-catalyst as well as slurry tech-
ically to the corresponding hydroquinone (AQH2 ); nologies using either supported Pd-catalyst or unsup-
2. an oxidizer where oxygen/air reacts with the AQH2 ported Pd-black. Recently, the use of various forms
to form H2 O2 and the original AQ; of structured reactors has been reported as well (see
3. an extraction step where water is used to remove [1,3,4] and references therein).
the H2 O2 from the working solution. Monolithic catalysts were developed to meet the
requirements for exhaust gas cleaning. The monolith
combines a large geometrical surface area with low
∗ Corresponding author. Tel.: +46-31-587908;
pressure drop and a high degree of uniformity which
fax: +46-31-982751.
makes it suitable for a process characterized by high
E-mail address:
[email protected] flowrates and a need for high selectivity. The idea of
(R. Edvinsson Albers). using a monolith for the AQ hydrogenation was put
0920-5861/01/$ – see front matter © 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 9 2 0 - 5 8 6 1 ( 0 1 ) 0 0 3 7 6 - 5
248 R. Edvinsson Albers et al. / Catalysis Today 69 (2001) 247–252
Fig. 1. The main steps of the AQ autooxidation process. The working solution is cycled between a catalytic hydrogenation, non-catalytic
oxidation and an extraction step.
forward by researchers at the Chalmers University of combination of a backbone providing strength (metal
Technology, Göteborg and picked up by Eka Chemi- or ceramics) and a porous washcoat providing the
cals. Development starts in the early 1980s and the first surface enlargement needed for the catalysis. In this
large-scale reactor was brought on stream a decade case, the process is operated under relatively mild
later. Today the new type of reactor is used in plants conditions (few bars, 40–70◦ C) and it turned out to
with a total annual capacity of approximately 200 kt. be possible to use a simpler approach in which a re-
The aim of the present paper is to give an idea of what inforced amorphous silica is used throughout the wall
was involved in this development. [5]. Such monoliths were not available on the market
and a method for making them became an important
part of the initial development.
2. Development of a monolith technology Once we have the monolithic support, we need to
deposit the catalytically active phase, i.e. the Pd, on
We may distinguish two major lines of develop- it. This is complicated by the fact that a large piece
ment: the development of a monolithic catalyst and of catalyst has to be handled. One useful method for
the development of an appropriate reactor technology. preparing large pieces of catalysts is outlined below:
Needless to say these are not independent. The procedure is called electroless deposition and
involves chemical reduction of a Pd-containing solu-
2.1. Catalyst support tion. It actually consists of three steps where the first
two can be viewed as a preconditioning of the support.
There are a number of requirements affecting the A more-detailed account can be found in [6].
choice of support, some of which are independent of
the shape of the catalyst. An example is the chemi- 1. Sensitizing:
cal compatibility with the process chemicals. Among 1.1. Contacting the porous silica support with an
useful support materials reported in literature we find aqueous acidic SnCl2 solution for 2 min.
carbon, Al2 O3 , SiO2 , where the latter was chosen. 1.2. Washing with water.
Next, it must be possible to shape it as a monolith 2. Activation:
and it must have sufficient mechanical strength in 2.1. Immersing the sensitized support in an acidic
order to allow large structures which are capable of aqueous PdCl2 solution for 2 min.
carrying the load from the catalyst, the liquid held-up 3. Electroless deposition:
and possible pressure drop. In catalysts for gas treat- 3.1. Immersing the activated support in an aqueous
ment at high temperature, this is typically done by a mixture of NH3 , NH4 Cl and PdCl2 .
R. Edvinsson Albers et al. / Catalysis Today 69 (2001) 247–252 249
3.2. Slow addition of an aqueous solution of number of years. The loss of activity is a slow process
NaH2 PO2 . and can be attributed to fouling.
3.3. Washing in acidic water solution and then dry-
ing at 50◦ C overnight. 2.3. Catalyst regeneration
2.2. Catalyst testing A procedure involving washing with an acidic so-
lution has been developed and shown to restore a sig-
In preparation of the support as well as in the metal nificant part of the activity. The procedure consists of
deposition, there are a fair number of preparation pa- the following steps:
rameters that influence the properties of the final cat- 1. treatment with a non-polar organic liquid, e.g.
alyst, such as pore volume and size, the total metal ShellsolTM , for 48 h;
loading, dispersion and distribution of metal, both lo- 2. treatment with acetone (1 h) and drying (2 h),
cally (within the porous wall) and globally (between optional;
various parts of a larger piece). In order to optimize 3. two successive treatments with a mineral acid for
these parameters, it is necessary to test the catalyst. 3 h;
Although some aspects of preparation can be tested 4. rinsing in de-ionized water.
using ground catalyst, testing of larger pieces of mono-
liths remains essential. This necessitated the develop- The effect of treatment with different acids is illus-
ment of bench and pilot scale test reactors. A bench trated in Fig. 2. More details on this procedure can be
scale batch reactor is sufficient for testing the activity found in [7].
of the catalyst as a first test. This scale is also appro-
priate for deactivation tests. A larger scale is needed 2.4. Reaction engineering
in order to verify the function from a hydrodynamic
point of view. The second large line of development focuses on
A complicating factor here is actually the excellent the reaction engineering aspects, by which we include
selectivity needed and achieved. It is so close to unity the design of the reactor and the mode of operation.
that it is difficult to measure accurately in a single At an early stage of the development, a comparative
pass. In practice, a continuous pilot plant including the study was made between a slurry reactor, a fixed-bed
whole cycle of the autooxidation process is needed. (trickle-bed) reactor and the monolith reactor [5]. The
The two main aspects of aging that need to be tests were carried out using a loop reactor illustrated
checked is the loss of catalytic activity and the pos- in Fig. 3. This pilot reactor was part of a pilot system
sible loss of mechanical strength. As it turns out, the including all the main stages of the autooxidation
mechanical strength can be maintained for a large process. It could be fitted with either a monolithic
Fig. 2. The relative productivity of regenerated catalysts depending on the mineral acid used. Relative productivity is defined as the
productivity (in kg(H2 O2 )/m3 h) of the catalyst relative to a virgin catalyst used for the same number of days. The optional acetone
treatment was applied only in the case of orthophosphoric acid (from [7]).
250 R. Edvinsson Albers et al. / Catalysis Today 69 (2001) 247–252
reactor, this can be done quickly since only a few large
assemblies need to be handled. A second advantage is
that this opens up the possibility to keep track of the
individual pieces of catalyst.
2.6. Hydrodynamics
The problem of obtaining hydrodynamically good
conditions is critical to all gas/liquid/solid reactors and
the most important challenges are listed below.
1. create good mass transfer from both the gas and
the liquid phases to the solid catalyst;
2. avoid the formation of local hot spots;
3. generate a uniform flow;
Fig. 3. Loop reactor set-up used for comparing different hydro-
genators (from [5]).
4. allow an up-scalable flow;
5. permit removal of the heat of reaction.
catalyst or with a fixed-bed catalyst. By using an A finding, in the initial testing using a small
empty pipe instead, it could be operated as a slurry diameter monolith and fairly high flow rates, was that
reactor. The results are summarized in Table 1. Note- the direction of flow made little or no difference and
worthy are the high selectivity (low AQ losses) and the up-flow was chosen. When moving to larger scale, it
high rates obtained even after a long time on stream. became difficult to obtain good distribution over the
whole reactor cross-section. The gas was introduced
2.5. Catalyst handling as small bubbles through a poral filter and as they
rose through the submersed bed, channeling occurred
leading to a low utilization of the catalyst. It was then
The monolith pieces are large, with dimensions in
found that down-flow offered several advantages:
the order of meters, which makes handling relatively
easy. The large pieces are mounted in frames, which • Uniform distribution of the liquid could be
turns out to be very practical when loading/unloading a achieved.
Table 1
Comparison of monolithic, slurry and packed-bed reactorsa
Type Slurry Monolith Packed bed
Catalyst type Pd-black Pd on porous silica gel ap- 0.3 wt.% Pd on alumina,
plied to a glass-fiber matrix 2 mm spherical
Catalyst volume (dm3 ) Very small 1 0.5
Bed diameter (mm) – 80 80
Flow direction – Mostly upwards Downwards
Circulation rate (l/min) Up to 300 30
Duration (h) 1000 1250 1000
AQ conversion (%) 45 ∼60
Productivity
g(H2 O2 )/g(Pd) (h) 16 80 (start) 105 (start)
g(H2 O2 )/g(cat) (h) 16 533 314
97% (after 1250 h) 85% (after 1000 h)
Selectivity
mmol(AQ-lost)/mol(H2 O2 ) 0.18 0.10 0.25
a Pressure was 2 bar and temperature 55◦ C (from [5]).
R. Edvinsson Albers et al. / Catalysis Today 69 (2001) 247–252 251
Fig. 4. Comparison of flow directions for two sizes of pilot reactors.
An AQ-containing working solution was hydrogenated at 52◦ C and
a (mean) pressure of 400 kPa. Liquid flow rate was 0.16 m3 /m2 s
and the test was run for 10 days (50 dm3 catalyst) or 30 days
(1000 dm3 catalyst) (from [8]).
• Taylor flow was obtained over a wide range of
flowrates.
• Essentially constant pressure could be established,
since flow is gravity driven, this allows spontaneous Fig. 6. Flow of AQ-containing working solution through a glass
tube (d = 1.5 mm, L = 200 mm) coated internally with silica
internal or external gas recirculation provided that
(from [8]).
there is an open connection between the top and
bottom of the reactor. mode. It is also worth noting that the scale effect is
A comparison between up-flow and down-flow much larger in up-flow mode.
mode of operation was done and is illustrated in Fig. 4. The desired flow pattern is Taylor flow (see Fig. 5)
We see that the productivity is higher in down-flow which provides excellent mass transfer due to the large
gas–liquid contact area, the short diffusion distance to
the channel wall and the internal circulation induced
in the liquid plug. In order to determine the flow rates
under which this flow pattern is realized, a simple
test was done in a glass tube (d = 1.5 mm, L =
200 mm) coated internally with silica [8]. The results
are shown in Fig. 6. We note that the falling rate was
ca 0.4 m3 /m2 s and that the flow became unstable at
flow rates below 0.1 m3 /m2 s.
Having established flow direction and flow rate, we
still need a way of distributing the liquid over the
whole cross-section. It turned out that a perforated
plate was adequate for this as is discussed in [9].
3. Conclusions
The concept of using a monolithic catalyst on a
large scale is a successful one. AQs dissolved in an
ordinary organic solvent can be hydrogenated with
Fig. 5. Taylor (or plug) flow of gas and liquid through a narrow extremely high selectivity at a high productivity.
channel. A chemically and mechanically stable monolithic
252 R. Edvinsson Albers et al. / Catalysis Today 69 (2001) 247–252
catalyst can be produced in large quantities and han- References
dled in a very simple way. Next to the mechanical
design of a hydrogenation reactor, it is possible to find [1] Ullmann, 6th Edition, Electronic Release, 1999.
a window of operating conditions that give steady and [2] Kirk-Othmer, Vol. 13, 4th Edition, 1995, pp. 961–995.
[3] I. Turunen, EP 672 617.
satisfactory operation over prolonged periods of time.
[4] I. Turunen, A. Pukkinen, E. Mäkiniemi, T. Maunula, EP
It should also be realized that the road from idea 596 938 B1.
to large-scale implementation is a long one. Next to [5] Th. Berglin, W. Herrmann, EP 102 934.
the process development itself, it is often necessary to [6] A. Sellin, M. Nyström, EP 878 235 A1.
develop the tools needed for the development. Prime [7] M. Siverström, M. Nyström, EP 778 085 B1.
[8] E. Bengtsson, EP 384 905 B1.
examples are the lab and pilot reactors used for test-
[9] S. Irandoust, B. Andersson, E. Bengtsson, M. Siverström, Ind.
ing the catalyst and the tools for manufacturing the Eng. Chem. Res. 28 (10) (1989) 1489–1493.
catalyst.
