The adsorption amount of Cd(II) per unit mass of the diatom biomass at equilibrium (q _e, mg g⁻¹) and the adsorption amount of Cd(II) per unit mass of adsorbent at time t (q _t, mg g⁻¹) were determined as follows: q t = C 0 − C t V W , (1) q e = C 0 − C e V W , (2) where C ₀ (mg L⁻¹), C _t (mg L⁻¹), and C _e (mg L⁻¹) are the heavy-metal concentrations at time zero, time t, and at equilibrium, respectively; V is the volume of the solution (L); and W is the mass of dry adsorbent used (g).

Pseudo-second-order model is widely used for bio-adsorbent to predict the adsorption kinetics. The non-linear form of the pseudo-second-order equation is given by q t = k 2 q e 2 t 1 + k 2 q e t , (3) where q _e and q _t are the amounts of solute adsorbed (mg g⁻¹) at equilibrium and at any time t (min), respectively. k ₂ is the rate constant for pseudo-second-order reaction (g mg⁻¹ min⁻¹). The straight line plots of 1/q _t versus 1/t are used to obtain the constants for the pseudo-second-order reaction.

The most common adsorption isotherm models, Langmuir and Freundlich, were applied to predict the equilibrium parameters of the Cd(II) adsorption process on the diatom biomass.

The non-linear form of the Langmuir isotherm model can be expressed as q e = q m K L C e 1 + K L C e , (4) where q _e is the amount of Cd(II) adsorbed at equilibrium (mg g⁻¹), q _m is the maximum adsorption capacity (mg g⁻¹), C _e is the equilibrium concentration of metal ions in solution (mg L⁻¹), and K _L is the Langmuir constant that is related to the free energy of adsorption (L mg⁻¹).

The non-linear form of the Freundlich isotherm model is given as q e = K F C e 1 / n , (5) where q _e is the amount of Cd(II) adsorbed at equilibrium (mg g⁻¹), C _e is the equilibrium concentration of metal ions in solution (mg L⁻¹), K _F is the Freundlich constant (mg L⁻¹) that is related to the adsorption capacity of the sorbent, and n is the Freundlich constant related to the adsorption intensity (dimensionless) of adsorbent.

The sorption isotherms of Cd(II) on diatom biomass at 288 K, 298 K, and 308 K were obtained to determine the thermodynamic parameters. The standard Gibbs free energy change (△G), standard enthalpy change (△H), and standard entropy change (△S) were determined from K _L and T by the following equations: ∆ G = − R T ⁡ ln ⁡ K 0 , (6) ln ⁡ K 0 = ∆ S R − ∆ H R T , (7) ∆ S = ∆ H − ∆ G T , (8) where R is the universal gas constant (8.314 × 10⁻³ kJ mol⁻¹ K⁻¹), and T is the absolute temperature (K). K ₀ is the thermodynamic equilibrium constant. Values of K ₀ are obtained by plotting ln (q _e/C _e) versus q _e. The intercept of the fitting line with the vertical axis gives the values of K ₀ (Niwas et al., 2000). △H was obtained from the slope of the linear plot of lnK _L against 1/T, and △S was obtained from Eq. 8.

The cell morphology of Chaetoceros sp., Nitzschia sp., and Thalassiosira sp. strains were measured using a fluorescent microscope. It can be seen from Figure 1 that the cell of Chaetoceros sp. was almost round or cylinder-shaped with several strip-shaped matters attached to the cell. In comparison, the cells of Nitzschia sp. and Thalassiosira sp. were rhombus- and cylinder-shaped, respectively.

As shown in Figure 2A, both the non-living biomass of Chaetoceros sp. and Thalassiosira sp. consisted of C, O, N, Si, Na, Cl, and S, with a little difference in the atomic ratio of these elements. In comparison, the non-living biomass of Nitzschia sp. contained Ca and Mg elements except for C, O, N, Si, Na, Cl, and S, which suggested that the chemical composition on the surface of Nitzschia sp. was significantly different from that of Chaetoceros sp. and Thalassiosira sp.

FTIR spectra of the non-living diatom biomass were visualized to obtain information about the chemical bonds presented on the surface of the biomass. The chemical bond information corresponding to each FTIR peak was obtained mainly according to published references (Zhu, 2005; Robert et al., 2007). As shown in Figure 2B, the shapes of FTIR spectra for non-living Chaetoceros sp. and Nitzschia sp., and Thalassiosira sp. biomass were similar, which implies that these three diatom biomasses possess similar chemical groups on the surface. A broad and strong peak (3,308 cm⁻¹ for Chaetoceros sp., 3,325 cm⁻¹ for Nitzschia sp., and 3,296 cm⁻¹ for Thalassiosira sp.) assigned to the overlapping of –OH and –NH stretching bands of polymeric compounds. The peaks at around 2,923 cm⁻¹ represented the stretching vibration of –CH₃. The peaks at around 1,653 cm⁻¹ were due to the overlapping of C=O stretching vibration and C=C bonds, respectively. The peaks at around 1,550 and 1,450 cm⁻¹ could be assigned to the stretching vibration of N–H and variable angle vibration of–CH₂–, respectively. The peak at around 1,080 cm⁻¹ represented the overlapping of C–O vibration in the alcoholic hydroxyl group and Si–O–Si stretching vibration. These results indicated that the non-living biomass of Chaetoceros sp., Nitzschia sp., and Thalassiosira sp. contained diverse chemical groups, such as carbonyl, hydroxyl, and amine groups, on its surface, which were potential binding sites for the heavy-metal ion (González et al., 2011; He and Chen, 2014).

The BET surface area of non-living Chaetoceros and Nitzschia were 0.7 and 4.6 m² g⁻¹, respectively, while the BET surface area of non-living Thalassiosira was not detected under the same weight condition as the other two species. This result suggested that Nitzschia possesses the highest surface area, followed by Chaetoceros, and Thalassiosira possesses the lowest. The cumulative pore volume of non-living Chaetoceros and Nitzschia were 0.0048 and 0.0293 cm³ g⁻¹, respectively. Moreover, it is clear from Figure 2D that mesopores took up the majority of pore volume on the surface of non-living Nitzschia, followed by macropores, and micropores took up the least. In comparison, the surface of non-living Chaetoceros was occupied by macropores, followed by mesopores, with no micropores on it (Figure 2C). These results indicated that non-living Nitzschia biomass possesses a more abundant pores structure than the other two diatom species.

As shown in Figure 3 and Table 1, the results show that Cd(II) adsorption by all the diatom biomass fit the pseudo-second-order model very well, with a correlation coefficient R ² > 0.88. For initial Cd(II) concentration of 10 and 120 mg L⁻¹, the equilibrium adsorption amount of Cd(II) by Chaetoceros sp., Nitzschia sp., and Thalassiosira sp. were calculated to be 6.14, 6.14, and 5.53 mg g⁻¹ and 111, 106, and 103 mg g⁻¹, respectively, which were in good agreement with the experimental capacities of 6.25, 6.15, and 5.69 mg g⁻¹ and 113, 114, and 107 mg g⁻¹, respectively. The suitability of the pseudo-second-order model indicates that chemical interaction is the rate-controlling step in the whole adsorption process, and the adsorption behavior involves valence forces through sharing of electrons between metal ions and the diatom biomass. Furthermore, the tendency of the adsorption rate constant (k ₂) was calculated according to the model as follows: Nitzschia sp. > Chaetoceros sp. > Thalassiosira sp., suggesting that Cd(II) was adsorbed fastest by non-living Nitzschia, followed by non-living Chaetoceros, and the slowest by non-living Thalassiosira.

The Cd(II) equilibrium adsorption data were fit to the non-linear Langmuir and Freundlich adsorption isotherm models (Figure 4). The calculated parameters corresponding to the two models were summarized in Table 2. For all temperature conditions (T = 288, 298, and 308 K), based on the values of correlation coefficient (R ²), the adsorption data of Cd(II) by three diatom biomass fit both the Langmuir and Freundlich models well. Furthermore, for every non-living diatom biomass, the Langmuir constant (K _L) generally increased with temperature, which indicates that the adsorption of Cd(II) by the diatom biomass was an endothermic reaction process and higher temperature within some range favors the adsorption. The generally higher values of Freundlich constants (K _F and n) under higher temperatures also indicated that the adsorption processes of the three diatom biomass can be promoted by temperature.

As shown in Table 2, for the temperature ranging from 288 K to 308 K, the theoretical q _max of Chaetoceros sp., Nitzschia sp., and Thalassiosira sp. were 238–356 mg g⁻¹, 289–430 mg g⁻¹, and 267–336 mg g⁻¹, respectively, which indicates that Nitzschia sp. possesses better Cd(II) adsorption capacity than Chaetoceros sp. and Thalassiosira sp. The higher BET surface of non-living Nitzschia was probably one of the reasons for its higher adsorption capacity compared with the other two algae species. The highly efficient and newest adsorbents reported elsewhere were selected to compare with the non-living diatom biomass in the present study. As shown in Table 3, in the absence of competitive cations and under low turbidity conditions, the non-living marine diatom biomass in the present study possess much higher Cd(II) adsorption capacity than all the other reported algae species, and also most of the other kinds of bio-adsorbents. In comparison with other adsorbents except for bio-adsorbents, the dry marine diatom biomass still shows better Cd(II) adsorption capacity. Although some reported adsorbents, such as some clay-based materials, nanoscale particles, modified bio-adsorbents, and polymeric substances, showed higher Cd(II) adsorption capacity than the diatom biomass, there is still a long way to go to use these adsorbents in heavy-metal wastewater treatment in future because of complicated modification/preparation processes applied for those adsorbents (Uddin, 2017; Chen et al., 2018). In addition, nano-materials always face difficult separation and potential biological safety risk problems, which make it hard to apply in wastewater treatment. In comparison, marine diatom biomass was not only highly efficient in adsorbing cadmium ions from an aqueous solution, but it is also easy to obtain and eco-friendly. From these aspects, the dry marine diatom biomass would be a competitive adsorbent used for cadmium wastewater in the future.

The thermodynamics parameters, the enthalpy change (△H) and entropy (△S), were calculated from the linear plot of ln K ₀ versus 1/T and are given in Table 2. The values of △H were positive for all adsorption cases, which indicates that the adsorption process was endothermic and Cd(II) ions were adsorbed more efficiently on the diatom biomass under higher temperatures. The value of △H can be used to differentiate chemisorption and physisorption. It was reported by Wang et al. (2012) and Sun et al. (2015) that the adsorption process was mainly physisorptive when △H < 20 kJ mol⁻¹ and was mainly chemisorptive when 80 <△H < 400 kJ mol⁻¹ or 40 <△H < 120 kJ mol⁻¹. Therefore, it can be inferred from the △H values for the three diatom biomass that the Cd(II) adsorption processes of non-living Chaetoceros sp., Nitzschia sp., and Thalassiosira sp. were dominated by chemisorption.

The positive value of △S indicates that there is an increase in the randomness in the system solid/solution interface during the adsorption process (Xie et al., 2011). The positive entropy change was possibly stemming from the changes taking place in the structure of the adsorbent and dispersing of water molecules from Cd(II) ions surrounded by water molecules during the adsorption process (Ozer and Pirinççi, 2006). Negative values of △G for three non-living diatom biomass indicated the spontaneous nature of the reaction between Cd(II) and the biomass. The values of △G became more negative with temperature, which meant that the adsorption process was more favorable at high temperatures.

Investigation of the influences of Na⁺, Ca²⁺, and Mg²⁺ on Cd(II) removal is of great importance due to their general existence in natural water and wastewater. As shown in Figure 5A, the amount of Cd(II) adsorbed by non-living Nitzschia decreased sharply from 108.1 mg g⁻¹ to below 30 mg g⁻¹ when Na⁺, Ca²⁺, and Mg²⁺ at 0.05 mol L⁻¹ were, respectively, added into the adsorption system. As the concentration of Na⁺, Ca²⁺, and Mg²⁺ increases to 0.05–0.2 mol L⁻¹, the amount of Cd(II) adsorbed by non-living Nitzschia decreased by 68.7%–88.95%, 83.3%–95.3%, and 75.1%–92.2%, respectively. According to the significance analysis results by SPSS software, the decreased extent of the amount of Cd(II) adsorbed by the biomass shows no significant correlation to the type of foreign ions (p > 0.05), suggesting similar influences of Na⁺, Ca²⁺, and Mg²⁺ in the range from 0.05 to 0.2 mol L⁻¹ on the amount of Cd(II) adsorbed by non-living Nitzschia. The effects of foreign ions (Na⁺, Ca²⁺, and Mg²⁺) on the adsorption of cadmium ions are mainly ascribed to 1) foreign ions competing with Cd(II) for the limited reactive sites on the surface of non-living Nitzschia, which reduces the adsorption amount of the adsorbates; 2) ionic strength of Na⁺, Ca²⁺, and Mg²⁺ impacting non-living Nitzschia particles aggregation by influencing electrostatic interactions; and 3) ionic strength affecting the activity coefficient of Cd(II) and thus limiting the cadmium ions transfer from the solution to the surface of adsorbent (Peng et al., 2017). Another explanation for the negative effects of Ca²⁺ and Mg²⁺ on the adsorption capacity of adsorbents was that both Ca²⁺ and Mg²⁺ in high concentration (>400 mg L⁻¹) can form hydrates with water “Ca²⁺/Mg²⁺-(H₂O)_n”, and these complexes could cover the surface of the adsorbents, making it less available to adsorbates, and resulting in a decreased number of heavy-metal ions adsorbed (Cheng et al., 2014). It is also noted from the results that the decreased extent of the amount of Cd(II) adsorbed by non-living Nitzschia shows no significant correlation to the concentration of foreign ions from 0.05 to 0.2 mol L⁻¹ (p > 0.05). This would be explained by the possibility that the minimum concentration of the foreign ions (0.05 mol L⁻¹) was high enough to make the adsorption capacity decrease to the lowest level.

Although Ca²⁺ and Mg²⁺ showed negative effects on the amount of Cd(II) adsorbed by the diatom biomass, the concentration of Ca²⁺ and Mg²⁺ is detected in the range of 10–80 mg L⁻¹ (Ca²⁺: 0.00025–0.0020 mol L⁻¹, Mg²⁺: 0.0004–0.0033 mol L⁻¹) in the surface freshwater (Cheng et al., 2014), which was far lower than the concentration set in this study. Therefore, it is likely that the adsorption of Cd(II) would be far less affected than the extent shown in the present study when using non-living Nitzschia for Cd(II) treatment in surface freshwater.

Figure 5B shows the effect of turbidity on the amount of Cd(II) adsorbed by non-living Nitzschia. The amount of Cd(II) per unit mass of non-living Nitzschia sharply decreased by 62.2% when the turbidity increased to 2 NTU and tended to be stable afterward. The suspended particles might have blocked the micro-pores on the surface of non-living Nitzschia and occupied the active sites for Cd(II) adsorption, therefore resulting in the decreased adsorption amount of Cd(II).

Humic acid is a widespread organic macromolecule in the natural environment. It is usually applied in agricultural feed, the petrochemical industry, battery production, etc. It has been reported that humic acid can combine with cations through ionic bonding force or coordination interactions (Kinniburgh et al., 1999). As shown in Figure 5C, the amount of Cd(II) adsorbed by non-living Nitzschia increased evidently with the concentration of additive humic acids. A similar result was reported by Zeng et al. that the adsorption capacity of Cd(II) by the bentonite could be enhanced by humic acids (Zeng, 2008). Meanwhile, it can be seen from Figure 5C that humic acid was capable of adsorbing Cd(II) to some extent (around 25 mg g⁻¹) in the absence of an adsorbent, which was similar to the results reported elsewhere. However, the significantly increased amount of Cd(II) adsorbed by non-living Nitzschia when humic acid was introduced cannot be fully explained by the small adsorption amount of Cd(II) by humic acid itself. As the surface of non-living Nitzschia contains various cations, such as Na⁺, Ca²⁺, Mg²⁺, and Cu²⁺, it can be inferred that the interactions between humic acid and those cations probably simultaneously occurred during the Cd(II) adsorption process. Therefore, the positive effect of humic acid for Cd(II) adsorption by non-living Nitzschia might be ascribed to 1) humic acid combining with some other cations (e.g., Na⁺, Ca²⁺, Mg²⁺, etc.) on the surface of non-living Nitzschia or in the solution, which makes more active sites available for Cd(II) adsorption. 2) Cd(II) was directly complexed by humic acid to form organic bonds with cadmium; 3) humic acid was first adsorbed onto the surface of non-living Nitzschia, and then complexation or ion exchange of Cd(II) with abundant functional groups of humic acids (amines, carboxylic acid, and carbonyl) took place.

The surface morphologies of non-living Nitzschia biomass before and after Cd(II) adsorption were examined by SEM (Figure 6A). The non-living Nitzschia biomass shows an unconsolidated and flocky surface texture. After Cd(II) adsorption, the loose surface becomes caked and lumpish, which might be due to the complexation of cadmium ions with the functional chemical groups on the surface of non-living Nitzschia.

As shown in Figure 6B, after being washed with distilled water, the overall FTIR spectrum of Nitzschia biomass was similar to raw non-living Nitzschia, with an evident shift of the peak assigned to the overlapping of –OH and –NH stretching bands of polymeric compounds. The peak representing the overlap of C=C/C=O became a little narrow. These results suggested that the distribution of chemical groups such as –OH, –NH, C=C/C=O could also be affected by water washing, which was hardly mentioned in most of the reported studies (Yang et al., 2011; Zheng et al., 2011). In comparison, after Cd adsorption, an evident shift of the peak assigned to the overlapping of –OH and –NH stretching bands of polymeric compounds was observed, which was similar to the biomass sample being washed by distilled water. In contrast, the intensity of the peak assigned to –NH (1548 cm⁻¹) became much stronger after Cd adsorption. This result implied that –NH bonds might be involved in the interaction with Cd(II) on the surface of the non-living Nitzschia.

A comparison of XPS spectra survey scans among non-living Nitzschia, distilled water-washed Nitzschia, and Cd(II)-loaded Nitzschia is illustrated in Figure 6C. It can clearly be seen from the figure that the peaks representing the Cd chemical state occurred only in the XPS spectra of Cd(II)-loaded Nitzschia, which demonstrated that Cd(II) was adsorbed onto the surface of non-living Nitzschia. Additionally, it is also noted that the atomic ratio of C 1s on the surface of the non-living Nitzschia decreased from 76.5% to 56.5% after being washed by distilled water, indicating that the organic leaching happened when non-living Nitzschia mixed with water. A similar decrease of C 1s in Cd-loaded Nitzschia was also observed. In order to further understand the functional bonds responsible for Cd ion adsorption, binding energy (BE) profiles of carbon (C 1s), oxygen (O 1s), and nitrogen (N 1s) atoms on non-living Nitzschia, distilled water-washed, and Cd(II)-loaded Nitzschia are shown, respectively, in Figure 7, and the ratio of various C 1s, O 1s, and N 1s bonds which were calculated by peak area is shown in Table 4.

The C 1s spectra of the non-living Nitzschia comprised five peaks with differentiated BE values (284. 6, 285.3, 286.3, 287.9, and 289 eV) via deconvolution. These peaks can be assigned to C=C, C–C, C–O, O=C, and O=C–O bonds, respectively (Beamson and Briggs, 1992; Yang et al., 2011; Deng et al., 2017). After Cd(II) adsorption, the ratio of C=C and C–C decreased, whereas the ratio of C–O and O=C increased, with no notable shifts of binding energy for the five carbon bonds. Meanwhile, after being mixed with distilled water, the changing trend of the carbon bonds was found to be very similar to the diatom sample after Cd(II) adsorption. Consequently, taking organics leaching into account, the decreases of C–C and C=C bonds after adsorption were due to organics leaching. In comparison, the decrease of C–C/C=C was also found for some bio-adsorbents after heavy-metal adsorption (Yang et al., 2011; Zheng et al., 2011). However, the decrease in the C–C/C=C ratio was attributed to partial breakage of the C–C/C=C bonds oxidized by heavy-metal ions (Zheng et al., 2011) or confirming the functional carbon bonds directly based on the change of C 1s after adsorption (Yang et al., 2011), without unveiling the effects of organics leaching on the chemical bonds of the adsorbent surface.

The O 1s spectra of the non-living comprised two peaks with differentiated BE values via deconvolution, which were assigned to O=C and C–O, respectively, with a higher ratio of O=C and a lower ratio of C–O bonds. After being washed with distilled water, the ratio of O=C was found to decrease by 18%, whereas the ratio of C–O increased correspondingly by 18%. Moreover, the binding energy (BE) value of O=C and C–O shifted to a higher level by 0.2 eV. This result suggested that organics leaching could also affect the ratio of oxygen bonds and induce BE shifts of oxygen bonds on Nitzschia. In previous studies, the complexation between O=C and heavy metal ions was usually confirmed by the results that O=C bonds on the bio-adsorbent decreased and C–O bonds increased after heavy-metal adsorption, in which the formation of O-heavy-metal bonds was not observed, and the effect of organics leaching on oxygen bonds was hardly mentioned (Chen and Yang, 2006; Yang et al., 2011; Li et al., 2012). Based on the present results, the involvement of O=C in heavy-metal complexation became ambiguous in those previous studies because organics leaching could also lead to the decrease of O=C and increase of C–O bonds. After Cd(II) adsorption, a decrease of O=C and an increase of C–O bonds were found compared with raw non-living Nitzschia with no new peak representing metallic oxygen observed, which was similar to the result of distilled-water-washed Nitzschia. Moreover, the binding energy (BE) value of O=C and C–O also shifted to a higher level by 0.2 eV compared with raw non-living Nitzschia. Therefore, it can be inferred that the change of oxygen bond distribution on Nitzschia after Cd(II) adsorption was due to organic leaching rather than the complexation of C=O with Cd(II).

The N 1s spectra of the non-living Nitzschia comprised three peaks with differentiated BE values via deconvolution, which can be assigned to pyridinic nitrogen, amine nitrogen, and pyrrolic nitrogen, respectively (Wang et al., 2017; Ji et al., 2018). After being washed with distilled water, the quantities of pyridinic nitrogen, amine nitrogen, and pyrrolic nitrogen decreased by 9.3, 3.8, and 4.8, respectively, and a new peak representing quaternary nitrogen of 18.0% occurred. This result indicated that some of the pyridinic nitrogen, amine nitrogen, and pyrrolic nitrogen were protonated in distilled water. In comparison, after Cd(II) adsorption, the ratio of pyridinic nitrogen decreased by 9.2%, which was similar to the sample after being washed with distilled water. However, different from the sample after being washed with distilled water, the ratio of amine and pyrrolic nitrogen groups after Cd(II) adsorption showed no notable change compared with raw non-living Nitzschia. These results implied that Cd(II) ions interacted with both amine and pyrrolic nitrogen during adsorption, which inhibited the protonation of these two nitrogen bonds. Moreover, the binding energy of amine nitrogen and pyrrolic nitrogen increased from 399.5 and 400.0 to 399.7 and 400.2, respectively, after Cd(II) adsorption (Table 4), which further confirmed the coordination of amine nitrogen and pyrrolic nitrogen with cadmium ions. The increased BE value of the nitrogen atom was because the nitrogen atom donates electrons to cadmium ions, and thus, the electron density around the nitrogen atom decreases (Yang et al., 2011). The greater affinities of amine and pyrrolic nitrogen to Cd ions than pyridinic nitrogen might be explained by the hard and soft acids and bases theory (HSAB) that the soft acid (Cd²⁺) interacts preferentially with the soft base (pyrrolic nitrogen and amine) than the hard base (pyridinic nitrogen) and the soft acid interacts with the soft base, usually via coordination. Similarly, Ji et al. reported a higher affinity of pyrrolic nitrogen to Pb²⁺ than pyridinic nitrogen (Ji et al., 2018).