A novel effective bio-originated methylene blue adsorbent: the porous biosilica from three marine diatom strains of Nanofrustulum spp. (Bacillariophyta)

In the present paper, for the first time the ability of the porous biosilica originated from three marine diatom strains of ‘Nanofrustulum spp.’ viz. N. wachnickianum (SZCZCH193), N. shiloi (SZCZM1342), N. cf. shiloi (SZCZP1809), to eliminate MB from aqueous solutions was investigated. The highest biomass was achieved under silicate enrichment for N. wachnickianum and N. shiloi (0.98 g L−1 DW and 0.93 g L−1 DW respectively), and under 15 °C for N. cf. shiloi (2.2 g L−1 DW). The siliceous skeletons of the strains were purified with hydrogen peroxide and characterized by SEM, EDS, the N2 adsorption/desorption, XRD, TGA, and ATR-FTIR. The porous biosilica (20 mg DW) obtained from the strains i.e. SZCZCH193, SZCZM1342, SZCZP1809, showed efficiency in 77.6%, 96.8%, and 98.1% of 14 mg L−1 MB removal under pH 7 for 180 min, and the maximum adsorption capacity was calculated as 8.39, 19.02, and 15.17 mg g−1, respectively. Additionally, it was possible to increase the MB removal efficiency in alkaline (pH = 11) conditions up to 99.08% for SZCZP1809 after 120 min. Modelling revealed that the adsorption of MB follows Pseudo-first order, Bangham’s pore diffusion and Sips isotherm models.

Diatoms (Bacillariophyta), representing a major group of photosynthetic microorganisms, are unicellular eukaryotic microalgae that live within cell walls composed of 3D structured porous biosilica (SiO 2 ). They play an essential role in global carbon and silicon cycles in the ocean and their photosynthetic activity accounts for almost one-fifth of the Earth's primary productivity 1,2 . Diatoms attract increasing attention in the applied sciences due to their potential for producing a variety of bioactive compounds and fine chemicals for industrial applications: fucoxanthin is known for its antioxidant effect and can be used in pharmaceuticals and cosmetics 3 ; unsaturated fatty acids have been used as food supplements 4 ; triacylglycerols (TAG) provide a carbon feedstock for conversion

Results
Batch cultivation of diatom cultures. Three strains of Nanofrustulum spp. showed distinct lag, exponential, and stationary phases during batch culture (Fig. 1a,b). The lag phase of the strain SZCZCh193 N. wachnickianum was observed until the 5th day of cultivation, with exponential growth for 7 days from the 5th to 12th day, and a stationary phase starting from the 12th day of growth. The maximum specific growth rate of the exponential phase was calculated as 0.115 d −1 (R 2 = 0.95). The growth of strain SZCZM1342 N. shiloi showed a lag (until the 4th day), exponential (for 6 days, from 4 to 10th day), and stationary (from 10th day) phases, and the maximum specific growth rate of the exponential phase was calculated as 0.270 d −1 (R 2 = 0.97) 6 days after inoculation, SZCZP1809 N. cf. shiloi showed exponential growth for 10 days (to the 16th day of cultivation), with a following stationary phase of growth. The maximum specific growth rate during the exponential phase was calculated as 0.513 d −1 (R 2 = 0.99). Therefore, it can be said that the maximum biomass yields for SZCZCh193 N. wachnickianum (0.39 ± 0.039 g L −1 DW), and SZCZM1342 N. shiloi (0.47 ± 0.033 g L −1 DW) were obtained on the 18th day, and for SZCZP1809 N. cf. shiloi (0.63 ± 0.013 g L −1 DW) on the 16th day of cultivation.
The UV-vis spectroscopy of sonicated biomass showed a distinct peak at 230 nm and small peaks at 270, 430, 495 and 676 nm ( Fig. 3d-f, green line). The pure, sonicated biosilica showed only one distinct peak at 230 nm ( Fig. 3d-f, gray line).
The biosilica Tauc's plot (Fig. 4a)  ATR-FTIR spectra revealed distinct peaks ( The PZC was calculated as the intercept between initial pH and change in pH value after 24 h of incubation and was estimated as pH pzc = 6.0 for SZCZCH193 N. wachnickianum and pH pzc = 5.3 for SZCZM1342 N. shiloi and SZCZP1809 N. cf. shiloi (see Supplementary Figure S2a).  Figure S2b). In the pH range of 4.0 to 12.0 an increase of zeta potential is observed (zeta potential > -25 mV). Figure 5a-c presents the results of TGA analysis for all three biosilica samples. The plot of the mass loss against temperature did not allow us to derive more specific information about individual stage of the process. Thus, the differential thermal analysis (DTA) was performed which showed three main stages of the process   www.nature.com/scientificreports/ ( Fig. 5a-c). SZCZCH193 N. wachnickianum biosilica sample (Fig. 5a) showed three distinguished peaks on the DTA analysis plot appeared with maxima at 59.  shiloi and 98.1% for SZCZP1809 N. cf. shiloi (Fig. 6a). Furthermore, the effect of different pH of MB solutions was investigated with exposure of biosilica to acidic (pH = 3), neutral (pH = 7), and alkaline (pH = 11) aqueous solutions of MB as shown in Fig. 6b. In acidic solutions, the removal of MB was lower than in neutral and alkaline conditions: 71.33% for SZCZCH193 N. wachnickianum, 91.90% for SZCZM1342 N. shiloi, and 97.23% for SZCZP1809 N. cf. shiloi. An increase in pH resulted in higher removal efficiencies: up to 77.27% for SZCZCH193 N. wachnickianum, up to 96.03% for SZCZM1342 N. shiloi, and up to 99.08% for SZCZP1809 N. cf. shiloi.
The influence of different MB concentrations in the initial solution was also examined. Figure 6c shows that, for SZCZCH193 N. wachnickianum, an increase in MB concentration resulted in a significant decrease in the removal efficiency-from 79.48% at 5 mg L −1 to 10.69% at 100 mg L −1 . Frustules of SZCZM1342 N. shiloi, exposed to a 100 mg L −1 solution of MB, showed an hyperbolic increase with concentrations almost of the same efficiency (89.82%) as at lower concentrations-88.  Figure S4a) model showed higher correlation coefficients for SZCZM1342 N. shiloi and SZCZP1809 N. cf. shiloi than Elovich (Supplementary Figure S4b) and pseudo-second order (Supplementary Figure S4c) kinetic models and the normalized standard deviation value for pseudo-first order was the lowest. For SZCZCH193 N. wachnickianum, the correlation coefficients for pseudo-second and Elovich models were higher than for the pseudo-first order model, but not close to unity and with very small constant values. The standard deviation value was higher than 50% for all three models ( Table 2).
According to the isotherms models' constants and coefficients of correlation, the Sips Isotherm model is the most suitable for SZCZCH193 N. wachnickianum, SZCZM1342 N. shiloi, and SZCZP1809 N. cf. shiloi (Table 2

Discussion
Several studies reported specific growth rate and biomass productivities for Nanofrustulum shiloi strains as 0.099 d −138 , 0.305 d −139 , 0.457 d −140 , which correspond to 12.8 mg L −1 d −138 , 31.29 mg L −1 d −140 , respectively. In the present study, the specific growth rate and biomass productivity of SZCZM1342 N. shiloi and SZCZP1809 N. cf. shiloi were observed as 0.270 d −1 , 0.513 d −1 , which correspond to 26.12 mg L −1 d −1 , 39.37 mg L −1 d −1 , respectively. Thus, SZCZP1809 could be considered the most productive strain of N. shiloi. To the best of our knowledge no work has been done on the growth investigation of Nanofrustulum wachnickianum. A significant increase of biomass yield at lower temperatures for SZCZP1809 N. cf. shiloi and higher temperature for SZCZCH193 N. wachnickianum could be explained by the origin of the strains: SZCZP1809 was collected from the Atlantic Ocean, Sea Point, Cape Town, South Africa, where the average sea water temperature amounts to 17 °C, and SZCZCH193 N. wachnickianum-from the Gulf of Mexico, Marquesas Keys, Florida, USA (an average sea water temperature of 28 °C). Silica is a major limiting nutrient for diatom growth because of their need to build siliconbased frustules. The strains of Nanofrustulum spp. showed that biomass yield can be significantly enhanced with an increased silicon source up to almost 1 g L −1 DW, which correlated with previous studies 41 .
Pictures from scanning electron microscopy revealed that the valve faces of strains SZCZP1809 N. cf. shiloi and SZCZM1342 N. shiloi contain more than one row of areolae whereas the valves of SZCZCH193 N. wachnickianum contain only one row. The differences in areolation of the valve face make frustules of SZCZM1342 N. shiloi and SZCZP1809 N. cf. shiloi more porous than those of SZCZCH193 N. wachnickianum. Li et al. 42 described a new species of N. wachnickianum and observed that this species could be differentiated from N. shiloi by the number of areolae on the valve face. In the case of MB removal, the pore size and number are crucial. Thus, SZCZCH193 N. wachnickianum has less potential as an adsorbing agent of MB due to a lower number of areolae, unlike SZCZM1342 N. shiloi and SZCZP1809 N. cf. shiloi. The areolae (width 80 nm and higher) of Nanofrustulum spp. strains could be considered macropores 43 . However, the N 2 adsorption/desorption revealed presence of micro and mesopores, with average diameter 4.217 nm (SZCZCH193 N. wachnickianum), 2.073 nm (SZCZM1342 N. shiloi), and 1.971 nm (SZCZP1809 N. cf. shiloi). Previous studies reported the similar values of average pore diameter of diatomaceous frustules: 3.93 nm for Pseudostraurosira trainorii 44 , 3-10 nm for Thalassiosira punctigera and 3.6-3.7 nm for T. weissflogii 45 , and 4.61 nm for Navicula australoshetlandica 46 . Additionally, the porous nature of material could be described by the shape of the N 2 adsorption/desorption isotherm. According to IUPAC classification the isotherms of Nanofrustulum spp. biosilica follow the Type I (microporous nature) and Type II (macroporous material), with combination of Type H3 and H4 hysteresis loop, where Type H3 shows presence of the macropores network and Type H4 indicates the existence of slit-like micropores in sample 43 .
Specific surface area of different strains of NanofrusItulum spp. differs slightly. Previous reports showed that different diatomaceous frustules have diverse S BET value: from 2 m 2 g −1 for Skeletonema sp. 47 and 30 m 2 g −1 for P.  48,49 , confirmed the presence of chlorophyll and carotenoids. A peak at 270 nm could indicate the presence of polysaccharides as reported by Trabelsi et al. 50 . Sonicated, purified frustules showed only one peak at 230 nm, revealing the presence of silica in the samples 51,52 .
First stage of thermal analysis is connected with water desorption, second stage-could be connected to oxidative degradation of organic compounds (i.e. lipids), while third stage revealed degradation of inorganic matter (i.e. magnesium and sodium salts) [53][54][55] . The highest mass loss (almost 75%) of the sample during the thermal analysis (TGA/DTA) was observed for the sample of SZCZM1342 N. shiloi. In comparison for samples SZCZCH193 N. wachnickianum and SZCZP1809 N. cf. shiloi the overall mass loss was 45.8% and 43.8%. This could be explained by presence of high amount of organic matter in sample SZCM1342 N. shiloi, considering that this sample had slightly brown color in comparison to other samples (see Supplementary Figure S6).
The EDS study confirmed the presence of silica and oxygen in the frustules. The atomic ratio of O:Si was calculated as 1.69:1, 2.39:1, and 2.27:1 for SZCZCH193, SZCZM1342, and SZCZP1809, respectively, which confirmed that diatom frustules are made from silica mostly in the form of SiO 2 nH 2 O-very similar to opal 56 . The distinct peak of carbon (C) could be explained by the presence of the remaining organic compounds surrounding the frustules and chitin template within the diatomaceous silica 57 . The ooccurrence of copper (Cu) is the result of using copper TEM grids.
X-ray diffraction analysis for all strains showed the broad peak indicating the presence of amorphous from of the silica. Thus, it can be concluded that studied samples mostly comprised from amorphous silica. For SZC-ZCH193 N. wachnickianum intense signals are most likely originating from quartz and cristobalite 58 . Alternatively, these signals could indicate the inclusions of magnesium inorganic salts 53 . For SZCZP1809 N. cf. shiloi the low intensity signals may come from the inclusions of alpha-quartz.
The calculated Tauc plot energy bandgaps (4.40 eV for SZCZCH193 N. wachnickianum, 4.05 eV for SZCZM1342 N. shiloi, and 4.10 eV for SZCZP1809 N. cf. shiloi) showed that pure porous biosilica has the properties to be an ultrawide semiconductor 59 . Analysis for amorphous silica 60 revealed a band gap of 3.35 eV. A previous study confirmed that amorphous and crystalline silica has a lower band gap than a porous one 61 and the difference in values for strains could be explained by their differences in pore sizes.
In order to identify the potential functional groups on the surface of the frustules, ATR-FTIR analysis was performed. According to several reports 12,62 , the appearance of strong bands at 1150, 1050, 805 and 442 cm −1 presented asymmetrical stretching, symmetrical stretching, and bending vibrations of Si-O-Si groups, respectively. There are strong bands recognised by several studies 12,62,63 as C-H stretching at 2981, 2922 (CH 3 ), 2851 (CH 2 ), and 854 cm −1 (Si-(CH 3 ) 2 ), and C=O stretching at 1748 cm −1 . Four strong peaks at 1630, 1540, 1450, and 1395 cm −1 showed the bending vibrations of amino groups, corresponding to a previous study 64 . According to Otzen 65 , these peaks indicate the presence of organic compounds surrounding the biosilica. The band at 3660 cm −1 is related to a stretching vibration of O-H from SiOH 62 , moreover, the broad band from 3500 to 3000 cm −1 is related to molecular water 66 .
The surface charge of the particles determines the ability of the particles to aggregate. The value of zeta potential is dependent on the properties of the given particles as well as from the pH and ionic strength of the solution. Particles with zeta potential close to zero will aggregate. In turn, stable and non-aggregating systems are characterized by absolute zeta potential values greater than +/− 25 − + /− 30 mV 67,68 . The results of zeta potential measurments for pure biosilica samples indicate that at pH greater than 4 aggregation is not occurring and suspensions are stable. The respective results can be confirmed by the photos taken of diatom solutions at different pH (2.0, 3.0, > 4.0) (see Supplementary Figure S7a-c). In the lowest pH a significant aggregation and sample precipitation can be observed. At pH around 3.0 the aggregation is still visible but in lower extend. At pH higher than 4.0 no visible aggregation occurs, the suspension is stable. It is noteworthy to mention, that obtained results differ from the results obtained for pure synthetically prepared silica described by Xu et al. 69 . The different shape of zeta potential plot of examined samples in comparison to Peng Zu's can be explained by the presence of carboxyl (COOH) and amine (NH 2 ) groups on surface of the biosilica. The presence of respective functional groups was confirmed by FTIR analysis. Moreover, TA/DTA analysis also revealed the presence of high amount of organic matter on the surface of the biosilica. The difference is also notable between the samples of biosilica, for instance, for SZCM1342 N. shiloi the positive charge of the surface was observed. The respective difference more likely is due to higher amount of organic matter on the surface of SZCZM1342 N. shiloi sample, (i.e. proteins).
The point zero charge (PZC) allows us to find the pH at which the charge at the surface is neutral, therefore in pH less than pH pzc the surface is charged positively, and at pH higher than pH pzc the charge of the surface is negative 70 . Pure biosilica showed a PZC value of pH pzc = 6.0 (SZCZCH193 N. wachnickianum), and pH pzc = 5.3 (SZCZM1342 N. shiloi, SZCZP1809 N. cf. shiloi), which corresponds with several studies of diatomaceous earth: the residual (RDE) and pure (PDE) from Brazil (pH pzc = 6.75 and pH pzc = 6.59 respectively) 70 , from the mine El Pino (pH pz c = 5.0) 71 , from East Jordan (pH pzc = 5.4) 72 , from Egypt (pH pzc = 6.21) 73 . The results from the PZC study of pure diatomaceous biosilica confirm the presence of O-H groups on the surface and suggest that these hydroxyl groups can gain or lose a proton by changing the pH. Therefore, in acidic media (pH < pH pzc ), the Si-OH group of biosilica gains a proton and produces Si-OH 2 + , and in basic media (pH > pHpzc) the Si-OH group loses a proton and produces Si-O −74 .
The percentage of MB removal from aqueous solution in all experiments for SZCZCH193 N. wachnickianum was lower than for the two remaining strains of N. shiloi, these differences could be explained by differences in morphology, pore density, specific surface area, and pore diameter between the Nanofrustulum spp. strains.  80 noticed a decrease in the removal of MB under increased concentrations for metal-silica amorphous adsorbents, similar to the removal observed for SZCZCH193 N. wachnickianum. However, the hyperbolic increase with concentration for SZCZM1342 N. shiloi and SZCZP1809 N. cf. shiloi was noticed as for MB absorbance onto fava bean peel 81 .

Scientific Reports
Even though the frustules of Nanofrustulum spp. showed a high percentage of MB removal, comparable to the efficiency of well-known adsorbents such as natural diatomite and activated carbon, the MB adsorption capacity of Nanofrustulum spp. frustules was reported almost 10 times less than diatomite 32 , algae Gellidium sp. 22 , little less than amorphous silica 82 , brown algae biomass 83 , and almost the same as some zeolites 26 and dead biomass of Aspergillus niger 84 (see Supplementary Table S2). Pre-treated frustules of Pinnularia showed a higher adsorption capacity 28 than frustules of Nanofrustulum spp., which could be explained by morphology and specific surface area of Pinnularia and Nanofrustulum, as well as by the cleaning method.
The pH of adsorbate solution is considered to be one of the most important parameters in water adsorption processes 74 . The pKa of MB is reported to be 3.8, thus at pH less than 3.8 the surface of MB molecule is not charged, and at pH higher than 3.8-positively charged 85 . Therefore, in pH less than pH pzc of biosilica, the main interaction between MB and frustules are hydrogen bonds and the adsorption process is slower because of repulsive forces between positive sites on the diatomaceous biosilica surface and the cationic dye. In pH higher than pH pzc of biosilica, the positively charged MB and negatively charge frustules interact electrostatically, and the adsorption process is stronger than in acidic media 73 . In our study, an increase in pH resulted in increased removal up to 99.08% for SZCZP1809 N. cf. shiloi. Similar results reported for diatomite from China 79 , diatomaceous earth from Egypt 80 , East Jordan 72 and Brazil 71 , palygorskite 74 . To achieve desirable pH, trace amounts of 1 M HCl and 1 M NaOH solutions were added and could negatively affect the adsorption as an ionic species (Na + and Cl − ), because in presence of inorganic salts the adsorbent surface becomes not easily accessible for MB. However, there were no influence of these ionic species on the dye uptake observed, presumably due to their low concentrations (less than 0.005 M) in dye solution 86 .
Moreover, several kinetic, diffusion, and equilibrium models were applied for better understanding of the possible MB sorption mechanism. The kinetic curve fitted the pseudo-first order model better (with a higher correlation coefficient and lower standard deviation value), therefore we can conclude that the adsorption behaviour of MB on biosilica predominantly followed the pseudo-first order kinetic model: the overall rate of adsorption process was controlled by physisorption, meaning that the molecular interaction between MB and biosilica is governed by van der Waals froces 87 . In order to determine whether the main resistance to mass transfer was in the thin film (boundary layer) surrounding the adsorbent particle, or in the resistance to diffusion inside the particles, Boyd's model was applied. The straight line passing through the origin indicated that the MB adsorption rate is governed by diffusion inside the particles 88 . The high correlation coefficients for Bangham's pore diffusion model indicated that the diffusion of MB molecules onto pores inside biosilica is a rate-controlling step in mass transfer for adsorption process 89 . The adsorption isotherms of SZCZM1342 and SZCZP1809 were characterized as slightly LS-shaped and the isotherm of SZCZCH193 as L-shaped 90 , which suggests that the studied diatomaceous biosilica exhibit a high adsorption affinity towards MB dye. Isotherms for Nanofrustulum spp. showed a higher affinity to the Sips equation due to high correlation coefficients and low standard deviation values. Based on that, we can conclude that the mechanism of MB sorption onto biosilica predominantly follows the Sips model, which combines the Freundlich and Langmuir isotherms and describes monolayer MB formation onto homogenous and heterogenous sites on biosilica surface 90 .
The present research introduced a novel environmentally friendly adsorbent of basic dyes from wastewatersthe porous biosilica originated from marine diatom Nanofrustulum. The diatom genus could be considered the most productive microalgae with the highest biomass productivity. The biomass could be further used for extraction of bio-active molecules, e.g. fucoxanthin, known by its antioxidant activity, polyunsaturated fatty acids, with anti-inflammatory properties, and neutral lipids for biodiesel production, while unused diatomaceous biosilica can be purified and efficiently applied in wastewater treatment, due to its porous architecture, the negative surface charge, and relatively high the specific surface area. Several researchers reported the ability of diatomaceous silica to remove heavy metals from aqueous solutions, although the present paper for the first time introduced biosilica as a novel efficient adsorbent of basic dyes from wastewaters. In the future, more intensive research should focus on evaluation of biorefinery potential of these three species, with special focus on co-production of fucoxanthin, fatty acid, and biosilica in economically desirable and eco-friendly way, and adsorption capacity to remove different types of organic pollutants in presence of inorganic salts, which could decrease dye uptake.
Characterization of the diatomaceous biosilica. Diatoms were harvested in the late exponential phase, centrifuged at 3000 rpm for 15 min, and pellets were purified by 30% H 2 O 2 solution at 110 °C for 2 days following a thorough cleaning with ddH 2 O. The colorless, cleaned silica was dried at 50 °C for 3 days and later used for characterization experiments. The morphology of the clean diatom frustules was imaged by scanning electron microscopy (SEM) using a Hitachi SU8000 (Hitachi, Tokyo, Japan). For the SEM study, 40 µL of frustules were dried on a Nuclepore™ 5.0 µm Track-Etch Membrane (Whatman™, Cytiva, Germany) at room temperature, later mounted on a M4 cylinder SEM holder, and sputtered with a 10 nm thick gold layer. For the elemental analysis of biosilica, a drop of frustules was placed on a carbon-coated copper grid (Sigma-Aldrich, USA), and the analysis was carried out by a Hitachi STEM S5500 equipped with an EDS detector (Hitachi, Tokyo, Japan). The measurements were performed with an accelerating energy of 30.0 kV and analyzed using NSS ThermoScientific software.
After sonication of the biomass and the cleaned frustules (using Hielscher UP100H ultrasonic processor (Teltow, Germany) for 20 min at 60% amplitude), the suspensions were subjected to the UV-Vis DR 6000 spectrophotometer (HACH-Lange) for optical measurements in the wavelength range of 200-900 nm in 10 mm cuvettes against ddH 2 O (blank). For absorbances higher than 2, the solution was diluted with ddH 2 O, the dilution factors were considered in the presented graphs.
FTIR spectra of dried biosilica were obtained in the mid-infrared range (4000-400 cm −1 ) with the utilization of attenuated total reflection (ATR) mode on an Alpha FTIR spectrometer (Bruker Daltonics, Bremen, Germany).
The low-temperature nitrogen adsorption/desorption isotherms were recorded on Quantachrome Autosorb iQ at 77.35 K.
Zeta potential measurements were carried out with Malvern Zetasizer NanoZS (Malvern) using DTS1070 cuvette (Malvern). The analysis was performed in the automatic selection mode of voltage and number of runs. Each measurement was repeated three times. Zeta potential was measured in 2.0-12.0 pH range. To maintain the pH 0.1 M NaOH and 0.1 M HCl solutions were used. FiveEasy Plus pH-meter (Mettler Toledo) with a combined electrode with glass membrane and Ag/AgCl reference system (Mettler Toledo) was applied to measure pH of suspensions. The pH-meter was calibrated using standard buffered solutions with pH of 4.0, 7.0, 10.0 before carrying out measurements.
The X-ray diffraction (XRD) spectra were recorded with an X'Pert Pro Analytical X-ray diffractiometer (Phillips, Würzburg, Germany) with Cu-Kα radiation (λ = 0.1541 nm, 40 kV, 30 mA); 1 mL of dried on glass slide sample was scanned in the 2θ range between 5° and 120° with step sizes of 0.0167.
The salt addition method was used to find the PZC of dried biosilica. An aliquot of 20 ml of 0.1 M NaCl was collected in 6 Erlenmeyer flasks and a 2-10 pH range was set by adding 1 M HCl and 1 M NaOH solutions using a pH meter (Voltcraft PH-100ATC). Then in each flask 20 mg of the dried biosilica were added and shaken at speed 100 rpm on an orbital shaker at temperature 23 °C for 24 h. After equilibrium, the contents were filtered, www.nature.com/scientificreports/ and the pH values of filtrates were recorded. The PZC value was determined by plotting the graph of initial pH against the change in pH.

Methylene blue (MB) batch sorption.
The cleaned frustules (20 ± 0.5 mg DW) were exposed to 10.0 ml of 14.0 mg L −1 MB in a 15 ml Falcon tube. The mixture (pH = 7) was mechanically stirred at 3000 rpm at 23 °C for 3 h. The removal of dye was recorded by a UV-Vis DR 6000 (HACH-Lange) spectrophotometer in the wavelength range from 500 to 800 nm at different time points : 5, 15, 30, 45, 60, 75, 90, 105, 120, and 180 min. Similarly, isotherm studies were performed by measuring 20 ± 0.5 mg of cleaned frustules into 15 ml Falcon tubes containing varying initial concentrations (5,10,15,20,50, and 100 mg L −1 ) of MB. The mixture was stirred at 3000 rpm under 23 °C for 120 min, which is necessary to attain equilibrium. The effect of different pH of the initial MB solution (14.0 mg L −1 ) was recorded for pH = 3, 7, and 11 after 120 min of exposure. The initial pH of the solution was adjusted with 1 M HCl and 1 M NaOH solutions. The absorbance value at 665 nm was used for further calculations.
The quantity of adsorbed MB by biosilica was calculated as follows: where q t is the MB adsorbed on the biosilica (mg g −1 ) at a given time (t); C 0 and C t are the concentrations of the MB at the start and at the given timepoint (mg L −1 ), respectively; V is the solution volume (L); m is the biosilica dosage (g). The percentage of MB removal (%) was calculated by Eq. (3): where C 0 and C t are the concentrations of the MB at the start and at the given time (t), respectively (mg L −1 ). To understand the possible mechanisms and rate controlling steps of adsorption, several kinetic, diffusion and isotherm models were applied (see Supplementary Table S3).
Furthermore, the applicability of the kinetic and isotherm models was validated by the normalized standard deviation, Δq (%), Eq. (4) where N is the number of data points, q exp and q cal (mg g −1 ) are the experimental and calculated adsorption capacity value, respectively. Data analysis. The batch cultivation experiments were conducted in duplicate. The figures show mean values and standard errors. The significance of differences between different groups was analyzed using a one-way ANOVA analysis with Tukey's post-hoc test, the alpha level 0.05. The batch growth experiment figures, UV-Vis, FTIR spectra, batch adsorption spectra were plotted using MS Excel software. The EDS spectra were obtained using NSS ThermoScientific software. The zeta potential results take into account the Smoluchowski approximation. The analysis data for TA/DTA was proceed with the use of TA Universal Analysis software (TA Instruments, New Castle, DE, USA). The X-ray diffraction pattern analysis was obtained from XRD Malvern Panalytical software (version 1.5a, Almelo, The Netherlands). Modeling of adsorption were performed in OriginPro 2022 software. (2)  www.nature.com/scientificreports/