Photoactive Zr-aromatic hybrid thin films made by molecular layer deposition

The principle of antimicrobial photodynamic therapy (PDT) is appealing because it can be controlled by an external light source and possibly the use of durable materials. However, to utilise such surfaces requires a process for their production that allows for coating on even complex geometries. We have therefore explored the ability of the emerging molecular layer deposition (MLD) technique to produce and tune PDT active materials. This study demonstrates how the type of aromatic ligand influences the optical and antimicrobial properties of photoactive Zr-organic hybrid thin films made by MLD. The three aromatic dicarboxylic acids: 2,5-dihydroxy-1,4-benzenedicarboxylic acid, 2-amino-1,4-benzenedicarboxylic acid and 2,6-naphthalenedicarboxylic acid have been combined with ZrCl4 to produce hybrid coatings. The first system has not been previously described by MLD and is therefore more thoroughly investigated using in situ quartz crystal microbalance (QCM), Fourier transform infrared (FTIR) and UV-Vis spectroscopy. The antibacterial phototoxic effects of Zr-organic hybrids have been explored in the Staphylococcus aureus bacteria model using a UVA/blue light source. Films based on the 2,6-naphthalenedicarboxylic acid linker significantly reduced the number of viable bacteria by 99.9%, while no apparent activity was observed for the two other photoactive systems. Our work thus provides evidence that the MLD technique is a suitable tool to produce high-quality novel materials for possible applications in antimicrobial PDT, however it requires a careful selection of aromatic ligands used to construct photoactive materials.


Introduction
An increasing range of the functional materials used in our daily life are based on interactions with optical radiation, both from a practical and sustainable approach. Photoactive materials and compounds are used in optoelectronic devices such as light-emitting diodes and solar cells, sensors, photocatalysts, and in biological systems. 1 One clinically established lightinduced method is photodynamic therapy (PDT). This method utilises compounds called photosensitisers that, in the presence of oxygen, are exposed to optical radiation of wavelengths matching the absorption spectrum of the target molecule. Following excitation, photochemical reactions take place generating reactive oxygen species (ROS) that can damage biomolecules. PDT is approved for skin pre-cancer and cancer treatments, and is also used as a diagnostic tool. 2 Moreover, this therapeutic procedure is applied clinically and experimentally to kill or damage Gram-positive and Gram-negative bacteria, viruses, fungi and parasites, in particular for dental periimplant infections. 3,4 Light-induced antimicrobial treatment could also be applied for decontamination and disinfection of various environmental surfaces to reduce risk of infection and transmission of pathogens. 5 Photoactive materials and surfaces containing immobilised photosensitizers can be designed with ultra-high precision on the atomic/molecular level through the atomic/molecular layer deposition (ALD/MLD) technique. ALD is an established method for coating substrates with high-quality uniform and conformal inorganic materials for microelectronics, photovoltaics, electroluminescent displays, among others. 6 It is based on sequential self-limiting gas-surface reactions that enables precise thickness and composition control of the thin lm. The MLD method expands on the ALD approach by allowing growth with larger building blocks to produce entirely organic or hybrid organic-inorganic lms. 7 Aromatic dicarboxylic acids are popular as organic linkers in MLD due to their rigid structure and high reactivity. In addition, they provide interesting optical properties such as absorption in the UV and visible light region. [8][9][10][11] Furthermore, by combining various organic and inorganic components we can design the optical absorption and luminescence of such hybrid materials. We recently reported how the type of transition metal affects the optical properties of MLD hybrid materials based on 2,6naphthalenedicarboxylic acid as the linker. 12 For instance, the presence of d 0 -metals such as Ti, Zr or Hf shis the light absorption towards the visible part of the spectrum and increases the potential application of such hybrid materials as photoactive agents. Moreover, the light absorption properties can be further tuned by increasing the size of the aromatic backbone or by adding various functional groups. 13,14 In this work, we demonstrate a successful MLD growth of a novel photoactive hybrid thin lm material based on surface reactions of zirconium (ZrCl 4 ) and 2,5-dihydroxy-1,4benzenedicarboxylic acid (abbreviated here as 25D-BDC). The 25D-BDC aromatic linker has high near-UV absorption, which can be benecial for application in certain types of antimicrobial PDT where shorter wavelengths are needed. Zirconium, on the other hand, is a non-toxic biocompatible element, 15 while zirconium dioxide (ZrO 2 ) thin lms fabricated by ALD are shown to promote cell adhesion and spreading, but also some antibacterial activity. 16 We further explored the antibacterial properties of our photoactive hybrid lms based on Zr and three different aromatic dicarboxylic acids, i.e. the 25D-BDC linker described above, 2amino-1,4-benzenedicarboxylic acid (2A-BDC) and 2,6-naphthalenedicarboxylic acid . The experiments were performed in vitro using Staphylococcus aureus as a model organism. This Gram-positive bacterium is a major human pathogen and a leading cause of healthcare-associated infections. 17 S. aureus infections may be challenging to treat due to biolm formation, persister cells formation and antibiotic resistance, such as the methicillin-resistant S. aureus (MRSA). 18 PDT has been reported to have appreciable killing effectiveness on drug-resistant bacteria and claimed to be associated with a potential low risk of drug resistance development. 19 We foresee that uniform, photoactive antimicrobial coatings prepared by the MLD technique could be applied in PDT on small medical devices for external use and/or surgical instruments to prevent and treat bacterial infections such as those caused by the S. aureus bacteria. Furthermore, in the current work we show how the type of organic linker affects the light absorption properties of Zr-organic hybrid materials and thus their antimicrobial phototoxicity.

Characterization of the lms
In situ quartz crystal microbalance (QCM) analysis of the Zr-25D-BDC system was conducted using two 6 MHz gold-coated quartz crystals (Incon) placed ca. 5 cm apart along the ow direction using a homemade holder. The crystals were used to monitor the mass increase during the deposition and to optimize the precursor pulse and purge lengths. The signal was recorded using a Colnatec Eon-LT unit and processed by averaging over 16 consecutive cycles. The temperature was stabilized for 2 h before the experiments began to minimize any effects of uctuating temperatures. The thickness and refractive index of the lms were determined with an alpha-SE spectroscopic ellipsometer (J. A. Woollam) in the wavelength range of 390-900 nm at an incident angle of 70 . The data were tted to a Cauchy model using the CompleteEASE soware. The thickness of the native oxide layer was determined before the deposition process and added as an individual layer in the interpretation of lm thickness. Fourier transform infrared spectroscopy (FTIR) measurements of the Zr-25D-BDC lm were performed using a Bruker Vertex 70 spectrometer equipped with a Pike VeeMAX III specular reection accessory with an angle of incidence at 75 . Data were acquired in the reection mode using an average of 64 scans in the wavenumber range of 4000-370 cm À1 and 4 cm À1 resolution. Films for FTIR measurements were deposited on electropolished steel substrates. A spectrum obtained from the uncoated electropolished steel substrate was used as the background during the measurements. FTIR measurement of the organic compound was performed using the same instrument in the transmission mode. The pellet was prepared by mixing approximately 1 mg of the sample with 100 mg KBr and pressed into a pellet in a hydraulic press. UV-Vis transmittance spectra of the lms were collected with a Shimadzu UV-3600 UV-Vis-NIR spectrophotometer in the wavelength range of 200-850 nm with 1 nm resolution. Water contact angle measurements were performed using a Theta Lite optical tensiometer (Biolin Scientic). For each measurement, a 4 ml droplet of distilled water was administered by a syringe and placed on the surface of the lm. The contact angle was measured for 60 s at room temperature.

Antibacterial phototoxicity experiments
Light source. The irradiation chamber (Polylux PT, Dreve) was equipped with two compact uorescent tubes emitting in the UVA (Ralutec 9W/78, Radium) placed on each side of one blue light tube emitting blue light (Osram 9W/71). The combined emission wavelength range was 350-500 nm, with the maximum emission at around 368 nm (UVA) and 447 nm (blue) (irradiation chamber conguration: UVA-blue-UVA). Emission spectra were determined with a calibrated double monochromator spectroradiometer (model DTM300, Bentham Instruments Ltd) and the irradiance was monitored with regular intervals with a UDT 271 radiometer (United Detector Technology) calibrated towards the spectroradiometer. The radiometer was equipped with probes sensitive in the blue (268BLU S/N 23476, calibrated at 450 nm) and UVA (268UVA S7N 8U021, calibrated at 365 nm) parts of the spectrum. The average irradiance of the light tubes (stable aer 20 min) was 9 mW cm À2 (AE5%).
Phototoxicity on planktonic bacteria. Stock cultures of Staphylococcus aureus (strain Newman) were stored at À80 C in brain heart infusion (BHI, CM1135, Oxoid Ltd) supplemented with 30% glycerol. The fresh cultures were prepared by growing S. aureus overnight (ON) for 18 h in BHI in an incubator (Panasonic MCO-19M) at 37 C and 100% humidity under normal atmospheric conditions supplemented with 5% CO 2 . For ON culture preparation, the sterile inoculating loop was dipped in the stock cultures and transferred into 10 ml BHI. The ON cultures were centrifuged at 5000 g for 5 min and resuspended in 0.9% saline solution (NaCl, Sigma-Aldrich, >99%) to an optical density at 600 nm (OD 600 ) of approximately 1.0 using a spectrophotometer (Thermo Scientic Spectronic 200E spectrometer). OD 600 of 1.0 corresponds to about 10 8 colony forming units (CFU) ml À1 . The bacteria suspension was then diluted 1 : 10 in 0.9% NaCl to achieve a desired bacteria concentration of approximately 10 7 CFU ml À1 . Thin lms for antibacterial assays were deposited on microscope cover glasses (ø ¼ 20 mm, VWR) precleaned with soap, water and ethanol followed by 8 min plasma treatment (PDC-002-CE plasma cleaner, Harrick Plasma). Films and uncoated cover glass controls were placed in two 12-well plates (Costar, Corning Inc.) and exposed to 1 ml of prepared bacteria suspension followed by 90 min incubation (37 C, 100% humidity and 5% CO 2 ). During the procedure, the samples were protected from light using aluminium foil. Half of the samples were placed in the irradiation chamber and irradiated with UVA/blue light for 20 min with a light dose of 11 J cm À2 . A 30 min post-irradiation incubation time was applied to ensure completion of dark toxicity reactions following phototoxicity reactions. The samples were removed from the wells and transferred to 50 ml Falcon tubes containing 10 ml glass beads 3 mm in diameter (Merck) and 5 ml of 0.9% NaCl and vortexed for 30 s. The suspension was serial diluted in phosphate buffered saline (PBS, Bio-Whittaker, Lonza) and plated onto BHI agar plates (Agar bacteriological No. 1, Oxoid Ltd). The supernatant from the wells was also serial diluted and plated onto BHI agar plates. Aer approximately 24 h incubation (37 C, 100% humidity and 5% CO 2 ) the colonies were counted, and CFU ml À1 was calculated. For the Zr-26-NDC samples, three independent experiments were performed in triplicate (n ¼ 9), whereas for the Zr-25D-BDC and the Zr-2A-BDC samples, two independent experiments were performed in triplicate (n ¼ 6). To verify the inoculum concentration, the OD 600 and the diluted 1 : 10 bacteria suspension were plated onto BHI agar plates during each experiment.
Phototoxicity on biolm. The Zr-26-NDC samples were also tested against bacteria established as a 24 h biolm. For biolm assays, ON culture of S. aureus was diluted 1 : 100 in BHI. Films and uncoated cover glass controls were placed in two 12-well plates and exposed to 1.5 ml of prepared bacteria suspension followed by 24 h incubation (37 C, 100% humidity and 5% CO 2 ). The BHI was then removed from the wells and replaced by 1 ml of 0.9% NaCl. Half of the samples were placed in the irradiation chamber and irradiated with UVA/blue light for 20 min with a light dose of 11 J cm À2 . Aer irradiation, the samples were placed back in the incubator for 30 min. The samples were removed from the wells and transferred to 50 ml Falcon tubes containing 10 ml glass beads 3 mm in diameter and 5 ml of 0.9% NaCl and vortexed for 30 s. The suspension was serial diluted in PBS and plated onto BHI agar plates. The experiment was performed in duplicate with three independent parallels (n ¼ 6). The established biolm was veried with a confocal laser scanning microscopy (CLSM, Olympus FluoView FV1200) using a diode laser emitting at 473 nm. The SYTO 9 staining (LIVE/DEAD™ Viability Kit, Thermo Fisher Scientic) was used to colour the biolm on the surface. The scanning was performed using a 60 Â water lens, 0.5 mm step size and a format of 512 Â 512 pixels corresponding to an area of 88 Â 88 mm.
The antibacterial properties were analysed using one-way analysis of variance (ANOVA) method and post-hoc t-test with Bonferroni correction. The signicance level was set to p < 0.05. The bacterial reduction R was calculated in percent from treated sample (A) compared to untreated control (B): (1)

Film growth
The growth dynamic and the saturation behaviour of the new Zr-25D-BDC system based on zirconium tetrachloride and 25D-BDC was initially investigated at 260 C using an in situ QCM technique. Fig. 1 shows the growth rate as a function of inorganic and organic precursor pulse and purge length. The QCM experiment was performed by sequentially changing the pulse and purge length one at a time while keeping the other parameters constant at the standard pulsing sequence. The standard pulsing sequence used throughout the QCM experiment was 4 s ZrCl 4 pulse, 3 s N 2 purge, 5 s 25D-BDC pulse and 3 s N 2 purge (termed here 4-3-5-3 s). We observed that for both organic and inorganic precursors, a 4-5 s pulse is sufficient to reach a complete saturation, and moreover, this experiment proves a self-saturating growth for both precursors. Fig. 2 The growth rate of the Zr-25D-BDC system was further studied in the temperature range of 240-320 C using 140 MLD cycles and the standard pulsing sequence of 4-3-5-3 s, see Fig. 3a. The highest growth-per-cycle (GPC) value of 9.6Å per cycle was observed at 240 C. The GPC then decreases with increasing reaction temperature to 7.1Å per cycle at 260 C and remains nearly constant up to 320 C showing an apparent MLD temperature window for this process between 260-320 C. The refractive index slightly decreases with deposition temperature from 1.848 at 240 C to 1.807 at 320 C. Furthermore, the depositions within the entire studied temperature range yielded visually smooth and homogeneous thin lms. For the rest of the experiments, we decided to x the deposition temperature at 260 C. Finally, we conrmed that the lm thickness increases linearly with increasing number of deposition cycles as expected for an ideal ALD/MLD growth (Fig. 3b). The extrapolated graph in Fig. 3b intersects the x-axis at around 2.57 nm, supporting a near ideal growth without nucleation barriers. The Zr-25D-BDC lm is amorphous as conrmed by X-ray diffraction (XRD) and exhibits low surface roughness as measured by atomic force microscopy (AFM) with the root-mean-square  (RMS) roughness of 0.2 nm. The density of the lm was estimated to be 2.04 g cm À3 from X-ray reectivity (XRR) measurements.

Bonding characteristics
Fourier transform infrared spectroscopy (FTIR) was used to conrm the presence of the organic moiety in the deposited Zr-25D-BDC lm and its coordination to the metal. In Fig. 4 we show FTIR reection spectra of the as-deposited Zr-25D-BDC lm and the same lm aer 5 days of air-storage at ambient atmosphere of 25 C and around 70% humidity. For comparison, the spectrum of the corresponding carboxylic acid precursor is included as reference. The dominant absorption peaks seen at 1568 and 1448 cm À1 at the spectra of the lms correspond, respectively, to asymmetric (n as ) and symmetric (n s ) stretching vibrations of the carboxylate group COO À . 8 These peaks are slightly shied towards the higher wavenumbers as compared to the pure organic precursor, which suggests a reaction with the metal component. The separation between asymmetric and symmetric stretching of the COO À varies depending on how the carboxylate group coordinates to a metal atom. A splitting value, D, in the range between 50 and 150 cm À1 is typical for bidentate (chelating) complexes, monodentate complexes show a separation of D > 200 cm À1 , and bridging complexes have D values between 130 and 200 cm À1 . 21 Given this information, the D ¼ 120 cm À1 for Zr-25D-BDC lm indicates a bidentate type of coordination, and a possible structure of this Zr-hybrid is shown in the inset of Fig. 4. A similar coordination mode was previously observed for the Zr-2A-BDC and Zr-26-NDC lms. 12,20 The absence of a sharp intense peak at around 1650 cm À1 from the C]O stretching vibrations of free COOH groups in the lm proves that it coordinates to the zirconium atom during formation of the hybrid structure. 10 Similarly, from the spectrum of as-deposited lms, the absence of the broad absorption between 2400 and 3500 cm À1 with the maximum at around 3080 cm À1 arising from stretching vibrations of OH groups from COOH indicates that the oxygen is deprotonated and bonded to the metal. 22 The lms, however, absorb water into the structure as it can be seen as a broad band between 3670 and 2940 cm À1 in the spectrum of the lm aer 5 days of storage in ambient atmosphere. Simultaneously, the bands from the carboxylate group are broadened without any apparent changes in the coordination mode.
The spectra of the Zr-25D-BDC lm also show the absorption at around 1210 cm À1 , which corresponds to C-O stretching vibrations of carboxylic acid, the absorption at around 910 cm À1 is due to the out-of-plane bending vibrations of OH groups, and the double absorption peak at around 820 cm À1 is due to the  out-of-plane bending vibrations of C-H bond in the aromatic ring of the 25D-BDC molecule. 23,24 The broad character of the peaks from asymmetric and symmetric stretching vibrations of the carboxylate group can also suggest an amorphous character of the deposited lms. Furthermore, no changes in the type of coordination mode nor possible decomposition of the lms were observed in the FTIR spectra of the Zr-25D-BDC lm deposited at different temperatures from 240 C to 320 C (data not shown).

Photoabsorption characteristics
The transmission spectra obtained by UV-Vis spectroscopy of the hybrid lms with different thicknesses, corresponding to the number of MLD cycles (n ¼ 70, 140, 170 and 210), and the solution spectrum of the 25D-BDC precursor are shown in Fig. 5. The strong band located at around 375 nm for 25D-BDC precursor is attributed to p-p* intraligand transition in the aromatic ring of the molecule. For the hybrid Zr-25D-BDC thin lms the absorption peak has been redshied toward the longer wavelengths with about 40 nm due to interactions between the organic ligand and a Zr atom. The lms show strong absorption in the visible region with the absorption onset occurring below 500 nm and l max ¼ 410 nm for thin lm with n ¼ 210.
The optical properties of the Zr-25D-BDC lm were further compared with systems based on alternative aromatic dicarboxylic acids, i.e. Zr-BDC (1,4-benzenedicarbozylic acid), Zr-2A-BDC (2-amino-1,4-benzenedicarboxylic acid) and Zr-26-NDC (2,6-naphthalenedicarboxylic acid). The possible structures of the Zr-organic hybrid lms evaluated in this work are presented in Fig. 6. The UV-Vis transmission spectra of the Zr-organic hybrid lms show that the position of characteristic p-p* transition for aromatic rings strongly depends on the choice of the organic component in the hybrid lm (Fig. 7). The Zr-BDC lm is fully transparent in the visible and near-UV region, showing the absorption onset at around 315 nm. The presence of amino and hydroxyl groups in the aromatic linker alters the absorption properties of the lms. These functional groups inuence the conjugated system of 1,4-benzenedicarbozylic acid and cause a shi of the absorption peaks towards longer wavelengths. This is particularly evident when two OH groups are present. The Zr-2A-BDC system also shows absorption in the visible region, with the onset around 450 nm. In contrast, the naphthalene ligand with an extended conjugated system causes much weaker absorption in the visible region with the onset around 400 nm and l max ¼ 360 nm. Most importantly, the presence of the intense absorption in the UVA and visible region of the complexes suggests that these materials can be potentially used in antimicrobial photodynamic therapy, depending on the optical penetration depth required for various treatments.

Wetting characteristics
Wetting properties of the surfaces can inuence antibacterial behaviour and cell attachments to a certain degree, but its measured value can oen be precluded by a slowly dissolving lm. 26,27 The effect of water on surface properties was studied by submerging the lms in distilled water for 3 h (Table 1), the same duration as required for antibacterial assays and bacteria adhesion to the surfaces. The thickness of the lms slightly decreases aer 3 h of water treatment of about 6% for the Zr-26-NDC, 3% for the Zr-2A-BDC, and 1% for the Zr-25D-BDC lm. This reduction is rather low and proves that the lms are sufficiently stable in aqueous solutions. Furthermore, no visible changes were observed on the lm surfaces aer water treatment.
The wettability of the Zr-25D-BDC, Zr-2A-BDC and Zr-26-NDC lms was further investigated by measuring the contact angle of water on the surfaces (Fig. 8). The wetting on a silicon substrate with a native oxide layer was measured for comparison. All lms exhibit a rather hydrophilic nature, with contact angles of 58.5 , 63.7 and 79.0 for Zr-25D-BDC, Zr-2A-BDC and Zr-26-NDC, respectively. The wetting of the Zr-26-NDC system appears to be more shied towards a hydrophobic character as compared to the other systems. This indicates that the surface wettability depends on the surface chemistry and organic component in the hybrid lms. This can be related to different polarizability of the functional groups and thus the surface free energy, but also to the size of the conjugated system. The long 26-NDC linker decreases the wettability of the surface (higher contact angle). Furthermore, the presence of -NH 2 , and -OH groups can increase the polarizability and thus its wettability, 28,29 which is consistent with the measured contact angles for Zr-2A-BDC and Zr-25D-BDC. Additionally, we show an evolution of the contact angle versus time. The contact angle for each surface exhibits a rather small change of about 2.4 and reaches equilibrium aer around 45 s, supporting that the lms are stable in water solution.

Antibacterial properties
The bacterial phototoxicity of photoactive Zr-2A-BDC, Zr-25D-BDC and Zr-26-NDC samples with relatively similar   thicknesses of about 110 nm was tested against S. aureus in planktonic form. The number of viable bacteria was measured both for surface-attached bacteria aer 90 min of incubation (Fig. 9a) and for free-oating bacteria in the supernatant above the lm surface from the same experiment (Fig. 9b). The phototoxicity studies were performed using a combination of UVA and blue light irradiation sources with the emission that overlaps to a great extent with the absorption spectra of all tested lms, see Fig. 7. In addition, the hybrid thin lms show high stability under UVA and blue light irradiation (20 min) as proved by the UV-Vis measurements and thickness analysis by spectroscopic ellipsometry (data not shown).
The irradiated Zr-26-NDC surface reached around 99.91% reduction of the attached bacteria and 99.97% reduction of free-oating bacteria in the solution above the surface as compared to non-irradiated control samples. Surprisingly, no signicant reduction was observed for Zr-25D-BDC and Zr-2A-BDC samples both for surface-attached and free-oating bacteria. This indicates that the longer 26-NDC molecule with a larger conjugated system has superior antibacterial activity than the shorter, one ring 2A-BDC and 25D-BDC molecules, even though the absorption is brought into the visible range. Two possible factors may impact the PDT efficacy, the reactive oxygen species quantum yield and/or type of generated species that enters the antibacterial process. 30,31 ROS can be generated via type I (superoxide anion radical O 2 c À , hydrogen peroxide H 2 O 2 or hydroxyl radical HOc) or type II (singlet oxygen 1 O 2 ) reaction mechanism, whereas most photosensitizers used in PDT are believed to operate via type II rather than type I mechanism. 2 We have not identied the type of generated ROS in our Zr-26-NDC system, but we can assume the formation is mostly of singlet oxygen, which is generally considered as the primary damaging species in PDT. 32 This implies a higher 1 O 2 quantum yield, which means the higher number of singlet oxygen molecules produced per absorbed photon for the Zr-26-NDC system as compared to the Zr-2A-BDC and Zr-25D-BDC systems. On the other hand, the photosensitization process in Zr-26-NDC system may also involve mixed type I and type II reactions with formation of highly reactive hydroxyl radicals. To pinpoint these mechanisms remains for further studies.
It is also worthwhile noting that not only the size of the conjugated system but also the type of the functional group and the polarity characteristic of the ligand can impact interaction and attachment of the molecule to the Gram-positive bacteria, and thus the antibacterial activity of the created systems. Moreover, the 2A-BDC and 25D-BDC light-absorbing molecules can be ineffective photosensitizers in PDT due to the possible internal conversion and radiationless decay of these molecules to the ground state without emitting a photon.
Another possible explanation for lack of phototoxic activity of the Zr-2A-BDC and Zr-25D-BDC lms is the level of overlap between the light absorbance of these lms and emission from the lamp that can affect the ROS quantum yield (Fig. 7). The Zr-NDC lm overlaps with the irradiation source only in the UVA region, while the Zr-2A-BDC and Zr-25D-BDC lms both in the UVA and visible part of the spectrum. However, no phototoxicity of the Zr-25D-BDC complex was observed when a different light source conguration was used consisting of two side blue light tubes and one middle UVA tube giving the average irradiance of 12 mW cm À2 (AE5%) (data not shown), with relatively similar emission spectrum. The light dose of this conguration was higher than in the previous conguration giving about 14 J cm À2 aer 20 min of irradiation. Increasing the irradiation time and thus radiant exposure even higher could potentially increase the phototoxic effect of these samples. Importantly, no signicant antibacterial effect was observed for the samples without light irradiation (dark toxicity), pointing to high stability and apparent lack of toxicity of the created thin lms.
The antibacterial activity of our Zr-26-NDC sample was higher than those reported by Lausund et al., 11 where they show about 50% reduction of viable bacteria. It should be noted, Fig. 10 (a) Confocal laser scanning microscopy image of S. aureus biofilm grown for 24 h on Zr-26-NDC film. Biofilm was stained with DNAbinding dye SYTO 9 and cells are green. (b) Phototoxic effect against S. aureus biofilm after exposure to Zr-26-NDC film (n ¼ 6) and uncoated coverslips as controls (n ¼ 4) combined with (+) UVA/blue light irradiation (11 J cm À2 ). A statistically significant difference from all other samples is marked by a red asterisk, p < 0.05. however, that in the present study a different bacterial species and a different UVA light source were used.
The antibacterial activity of the Zr-26-NDC sample was further investigated against S. aureus bacteria in a biolm. The formation of a 24 h biolm on the surface was rst veried with confocal laser scanning microscopy (CLSM) (Fig. 10a). The CLSM images conrmed the formation of a relatively dense biolm on the lm surface. The Zr-26-NDC surface in combination with UVA/blue light irradiation (11 J cm À2 ) was effective against established biolm (p < 0.05) (Fig. 10b). Around 57% reduction of viable bacteria was observed for irradiated Zr-26-NDC (+) sample as compared to the irradiated controls, and around 48% reduction as compared to the non-irradiated controls. Increasing the irradiation dose to about 16 J cm À2 (30 min of irradiation) could potentially increase the phototoxic effect against created biolm. Moreover, no reduction of viable bacteria was observed for non-irradiated Zr-26-NDC (À) samples, which proves lack of cytotoxicity and high stability of this system in bacteria solution.

Conclusions
In this work, we have developed a new MLD process for fabricating photoactive hybrid thin lms based on Zr clusters and aromatic 25D-BDC organic linkers. The amorphous and uniform lms display high absorption in the visible part of the spectrum. Moreover, we show that the type of the organic ligand and its functional groups can be used to tune the optical properties of Zr-organic hybrids, as shown by the Zr-2A-BDC and Zr-26-NDC systems in this work. The presence of amino -NH 2 and especially hydroxyl -OH functional groups on the aromatic linker broadens and shis the optical absorption of the lms towards the visible part of the spectrum. Our results also indicate that the type of the organic ligand changes the wetting properties of the surfaces. All lms are hydrophilic, while the Zr-26-NDC lm is more towards a hydrophobic nature. The antibacterial properties of the lms are dependent on the type of ligand used. The phototoxicity experiments against S. aureus Gram-positive bacteria reveals a lack of activity for the Zr-25D-BDC and Zr-2A-BDC systems, whereas a high activity is observed for the Zr-26-NDC systems, both against planktonic bacteria and bacteria in biolms. Antibacterial activity, absence of dark toxicity and high stability in contact with water of Zr-26-NDC system indicates potential for application of this hybrid material in photodynamic therapy.

Conflicts of interest
There are no conicts to declare.