Submerged surfaces exposed to marine biofouling – Experimental investigation of cleaning parameters effectiveness

Physical cleaning, an important factor in maintaining ships' hulls free from biofouling , can help to prolong the lifetime of fouling-control coating systems. This study investigated how mechanical cleaning parameters (cleaning force, duration and frequency) affected antifouling coatings' performance including fouling resistance property and cleaning efficiency using a specially designed manual cleaning system. Meanwhile, coating surface conditions, including visual appearance condition, roughness, contact angle were evaluated in detail. Biocide release rates of copper-based soluble antifouling paint (Cu-AF) under different cleaning conditions were also assessed by Scanning Electron Microscopy - Energy Dispersive X-ray (SEM-EDX) analysis. Results from the field testing in Danish sea area showed that cleaning frequency affected more than cleaning force and duration on antifouling coating performance. Algae glue attachment increased the difficulty in the cleaning of non-coated acrylic surface (A-BS) and fluorine-based copper free insoluble antifouling paint (F-AF). For Cu-AF coated panels, no significant mechanical damages were observed after cleaning. For F-AF, scratches formed on the coating surface after cleaning were more related to cleaning force and cleaning frequency. The biocide release rates of Cu-AF under various cleaning condition were between 15.7 and 24.6 μ g Cu cm (cid:0) 2 day (cid:0) 1 . Gentle cleaning of Cu-AF with proper frequency has little influence on the copper release rate. This study provides guidance on cleaning strategies for antifouling coatings.


Introduction
Marine biofouling, which refers to the undesirable accumulation of organisms on underwater structures, has been a severe problem around the world. Biofouling on ship hulls will result in high frictional resistance, higher emissions of exhaust gases, need of surface cleaning and maintenance costs, and other economic losses on fuel consumption [1,2]. In addition, some invasive or non-native species may be introduced to new sea areas by attaching to the surface of ship hulls, causing challenges to the global marine ecology [3]. The application of foulingcontrol coatings is so far the simplest and most widely used method for biofouling prevention [4][5][6][7][8][9]. Besides, underwater cleaning, which is the physical removal of biofouling done by divers or remotely operated vehicles (ROVs) from the underwater structures, has also been advised as an important part to prevent biofouling [10][11][12]. During the cleaning process, coating damages, wears or coating detachment from the substrate are normally not avoidable resulting in adverse effects on the service performance of coatings systems. Therefore, deeply understanding the interaction between the underwater cleaning process and different coatings' performance is propitious to give guidance for a cleaning strategy.
Physical removal of biofouling by underwater cleaning methods is efficient and cost-effective so mechanical cleaning is usually considered for long-term static underwater structures [13]. Various methods and technologies for underwater cleaning have been reported, which can be divided into mainly three categories: manual hull cleaning (scrubbing by divers), powered rotary brush cleaning systems, and noncontact cleaning technology, such as water jetting, ultrasonic cleaning and laser cleaning [10]. Practical hull cleaning is conducted either by divers using rotating-brush carts, or using ROVs equipped with rotating brushes or waterjets as cleaning devices [14]. Among these techniques, a powered rotary brush cleaning system is the most widely used method and easier to achieve and operate except that it may cause damage to coating surface. Water jetting causes little damage to the coating system but it needs higher operating pressure, the costs are higher and it is difficult to control for cleaning. Ultrasonic cleaning and laser cleaning have no large-scale commercialization yet at present. Although underwater vehicle can provide high cleaning efficiency without potential personal injury, its price is prohibitive for many applications [10,15]. Taking all aspects into considered, powered rotary brush system is suitable for underwater cleaning for laboratory investigations.
Hull grooming, defined as frequent and gentle mechanical wiping of the hull, has been suggested as a viable maintenance approach to decrease the harmful effects on removing fouling organisms [16][17][18]. Previous research on underwater cleaning and hull grooming on marine coatings mainly focused on the investigation of effects of hull grooming on different coating types both in short-term small-scale panels and long-term large-scale panels, fouling adhesion studies among fouling release coatings, and biocide release rating of antifouling coatings. Researchers in Florida Institute of Technology have done a series of work on underwater cleaning and hull grooming. Tribou and Swain [16] performed 3, 6, 12, and 24-day grooming intervals on an antifouling coating, a fouling release coating, a two-component epoxy coating, and polytetrafluoroethylene plastic using cleaning sponge for 120 days, and then proposed that frequent grooming could prolong coatings' lifetime. Long-term mechanical grooming on large-scale test panels coated with antifouling and fouling release paints by a five-head rotating brush was conducted subsequently [17][18][19]. Besides, leaching rate of copper from a copper ablative paint after 6-year grooming and grooming effects on biofilm community and hydrodynamic drag were also investigated, which suggested that grooming is a viable method for maintaining copper ablative coatings in a fouling-free condition without adverse increases in the total copper output [20,21]. Oliveira and Granhag [22] determined minimal cleaning forces on antifouling and fouling release coatings without damage or wear on coating surface using a specially designed immersed waterjet. However, there is no report yet on how exactly mechanical cleaning parameters affect coating performance and coating surface condition during and after cleaning process.
In this study, effects of mechanical cleaning parameters, including cleaning force, cleaning duration and cleaning frequency on a soluble copper based antifouling coating (Cu-AF), an insoluble fluorine-based copper free antifouling coating (F-AF) and the non-coated acrylic surface (A-BS) were investigated in detail using a self-designed bench drill rotating brush system at the CoaST Maritime Test Center (CMTC), the Technical University of Denmark (DTU). At the end of study, the coating surfaces of the two antifouling paints were characterized and the release rate of copper from Cu-AF was evaluated. This study is dedicated to better comprehending how mechanical cleaning parameters influence the performance of different coating systems during and after cleaning process, which will put forward a guidance to industrial hull cleaning development and cleaning strategy making.

Test panels preparation
Acrylic panels (10 × 20 × 0.3 cm), from RIAS, Denmark, were used as substrates in this study. Three sets of test panels were prepared: the first set of panels was coated with a soluble copper antifouling coating product, marked as Cu-AF panels; the second set of panels was coated with an insoluble fluorine-based copper free antifouling paint, marked as F-AF panels; The final set of panels was non-coated acrylic panels, marked as A-BS panels. In order to prevent fouling settling on the back side of acrylic panels, a three-layer commercial coating system was applied on the rear side by a spiral film applicator: a two-component polyamide adduct cured epoxy primer at the DFT of ~50 μm, a tiecoat at the DFT of ~50 μm and a silicone-based fouling release top coating at a specified dry film thickness (DFT) of ~100 μm.
Cu-AF panels were coated with a two-layer coating system by airless spray. The top coat was a red colored self-polishing coating, which contains cuprous oxide (Cu 2 O) as the main biocide and zinc oxide as the co-biocide, at a specified DFT of ~200 μm, over an epoxy primer at the DFT of ~50 μm. F-AF panels were coated with a two-layer coating system by roller. The top coat was brown colored at a specified DFT of ~50 μm, with an underlying epoxy primer at the DFT of ~50 μm. A-BS panels were blank acrylic surface without any coating systems on the front side. All coating systems were applied according to the manufacturer's specifications. Two panels were prepared for reproducibility check under each cleaning treatment. Details of paint specifications are listed in Table 1. The composition of test paints is listed in Table S1.

Field testing
The field exposure and testing was carried out at the CoaST Maritime Test Centre (CMTC), as shown in Fig. 1. CMTC is located in Hundested harbour, north coast of Denmark (55 • 57 ′ 57.8 ′′ N, 11 • 50 ′ 33.8 ′′ E). All test panels were deployed at the static floating raft under similar light exposure.
The seawater depth at the field site ranged between 6.5 and 7 m, depending on the tide (± 0.3 m, data from https://hundestedhavn.dk/). In order to eliminate the deployment depth effect, test panels under the same condition were placed with the depth varying from 0.15 m to 0.65 m under the waterline. During the deployment period (from May to November 2021), environmental parameters including seawater temperature, salinity and pH were monitored weekly by a pH meter equipped with an electrode (MU 6100H, VWR®, Germany, 0.01). The average values were: temperature 17.9 ± 3.2 • C, salinity 17.4 ± 0.7 ppt and pH 7.8 ± 0.1.   Fig. 2 shows a schematic diagram of the cleaning system. It is a selfdesigned bench drill brush system comprised of a bench drill (PBD 40, BOSCH, Germany), a rotating drill brush and an energy reader (EMD-3, SARTANO®, Italy). The drill brush is a general-purpose nylon brush. The brush was 45 mm in diameter and 25 mm in length with vertical nylon bristle tufts. The original photo is shown in Fig. S2.

Cleaning system and time schedule
Cleaning with controlled cleaning parameters of cleaning force, cleaning duration and cleaning frequency was performed on testing panels. In order to quantify the cleaning force, normal force related to bristle depth and shear force related to rotating speed were converted on the testing panels. Weekly cleaning was performed on the left-half side of each panel and bi-weekly cleaning was done on the right-half side. The cleaning duration was the length of time staying on one cleaning area. The panel was moved at a step size of ~2.4 cm (radius distance of cleaning area) horizontally for continuous cleaning.
In this study, different coating systems and cleaning treatments are identified by codes as: top coat -Cu-AF: soluble copper-based antifouling, F-AF: insoluble fluorine-based copper free antifouling, A-BS: non-coated acrylic surface; cleaning force -F 1 : normal force 7.8 N at 740 rpm, F 2 : normal force 7.8 N at 240 rpm, F 3 : normal force 0.4 N at 240 rpm, F 4 : normal force 2.5 N at 240 rpm; cleaning duration -T 1 : 5 s, T 2 : 10 s and cleaning frequency -W 1 : weekly, W 2 : bi-weekly. For instance, case Cu-F4T1W1 corresponds to Cu-AF panels, cleaned with a normal force of 2.5 N at 240 rpm (F 4 ) at a cleaning duration of 5 s (T 1 ) on a weekly frequency (W 1 ). The codes of testing panels corresponding to controlled cleaning parameters were listed in Table 2. The time schedule for cleaning on testing panels was listed in Table 3. Fig. 3 shows the monthly fouling pressure on acrylic panel at the testing site -CMTC. The fouling pressure was monitored by observing organisms coverage attached to a clean acrylic panel through visual assessments. The acrylic panel was replaced monthly at the same position of the testing cartridge. The fouling coverage evaluation followed the ASTM D6990 [23] where the coverage of organisms in direct attachment to the acrylic panel surface was calculated. The fouling species detected in CTMC were biofilm, algae (Pilaiella littoralis and  Ectocarpus siliculosus, Ceramium sp. and Vertebrata sp., Ulva sp., etc.), barnacles (Balanus improvisus), tunicates (Ciona intestinalis and Botryllus sp., etc.), mussels (Mytilus edulis) and hydroids. The brown algae (Pilaiella littoralis and Ectocarpus siliculosus) was the dominated one in May. The brown algae turned to green (Ulva sp.) and red algae (Ceramium sp. and Vertebrata sp.) from June. Barnacles (Balanus improvisus) appeared at the end of June. Tunicates (Ciona intestinalis and Botryllus sp.) appeared in the middle of July and mussels (Mytilus edulis) appeared at the end of July. Hydroids appeared in October.

Fouling rating calculation
The level of fouling observed on immersed panels was rated based on the US Naval Ships' Technical Manual Fouling Rating [24], as shown in Table 4. This fouling rating (FR) relies on visual and photographic inspection, varying from 0 (foul-free surface) to 100 (all forms of fouling present).
The US Navy's FR was applied in this study, with the following two additions. The first addition was counting algae glue under the same category as tubeworms, FR 40. Algae glue is known as phenolic compounds produced by red and brown algae, which exhibit adhesive properties and extraordinarily high cohesive strength [25]. Such X X X X a All types of panels were deployed into the seawater on 6th May 2021. b 'X' stands for testing sample cleaning.
biological 'glues' binds non-specifically to both hydrophobic and hydrophilic surfaces in aqueous conditions to provide both temporary and more permanent adhesion [26][27]. The second addition was that juvenile barnacles less than 1 mm in diameter were considered as microfouling, and thus included under FR 20, as proposed by Dinis Reis Oliveira et al. [22]. Visual inspection of panels resulted in a variable number of FR values detected on each half panel. The percentage coverage was then estimated visually for each of these FR values, through comparison with standard extent diagrams from ASTM D6990-05. No overlapping was allowed, i.e. the maximum total percentage coverage was 100 %. Edge effects were discounted from the analysis by excluding 1 cm margins. Finally, FR values, and their corresponding percentage coverage were combined into a mean value, FR mean , which represented as an area mean for each half panel, as shown in Eq. (1) [22]: N stands for the total number of visible fouling ratings on a given half panel, FR i stands for fouling rating of a specific fouling organism, cover i stands for coverage of a specific fouling organism on testing area.

Physical condition evaluation
Coating surface was evaluated by visual inspection and Digital Microscope (Keyence, VHX-6000, Japan) on both newly applied and cleaned testing panels. Microscope images were gained through 20-200 x lens on each half side of testing panels.

Roughness analysis
Peak-to-valley height (R t, 50 ) of surface roughness of the testing panels was determined by 3D Optical Profilometer (Keyence, VR-3100, Japan) prior to deployment and at the end of study.

Wettability evaluation
Wettability evaluation of testing panels was achieved by water contact angle (WCA) using the sessile drop needle method on Drop Shape Analyzer (Krüss, DSA30, Germany) where a drop of MiliQ water (4 μL) was placed on the surface manually. All WCA reports are an average of at least three measurements on 13 different drop placements uniformly spaced on the testing surface.

Scanning Electron Microscopy -Energy Dispersive X-ray (SEM-EDX) analysis
Cu-AF coated panels were cut, embedded in epoxy resin, and polished for further observations of their cross-section by scanning electron microscopy (Thermo Scientific, Prisma E, UK) coupled with Energy Dispersive X-rays for imaging and chemical analysis. The size of withdrawn samples used for SEM observations is small (2 × 2 cm) and they could not reflect the whole coated panel. Therefore, three random samples were withdrawn from Cu-AF coated panel under each condition to obtain average values of the thickness of the coating before exposure or leach layer thickness or the thickness of the remaining coating after cleaning.
The embedding and polishing steps were shown as follows. The samples were set in a slow-curing transparent epoxy (CaldoFix-2 resin, Struers) in which the ratio of resin and hardener is 25:7 and cured for 48 h. The polishing step of the samples in cross-sections contained: (1) grinding with a wetted SiC Foil (US #500) for 2 min, (2) polishing with 9 μm water-based diamond suspension on polishing wheel for 3 min, (3) polishing with 3 μm water-based diamond suspension on polishing wheel for 18 min. After polishing, the specimens were sputtered by silver with 5 nm using Double-sputter Coater (Leica, EM ACE 600, Denmark) for SEM/EDX analysis.
SEM with an accelerating voltage of 20 keV in the high vacuum mode was used to take images. EDX line scan analysis from surface to substrate was used for selected elements. For each Cu-AF cross-section, three images were taken at random positions. Later, average values of DFT was analyzed by ImageJ.

Biocide release rate evaluation
Biocide release rates were determined by polishing rate massbalance method which is suitable for self-polishing coating [28]. In order to determine biocide release rates from the Cu-AF coated panels as a function of cleaning treatment, DFT was measured by SEM-EDX before and after deployment (DFT i and DFT e ). Biocide release rates are calculated as total biocide loss divide by exposure time.   The equivalent total biocide loss was determined by

Results and discussions
All test panels were immersed in seawater on 6th May 2021 before the fouling peak appeared at the testing site. The non-coated acrylic surfaces were given a conditioning (exposure to the sea) period of 4 weeks before the initiation of cleaning and the two kinds of antifouling surfaces were conditioned for 16 weeks because of the good antifouling property. The cleaning was performed on each testing surface for 12 weeks. During the cleaning process, the testing surfaces were evaluated in terms of their fouling resistance property (before cleaning) and cleaning efficiency (after cleaning). The fouling resistance property, described as the ability to prevent re-fouling, was semi-quantitatively displayed based on the mean fouling rating, FR mean (Eq. (1)). The higher FR mean , the worse fouling resistance property. Additionally, the cleaning efficiency, known as the cleaning ability of cleaning system, correlates to the remaining fouling coverage on the testing surfaces after each cleaning. The lower the remaining fouling coverage, the higher the cleaning efficiency. Results of the cleaning effect of the fouling condition on non-coated and coated acrylic panels are presented below.

Fouling resistance
The acrylic surface became significantly covered by biofilms within one week of exposure. After the conditioning period of four weeks, accumulations of small brown algae (Pilaiella littoralis and Ectocarpus siliculosus) appeared on the acrylic panels and barnacles (Balanus improvisus) started to grow four weeks later. Cleaning started after 4 weeks' exposure, and the week which the first cleaning took place was marked as week 1. Fig. 4 gives the cleaning parameters variation on the non-coated acrylic surfaces. FR mean values of control acrylic surface at week 1 were less than 20 but increased to more than 90 after week 11, as shown in Fig. 4 (a).
FR mean values on non-coated acrylic surfaces before cleaning with various cleaning parameters are shown in Fig. 4 (b)-(e). In the first four weeks' cleaning period, only soft fouling (e.g., slimes and algaes -Ceramium sp., Vertebrata sp. and Ulva sp.) was present on the acrylic surfaces and hard fouling (e.g. barnacles) started to grow in week 5. Comparing with the acrylic panel without cleaning, cleaning at both weekly and bi-weekly frequency resulted in a large decrease in FR mean values. In addition, weekly cleaning has a better fouling resistance property for acrylic surfaces than bi-weekly cleaning, suggesting that increasing the frequency of cleaning can help to prevent the buildup of fouling on acrylic surfaces.
As shown in Fig. 4 (b), there was not much difference in FR mean values when cleaning was performed with varying cleaning forces (F 1 and F 2 , F 1 > F 2 ) during the first 4 weeks of weekly cleaning. However, the FR mean values for acrylic surfaces cleaned with F 2 increased from week 5 compared with those cleaned with F 1 , indicating that acrylic surfaces cleaned with stronger force suggested better fouling resistance property when hard fouling appeared (week 5) at weekly cleaning. It showed the same for acrylic surfaces cleaned at bi-weekly frequency in Fig. 4 (d). Fig. 4 (c) showed that cleaning duration (T 1 , T 2 ) only had a slight effect on the fouling resistance properties of acrylic surfaces because of little differences between FR mean values with different cleaning duration at weekly intervals. However, for bi-weekly cleaning (Fig. 4 (e)), FR mean values for acrylic surfaces cleaned with T 1 increased from week 5 compared with those that were cleaned with T 2 . This suggested that longer cleaning duration enabled better fouling resistance properties of non-coated acrylic surfaces with longer cleaning intervals.

Cleaning efficiency
Cleaning efficiency of non-coated acrylic surfaces based on fouling coverage of different fouling types are shown in Fig. 5 (a)-(f). Biofilms, algae (Ceramium sp., Vertebrata sp. and Ulva sp., including algae glue), barnacles (Balanus improvisus) and mussels (Mytilus edulis) were present on non-coated acrylic surfaces after 12-week cleaning. It was found that the cleaning efficiency of the non-coated acrylic surfaces at weekly frequency was better than the efficiency of cleaning at bi-weekly frequency under any cleaning forces and duration, which indicated that cleaning non-coated acrylic surface weekly could improve the cleaning efficiency. After 12-week cleaning, algae glue was dominant on noncoated acrylic surfaces, which increased the difficulty of cleaning. Fig. S3 shows that algae glue still remains on some of A-BS and F-AF testing panels after cleaning. Therefore, algae glue was needed to be removed together with the algae at the early stage, or it would increase the difficulty on cleaning efficiency of mechanical cleaning with time.
With regard to weekly cleaning, a strong cleaning force F 1 for short cleaning duration T 1 can decrease the fouling coverage to less than 20 % at the end of 12-week cleaning, as shown in Fig. 5 (a). When the cleaning duration was increased to T 2 , the fouling coverage on non-coated acrylic surfaces fell below 10 % (Fig. 5 (c)). Therefore, longer cleaning duration had a positive effect on cleaning efficiency. It was noted that fouling coverage rapidly increased from week 7 with a small cleaning force F 2 due to higher water temperature, as shown in Fig. 5 (b). The coverage of algae (mainly algae glue) and barnacles sharply increased from week 7 and then remained unchanged after week 9. This can be explained by the fact that uncleaned algae glue and juvenile barnacles grew with the increasing temperature (16.7 • C to 22.5 • C) from week 1 to week 7 and occupied the surface after week 9 because of the strong adhesion. After that, algae glue and adult barnacles, which are difficult to remove, occupied the acrylic surfaces.
At bi-weekly cleaning frequency, the remaining fouling coverage of acrylic surfaces was around 60 % with a strong cleaning force F 1 for short cleaning duration T 1 after 12-week cleaning, as shown in Fig. 5 (d). When the cleaning duration was increased to T 2 , the fouling coverage on the acrylic surfaces decreased to 30 % (Fig. 5 (f)). However, the fouling coverage on acrylic surfaces was more than 90 % with a weak cleaning force F 2 (Fig. 5 (e)), which meant that the benefit of cleaning activity was lost.

Fouling resistance
Because of the good antifouling property of Cu-AF, only soft fouling (e.g. slimes and algae) accumulated on the Cu-AF control surface throughout the whole exposure period. FR mean values of Cu-AF control surface were between 20 and 30, as shown in Fig. 6 (a).
FR mean values for Cu-AF surfaces for various cleaning parameters are shown in Fig. 6 (b)-(e). The week which first cleaning of Cu-AF surfaces took place was marked as week 1. Compared with the control surface, cleaning at both weekly and bi-weekly frequency led to a decrease in the FR mean values to some extent. Additionally, weekly cleaning also resulted in better fouling resistance properties for Cu-AF surfaces than biweekly cleaning, especially after week 9.
In Fig. 6 (b) and (d), there was not much difference among the FR mean values for Cu-AF surfaces with different cleaning forces (F 3 and F 4 , F 4 ˃ F 3 ) for both weekly and bi-weekly cleaning, suggesting that the cleaning force only has a slight effect on the fouling resistance properties of Cu-AF surfaces compared with the cleaning frequency. The result is also the same with different cleaning duration (T 1 and T 2 ) on Cu-AF surfaces, as shown in Fig. 6 (c) and (e). In addition, FR mean values of Cu-AF surfaces decreased gradually from week 8 to the end of 12 weeks' cleaning with weekly cleaning. However, for bi-weekly cleaning, FR mean values were almost the same throughout the whole cleaning process. All these results indicated that in order to gain better fouling resistance properties of Cu-AF surfaces, cleaning interval is recommended to be kept shorter than 2 weeks.

Cleaning efficiency
Cleaning efficiency of Cu-AF surfaces is shown in Fig. 7 (a)-(f). According to the figure, fouling coverage on Cu-AF surfaces within the testing period was less than 3 % under weekly cleaning and appeared as fouling free surface after 12-week cleaning, as shown in Fig. 7 (a)-(c). The fouling coverage of testing panels with bi-weekly cleaning is lower than 5 % after 12-week cleaning. All these indicated the good cleaning ability of cleaning method. In addition, fouling coverage was not obviously increased with weak cleaning force or short cleaning duration with both weekly cleaning and bi-weekly cleaning, which proved that cleaning efficiency of Cu-AF was mainly related to cleaning frequency.

Fouling resistance
The F-AF control surface, which was covered by soft fouling (e.g. slimes and algae) and several barnacles during the whole exposure period, also had good antifouling property. F-AF surfaces and Cu-AF surfaces have the same starting date of cleaning process. FR mean values of F-AF control surface were around 30 throughout the whole cleaning process except for the period from week 5 to week 7, which may be related to the windy weather resulting to higher current speed to wash away soft fouling with less adhesion, as shown in Fig. 8 (a).
FR mean values for F-AF surfaces for various cleaning parameters are shown in Fig. 8 (b)-(e). Compared with the F-AF control surface, FR mean values decreased to below 20 for bi-weekly cleaning after 12-week cleaning, and even lower than 1.5 for weekly cleaning. The results suggested that F-AF surfaces for weekly cleaning showed excellent fouling resistance properties.
In Fig. 8 (b) and (d), F-AF surfaces were cleaned with different cleaning forces (F 3 and F 4 , F 4 ˃ F 3 ). It can be seen that FR mean values are almost the same before week 8 and has a slight difference thereafter, Fig. 8(b), indicating that stronger cleaning force F 4 led to a slight increase of fouling resistance of F-AF surfaces with weekly cleaning. However, FR mean values of F-AF surfaces cleaning with F 4 significantly decreased compared with those that cleaned with a weak force F 3 at biweekly frequency after week 8. It suggested that strong cleaning force increased the fouling resistance property of F-AF surfaces with the increase of cleaning intervals. A similar trend was observed with different cleaning duration (T 1 and T 2 ) on F-AF surfaces, as shown in Fig. 8 (c) and (e). All these results indicated that the advantages of stronger cleaning force and longer cleaning duration were present with the lower cleaning frequency to improve the fouling resistance of F-AF surfaces.

Cleaning efficiency
Cleaning efficiency of F-AF surfaces based on fouling coverage of remained fouling types after cleaning is shown in Fig. 9 (a)-(f). Biofilms, algae (including algae glue) and several barnacles were present after 12week cleaning. It was found that the efficiency of cleaning F-AF surfaces at weekly frequency was better than the efficiency of cleaning at biweekly frequency under any cleaning parameters, which indicated that cleaning F-AF surfaces weekly could improve the cleaning efficiency. Additionally, F-AF surfaces cleaned with F 4 showed better cleaning efficiency compared with those cleaned with F 3 , as shown in Fig. 9 (a)-(b) and (e)-(f), indicating that F 4 is more suitable for cleaning F-AF surfaces. On some F-AF surfaces, algae glue was also found after cleaning, decreasing the cleaning efficiency, as shown in Fig. 9 (a), (d)-(f), which resulted in the fact that longer cleaning duration even led to a decreased cleaning efficiency.

Comparison between Cu-AF and F-AF surfaces under different cleaning conditions
Cu-AF is a high performance, self-polishing antifouling paint which ensures efficient fouling protection and color retention all season. F-AF is high strength hard performance antifouling with a fluoro microadditive that gives ship hull a slick low friction surface for maximum speed and efficiency.
FR mean values of Cu-AF and F-AF surfaces under different cleaning conditions are shown in Fig. 10. For weekly cleaning, FR mean values are decreasing on both Cu-AF and F-AF surfaces despite some fluctuations, which suggested that weekly cleaning can help to improve the fouling resistance property of antifouling coatings, as shown in Fig. 10 (a)-(b). Moreover, it can also be found that FR mean values of Cu-AF surfaces were lower than F-AF surfaces' in the first 7 weeks during the weekly cleaning process, but the trends were in the opposite direction after that. The result showed that the Cu-AF performed better than F-AF at the early stage of cleaning schedule, but F-AF perform better after 7 weeks' cleaning and even became fouling free surface after 12-week cleaning.
The situation was different for Cu-AF surfaces and F-AF surfaces with bi-weekly cleaning, as shown in Fig. 10 (c)-(d). Cu-AF exhibited better fouling resistance property than F-AF. Moreover, fouling resistance for both Cu-AF and F-AF remained stable after 12-week cleaning.

Coating condition
Cu-AF and F-AF coating conditions were evaluated by optical microscope before deployment and after cleaning procedure drying for 1 month. The results are shown in Table 5 and Table 6. Before deployment, no physical damage was observed on any coating surface. When microscope images were taken at the end of the study, wearing, cracking and scratching were observed on both Cu-AF and F-AF surfaces. For Cu-AF, several scratches (˂1 cm long, ˂5 μm wide) were present on all panels except for Cu-F3T1W1. Additionally, more scratches were observed with stronger cleaning force and longer cleaning duration.
Only one crack (˂1 cm long, ˂5 μm wide) was observed on Cu-F3T2W2 surface, as shown in Fig. S1. Several minor wears with diameter less than  0.15 mm were observed on Cu-F3T2W1 and Cu-F3T2W2 surfaces. All these 'damages' cannot be seen by eyes since the physical appearance of the coating was visually even smoother, so all panels were considered no significant damages for Cu-AF after cleaning. For F-AF, scratches were more related to cleaning force and cleaning frequency. Stronger cleaning force and more frequent cleaning resulted to more scratches. More cracks were observed on F-F4T1W1 which were related to not only cleaning but also seawater environment. In addition, there was evidence of barnacle base plate footprints after removal, but these appeared not to affect coating performance.

Roughness test
Peak-to-valley roughness height (Rz) was measured on both Cu-AF and F-AF surfaces under different cleaning conditions, before deployment and at the end of the study (12-week cleaning). F-AF and Cu-AF testing samples after exposure were cleaned and rinsed with DI water and dried in the room temperature before doing the roughness measurement. Therefore, Cu-AF testing samples were free from dried biofilm or algae when doing the roughness measurement, though dried algae still remained on some of the F-AF testing samples. The surface area for roughness measurement on testing sample is 24 × 85 mm under each condition.
The results were presented in Fig. 11, where a significant difference between these two antifouling coatings before deployment and at the end of the study was shown. Before deployment, Rz of Cu-AF with 119.1 ± 15.4 μm, which was significantly smoother than the F-AF with Rz of  487.7 ± 137.3 μm. This difference in roughness height is primarily attributed to the variation of the paint formulation and especially the quality of paint application, since Cu-AF testing samples were applied by airless spray while F-AF testing samples were applied by roller following the technical data sheet. Then, the Cu-AF and F-AF became significantly smoother by the end of the study, dropping to 71.5 ± 5.4 μm and 273.1 ± 31 μm peak-to-valley roughness height, respectively, which is consistent with previous reported study [22]. For Cu-AF, this decrease in Rz may not solely be attributed to the effect of cleaning, but also due to its self-smoothing effect [29]. For F-AF, the combination of rotating brush abrasion on the polymer matrix and dried algae present on some of the testing surface may result in the decrease of Rz, as shown in Fig. S4. Moreover, by comparing different cleaning parameters, extending cleaning duration and increasing cleaning force can make the coating surface smoother.

Water contact angle test
Water contact angle (WCA) measurement for F-AF and Cu-AF testing samples were conducted before exposure and after exposure. After exposure, F-AF and Cu-AF testing samples were cleaned and rinsed with DI water before drying in the room temperature. There was no dried biofilm or algae on Cu-AF testing samples when doing the WCA measurement. For some of F-AF testing samples, dried algae still remained. However, the WCA measurement on F-AF avoided the area which covered by dried algae.
WCA was measured on both Cu-AF and F-AF surfaces under different cleaning conditions, before deployment and at the end of the study (12-week cleaning), as shown in Fig. 12. Before deployment, WCA of Cu-AF (~103 • ) was slightly smaller than WCA of F-AF (~110 • ). These differences in WCA are also primarily attributed to the differences in the paint formulation and the quality of paint application. At the end of the study, the Cu-AF got a slight decrease of WCA which was still over 90 • , as shown in Fig. 12 (a)-(b), indicating that mechanical cleaning hardly affected the wettability of Cu-AF surface. However, the WCA of F-AF largely decreased to below 90 • after 12-week cleaning, as shown in Fig. 12 (c)-(d), indicating that mechanical cleaning increased the wettability of F-AF. In addition, the WCA of F-AF decreased with the decreasing of the cleaning force and cleaning duration at the end of study, which corresponded to the fact that stronger cleaning force and longer cleaning duration led to better fouling resistance property of F-AF.

Biocide release rate of Cu-AF
Effect of cleaning on biocide release rate was tested on Cu-AF panels at the end of study, using the polishing rate mass-balance method, which relies on changes in paint thickness for calculating the amount of released biocide. This method was complemented with SEM-EDX imaging and chemical analysis for determining the extent of underestimation due to formation of a leached layer on the AF coating [22,30,31].
SEM-EDX images of Cu-AF with cross section analysis before deployment and after cleaning process were shown in Fig. 13 (a)-(f). After cleaning, the thickness of testing panels was decreased apparently compared with the initial thickness before deployment due to either selfpolishing or cleaning. The leaching layer (LL) could be seen clearly on Cu-AF control surface without cleaning in Fig. 13 (b).
Biocide release rate of Cu-AF under different cleaning condition, together with corresponding polishing thickness DFTi -DFTe is shown in Fig. 14. The biocide release rate values of Cu-AF after cleaning under different cleaning condition is shown in Table S2. Release rate of uncleaned Cu-AF surface was 14.4 μg Cu cm − 2 day − 1 and the values of different cleaning parameters were between 15.7 and 24.6 μg Cu cm − 2 day − 1 . The biocide release rate values of cleaning samples are all higher than the no-cleaning sample, especially Cu-F4T1W1. It can be found that the copper release rates of Cu-AF with weekly cleaning were higher than those that cleaned with bi-weekly frequency and even higher than Cu-AF control surface, suggesting that weekly cleaning could increase the release rate of Cu-AF. From the discussion of cleaning effect of the fouling condition on Cu-AF during the cleaning process, fouling resistance property and cleaning efficiency on Cu-AF were mainly depended on cleaning frequency. Comprehensive consider the copper release rate, cleaning effect and cost of cleaning, 0.4 N at 240 rpm, 10 s, bi-weekly frequency was recommended as the most suitable cleaning condition for Cu-AF.
The release rate is of the same order of magnitude as previously reported values obtained using the same method on erodible coatings: values ranging 7.6-18 μg Cu cm − 2 day − 1 for in-service conditions on US Navy vessels [32] (Haslbeck and Ellor 2005), an average copper release rate of ~13 μg Cu cm − 2 day − 1 for 1 year study with different treatments [22], and theoretical values 21.1 and 28.1 μg Cu cm − 2 day − 1 for monthly-and weekly-groomed panels [20]. However, it should be noted that the AF products used differ between studies, as well as environmental conditions and cleaning procedure. It was reported that temperature, pH and salinity may significantly impact biocide release rates [33]. Therefore, absolute comparisons are currently limited to a plausibility check on the order of magnitude.
In this study, all the cleaning parameters as recommended in this work were based on the evaluations of coating performance during the cleaning process and coating conditions after the cleaning process. Worth to mention that when it comes to ship hull cleaning under real conditions, economic cost for hull management (operating and support costs) together with environmental issues for sustainability (side effects of biocide release) should also be taken into consideration. With the focus of this study, the present work is targeted at assisting the cleaning strategy/guidance making towards a more sustainable ship hull management.

Conclusions
The current study used a specially designed bench drill brush system for testing the effects of varying cleaning force, duration and frequency on different antifouling coatings' performance and surface condition. Coatings were immersed under local almost-static flow conditions at the CMTC in Denmark. Different cleaning parameters on fouling resistance property and cleaning efficiency were investigated on a soluble copper based antifouling coating (Cu-AF), an insoluble fluorine-based copper free antifouling coating (F-AF) and the non-coated acrylic surface (A-BS). Furthermore, coating surface condition evaluations, including surface roughness, contact angle and SEM-EDX characterization as well as copper release rate were also conducted. The following conclusions can be drawn from this work: • Cleaning frequency affects more than cleaning force and cleaning time on three testing surfaces. Additionally, longer cleaning time and stronger cleaning force showed a positive effect on coatings' performance with longer cleaning interval. • Algae glue increased the difficulty on cleaning efficiency with time going by. Therefore, it is recommended that it is better to remove the algae glue when cleaning algaes. • Extending cleaning duration and increasing cleaning force can make two antifouling coating surface smoother. Compared with Cu-AF, the wettability of F-AF largely increased after cleaning and scratches were more related to cleaning force and cleaning frequency. • Biocide release rates of Cu-AF under different cleaning parameters were between 15.7 and 24.6 μg Cu cm − 2 day − 1 and bi-weekly cleaning was more suitable for Cu-AF.
Finally, based on the evaluation results, the following cleaning parameters are suggested for the three surfaces in order to achieve better fouling resistance and cleaning efficiency: a perpendicular force of 7.8 N at a rotational speed 740 rpm with 5 s, weekly frequency for the noncoated acrylic surface; a perpendicular force of 0.4 N at a rotational speed 240 rpm with 10 s, bi-weekly frequency for copper based antifouling paint; and a perpendicular force of 2.5 N at a rotational speed 240 rpm with 5 s, weekly frequency for fluorine-based copper free antifouling paint.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability
Data will be made available on request. The authors regret an error in the composition of F-AF in Table S1 of Appendix A. Supplementary data when checking the published article, which should be xylene, ethylbenzene, pyrithione zinc and 1-methoxy-2-propanol. The authors note that the correct coating type was used in the experiment which does not affect the conclusions of the article. The corrected Table S1 is as follows: