1 Introduction

In the resent years there is observed a great progress in the field of photocatalytic studies (Friedmann et al. 2010; Herrmann 2010) and the application of Advanced Oxidation Processes (AOPs) in the remediation of contaminated air, water and soil. Beside AOPs, heterogeneous photocatalysis has gained the attention as promising technique that can be used as alternatives or as a complement to conventional wastewater treatment (polishing step). It enables the mineralization of completely recalcitrant contaminants (including organics, inorganics, metals and pathogens) into CO2, H2O and small amount of simple inorganic compounds. The photocatalysis is described as the generation of *OH radicals. The main advantage of this system is that radicals can be generated chemically, photochemically, and photocatalytically, depending on the treatment condition and requirements.

TiO2 is the mostly studied photocatalyst. The application of suspended TiO2, beside its unquestionable advantages such as biological and chemical resistance, safety, low price and affordability, is connected however with expensive recycling and recovery of the photocatalyst (Dolat et al. 2012).

One of the factors affecting the photocatalytic oxidation over TiO2 is the initial adsorption of the contaminants on the TiO2 surface. Solution can be supported TiO2 photocatalysts prepared using different adsorptive materials (Herrmann 2010). Many affords have been taken to enhance the adsorption and different types of supports for TiO2 have been tested including silica, perlite, activated carbon (Yao et al. 2010), glass (Fernandez et al. 1995), clay, zeolites and many others. The supported porous catalysts with higher surface area are often being chosen, since, apart from easier disposal, they ensure a high density of active centers for reactions and an enhanced light harvesting because of light reflection and scattering by the pores (Friedmann et al. 2010).

The successful application of TiO2 in the removal of different organic pollutants such as dyes (Yao et al. 2010) or phenols (Emeline et al. 2012) has encouraged the studies dealing with the removal of recently recognized wastewater pollutants—Pharmaceuticals and Personal Care Products (PPCPs). PPCPs in the environment are derived from usage by individuals and pets (in the households and intensive animal breeding farms as growth enhancers), disposal of outdated medications in sewage systems, release of treated and untreated hospital and veterinary wastes to domestic sewage systems, transfer of sewage solids (biosolids) to land, industrial waste streams and releases from aquaculture of medicated feeds, etc. (Ellis 2006; Fatta-Kassinos et al. 2011; Qin et al. 2012). A total of 170 pharmaceutical chemicals are estimated to be used in excess of 1 tone per year (Ellis 2006). Many of PPCPs are not new chemicals and have been present in wastewater for many decades but only now they are being recognized as potentially significant water pollutants.

Depending on their properties, PPCPs can enter the aquatic environment or remain adsorbed on solid particles through electrostatic forces. Many PPCPs are highly bioactive, most are polar, optically active, and all are present in the environment at extremely low concentrations ranging from ng dm−3 to μg dm−3 (e.g. in natural waters, wastewater, sediments and sludge) (Ellis 2006).

Conventional treatment methods are not able to degrade them completely because they are persistent against biological degradation and natural attenuation. It has resulted in their frequent presence in the wastewater. In 2004 there was identified 2986 PPCPs in treated wastewater (Sanderson et al. 2004). PPCPs introduced into the aquatic environment are considered to be that ensures a vacuum of as Endocrine Disrupting Compounds, EDCs that mimic, inhabit or enhance the action of hormones (Gultekin and Ince 2007).

UV irradiation (Lopez et al. 2003; Kasprzyk-Hordern et al. 2009; Kim et al. 2009; Jelic et al. 2010; Fatta-Kassinos et al. 2011) and ozonation (Ternes et al. 2003; Álvarez et al. 2011), were successfully used to remove different PPCPs. β-blockers (metoprolol, propranolol, atenolol) (Álvarez et al. 2011; Ioannou et al. 2011; Hapeshi et al. 2010), antibiotics (ofloxacine, amoxyciline) (Hapeshi et al. 2010), psychiatric drugs (carbamazepine) (Kim et al. 2009), and analgesics and anti-inflammatory drugs (diclofenac) (Kim et al. 2009; Achilleos et al. 2010) were removed from water and wastewater using TiO2-assisted photocatalysis.

There is interesting to study the photocatalytic removal of some typed representatives of PPCPs from water using commercially available TiO2 anchored to reactor walls. The main aim of the studies though was the determination of the catalytic activity of wall-deposited TiO2 photocatalysts in the removal of PPCPs belonging to different therapeutic class of pharmaceuticals.

2 Experimental

2.1 Chemicals

As the photocatalyst were used 4 commercially available TiO2 photocatalyst: TiO2—Tytanpol, Police Z.A. Poland (catalyst T1), and TiO2—Sigma-Aldrich (T2), Hombikat—Sigma-Aldrich (T3) and TiO2-S21 Sigma-Aldrich (T4). As the substrates analgesics and anti-inflammatory drug: diclofenac sodium salt (DFC), antibiotic: chloramphenicol (CPL), β-blocker: metoprolol titrate salt (Mt), natural estrogen: estrone (E1) were supplied from Sigma-Aldrich. Ethanol was used as co-solvent for E1 in aqueous solution.

2.2 Photocatalysts preparation

14.6 g of proper TiO2 was mixed with aluminum cement (142 g) and 40 cm3 of liquid glass and 40 cm3 (or 80 cm3 in case T3) of water. The 25 cm long tube with the internal diameter 48 mm was fulfilled with the mixture for 10 min to make the uniform layer on the aluminum tube. The excess of the mixture that did not adsorbed after 10 min was thought out. Prepared tubes were dried for 24 h at room temperature before use.

2.3 Photocatalysts characterization

The specific surface area of the catalysts was determined on the basis of low-temperature nitrogen adsorption–desorption method in a volumetric analyzer AUTOSORB-1CMS, Quantachrome Instruments, (USA) that ensures a vacuum of at least 2 × 10−6 kPa. Standard Brunauer-Emmett-Teller (BET) method for nitrogen adsorption data was used to calculate SBET of the photocatalysts. Consequently, the total pore volume (Vp) of mesopores was estimated from a single point adsorption at a relative pressure of 0.985 and the pore size distribution was obtained from desorption branch of the isotherm according the Barrett-Joyner-Halenda (BJH) procedure.

SEM studies were conducted on a Tesla BS-301 microscope Quanta 3D FEG operating at 15.0 keV. X-ray diffraction (XRD) on Difractometer HZG-4 (Carl Zeiss Jena) using monochromatic Cu Kα radiation was used to determine the phase composition of the catalysts and estimation of the crystal size of TiO2. The X-ray patterns were scanned in 2θ range of 20o–80o with a step 0.05o.

The quantitative property of treated wastewater was determined using COD and UV–Vis spectroscopic measurements. COD/COD 0 and c/c 0 have been used to calculate the conversion, where c 0 or COD 0 was the initial concentration or initial COD of a pollutant, respectively and c or COD was the concentration of the compounds or COD under study that did not react at a steady state.

2.4 Photocatalytic studies

The photocatalytic oxidation of 50 mg dm−3 DFC, CPL, E1 and Mt was conducted in the tube reactor, equipped with the UV lamp (254 nm, 0.7 kW) emitting 95 % UV, with the light intensity 2 mW cm−2. Such concentrations were chosen to allow assessing the process efficiency within a measurable time scale, and determining the residual concentrations with the analytical techniques (UV–Vis spectroscopy, COD measurements) employed in this work.

The lamp was placed vertically inside the reaction vessel and housed in an inner glass tube (Fig. 1). The tube with photocatalyst was placed in the reactor in the central part and was connected at the top and at the bottom with two other tubes without photocatalyst. The reactor volume actively irradiated by UV was 0.3 dm3. The reactor was fulfilled at the volumetric flow of 4.45 cm3 min−1 before the lamp was turned on for 24 h to maintain the equilibrium and in order to know the adsorption into the samples.

Fig. 1
figure 1

Reactor scheme

Full size image

The UV–Vis absorption spectra were recorded in a wavelength range of 190–300 nm in a 10 mm quartz cell in Varian Cary 4000 Spectrophotometer, in the following conditions: scan rate 600 nm·min−1; time response: 0.1 s; spectral band 2 nm. The concentration of studied pollutants was measured from the absorbance at the proper wavelength (Table 1; Fig. 2) by using a calibration curve. The COD was determined using bichromate method.

Table 1 The properties of studied PPCPs
Full size table
Fig. 2
figure 2

UV–Vis spectra of studied pollutants (Kumar et al. 2010; Kasperek et al. 2011) (Color figure online)

Full size image

3 Results and discussion

3.1 Characterizations of the samples

3.1.1 BET

The applied materials are mesoporous with the presence of mesopores with uniform pore size distribution centered at 10 nm (Table 2).

Table 2 The physicochemical properties of the studied photocatalysts
Full size table

The BET surface area of T1 and T2 was ca. 12 m2 g−1 and T3 and T4 was 55-61 m2 g−1. The pore volume of the photocatalysts was 0.025 cm3 g−1 for low surface area photocatalysts and 0.13–0.15 cm3 g−1 for its counterparts with higher surface area. However, there was no hysteresis loop at relative pressures of 0.4–0.99 in the sample T1 and its BET surface area was 12 m2 g−1, further confirming the absence of mesopores in the T1.

3.1.2 SEM

The scanning microscopy was applied to study the surface morphology of the samples. The morphology of the samples presented in Fig. 3 looks similarly.

Fig. 3
figure 3

SEM micrographs of the studied photocatalysts: a T1, b T2, c T3, d T4

Full size image

The photocatalysts are in the form of small agglomerates of crystals with the diameter of ca. 0.2–0.25 μm. The surface is full of voids and pores. The SEM micrographs were used to estimate the diameter of the primary structure of TiO2 and obtained value 20 nm is in good agreement with XRD data.

3.1.3 XRD

XRD was applied to determine the phase composition of studied photocatalysts. All XRD spectra presented in the Fig. 4 looks similarly and the peaks confirming anatase structure are directly observed. According to JCPD 21-1272, the peaks at 2Θ of about 25.2, 37.9, 48.3, 53.8, 62.7 and 68.9 (depicted in Fig. 4 as A) are representing the presence of anatase structure (planes 1 0 1, 1 0 3, 2 0 0, 1 0 5, 2 1 3 and 1 1 6, respectively). In the XRD spectra of T2 and T3 the presence of rutile crystallites was evidenced at 2Θ of 27.4 and 37.1 (JCPDS 21-1276) depicted in Fig. 5 as R. The crystalline sizes of samples (Table 2) were calculated applying Scherrer equation (Eq.1):

$$ D = \frac{K \cdot \lambda }{\beta \cdot \cos \theta } $$
(1)

where D is the crystal size in nanometer, K is the factor of the crystal shape (0.94), λ is the wavelength of X-Ray radiation (0.1548 nm for Cu Kα radiation), β is the full width at half-maximum and θ is the peak position.

Fig. 4
figure 4

XRD spectra of studied photocatalysts, A anatase, R rutile (Color figure online)

Full size image
Fig. 5
figure 5

The DFC photocatalytic removal from water a DFC concentration determined by UV–Vis spectrophotometer, b COD of treated wastewater containing DFC (Color figure online)

Full size image

Concluding the data obtained by XRD, SEM and BET TiO2: T2 and T3 possessed rutile phase and differed in SBET and Vp. The anatase photocatalysts also varied in SBET and Vp. T1 and T2 that beside the difference in rutile presence had the similar value of the other parameters. The same was observed for T2 and T4.

3.2 Photocatalytic test

3.2.1 DFC photoremoval

It can be observed (Fig. 5) that DFC was effectively removed from water using photocatalytic oxidation. The best changes in the water quality were observed during first 20–40 min of treatment and resulted in at least 65 % decrease of the concentration of studied pollutant.

The highest DFC c/c 0 decrease was determined using T1 photocatalyst but generally all anatases revealed similar activity. The presence of rutile in TiO2 slightly diminished the photocatalytic efficiency. COD of treated DFC water, however, revealed that mineralization of DFC over T3 preceded almost at the same level that photocatalytic oxidation during first 2 h of treatment. It may indicate that the mineralization was as fast as the DFC photoremoval. The general tendency of the removal of DFC and its photooxidation products over all studied photocatalysts was similar—the highest decrease was observed during first 40 min of treatment and the further photocatalytic oxidation did not result in the total elimination of studied pollutant. Taking into consideration the type of the photocatalysts the efficiency of DFC removal over the studied photocatalysts was decreasing in the following order: T1 > T4 > T2 > T3.

Some of end-products are expected to be quite recalcitrant to further decomposition. It may result from the presence of intermediates that absorb at studied range and, thus, misleadingly contribute to the organic content of the Dc curve. They may also adsorb onto the catalyst surface and, thus, do not contribute to the measured COD (He et al. 2011).

The tendency of DFC removal (described as COD decrease) over T2 was the deepest during 60–80 min and then the COD reduction was maintained at the same level indicating that produced by- and end-products of DFC photooxidation were stable and resistant to further oxidation. Although the tendency of COD reduction using T1, T3 and T4 was similar, the photocatalysts differed in final effectiveness—the results obtained over T1 (with the smallest crystals) were almost two times better then over T3. The presence of rutile in comparison to anatase expectedly diminished the final results however; the small fraction of rutile significantly accelerated the DFC decomposition. T1 that is described by the producer as anatase contains however a small addition of rutile, (XRD peak at 27.4) revealed the highest photoactivity. The key is though the small amount of rutile, not pure anatase structure.

The studies of Zhang et al. 2011 indicated that the self-sensitized photooxidation of pollutants may be observed. *OH, 1O2 and O *−2 were involved in the DCF photooxidation. The photodegradation of DCF was then the sum of the degradation by direct photolysis and self-sensitization.

The primary transformation product of diclofenac under the studied conditions was (8-chloro-9H-carbazol-1-yl)acetic acid (λmax = 254 nm). That was in agreement with the results obtained by Zhang et al. (2011), Martínez et al. (2011) or Qin et al. (2012). (8-chloro-9H-carbazol-1-yl)acetic acid may be obtained as the result of *OH addition to 2-(8-chloro-9H-carbazol-1-yl)acetic acid, the product of photocyclization to the monohalogenated carbazole, involving loss of chlorine and hydrogen atoms, followed by structural rearrangement resulting in the formation of a new ring.

The pH of DFC water was 7.34 and taking into consideration the pHpzc of different TiO2 that lies in the pH range of 4.5–7.0 (Chong et al. 2010), it can be observed that at pH > PZC(TiO2), the catalyst surface is negatively charged and repulse the anionic compounds in water. In the studied region of pH the surface charge density distribution for these TiO2 catalyst clusters TiOH is ≥80 % at 3 < pH < 10, according to the following water equilibrium equations (Eqs. (2) and (3)):

$$ {\text{At pH }} < {\text{ PZC}}:{\text{ TiOH }} + {\text{ H}}^{ + } \leftrightarrow {\text{ TiOH}}_{ 2}^{ + } $$
(2)
$$ {\text{At pH }} > {\text{ PZC}}:{\text{ TiOH }} + {\text{ OH}}^{--} \leftrightarrow {\text{ TiO}}^{--} + {\text{ H}}_{ 2} {\text{O}} $$
(3)

The solution pH affects the degree of dissociation of organic, the extent of adsorption in their ionized form and hence the observed reaction rates. Electrostatic favored adsorption results in sensitivity of TiO2 to pH adjustments. Electroneutral adsorption cannot be neglected being the dominant manner of adsorption of organic acids in solution (Friedmann et al. 2010). A rapid and simultaneous chemisorption of dissociated H+ and the deprotoned organic acid in the very close vicinity of the surface takes place. The exposed facets either of rutile (110) or anatase (100) containing dangling Ti and O atoms closed to each other make titania an optimum surface for electroneutral adsorption. The pH of the solution after treatment finally increased to the value 8.

3.2.2 CPL photooxidation

The photocatalytic photooxidation of CPL as the representatives of antibiotics is presented in Fig 6.

Fig. 6
figure 6

The photocatalytic removal of a CPL and b COD of treated water containing CPL (Color figure online)

Full size image

It can be observed that the best results were again obtained during first 40-60 min of treatment. The further irradiation still did not result in the total removal of studied pollutant from water. The decrease of the CPL concentration using anatase TiO2 was similar. During the first 60 min the removal of CPL proceeded almost linearly and then there were observed a small decrease of CPL concentration. It indicates that the products of photooxidation of CPL (glycolic aldehyde, dichloroacetamide and 4-nitrobenzaldehyde) are resistant to further photooxidation.

The c/c 0 reduction was the most effective over all photocatalysts except T2. It can be observed that the photooxidation of CPL proceeded faster over the smaller crystals of TiO2. COD of treated water containing CPL was also significantly decreasing during first hour of treatment. In case of anatase containing TiO2 there can be observed a change in the shape of the curve: after 45 min the COD reduction tendency changed from linear to logarithmic dependency indicating that in the first minutes the COD reduction was connected with CPL removal and then the COD decrease was combined with by-products photooxidation. The photocatalysts with enhanced surface area were more effective than their low surface area counterparts.

The lowest COD reduction however was determined using T1 and T2 photocatalysts indicating that the key role in the removal of CPL and its oxidation products plays the surface area. pH of the solution was 6.888 and at pH below pHpzc the surface of TiO2 is positively charged and adsorption by electrostatic forces occurs. After the treatment the pH value 7 was maintained.

3.2.3 E1 photoremoval

The photocatalytic removal of E1 was observed over all studied photocatalysts (Fig 7).

Fig. 7
figure 7

The photocatalytic removal of a E1 and b COD of treated water containing E1 (Color figure online)

Full size image

The lowest activity in E1 removal from water was determined for T2 photocatalyst. There was more than 60 % of initial concentration of E1 left without decomposition. The highest decrease of the E1 concentration during treatment was observed over T4. In case of this pollutant it can be also noted that the similar general tendency for low and high surface area photocatalysts can be observed. Although E1 was photooxidized, its oxidation products were stable and resistant to further photooxidation. Again there can be observed the two steps oxidation over all photocatalysts. The key time of photocatalytic oxidation was first 45-60 min. The obtained results may indicate that the decomposition of E1 was fast but the products of the photooxidation (e.g. ketohydroperoxide) are more stable for the further photooxidation.

The pH of E1 water was 7.798 and taking into consideration the pHpzc of studied materials adsorption of the pollutant possessed electroneutral characteristics. The pH value of finally treated wastewater dropped to 7. It was observed by Liu and Liu 2004, that pH = 7 was the most suitable for the photolysis of E1. The mechanism of photooxidation is described as the breakage and oxidation of benzene rings to produce compounds containing carbonyl groups.

3.2.4 Mt photoremoval

The presented in Fig 8 photocatalytic oxidation of Mt proceeded the most effectively using T3. This photocatalyst was the most suitable both for Mt photoremoval and COD reduction of water containing Mt. The lowest activity was obtained for T2. The presence of rutile and the highest surface area favored the Mt photooxidation. The curves describing the Mt removal using both anatases proceeded similarly. The photocatalysts with lowest SBET value revealed the lowest activity in Mt photocatalytic oxidation. The results of COD reduction indicated that the process proceeded with almost the same activity for all studied photocatalysts but during initial 60 min of treatment, still, T2 did not enhanced the photoremoval.

Fig. 8
figure 8

The photocatalytic removal of a Mt and b COD of treated water containing Mt (Color figure online)

Full size image

The pH of the water with Mt was 6.888 indicated that adsorption occurred via electrostatic forces. Final pH of treated water was 7.5.

3.3 Kinetics of photooxidation

For the determination of the kinetic data of Dc photooxidation a preliminary power law kinetic model was applied (Eq. 4) (Herrmann 2010):

$$ r = - \frac{dc}{dt} = kc^{n} $$
(4)

where r, c and t represent the rate of degradation, pollutants concentration and time respectively. Also k and n are the rate constant and reaction order.

From the dark phase measurements of different TiO2 photocatalysts (Martínez et al. 2011), the photocatalytic degradation of DFC using UV–Vis can be adequately described using a modified Langmuir–Hinshelwood kinetic model that when the surface of the catalyst is homogeneous, can be simplified to the first order kinetic model (Eq. 5):

$$ r = k_{app} \cdot c $$
(5)

where k app would be the apparent pseudo-first order kinetic rate constant. The initial concentration of the organic substrate c 0 was kept constant to simplify the heterogeneous photocatalytic process.

The rate constants per unit reactor volume k 1 , which includes the effect of radiation absorption, were obtained by fitting the experimental data from the relation (Eq. 6):

$$ - \ln (c/c_{0} ) = k_{1} t $$
(6)

where c 0 and c are initial and determined at time t (min) concentration, k 1 is the pseudo first-order rate constant (min−1). The k 1 was determined from the slope of the linear regressions obtained by plotting −ln(c/c 0 ) versus time. Since all experiments were carried at equal level of photon absorption this effect was the same for all photocatalysts. The apparent rate constants of the catalyst per mass unit (k w ) and per surface area unit (k s ) were estimated from Eqs. 7 and 8 (Dolat et al. 2012):

$$ k_{w} = \frac{{k_{1} }}{{c_{w} \cdot V_{r} }} $$
(7)
$$ k_{s} = \frac{{k_{1} }}{{S_{BET} \cdot c_{w} \cdot V_{r} }} $$
(8)

where c w was the concentration of the photocatalysts, S BET BET surface of photocatalyst and V r —reactor volume (0.3 dm3). Consequently, half-lives were calculated using Eq. 9:

$$ t_{1/2} = \frac{0.6931}{{k_{1} }} $$
(9)

According to the obtained results it can be deduced that the degradation of studied pollutants may be generally described by pseudo-first order kinetics (Table 3). That is with agreement with data obtained by Zhang et al. (2011) for DFC, Liu and Liu (2004) for E1.

Table 3 The kinetic data obtained for the studied photocatalysts: apparent rate constants of DFC, CPL, Mt and E1 degradation per unit reactor volume (k 1 ), per gram of catalyst (k w ) and per unit surface area of catalyst (k s ) and initial reaction rate (r 0 )
Full size table

The closer analysis revealed, however, that two steps during the process may be observed. The identification of the process during first 45 min of treatment needs further studies.

The determined k 1 values are changing in the following order:

DCF:

T1 > T4 > T2 > T3 with the difference between the lowest and the highest value about 100 %

E1:

T4 > T3 > T1 > T2 with almost three times higher k 1 value for T4 than T1

CPL:

T1 > T3 > T2 > T4 with k 1 determined for T4 was two times higher than k 1 for T2

Mt:

T4 > T1 > T2 > T3 with the similar value obtained for anatase TiO2

The presence of rutile has not always the detrimental effect on the removal of studied pollutants. For DFC and Mt removal the anatase structure favored the photocatalytic oxidation. The intermediate effect was observed for CPL and E1. The key however was not only the presence of rutile but rather surface properties of studied photocatalysts.

Taking into consideration that the process of heterogeneous photocatalysis depends on the total surface area and affordability of active centers on the photocatalyst surface, the relation of initial reaction rate (r 0 ) versus SBET was presented in Fig. 9.

Fig. 9
figure 9

The relation of initial reaction rate with the type of the photocatalysts (Color figure online)

Full size image

It can be observed that over T4 almost all pollutants were removed with similar initial reaction rate. It is worth noting that generally the initial reaction rate of the removal of DFC and CPL did not differ significantly. However, the linear dependency of r 0 with the SBET was not observed for any of the photocatalysts.

4 Conclusions

According to the obtained results it may be concluded:

  • The time dependence of treatment of studied pollutants revealed that the best changes in the water condition were observed during first hour of treatment and further prolongation of treatment time till 4 h. did not result in the total removal of pollutants because produced by- and end-products of photooxidation were stable and resistant to further oxidation what may suggest that, the photocatalytic treatment using described photocatalysts and rector configuration can not totally solve the environmental hazards arising from PPCPs;

  • The presence of rutile and large TiO2 crystals in TiO2 slightly diminished the photocatalytic efficiency of DFC removal;

  • The mineralization of DFC over T3 was as fast as the DFC photoremoval;

  • The photooxidation of CPL proceeded faster over smaller crystals of TiO2 and the photocatalysts with enhanced surface area were more effective than their low surface area counterparts;

  • The similar general tendency for low and high surface area photocatalysts can be observed during E1 removal;

  • The presence of rutile and the highest surface area favored the Mt photooxidation. The photocatalysts with lowest SBET value revealed the lowest activity in Mt photocatalytic oxidation.

  • The photooxidation of all studied pollutants followed the pseudo-first order kinetics.