In the domain of photochemical processes for water treatment much of t
he technology has been developed using u.v.-radiation in order to degr
ade the microbiological and chemical contaminants in drinking water (L
egrini et al., 1993). With regard to the removal of microbiological po
llutants, the u.v.-disinfection of water is discussed as a promising a
lternative to the use of chemical disinfectants (Von Sonntag and Schuc
hmann, 1992). In addition, a number of papers have reported on advance
d techniques for the oxidation of organic pollutants by the combinatio
n of u.v.-light and oxidants such as O-3 and/or H2O2 (Von Sonntag et a
l., 1993; Paillard et al., 1992; Hessler et al., 1993). These advanced
oxidation processes (AOP) are particularly oriented towards commercia
l application and have stimulated interest in the design of efficient
reactors and construction of new light sources. Thus, the formulation
of the photokinetic rates of a photochemical process is one of the mos
t difficult parts of reactor design and involves determination of the
absorbed light intensity causing chemical conversion, e.g. the generat
ion of highly reactive hydroxyl radicals. Actinometric measurements ma
y often cause problems due to the polychromatic emission of medium-pre
ssure Hg-arcs, which are used in oxidative degradation procedures.Refe
rring to polychromatic actinometry, this paper deals with the practica
l use of filter solutions as an appropriate method for approaching mon
ochromatic conditions with polychromatic radiation. The experiments we
re carried out with the annular photoreactor of a pilot plant for drin
king water treatment (Fig. 1). For evaluation of the u.v.-irradiance o
f the medium-pressure Hg-arc (Fig. 2) in the 240-470 nm range, the azo
benzene photoisomerization was used as a chemical actinometer (Gauglit
z and Hubig, 1985; Gauglitz and Hubig, 1981; Kuhn et al., 1989) and fi
lter solutions were taken to isolate certain parts of the radiation sp
ectrum. The general problem of preparing solution filters is essential
ly an empirical one of combining inorganic salts and organic, mainly a
romatic, compounds in appropriate solvents like water. The selection r
equirements for choosing a proper solution filter are as follows: spec
ific transmittance to obtain wavelength bands or strong emission lines
, which are within the range of the absorption of pollutants; removal
of unwanted spectral bands for wavelength selective actinometry; photo
chemical stability and no dark reactions. With respect to the determin
ation of light fluxes over the entire irradiated volume, which could b
e carried out inside the irradiation unit as internal actinometry, exp
eriments with liquid filters are not practicable because filter and ac
tinometric solutions must be separated (Table 2). Thus, only external
actinometry was possible and cuvettes containing a filter and the acti
nometer solution were placed outside the vessel (Fig. 3). As a consequ
ence, the reactor geometry must be taken into account by theoretical e
quations concerning the radiation field (Alfano et al., 1986). The ban
d pass filters, which were used for actinometric measurements, are com
posed of various filter components (Table 1) and gave a light distribu
tion enriched with narrow bands of radiation [Fig. 4(a) and (b)]. The
spectral overlapping of distinct absorption bands of the azobenzene ac
tinometer (Fig. 5) with non-attenuated wavelengths of the are-spectrum
[Fig. 6(a)(c)] is required for obtaining selectivity in actinometric
measurements. However, the calculation of intensities by photokinetic
rate equations is limited to monochromatic light and influences of sli
ghtly polychromatic irradiation on fundamental parameters, such as qua
ntum yields, still exist. These difficulties could be overcome if the
photokinetic quantities, particularly the molar absorptivities of tran
s- and cis-azobenzene and thus the pseudo quantum yield, were by weigh
ted averages. Irradiation intensities in the range of 290-360, 335-400
and 375-470 nm could be measured with the filters I-III by assuming t
hat the absorption of the actinometer solutions is negligible. Measure
ment of the light intensity in the u.v.-C region could be achieved by
using the filter solutions IV and V with different transmission charac
teristics (Fig. 7). In the case when total absorption of the actinomet
er is assumed, a linear relationship between Delta E and the irradiati
on time is obtained for each filter respectively. The intensity of the
irradiation source can finally be calculated by the difference of the
slopes. In general, radiation powers of light sources measured by ext
ernal actinometry must be related to the total intensity at the outsid
e wall of the inner quartz tube of the u.v.-arc Experimental quantitie
s, which decrease the u.v.-radiation in the actinometric solution like
the transmittance of the filter solutions and the cooling water must
therefore be considered.