Separation and Purification Technology 28 (2002) 125– 140

www.elsevier.com/locate/seppur

Separation and oxidative degradation of organic pollutants

in aqueous systems by pervaporation and
vacuum–ultraviolet-photolysis
Christine Baus a,*, Karlheinz Schaber a, Irina Gassiot-Pintori b, André Braun b
a
Institut für Technische Thermodynamik und Kältetechnik, Uni6ersität Karlsruhe (TH), 76128 Karlsruhe, Germany
b
Lehrstuhl für Umweltmesstechnik am Engler-Bunte-Institut, Uni6ersität Karlsruhe (TH), 76128 Karlsruhe, Germany
Received 28 July 2001; accepted 20 March 2002

Abstract

This paper presents a new hybrid process for the treatment of industrial waste water. It combines pervaporation
as a means of separating the pollutant from the condensed aqueous phase by phase transfer and vacuum– UV (VUV)
gas phase photolysis for the oxidative degradation of the organic pollutant. Organic substances dissolved in water are
separated and concentrated in a multistage pervaporation process. The gaseous permeate is subsequently oxidised and
mineralised in a VUV-photochemically induced oxidation process. In contrast to the VUV-photolysis of aqueous
solutions, gas phase photolysis is highly energy efficient. The process combination was developed from bench to pilot
scale. The results show that industrial waste water containing organic pollutants can be purified to very low residual
concentration permitting recycling, re-use (irrigation) or release into the environment. © 2002 Elsevier Science B.V.
All rights reserved.

Keywords: Pervaporation; VUV-photolysis; Waste water purification; THF; Photooxidation

1. Introduction which has to be treated before it can be released

The growing concern about emissions of haz-
ardous Volatile Organic Compounds (VOCs),
particularly in urban areas, has strengthened the
requirements for purification of exhaust gases and
waste effluents. Industrial processes often create
wastewater contaminated with organic substances
* Corresponding author. Tel.: + 49-721-608-2322; fax: +
49-721-607-102.
E-mail address: sekretariat@ttk.uni-karlsruhe.de (C. Baus).
without causing damage to the environment. An
efficient and sustainable process for this purifica-
tion or for the remediation of ground water has to
meet the law-enforced limits reliably and at the
same time operate economically.
Conventional processes separate the contami-
nants by a phase transfer followed by a concen-
tration or/and elimination of the pollutants,
mostly by incineration. More stringent regulations
call for processes which do not entail secondary
problems such as air pollution or the secondary
formation of new hazardous substances.

1383-5866/02/$ - see front matter © 2002 Elsevier Science B.V. All rights reserved.
PII: S 1 3 8 3 - 5 8 6 6 ( 0 2 ) 0 0 0 4 4 - 8
126 C. Baus et al. / Separation and Purification Technology 28 (2002) 125–140
Improvements made in the field of membrane
techniques and VUV-induced photooxidative
degradation establish realistic prospects to meet
these requirements by a combination of the two
techniques.
The main focus of this work is the presentation
of a new hybrid process for the purification of
wastewater containing organic substances at levels
below 1 g/l. In the first two chapters the process
steps are described separately; the third chapter
discusses the results of the combination of the two
processes which have been obtained in a semi-in-
dustrial plant.
Finally a proposal is made for a plant on an
industrial scale. The chapter finishes with a short
economic evaluation.
2. Basic principles
2.1. Per6aporation
Pervaporation is a membrane process in which
a mixture of two or more components is separated
by evaporation through a dense polymeric layer.
The process is often described by a solution– dif-
fusion-mechanism. The permeating compounds
dissolve into the polymer layer of the membrane,
diffuse through it and evaporate at the back of
the membrane.
The driving force for this process is a diminu-
tion in the chemical potential vi of the permeating
component. During pervaporation it can be as-
sumed that the temperature remains constant
across the membrane. If the influence of the mod-
erate pressure on the liquid side on the chemical
potential is neglected the difference in the chemi-
cal potential can be expressed by the diminution
of the fugacities f of the permeating compound i
between liquid and gaseous phase (TL being the
liquid side temperature and R the universal gas
constant).
f Li
Dvi =RTL ln
fG
i applied to obtain technically relevant results.
In a first approach the fugacity of i in the liquid
f Li equals the partial pressure of i above the
solution. This can be described using Raoult’s law
with the activity coefficient ki, the mole fraction xi
and saturation pressure of component i at a given
temperature p0i,sat (T):
f Li = ki xi p0i,sat(TL) (2)
If the pressure is low enough the gaseous per-
meate can be treated as a mixture of ideal gases
obeying Dalton’s law and the fugacity in the gas
phase simplifies to:
fG
i = yi p
P
(3)
In this equation yi is the mole fraction of the
permeating compound in the permeate and p P is
the total pressure on the permeate side.
These equations combined with Eq. (1) yield:
Dvi = RTL ln
ki xi p0i,sat(TL)  (4)
yi p P
Conclusively the driving force is mainly influ-
enced by feed concentration, feed temperature
and permeate pressure. The required lower partial
pressure on the permeate side can be evoked
either by applying a vacuum—i.e. by lowering the
total pressure of the system— or by flushing the
backside of the membrane with an inert gas, e.g.
nitrogen.
The condition for a component i to effectively
pervaporate is given by
Dvi \ 0 i.e.
xi
p P B ki p0i,sat(T) (5)
yi
The inverse ratio of mole fractions of i in the
feed and permeate is often called the enrichment
factor i0 = yi /xi ([1,2]). According to Eq. (5) a high
enrichment factor, which is strongly influenced by
the membrane material, results in a required very
low permeate pressure. High enrichment factors,
i.e. a high selectivity, are not always desirable
since a low vacuum requires high operating costs
and cooling temperatures below 0 °C if the per-
meate is to be condensed. Therefore, in this work
(1) a minimum permeate pressure of only 3 kPa is
The enrichment factor is generally used as a
characteristic parameter of the membrane perfor-
mance ([1,2]). In the work presented here, the
 C. Baus et al. / Separation and Purification Technology 28 (2002) 125–140
enrichment factor is defined by the ratio of mass
fractions in permeate and feed.
x Pi
i=
x Fi
Besides the enrichment factor the total and
partial permeate flux is another important
parameter of the pervaporation performance. This
flux comprises the mass transferred through the
membrane per time and square meter membrane
area.
2.2. VUV-photochemically induced oxidati6e
degradation
Within the domain of photochemical technol-
ogy, photochemically induced processes are of
primary industrial importance. Most Advanced
Oxidation Processes (AOP) are based on a photo-
chemical induction which initiates a manifold of
thermal chain reactions leading to the oxidation
and eventual mineralisation of a pollutant (mix-
ture). AOP are industrially implemented for the
treatment of industrial waste water, and for the
abiotic remediation of chemically contaminated
surface and ground waters, too [4].
The development of similar processes in the gas
phase is more or less restricted to photocatalytic
(TiO2) and photolytic methods. Among the latter,
the VUV-photochemically induced processes are
of particular interest. They make use of hydroxyl
radicals and/or atomic oxygen generated by the
photolysis of water vapour and molecular oxygen,
respectively. Both are natural components of a
normal atmosphere [5]. VUV-photolysis in the gas
phase has become technically and economically
feasible with the development of excimer light
sources and the corresponding power supplies in
the kW range [6].
VUV-photochemically induced oxidative degra-
dation consists of three steps:
“ Initiation: Generation of reactive intermediates,
e.g. hydroxyl radicals or atomic oxygen, which
initiate the oxidative degradation by reacting
with pollutant molecules.
“ Thermal chain starting reaction: Mostly trig-
gered by peroxyl radicals produced by the ad-
dition of molecular oxygen to C-centered
127
radicals formed during the step of initiation, or
by thermal decomposition of products of
atomic oxygen insertion into organic pollutant
(6) molecules.
“ Chain reproduction reaction: Leading to miner-
alisation of the organic material by repeating
thermal oxidative processes induced mainly by
peroxyl, alkoxyl and hydroxyl radicals.
Photooxidation of waste air has already been
studied to some extent ([7,8]). However, the inves-
tigated procedures do not allow generation of
hydroxyl radicals without the addition of corre-
sponding precursors. VUV photolysis of water
vapour occurs at wavelengths of less than 190 nm
Eq. (7).
h6
H2O“ HO’ + H’ (7)
The quantum yield of the production of hydroxyl
radicals in the gas phase is 1.0 (uexc 175–190 nm)
[9]. Less than 0.003 of the excited molecules de-
compose via H2 and O(1D), respectively ([10,11]).
The oxidation of VOCs in the gas phase by
HO-radicals may be initiated either by hydrogen
abstraction Eq. (8) or hydroxyl radical addition to
p-Systems. Subsequent trapping of the C-centered
radicals by molecular oxygen leads to oxidative
degradation (up to mineralisation) where in most
cases peroxyl radicals Eq. (9) serve as thermal
initiators.
RH +HO’ “ R’ +H2O (8)
R’ +O2 “ RO’2 (9)
In addition, (simultaneous) photochemical ho-
molysis of O2 leads to atomic oxygen and hence
to a different manifold of oxidation reactions.
3. Materials and methods —experimental set-up
3.1. Laboratory scale installations
Fig. 1 shows the flow sheet for the laboratory
scale photochemical reactor set-up. The photo-
chemical reactor is supplied with a stream of
oxygen saturated with water at different tempera-
tures and a second stream nitrogen saturated with
the contaminant. Via flow meters variable concen-
128 C. Baus et al. / Separation and Purification Technology 28 (2002) 125–140
trations of the reactants can be adjusted. The
photochemical reactor itself (cf. Fig. 2) consists of
a cylindrical glass tube which is closed by two
Teflon coated metallic flanges, and a coaxial
SUPRASIL protection tube separating the Xe2*-
excimer-light source (maximum 1 kW) and its
outer (grounded) electrode from the reaction sys-
tem. The results reported in this paper were ob-
tained by using 210 W of electrical power. The
gap between the light source and the protecting
tube (the inner wall of the reactor) is flushed with
nitrogen in order to prevent light absorption by
molecular oxygen as a component of air. The
lamp is cooled with deionised water.
Quantitative analyses of THF and the products
of its VUV photooxidation were made on-line by
means of a HP Series II GC equipped with two
six-way valves, a HP-Innowax column (30 m×
0.32 mm× 0.53 mm) and a FID detector. Analyses
were performed at an isothermal temperature of
70 °C or with a temperature programme (analysis
of g-butyrolactone, 70 °C, 30 min, 10 °C/min,
150 °C, 2 min) using He as a carrier gas.
During experiments in the batch mode, samples
were drawn into the GC by means of a vacuum
pump. The first six-way valve (V1) was used to
automatically open or close the reactor against
the analytic equipment. The second six-way valve
(V2) was used as an injection valve equipped with
a 250 ml injection loop. The reaction zone was
always kept closed, except for an interval of 15 s
Fig. 1. Design of the photochemical experimental set-up (laboratory scale).
before injection. During experiments in the con-
tinuous mode, the use of the vacuum pump was
not necessary, and V1 was always open. Under
these conditions, injections were automatically
made every 1.5 min.
Liquid phase standards for quantitative GC-
FID measurements were prepared daily using wa-
ter as a solvent. Injections of liquid samples were
made with an autosampler (HP 7673). Calibration
factors were calculated from peak areas.
Products formed during irradiation were iden-
tified by CG/FTIR/MS combined with literature
data. Product mixtures were accumulated in a
cold trap (liquid nitrogen) and dissolved in 2 ml
of p-xylene. Product separation was achieved in
this case with a HP-1 column (30 m× 0.32 mm×
0.25 mm) at an oven temperature of 30 °C. Quan-
tification of CO2 was performed using a BINOS
IR/VIS/UV gas analyzer (Leybold-Heraeus).
The set-up has been described in detail in a
previous publication [5].
3.2. Pilot plant
3.2.1. Design of the per6aporational part (Fig.
3-1)
In a tank containing 60 l, the contaminated
water (feed solution) is heated to the pervapora-
tion temperature. It is then pumped into the
membrane module and led back into the feed
container. The decrease of contaminant in the
 C. Baus et al. / Separation and Purification Technology 28 (2002) 125–140
Fig. 2. Design of a typical annular photochemical reactor (enlarged description of the reaction zone).
feed container is monitored by taking samples of
the feed. On the backside of the membrane a
vacuum is applied in order to extract the permeate
from the membrane. The permeate is condensed
using cooling water.
The membrane module used is a plate and
frame module by CM Celfa, Seewen, Switzerland,
built like a plate heat exchanger. It includes a
membrane area of 2 m2. The vacuum applied is
variable down to 2 kPa [3].
3.2.2. Combination of per6aporation and
photochemistry in the pilot plant (Fig. 3-2)
The feed container for the hybrid process in the
pilot plant holds 40 l. The applied membrane area
(again a plate and frame module by CM Celfa)
yields only 0.3 m2 since a reduced flow through
the photochemical reactor was desired. Further
more the feed concentration in the container can
thus be regarded constant over a limited experi-
mental time. The vacuum on the backside of the
membrane is adjusted to 10 kPa in order to
lengthen the residence time of the permeate in the
photochemical reactor. Additional oxygen may be
applied at the reactor inlet. The vacuum is main-
tained by a water jet pump with a closed water
cycle where by-products and end products of the
degradation are collected.
129
The photochemical reactor (Fig. 2) has a simi-
lar design as the one used in the laboratory instal-
lation but is powered by two excimer light sources
connected in series with an enhanced maximum
electrical power of twice 1.5 kW. The maximum
optical length is 72 mm, the total length of the
reactor is 900 mm. The gaseous permeate is fed
into the reactor through a perforated ring made
of Duran glass in order to obtain good mixing
with simultaneously fed molecular oxygen and
optimal distribution of the pollutant within the
reactor volume. The mixture can be analysed
before and after the reactor via computer-con-
trolled on-line gas chromatography ([3,5]).
3.3. Materials
3.3.1. Substances
The model contaminant used in this project was
tetrahydrofuran (THF), a versatile solvent for
synthetic purposes. It is extensively used in the
varnish and film industries, as a solvent for natu-
ral and synthetic resins and polymers (e.g. PVC)
as well as a cosolvent for printing inks and adhe-
sives [12]. It is easy to handle in terms of solubility
and therefore side effects brought about by im-
miscibility are reduced.
130 C. Baus et al. / Separation and Purification Technology 28 (2002) 125–140

3.3.2. Membrane layer made of poly-(dimethyl-siloxane) (PDMS) is

In the pervaporation step a composite membrane about 10 mm in thickness, the porous, supporting
consisting of two layers has been used: the active sublayer is made of polyether–copolymer.

Fig. 3. Flowsheet of the pilot plant: pervaporational part (1), hybrid process (2).

C. Baus et al. / Separation and Purification Technology 28 (2002) 125–140
4. Experimental results and discussion
4.1. Per6aporation
The enrichment factor i can be directly ob-
tained from experimental data if feed and perme-
ate are analysed simultaneously. In the pilot
plant, however, the feed concentration decreased
rapidly in the course of the experiment and not
enough permeate appeared in a short enough time
to allocate a specific feed concentration to the
measured permeate concentration. Therefore, an-
other technique had to be developed to calculate
the enrichment factor.
A total mass balance of the plant was used as a
basis for judging the membrane performance. The
enrichment factor was estimated analysing the
diminution of contaminant concentration since
additional experiments showed no overall loss of
THF.
The decrease of THF, from now on called
contaminant c, with time can be described as
dMc
dM dx
=xc +M c
dt dt dt
where M is the total mass in the feed tank and Mc
is the proportion of the contaminant therein (t
indicates the experiment duration time).
Coupling of Eq. (6) with the permeate flux
results in
ixc
M: c = M: w
1− ixc
Assuming that water flux remains constant
throughout (as shown at the laboratory scale) and
is much larger than the flux of the contaminant
(M: w =constant \ \Mc ) — which is given when
xc  1) the following equations can be deduced.
(M0 indicating the initial feed bulk)
M = M0 − M: wt− M: ct
dM
$ −M: w
dt
dMc
= − M: c
dt
Those immersed into Eq. (10) and integrated
over time yield
&
131
M0 − M: wt E xE
c 1− ixc
ln = dxc (15)
M0 − M: wt 0 x 0c xc(1−i(1+xc))
If the time interval is short enough, i may be
assumed to be constant. The right side of Eq. (15)
can then be integrated and yields
M0 − M: wt E 1− i(1+ x Ec )
ln = ln
M0 − M: wt 0 1− i(1+ x 0c )
1 1−i(1+x Ec ) x 0c
+ ln
i− 1 1− i(1+ x 0c ) x Ec
(16)
Since we are looking at very small concentra-
tions of xc, Eq. (16) can be simplified and easily
be resolved for the enrichment factor.
ln x 0c /x Ec
i$1− (17)
ln M0 − M: wt E/M0 − M: wt 0
This final equation has been used in the following
evaluation of experimental results.
Fig. 4-a shows the results of several experiments
performed with the 2 m2 membrane module at
different feed temperatures and a permeate pres-
(10) sure of 3 kPa. Feed bulk contained 60 l of con-
taminated water. With increasing temperature the
time required for the removal of THF from 1000
ppm (w/w) down to e.g. 10 ppm (w/w) decreased
considerably (from 7 h (approximately) at 40 °C
to 3 h (approximately) at 70 °C). Concurrently,
however, the amount of feed solution that has to
be evaporated increases unproportionally because
(11) the total permeate flux through the membrane is
strongly dependent on the temperature. It follows
an Arrhenius type law [3].
Other experiments carried out on a larger pilot
plant installation [3] with a membrane area of 8
m2 and a feed bulk of 530 l show the possibilities
of a pervaporative removal of organics from wa-
ter (Fig. 4–6). Starting from 90 ppm (w/w) a
concentration of less than 1 ppm (w/w) could be
(12)
achieved. However, a removal of 99% of the
contaminant (down to less than 1 ppm (w/w)
(13)
requires twice the energy in terms of evaporation
as the removal of 90% (almost 10% of the feed
(14) has to be evaporated for the latter case while only
5% will be evaporated if only 9 ppm (w/w) are
required). Purifying down to very low levels has
therefore to be considered very thoroughly.
132 C. Baus et al. / Separation and Purification Technology 28 (2002) 125–140

Fig. 4. (a) Depletion of THF in the feed container of the pilot plant (experiments at different feed temperatures, permeate pressure:
3 kPa, membrane area: 2 m2, initial feed bulk: 60 l). (b) Depletion of THF in the feed container of the pilot plant (different
experiments at feed temperature: 50 °C, permeate pressure: 3 kPa, membrane area: 8 m2, initial feed bulk: 530 l), also shown the
amount of feed solution that has to be evaporated.

The enrichment factor calculated for these ex-
periments Eq. (17) is shown in Figs. 5 and 6.
Fig. 5 shows the influence of the temperature
on the enrichment factor. With increasing temper-
ature the enrichment factor decreases. At 40 °C it
ranges between 80 and 95 while at 70 °C it
amounts to only 35– 40. This is mainly due to the
temperature dependence of the solution and the
diffusion of the permeating compounds through
the membrane. These properties of the different
compounds are unequally affected by tempera-
ture. The flux of water through the membrane
experiences a stronger enhancement with tempera-
ture than that of THF. That means the enrich-
 C. Baus et al. / Separation and Purification Technology 28 (2002) 125–140
ment factor decreases. This conforms with the
observation that the permeate flux strongly in-
creases with temperature (Arrhenius type law) [3].
If the pressure on the backside of the mem-
brane is increased, the transmembranal flux will
decrease because the driving force of the process
is decreased. This was observed in the experi-
ments, where flux decreased from 0.45 kg/m2 h at
50 °C and 3 kPa to 0.19 kg/m2 h at 50 °C and 10
kPa [3]. At 70 °C the impact is not so obvious
with flux declining from 1.25 kg/m2 h down to
0.94 kg/m2 h.
The influence of the permeate pressure on the
enrichment factor is shown in Fig. 6. At 50 °C
there is a noticeable rise in the enrichment factor
when the pressure is increased from 3 to 10 kPa.
The increase is not so obvious, however, when
raising the temperature to 70 °C. This can be
explained with the different affection of the driv-
ing force by the pressure at different tempera-
tures. If the driving force is reduced to Eq. (4) one
can see that the ratio of saturation pressure to
applied permeate pressure is the determining fac-
tor. Looking at the water compound the satura-
tion pressure at 50 °C accounts to 12.3 kPa [13].
An augmentation of the permeate pressure from 3
to 10 kPa results in an 85% reduction of the
driving force. At 70 °C the saturation pressure is
31.16 kPa [13] and the reduction is only 51%.
x PTHF
Fig. 5. Calculated enrichment factor i = in the pilot plant at different feed temperatures; permeate pressure 3 kPa.
xF THF
133
Since the temperature is higher, the total value of
Dvi is higher and therefore the impact on the
transmembranal flux of water is less distinct. The
influence on the driving potential of THF is less
striking, at 50 °C a reduction of approximately
65% is found while at 70 °C the reduction sums
up to only 44% (data according to [3,14,15]). So
the transmembranal fluxes of the compounds
change unproportionally with temperature and
pressure. The transmembranal flux of water is less
favoured at low temperature and high pressure
and thus a shift of the ratio of the permeating
compounds, i.e. the enrichment factor, is reached.
Another important observation to note is that
the enrichment factor does not decline when the
concentration is getting very low. There is no
evidence that the pervaporation does not work at
low concentrations.
It can be concluded that THF-contaminated
water can be purified down to very low concentra-
tions under pilot level conditions.
4.2. Photooxidati6e degradation
Fig. 7 shows the VUV-induced photooxidative
degradation of THF at 100 kPa. After approxi-
mately 15 min of continuous operation, no THF
was detected at the exit of the reactor. Mineralisa-
tion was achieved after 50 min operating time as
134 C. Baus et al. / Separation and Purification Technology 28 (2002) 125–140
x PTHF
Fig. 6. Calculated enrichment factor i =
xF
THF
feed temperatures.
can be concluded from the CO2 curves in Fig. 7.
After turning off the light source, the CO2 con-
centration in the off gas diminished, and the
substrate flowed through the reactor without
reacting.
The intermediates of the photooxidative degra-
dation have been isolated and identified by gas
chromatography, mass spectroscopy and IR spec-
troscopy from partly oxidised mixtures. The first
stable intermediate is g-butyrolactone. Its produc-
tion is most probably due to hydrogen abstraction
in 2-position by either hydroxyl radical or atomic
oxygen. Adding molecular oxygen to the interme-
diate C-centred radical will lead to the corre-
sponding peroxyl radical from which
g-butyrolactone genesis may occur either via 1,3-
hydrogen shift and subsequent hydroxyl radical
fragmentation or via hydroperoxide formation,
hydroxyl radical fragmentation and oxidation of
the resulting alkoxyl radical. However, this inter-
mediate alkoxyl radical will not only oxidise to
g-butyrolactone, but will also split the a-CC-
bond, in accord with Atkinson [16] yielding the
corresponding propyl formate which has also
been isolated and which is considered as the sec-
ond key intermediate of the oxidative degradation
of THF.
in the pilot plant - influence of permeate pressure on the pervaporation at different
g-Butyrolactone is photochemically unstable
and reacts from its electronically excited state in a
Norrish I-type reaction (a-splitting) to be fol-
lowed in this specific case by the fragmentation of
CO2 and ring closure of the remaining biradical-
type intermediate to cyclopropane. Cyclopropane
may react with the primary initiators of the oxida-
tive manifold by hydrogen abstraction, leading to
a C-centred radical which is further oxidised by
repeating series of hydroperoxyl and alkoxyl radi-
cal formation and subsequent formaldehyde frag-
mentation until complete mineralisation.
Formaldehyde as well as acetaldehyde have been
identified as intermediates of this manifold.
In a similar way, propyl formate is oxidatively
degraded, and formic acid esters of the progres-
sively degraded ester function (ethyl and methyl
esters) have been identified and support the work-
ing hypothesis.
No evidence has been found yet for a cleavage
of the the a-CO-bond of the 2-tetrahydrofuranyl
alkoxyl radical which would lead to 4-hydroxy-
butyraldehyde and to subsequent oxidation prod-
ucts including succinic acid.
The oxygen flow was varied in the laboratory
scale experiments in order to define the best oper-
ating conditions. Fig. 7 exemplarily shows the
 C. Baus et al. / Separation and Purification Technology 28 (2002) 125–140
course of degradation at two different oxygen
concentrations. In the first minutes the diminution
of the THF concentration follows a pseudo-first
order kinetics indicating that stationary absorp-
tion conditions in the irradiated reactor zone are
reached. After approximately 40 min CO2 concen-
tration remains constant, i.e. a constant amount
of THF is mineralised. At higher oxygen concen-
trations the rate of mineralisation is higher but
the oxygen concentration has no influence on the
rate of CO2 formation.
However, the rate of THF oxidation was sig-
nificantly enhanced by an increase of the H2O
concentration [5]. Since less oxygen molecules are
activated at a higher water content, there will be a
lower production of ozone, and, simultaneously,
there will be a lower amount of intermediates
because they react faster at higher HO-radical
concentrations.
4.3. Combination of per6aporation and
photooxidati6e degradation in the pilot plant
The experiments at the pilot plant were carried
out at a feed temperature of 50 °C and at a
permeate pressure of 10 kPa. Since the pollutant
Fig. 7. Degradation of THF in the laboratory installation (total flow 40.8 l/h at 100 kPa and 25 °C with 0.073 l/h THF and 0.54
l/h water flow).
135
is enriched in the gaseous vapour phase of the
permeate it is expected that the H2O alone will
absorb the UV irradiation and form highly reac-
tive OH- and H-radicals. These start the chain
reaction which leads to a destruction of the or-
ganic compound. For a complete mineralisation
(down to CO2 and H2O), however, the addition of
oxygen is necessary.
The residence time of the pollutant in the pho-
tochemical reactor was identified as one of the
major parameters effecting the efficiency of the
degradation process. The results obtained in the
laboratory installation have shown that the oxida-
tion of the pollutant is a very fast process whereas
the total mineralisation takes more time since a
vast number of reactions have to be completed.
The influence of the residence time is shown in
Fig. 8.
By means of a perforated plate the flow in the
reactor could be reduced, and the rate of de-
graded THF increases from 45% without flow
reduction to 85% with inserted flow reduction
diminishing the effective input area to 60 mm2. At
5 mm2 the permeate pressure could not be re-
duced to less than 18 kPa, taking into account the
addition of oxygen. No THF was then found in
136 C. Baus et al. / Separation and Purification Technology 28 (2002) 125–140

Fig. 8. Influence of the volume flow reduced by a perforated dressing with various cross-sections (ranging from 1200 to 5 mm2) on
the degradation of THF in the pilot plant (xTHF = 1.49 0.3%; O2 =2 l/min). Shown is the time from switching on the light sources
until a steady state degradation is achieved during steady state pervaporation.

the permeate. These results show that without
volume flow reduction the residence time of the
pollutant in the reactor is too short to realise
complete mineralisation of THF [17]. At laminar
gas flow conditions and a width of the irradiated
volume of 72 mm the possibility of molecules
entering the irradiated section of the reactor is
quite low.
The concept of two excimer light sources con-
nected in series allowed a more detailed observa-
tion of the gas distribution in the photochemical
reactor. If the reactor were designed under the
terms of optimal fluid dynamic conditions the
degradation rate achieved with both lamps should
be twice the degradation rate of each lamp alone.
Experiments were carried out with both lamps
and each lamp operating alone (Fig. 9). While a
total degradation of 82% was achieved with both
lamps, the upper lamp alone showed, however, a
degradation rate of 78% and the lower lamp
approximately 72%. The deviations between up-
per and lower lamp are in the range of the error
margin and therefore considered negligible. How-
ever, the theoretically possible degradation rate of
the two lamps connected in series is 94%— an
indication that the maximum of degradation po-
tential is not reached in the reactor.
For further experiments, a new geometry of the
photochemical reactor will have to be chosen, the
width of the irradiated volume will have to be
optimally adapted to the concentrations of water
vapour and molecular oxygen at a pressure of 10
kPa and a temperature of 50 °C. These parame-
ters will lead to even higher volume flows through
the reactor and hence to longer reactor tubes
ensuring minimum residence times for efficient
pollutant oxidation and/or mineralisation.
5. Industrial application— proposal of a technical
plant for the purification of contaminated waste
water
On the basis of the results obtained in the pilot
plant a concept of a technical plant has been
worked out (Fig. 10). The concept provides a
two-stage pervaporation step with a subsequently
connected photoreactor. The contaminated waste
water—the feed— is heated and then lead into the
first pervaporation stage. The membrane area is
 C. Baus et al. / Separation and Purification Technology 28 (2002) 125–140
distributed in several modules. Before each mod-
ule the feed is reheated in order to supplement the
energy consumed by the vaporisation of the per-
meate. The higher concentrated permeate of the
first modules can be directly led into the photore-
actor, the less concentrated permeate of the fol-
lowing modules is condensed and given as feed to
the second stage. 3 wt.% is set as the limit
concentration.
The permeate of the second stage is then mixed
with the permeate of the first stage and directed to
the photoreactor. The feed of the second stage
does not have to be purified down to the desired
retentate concentration but only to the concentra-
tion in the original feed. Then the retentate stream
of the second stage is mixed with the feed stream
of the first stage.
The modular design of the first stage permits an
individual assignment of small permeate streams
to the photoreactor or to the second stage and
thus allows a reaction to different inlet concentra-
tions while the photoreactor can still be supplied
with an optimal reaction mixture.
The main part of the operational costs is
brought up by the energy consumption for the
evaporation of the permeate. Therefore, an inter-
nal heat recycling is of great importance for the
Fig. 9. Influence of the lamp position on the degradation rate (initial concentration of THF = 1.7 v%, oxygen flow = 3 l/min).
137
profitability of the process. The feed is preheated
with a recooling of the hot retentate before it is
finally heated to the pervaporation temperature.
The results of the pilot plant show that at given
basic conditions the photodegradation is only de-
pendent on the electrical power if the residence
time of the gas in the reactor is high enough.
Therefore, based on results obtained with the
photoreactor in the laboratory installation the
electrical power consumption of the photoreactor
is assumed to be 3 kW.
Based on the experimental data, a first calcula-
tion of the process was made. Simple mass and
energy balances were used over the membrane
and modules— the temperature decline over the
modules was considered negligible, i.e. enrichment
factor and flux were assumed to be constant and
taken from the experiments. The electrical power
consumption was estimated by summing up pump
and photoreactor consumption.
Table 1 shows the results of the calculation for
THF as contaminant.
The membrane area required for a 99% removal
of THF out of 1000 kg/h contaminated water is
strongly temperature and pressure dependent as
expected from the experimental results. Increasing
feed temperature and decreasing permeate pres-
138 C. Baus et al. / Separation and Purification Technology 28 (2002) 125–140
Fig. 10. Concept of an industrial scale plant for the purification of waste water contaminated with VOCs.
sure result in smaller membrane areas, i.e. lower
capital costs. The impact of the pressure is strik-
ing at 50 °C while at 70 °C the increase in mem-
brane area is only marginally noticed. A maximal
operating pressure appears to lie at approxi-
mately 70% of the saturation pressure of water,
the main part of the transmembranal flux, at the
chosen feed temperature.
The operating costs specified by the sum of
required heating energy (steam), cooling water
and electrical power consumption, however, show
a controversial behaviour. They rise with increas-
ing temperature and decreasing pressure. There-
fore, a compromise in operating conditions has
to be found between operating and investment
costs.
Since the photoreactor is working more effi-
ciently at higher permeate pressures because of
the longer residence time of the gas phase in the
reactor a temperature higher than 50 °C has to
be considered. Taking into consideration the
above mentioned influences a possible compro-
mise could be a feed temperature of 60 °C and a
permeate pressure of 13 kPa.
These data, however, have to be validated ei-
ther by applying a more sophisticated modelisa-
tion of the process which allows an extrapolation
of the conditions or by carrying out experiments
at other experimental conditions since no experi-
ments have been made at these conditions yet.
6. Conclusion
Experiments carried out on a pilot plant have
shown that organic pollutants (represented by
THF as a model compound) can be removed by
pervaporation unto very low residual concentra-
tions using commercially available membranes
and modules.
Though many authors recommend other mem-
brane materials [18], the implemented membrane
was deliberately chosen because it is the most
common one that is commercially available. The
achieved selectivity is lower than with other ma-
terials, but selectivity and flux are still high
enough for an industrial plant being operated
economically.
 C. Baus et al. / Separation and Purification Technology 28 (2002) 125–140 139

Table 1
Results of the simulation for the purification of a waste water stream of 1000 kg/h contaminated with 1000 ppm (w/w) THF down
to 1 ppm (w/w)

50 °C/3 kPa 50 °C/10 kPa 60 °C/3 kPa 70 °C/3 kPa 70 °C/10 kPa

Membrane area (m2) (first stage) 327.7 613.7 258.6 202.9 214.5
Membrane area (m2) (second stage) 26.2 57.1 22.9 16.6 18.2
Flow through the photochemical reactor (kg/h) 33.8 32.6 30.8 32.9 32.6
Concentration in the photochemical reactor 8.3 9.5 9.4 7.33 7.84
[wt.%]
Required heating energy (steam (t/h)) 0.267 0.134 0.22 0.278 0.260
Required cooling water (m3/h) 7.3 5.11 9.73 12.44 11.59
Electrical power consumption by, photochemical 12 kW 12 kW 12 kW 12 kW 12 kW
reactor, pumps etc.

Minimal concentration for the direct inlet into the photochemical reactor: 3 wt.%.

The VUV photolysis equipment at laboratory
scale was successfully used to mineralise the pollu-
tant. The amount of oxygen added is of major
importance on the amount of degraded pollutant
but not on the rate of the degradation. The latter
is mostly influenced by the content of water in the
vapour phase. Experiments on pilot plant scale
with a continuous flow through the photochemi-
cal reactor have shown that the degradation is
mainly dependent on the residence time of the
gaseous mixture in the reactor.
A concept of an industrial scale plant for the
purification of wastewater has been developed.
The first approach estimation of capital and oper-
ating costs shows a minimum at medium tempera-
tures and permeate pressures lower than 70% of
the saturation pressure of the permeating com-
pound which contributes most to the transmem-
branal flux. More detailed investigations of the
pervaporation process have to be made preferably
with a more sophisticated modelling approach.
The results permit a positive balance for the
hybrid process. As the light sources were not used
to their full capacity, even higher pollutant con-
centrations could be mineralised with the existing
equipment after optimising the reactor geometry.
The new process is especially suited for the
removal of pollutants at low concentrations from
water. The combination of pervaporation and
photochemical degradation might present an in-
teresting and in many cases economical alterna-
tive to conventional processes, such as vapour- or
air-stripping or adsorption on active carbon.
Acknowledgements
The authors would like to thank the govern-
ment of Baden-Württemberg for financial support
of this research work through the foundation
BWPlus-PWAB.
References
[1] R.Y.M. Huang (Ed.), Pervaporation Membrane Separa-
tion Processes, Membrane Science and Technology Series,
vol. 1, Elsevier, 1991, ISBN: 0-444-88227-8.
[2] J.-L. Philippe, Thesis, University of Karlsruhe, VDI-
Fortschrittsberichte Reihe 15, Nr. 192, 1997.
[3] Ch. Baus, Thesis, University of Karlsruhe, VDI-
Fortschrittsberichte Reihe 3, Nr. 721, 2002.
[4] O. Legini, E. Oliveros, A.M. Braun, Chem. Rev. 93
(1993) 671.
[5] I. Gassiot-Pintori, Thesis, University of Karlsruhe, 2001.
[6] P. Schwarz-Kiene, Thesis, University of Karlsruhe, 2000.
[7] R. Kamps, H. Müller, M. Schmitt, S. Sommer, Z. Wang,
K. Kleinermanns, Chemosphere 27 (11) (1993) 2127.
[8] M. Bhowmick, M.J. Semmens, Water Res. 28 (11) (1994)
2407.
[9] D.L. Baulch, I.M. Campbell, Gas Kin. Eng. Transf. 4
(1981) 137.
[10] C.C. Chou, J.G. Lo, F.S. Rowland, J. Chem. Phys. 60
(1974) 1208.
[11] G. Calvert, J.N. Pitts Jr., Photochemistry, John Wiley
140 C. Baus et al. / Separation and Purification Technology 28 (2002) 125–140

and Sons, New York, 1966.
[12] Ullmann’s Encyclopedia of Technical Chemistry, vol. 6,
part 16, 1986.
[13] E. Schmidt (Ed.), VDI-Steam Tables, 6th ed., Springer
Verlag, 1963.
[14] Gmehling, Onken, Vapor Liquid Equilibrium Data Col-
lection, Aqueous – Organic Systems, Dechema Chemistry
Data Series, 2nd ed., vol. 1, part 1, 1991.
[15] NIST Chemistry WebBook, NIST Standard Reference
Database Number 69 — February 2000 Release, http://
webbook.nist.gov/chemistry.
[16] R. Atkinson, Chem. Rev. 86 (1986) 69 – 201.
[17] I. Gassiot, A.M. Braun, Ch. Baus, K. Schaber, 17th
IUPAC Symp. Photochem., Sitges, 1998, p. 38.
[18] G. Bengston, K.W. Böddeker, Chem.-Ing.-Tech. 62 (11)
(1990) 937 – 939.