Characterization
The morphology of adsorbent was understood using SEM micrographs. Figure 1(a, b) shows SEM image revealing the surface characteristics of TW and t-TW, respectively. Both images represent largely smooth surface of the TW and t-TW at same (3000 x) magnification. There were no observable differences before and after treatment with ascorbic acid except the surface roughness appear to slightly increase in the t-TW. The EDX analysis, however, exhibited a slight decrease (~ 2.2%) in oxygen content from 27.7% in TW to 25.5% in t-TW. This might be due to the reduction of some oxygen-rich functional groups33. Some studies suggest biosorption of heavy metals by carboxyl and amine groups present on similar material 22,34, therefore the FTIR analysis was performed as shown in Fig. 1c.
The TW and t-TW constitute numerous functional groups where Cr(VI) can be adsorbed 22,34−37. We observed that the absorbance at certain wavelengths varied by magnitude and/or shifted upon ascorbic acid treatment of TW to synthesize t-TW; the absorbance increased around 1040,1060, and 2873 cm− 1 wavelengths but decreased around 1500, 1800–3200, and > 3550 cm− 1 wavelengths. In Fig. 1 (c) the absorbance bands between 600–900 cm− 1 represent primary and secondary amines and amides; around 1040 cm− 1 represent polysaccharides and aromatic ethers (C-O-C) and silicates (Si-O-Si); at 1319 cm− 1 represent C-N stretch of primary and secondary aromatic amines; at 1420 cm− 1 represent C-O stretch of carbonates and/or O-H plane bend of carboxylic acid; at 1650 represent C = C stretching of carboxylic acid and amide I; between 1600–1750 cm− 1 represent C = O vibration of bonded conjugated aldehydes, ketones, esters, and ketones; between 2850–2920 cm− 1 represent C-H stretch of aliphatic groups; around 3400 cm− 1 represent either O-H stretching of hydroxyl groups i.e. carboxylic acids, alcohols, and phenols or N-H stretching in primary amines, secondary amines, and/or amides 35–37. The antioxidant activity of black tea enhances by ascorbic acid 38. We postulate lessening in carbonyl, epoxy, aromatic ethers, carboxyl, and hydroxyl functional groups due to partial reduction of t-TW by ascorbic acid as observed by other researchers 39,40. These variabilities in functional groups lead to the comparatively enhanced adsorption of Cr(VI) from water by t-TW, as shown in Fig. 2. The experimentally observed equilibrium adsorption capacity of t-TW was up to twenty percent higher than TW at identical experimental settings; shaking speed, temperature, pH, water quality, etc. We observed the maximum experimental adsorption capacities of TW and t-TW to be 193.45 mg.g− 1 and 235.95 mg.g− 1, respectively. These values are comparable to those reported in literature for the tea waste biochar 23 and mixed tea-waste 22. These results warranted further investigation into modeling and optimization of Cr(VI) adsorption on t-TW.
Model development, validation, and diagnostic analysis
The experimentally obtained Cr(VI) removal (%) at different pH, initial Cr(VI) concentration, t-TW dosage, and temperature settings are listed in Table 2. The Cr(VI) removal (%) was in the range of 0–93.3% corresponding to experimental run no. 1 and 22, respectively (Table 2). These two experiments were performed at same temperature (30°C) and initial Cr(VI) concentration (50.5 mg/L) but using different adsorbent dosages (0 and 2 t-TW (g.L− 1)) at variable pH (7 and 2) conditions. From the experimental results, an empirical model representing the relationship between the factors and response (Cr(VI) removal (%)) was developed, as shown below:
Cr(VI) removal (%) = 9.56–22.13 A − 15.02 B + 4.57 C + 1.86 D + 2.95 AB − 3.31 AC − 2.26 AD − 3.47 BC + 9.16 A² + 5.90 B² + 11.29 A²B + 5.40 A²C + 3.41 A²D + 12.40 AB² (1)
The coefficients in the equation represent the linear, quadratic, and cubic terms of the factors. A negative sign in the equation represents antagonistic effect, whereas, a positive term represents synergistic effect of a certain factor (or combination of factors) on the Cr(VI) removal (%) 41,42. The adequacy of the model (Eq. 1), to represent the experimental data, was tested by plotting the experimental values against RSM model predicted values. The RSM model exhibited satisfactory approximation of the actual Cr(VI) adsorption removal (%) as demonstrated by high correlation (R2) in Fig. 3. The correlation coefficient (R2) was calculated to be 0.99, implying that 99% of the variations in the results can be attributed to the studied factors 43. Also, the adjusted correlation coefficient (adj. R2 = 0.97) was very close to the R2. Adj. R2 is usually preferred over R2 because (unlike R2) it only increases upon the addition of statistically significant model term/s 41. The difference between adj. R2 and predicted R2 was 0.23 which was slightly higher than the desirable range (≤ 0.2) 44. This might be due to the complexity of the cubic model and suggests that the model should be used with caution for predicting above or below experimentally validated ranges.
The analysis of variance (ANOVA) was performed to statistically validate the model, as shown in Table 3. The model and all its terms (except “D”) were statistically significant at > 95% confidence level as indicated by p-value below 0.05. The linear term D, representing the effect of temperature on Cr(VI) reduction, was significant only at 85% confidence level. Despite its statistically low significance, it was kept to maintain the model’s hierarchy due to its interaction with the statistically significant terms (AD and A2D) 45. The adequate precision (signal to noise ratio) was 36.56 i.e., well above the required threshold value of four. This indicates the adequate response signals, thereby, expressing the suitability of model to navigate in the experimental design space 41,46. Consequently, the regression analysis and ANOVA validate the model to study the effect of factors on Cr(VI) adsorption removal in our system.
Table 3
Analysis of variance of a reduced cubic model representing adsorption removal of ascorbic acid treated tea-waste.
Source of
variation
|
Sum of
squares
|
Degrees of freedom
|
Mean
square
|
F value
|
p value
|
Significance
|
Model
|
13354
|
14
|
954
|
79
|
< 0.0001
|
significant
|
A:pH
|
3919
|
1
|
3919
|
323
|
< 0.0001
|
|
B:Cr-6
|
1804
|
1
|
1804
|
149
|
< 0.0001
|
|
C:t-TW
|
167
|
1
|
167
|
14
|
0.0021
|
|
D:Temp.
|
28
|
1
|
28
|
2
|
0.1523
|
|
AB
|
139
|
1
|
139
|
11
|
0.0041
|
|
AC
|
176
|
1
|
176
|
14
|
0.0017
|
|
AD
|
82
|
1
|
82
|
7
|
0.0202
|
|
BC
|
193
|
1
|
193
|
16
|
0.0012
|
|
A²
|
2389
|
1
|
2389
|
197
|
< 0.0001
|
|
B²
|
989
|
1
|
989
|
82
|
< 0.0001
|
|
A²B
|
680
|
1
|
680
|
56
|
< 0.0001
|
|
A²C
|
156
|
1
|
156
|
13
|
0.0027
|
|
A²D
|
62
|
1
|
62
|
5
|
0.0393
|
|
AB²
|
820
|
1
|
820
|
68
|
< 0.0001
|
|
Residual
|
182.07
|
15
|
12.14
|
182.07
|
|
|
R²
|
0.99
|
|
Adj. R²
|
0.97
|
|
|
Predicted R²
|
0.74
|
|
Adeq. precision
|
36.56
|
|
|
Effects of variables on Cr(VI) adsorption
To investigate the linear effects of changing the levels of a single factor on the response, one factor effects’ plots were generated as shown in Fig. 4. The figure shows the Cr(VI) removal as effected by pH, initial Cr(VI) concentration, adsorbent (t-TW) dosage, or temperature. The red circles represent experimental design points, black lines represent modeled prediction, and turquoise lines represent the least-significant-difference calculated at ninety-five percent confidence level. Figure 4a shows that Cr(VI) removal decreases with the increase of pH from 4.5 to 9.5 (Level ± 1 from Table 1). At pH 4.5, approximately 40% Cr(VI) removal was observed which decreased to 10 ± 2% as the pH approached to neutrality. The trend continued at basic pH conditions until no removal was observed at pH 8.5. The decrease in Cr(VI) adsorption with the increase in pH was reported in activated carbons 47,48, nanocomposites 49, and organic adsorbents 50 such as ours. The pH of solution effects speciation of metal ions as well as the surface charge of the adsorbent. The point of zero charge of tea waste is nearing pH 5.7 51,52. The Cr(VI) speciation in aqueous solution is driven by pH 49. At pH ≤ 1, Cr(VI) mostly exists as chromic acid (H2CrO4). At pH 1–7, hydrogen chromate (HCrO4 − 1) ion dominates, whereas, above pH 7 only chromate ion (CrO4 − 2) prevails 49,53,54. The gradual decrease of adsorption from acidic to neutral pH can be explained by the gradual conversion of predominant monovalent hydrogen chromate (HCrO4 − 1) to divalent chromate (CrO4 − 2, Cr2O7 − 2) ions. Since the free energy of adsorption of divalent ions is higher than monovalent ions, the divalent ions adsorbed less frequently on t-TW 49. Also, the surface of t-TW deprotonated with the increase in pH of the surrounding aqueous solution leading to the decreased positive surface charges. Consequently, the negatively charged chromate ions experienced electrostatic repulsion at higher pH which resulted in decreased adsorption of chromate ions with the increase of pH.
Figure 4b shows the decline in Cr(VI) removal (%) with the increase in initial Cr(VI) concentration from 25.5 to 50.5 (mg.L− 1). The trend is similar to that of pH, except that it is less pronounced (exhibited by less steep slope) when compared with pH. The decrease in adsorption removal - upon increasing initial Cr(VI) concentrations - can be attributed to the unavailability of sufficient adsorption sites (on t-TW) at higher initial Cr(VI) concentrations 53. This hypothesis was supported by the increase in Cr(VI) removal with the increase in t-TW dosage as shown in Fig. 4c where nearly linear increase in Cr(VI) adsorption occurred with the increase of t-TW dosage. Figure 4d indicates the insignificant increase in Cr(VI) removal with the increase in temperature from 20 to 50°C. A little (0.7%) increase in Cr(VI) removal was observed upon increasing temperature from 20 to 30°C which is difficult to attribute to increase in mass transfer rate with the increase of temperature 43. Also, the term is significant only at 85% confidence level (Table 3), therefore such a minor change can be deemed indifferent. To summarize, one factor at a time plots suggest the (i) highest impact of pH on Cr(VI) removal followed by initial Cr(VI) concentration, adsorbent dosage, and temperature, (ii) the increase in Cr(VI) removal from water with the decrease in pH and initial Cr(VI) concentration, and (iii) the increase in Cr(VI) removal from water with the increase in the adsorbent dosage.
Further analysis of the model parameters was performed using three-dimension response surface plots. RSM allows to investigate combined effects of factors on response with the aid of surface plots. Surface plots can be generated by varying two variables at a time - and observing their effect on the response - while keeping the others constant at a certain level (usually middle level) 55. Table 3 shows the significant interaction of the terms AB, AC, AD, and BD, therefore, response surfaces were generated to study these interactions. Figure 5 shows the combined effects of pH and initial chromium concentration (AB), pH and adsorbent dosage (AC), pH and temperature (AD), and initial chromium concentration and adsorbent dosage (BC). The decrease in Cr(VI) adsorption removal was noted with the increase in pH in combination with the initial Cr(VI) concentration (Fig. 5a), t-TW (Fig. 5b), and temperature (Fig. 5c). The pH dominated effects, like one factor effect of pH (Fig. 4a), can be explained by the prevalence of less affinitive chromate ions (at high pH) leading to the less frequent electrostatic interactions with adsorbent surface resulting in reduced adsorption removal of Cr(VI) 47,49. Figure 5d shows the decrease in adsorption removal with the decrease in adsorbent dosage and increase in initial Cr(VI) concentrations. This can be due to the unavailability of adsorption sites for Cr(VI) adsorption at lower t-TW dosages as discussed earlier 47. To conclude, one-factor plots and response surface graphs establish the most pronounced effect of pH on Cr(VI) removal which is consistent with earlier similar studies 48,49,53,56. These observations require the optimization of parameters to effectively remove Cr(VI) from water using t-TW in drinking water pH range.
Process optimization
The adsorption removal of Cr(VI) was optimized in drinking water pH range (6.5–9.5) using t-TW as adsorbent material. The initial Cr(VI) concentration was fixed at 1 mg.L− 1 and complete removal was targeted as shown in ramp plots (Fig. 6a). The model predicted 99% removal, thereby, limiting residual concentration of Cr(VI) at 0.01 mg.L− 1. It should be noted that the targeted residual Cr(VI) concentration - at optimized conditions - was well below the allowable concentrations recommended by US-EPA (0.1 mg.L− 1) and WHO (0.05 mg/L) 8,9. Experiments were conducted at prescribed settings and no residual Cr(VI) was detected. We assume ≥ 99% adsorption removal of Cr(VI) at optimized conditions owing to the method detection limit of 0.002 mg.L− 1. Flag point in Fig. 6b represents the location of optimized parameters in the experimental design space.
Adsorption equilibrium
The established isotherm models were fitted to the experimental data to comprehend the adsorption of Cr(VI) on t-TW. The mathematical expressions describing Langmuir, Freundlich, Temkin, and D-R models are:
$$Langmuir: {q}_{e}=\frac{{q}_{m}{K}_{L}{C}_{e}}{1+ {K}_{L}{C}_{e}}$$
2
$$Freundlich: {q}_{e}={K}_{f}{{C}_{e}}^{\frac{1}{n}}$$
3
$${Temkin: q}_{e}=\frac{RT}{{b}_{T}}\text{l}\text{n}\left({A}_{T}{C}_{e}\right)$$
4
$$Dubinin-Radushkevich :{q}_{e}={q}_{DR}{e}^{-{K}_{DR}{\epsilon }^{2}}$$
5
where \({C}_{e}\) (mg.L− 1) is the equilibrium concentration, \({q}_{m}\) (mg.g− 1) is the Langmuir maximum adsorption capacity, \({K}_{L}\) (L.mg− 1) is the Langmuir adsorption constant, \({K}_{f}\) ((mg.g− 1)(L.mg− 1) 1/n) is the Freundlich constant, n is the Freundlich exponent, \({A}_{T}\) (L.mg− 1) is the Temkin isotherm equilibrium binding constant, \({b}_{T}\) (J.mol− 1) is the Temkin isotherm constant, \({q}_{DR}\) (mg.g− 1) is the D-R maximum sorption capacity, \({K}_{DR}\) (mol2.kJ− 2) is the D-R constant related to sorption energy, and \(\epsilon\) (= RT ln (1/1+\({C}_{e}\))) is the Polanyi potential.
Figure 7 shows Cr(VI) adsorption isotherms and the corresponding equilibrium parameters are reported in Table 4. The adsorption equilibrium was best described by Temkin > Langmuir > Freundlich > D-R models, based on regression co-efficient values. The Temkin model assumes uniform distribution of binding energy sites on t-TW surface and linear decrease in heat of adsorption of Cr(VI) species as the adsorption progresses 57,58. Similarly, Langmuir also suggests monolayer adsorption 59. The separation factor was calculated which suggests Langmuir adsorption of Cr(VI) on t-TW to be “favorable” 58. Nevertheless, Freundlich and D-R fittings were quite significant as well which propose multilayer adsorption and pore-filling due to heterogeneous surface 58,60−63. Therefore, we assume the adsorption of Cr(VI) on t-TW to be largely monolayer but occasionally multilayer due to mixture of uniform and non-uniform surface as shown in SEM micrographs (Fig. 1).
Table 4
Adsorption equilibrium and kinetics parameters
Model
|
Parameters
|
R²
|
Langmuir
|
KL = 1.75 (L.mg− 1); qm = 210.97 (mg.g− 1);
|
0.95
|
Freundlich
|
KF = 91.39 (mg.g− 1). (L.g− 1)4.5; n = 4.5
|
0.94
|
Temkin
|
bT = 79.94 (kJ.mol− 1); AT = 39.9 (L.g− 1)
|
0.97
|
Dubinin- Radushkevich (D-R)
|
EDR = 266 (J.mol− 1);
qDR = 233.7(mg.g− 1); KDR = 7.05 (mol2.k− 1J− 2)
|
0.90
|
PFO
|
k1 = 0.12 (1.h− 1); qe = 159 (mg.g− 1)
|
0.95
|
PSO
|
k2 = 0.001 (g.mg− 1.h− 1); qe = 176 mg.g− 1
|
0.98
|
Intraparticle
|
KIP(1) = 34.52 (mg.g.h− 1), C(1) = 0.81 (mg.g− 1),
KIP(2) = 0.68 (mg.g.h− 1), C(2) = 111 (mg.g− 1)
|
0.97, 0.96
|
Adsorption kinetics
Adsorption kinetics describes the rate and mechanism of the adsorption process. Figure 8 shows the saturation of t-TW surface with Cr(VI) over time. It was observed that adsorption kinetics was rapid for first eight hours, probably due to the abundance of available adsorption sites for the initial adsorption. Afterwards, adsorption of Cr(VI) progressed at a relatively slower rate from 8–24 h and reached equilibrium after two days. The experimental kinetic data were fitted to the pseudo-first-order (PFO), pseudo-second-order (PSO), and intraparticle models 43,64,65:
where \({q}_{t}\) is the adsorption capacity at time t, \({k}_{1}\) (1.h− 1) is the PFO rate constant, \({k}_{2}\) (g.mg− 1.h− 1) is the PSO rate constant, \({K}_{ip}\) (mg.g− 1.h− 0.5)) is the intraparticle diffusion rate constant, and C (mg.g− 1) is the intercept of intraparticle diffusion plot.
Figure 8 shows PFO, PSO, and intraparticle diffusion kinetic models fitted to the experimental data and their corresponding kinetic parameters are tabulated in Table 4. PSO kinetic model appears to reasonably fit the data as evident from high regression coefficient value of 0.98. Furthermore, the PSO closely predicted the experimental equilibrium adsorption capacity. Therefore, we assume Cr(VI) adsorption on t-TW to be a physisorption phenomenon where the adsorption rate was proportional to the availability of adsorption sites 43,64. Intraparticle diffusion model postulates that the intraparticle diffusion would be the sole rate-limiting step if the plot of qt vs t0.5 crosses the origin 65. The adsorption of Cr(VI) on t-TW generated two step multi-linear intraparticle diffusion plots (Fig. 8b): step 1 from 0–8 h and the step 2 from 8-120 h. Nearly sixty percent of Cr(VI) was adsorbed on t-TW during the step 1. Also, the slope of the step 1 is steeper than that of step 2, depicting a rapid transport of Cr(VI) ions from bulk to the external surface of t-TW 66,67. The corresponding parameters of the intraparticle diffusion plots are provided in Table 4. The intraparticle model fits the experimental data reasonably well as evident from high regression co-efficient values. The intercept of the first step (C(1)) nearly passes the origin (C(1) = 0.81 (mg.g− 1)) suggesting that the initial adsorption was controlled by intraparticle diffusion. Therefore, the adsorption of Cr(VI) on t-TW surface was largely directed by boundary layer of the adsorbent and was independent of Cr(VI) mass transfer in the aqueous system 43,68.
Comparison with other adsorbents
The adsorption capacity of t-TW compared to other adsorbent materials, reported in literature, is presented in Table 5. The adsorption capacity of t-TW is 232 mg.g− 1 which is significantly higher than several low-cost and some advanced adsorbents. Moreover, the ascorbic acid treatment to synthesize t-TW is a simple process which didn’t require any pre-treatment of adsorbent surface. The pre-treatment step is usually an energy intensive process which requires hazardous and corrosive chemicals such as persulfates, mineral acids, peroxides, and alkalis 69. Therefore, ascorbic acid mediated t-TW synthesis can be classified as green route towards material functionalization.
Table 5
Comparison of adsorption of Cr(VI) with other adsorbents.
Sr. no
|
Adsorbent
|
Cr (VI)
(mg.g-1)
|
Ref.
|
1
|
Rice husk
|
52.13
|
(Sugashini, S., et.al, 2015)17
|
2
|
tamarind hull
|
81.1
|
(Verma, A., et.al, 2006)19
|
3
|
Industrial waste
|
15.24
|
(Gupta, V., et.al, 2010)20
|
4
|
peanut shell
|
12
|
(Al-Othman, Z. A., et.al, 2012)21
|
5
|
Ionic solid impregnated phosphate chitosan
|
266.67
|
(Kahu, S. S., et.al, 2016)70
|
6
|
KOH-activated activated carbon
|
315
|
(Khezami, L. & Capart, R., 2005)71
|
7
|
Chitosan/aluminum–lanthanum mixed oxyhydroxide (CSALMOH)
|
78.9
|
(Preethi, J., et.al, 2017)72
|
8
|
Mesoporous silica embedded with magnetite nanoparticles
|
50.51
|
(Hozhabr Araghi, S., et.al, 2015)73
|
9
|
Tea waste biochar
|
197.95
|
(Khalil, U. et al., 2020)23
|
10
|
Ascorbic acid treated tea-waste
|
232.2
|
This study
|