**Application of Pineapple Leaves as Adsorbents for Removal of Rose Bengal from Wastewater: Process Optimization Operating Face-Centered Central Composite Design (FCCCD)**

### **Siham S. Hassan, Ahmed S. El-Shafie , Nourhan Zaher and Marwa El-Azazy \***

Department of Chemistry and Earth Sciences, College of Arts and Sciences, Qatar University, Doha 2713, Qatar; s.hersi@qu.edu.qa (S.S.H.); aelshafie@qu.edu.qa (A.S.E.-S.); nm1601246@qu.edu.qa (N.Z.)

**\*** Correspondence: marwasaid@qu.edu.qa

Academic Editors: Chiara Bisio and Monica Pica Received: 4 July 2020; Accepted: 5 August 2020; Published: 18 August 2020

**Abstract:** Adsorptive removal of rose bengal (RB) from contaminated water samples was approached using pineapple leaves (PAL). Three adsorbents were utilized for that purpose; raw pineapple leaves (RPAL) and the thermally activated bio-waste leaves at 250 and 500 ◦C. Two measures were executed to evaluate the functionality of exploited biomasses; percentage removal (%R) and adsorption capacity (*qe*). Face-centered central composite design (FCCCD) was conducted to experiment the influence of variables on the %R. Dose of PAL as adsorbent (AD), concentration of RB (DC), pH and contact time (CT), were the inspected factors. Existence of functional groups and formation of activated carbon was instigated employing Fourier-transform infrared (FT-IR) and Raman spectroscopies. Scanning electron microscopy (SEM) and energy-dispersive X-ray spectroscopy (EDX) analyses were used to explore surface features. Thermal behavior of adsorbents was studied using thermogravimetric analysis (TGA). The surface area and other surface structural properties were established using the Brunauer Emmett-Teller (BET) analysis. An amount of 92.53% of RB could be removed with an adsorption capacity of 58.8 mg/g using a combination of pH 5.00 ± 0.20, RPAL dose of 0.05 mg/50 mL, and 10-ppm RB for 180 min. Equilibrium studies divulge a favorable adsorption that follows the Freundlich isotherm. Pseudo-second-order model explains the observed adsorption kinetics.

**Keywords:** green adsorbents; pineapple leaves; rose bengal (RB) dye; face-centered central composite design (FCCCD), percentage removal (%R); adsorption capacity (*qe*)

### **1. Introduction**

Water is perceived as the most important renewable source of life, where surface and ground water play major roles in agriculture, livestock production, hydropower generation, etc. The rate of growth of the world population is increasing day after day. This escalating growth is logically associated with several environmental concerns. Water pollution is one of the most serious apprehensions that living creatures have ever faced, if not the most challenging at all. The quality of water is particularly significant for human health. As per the World Health Organization (WHO) reports, poor water quality is responsible for 2.2 million deaths annually. Moreover, more than 2/3 of infant deaths stem from waterborne diseases [1–4].

Numerous contaminants contribute to water pollution. Among these pollutants, heavy metals, anions (sulfates, phosphates, fluoride, etc.), dyes, pesticides, fertilizers, and pharmaceuticals are the most common [4–10]. Dyes, the topic of the current investigation, are widely applied in various industries, e.g., paper, cosmetics, paint and textiles production, food processing, etc. Discharge of the industrial effluents into water bodies causes not only a direct mutilation of water physicochemical

features (such as color, pH, salinity, organic carbon content, etc.), but also instigates detrimental effects on the ecosystem and consequently the human health. This effect is exacerbated by the diverse chemical structure of these dyes and their resistance to biodegradation [11–13].

Rose bengal (RB), a basic xanthene dye, also known as 'C.I. 45440 and C.I. Acid Red 94′ , is chemically recognized as disodium-4,5,6,7-tetrachloro-3′ ,6′ -dihydroxy-2′ ,4′ ,5′ ,7′ -tetraiodo-3*H*-spiro [isobenzofuran-1,9′ -xanthen]-3-one (molar mass: 1017.64 g/mol). Sodium salt of RB is commonly used in diagnosing eye damage via staining the corneal and conjunctival cells. Other applications of RB include treatment of certain cancers (melanoma and breast cancers), skin conditions such as psoriasis and also as antibacterial. Moreover, RB is extensively used in fabric and photochemical manufacturing. Nonetheless, RB has shown serious impacts on human health, especially when it gets in contact with skin and eyes causing discomfort, irritation, redness and blistering [14–18]. Removal of RB from wastewater has been done implementing various methods such as photo-degradation, nanofiltration and adsorption [19–22].

Adsorption is one of the most promising strategies for wastewater treatment. On one hand, adsorption is convenient, easy to maneuver, and can be conducted using readily available materials. Conversely, dyes in specific are premeditated to be chemically stable with long–standing photolytical properties. Most of the used strategies for removal of dyes require pre- and post-treatment steps. In addition, majority of these methods are either impractical (request a tedious experimental setup) or expensive with reduced removing capabilities [6–11,18,23,24]. Moreover, some of these techniques might not be efficient at low pollutant concentrations. Adsorption is therefore a reasonable choice. Developing the model adsorbent and how the adsorption process is conducted are the keywords in managing the adsorption process. Agricultural as well as industrial wastes represent a real burden on the ecosystem if not appropriately recycled and reprocessed. Sources of agricultural wastes are variable. Yet, by-products of the agricultural processing such as peels, pits, shells, leaves, etc. represent important naturally occurring resources that are copiously available and should be thoroughly thought of for the production of value-added materials [24].

Pineapples (PA, *Ananas comosus*, Family: Bromeliaceae) is a perennial herbaceous plant. PA fruit is mostly planted in coastline and tropical areas. In India, for example, PA fruits are grown on around 2,250,000 acres of land. The first bud of the leaves looks attractive. Later on, leaves become stiff; sword shaped and spirally assembled around the fruit [25,26]. Leaves represent the waste biomass of PA fruits and are commonly used as a source of natural fibers. Amount of waste produced from PA (leaf waste) is worrying, where approximately 20,000–25,000 tons per acre are left out after the harvesting process [27]. Leaf fibers consist of mainly holocellulose and lignin with minor amounts of ash [28]. Raw and activated PAL have shown a promising removal potential for different kinds of contaminants. Table 1 shows an evaluation for the performance of PA through different studies with different adsorbates [29–35].

As previously mentioned, having an ideal adsorption process could be managed by not only developing the model adsorbent, but also by engineering the adsorption process and more specifically the influencing variables. Different parameters are known to affect the interaction between the adsorbent and the adsorbate such as adsorbent dosage (AD), concentration of the adsorbate (DC), contact time (CT), pH, surface area, as well as the nature of the adsorbent and the pollutant. The conventional strategy for investigating the influence of these variables on the adsorption capability of an adsorbent is to scrutinize the effect of a single variable per time (univariate analysis). This stratagem and in addition for being time and effort consuming, involves several experimentations, an issue that jeopardizes the method greenness. Moreover, this univariate-based strategy does not yield the adequate amount of data that enable the researcher to draw the full picture for the adsorption process. Nevertheless, and since the objective is to build a green bioremediation strategy, coupling of the adsorption process to factorial designs would overcome these concerns [36].

**Table 1.** Evaluation of the performance of pineapple leaf (PAL) processed in current work compared with other studies used PLP as adsorbent for removal different adsorbates.


\* ND: Not Determined.

Offering irresistible advantages including saving of time, efforts, and resources, a response surface methodological approach (RSM)–face-centered central composite design (FCCCD) will be utilized in the current approach to optimize the investigated responses. Factorial levels for four independent variables will be adjusted with the target being set to maximize the removal of the studied contaminant (RB) using PAL (raw and thermally treated at 250 and 500 ◦C, labelled as TTPAL250 and TTPAL500, respectively) as adsorbents. The amount of RB dye adsorbed will be analyzed using spectrophotometry. TGA, FT-IR, SEM, Raman, EDX, CHN, and BET analyses will be used to characterize the prepared adsorbents. To further study the nature of the adsorbents and adsorption process, both kinetic and equilibrium studies will be performed.

#### **2. Results and Discussion**

#### *2.1. Selection of the Best Performing Adsorbent*

Performance of the three prepared adsorbents was measured in terms of %R and the adsorption capacity (*qe*) and using Equations (1) and (2), respectively. Table 2 shows a comparison between the three prepared adsorbents under the same conditions. As per the results revealed in Table 2, RPAL had the highest %R and *q<sup>e</sup>* and therefore was further used in the subsequent studies:

$$\text{L}\,(\% \text{R}) = \frac{\text{C}\_0 - \text{C}\_{\varepsilon}}{\text{C}\_0} \times 100\% \,\text{\textdegreeline }\tag{1}$$

$$\mathbf{u}(q\_{\varepsilon}) = \frac{\mathbf{C}\_{0} - \mathbf{C}\_{\varepsilon}}{\mathbf{W}} \mathbf{V} \tag{2}$$

where *C*<sup>0</sup> (mg L−<sup>1</sup> ) denotes the initial concentration of RB solution, *C<sup>e</sup>* is the concentration of the RB solution at equilibrium, V stands to the volume of the solution (L), and W is the weight of the adsorbent used (g).

**Table 2.** Performance of PAL—based adsorbents in terms of %R and *qe*. Testing adsorption performance was conducted using a variable blend of pH = 7.00 ± 0.20, DC = 50 ppm, AD = 50 mg/15 mL, CT = 30 min. The responses shown were calculated using Equations (1) and (2).


#### *2.2. Response Surface Methodology (RSM): FCCCD*

As previously mentioned, the purpose of the current approach is to investigate and optimize the adsorption capability of PAL to RB dye from artificially contaminated water samples. The novelty of the current approach stems from using a multivariate platform that surmounts all the previous cons of the univariate approach. FCCCD, as mentioned, was the design of choice, where the impact of four variables on a single response was assessed and optimized. Central composite designs (CCD) usually contain built-in points from preceding full/fractional designs. In the current case, a full factorial design was the preceding design and the value of alpha (α) or the distance between the axial points and the center was equal to one, denoting a FCCCD [36]. The measured response (%R) was calculated using the formula shown in Equation (1). Conducted experimental runs (as executed by the design setup) as well as the working factorial limits accompanied by the observed and predicted responses are shown in Table 3.


**Table 3.** Independent factors and their levels together with the observed and predicted dependent variable and the FCCCD matrix.

\* Blk: Block; \*\* Obs: observed readings; \*\* Pred.: predicted readings; \*\*\* RE = Relative error = / (*Measured value* − *Actual value*)/Actual value /.

#### *2.3. Investigation of Statistically Significant Variables*

In order to investigate the statistical significance of tested variables, Pareto chart of standardized effects, normal and half-normal probability plots alongside with analysis of variance (ANOVA) were implemented. Pareto chart (Figure 1) shows that AD (B) is the most statistically effective factor, followed by the effect of pH (A). It can be also observed that the CT is not that much effective compared to the other factors, however the squared interaction (CT × CT) was the third most influencing variable. The interaction of the CT × AD was the least effective factor on the %R of RB dye. Similar conclusions were obtained using the analysis of variance test (ANOVA) at 95.0 confidence interval (95.0 CI). ANOVA results are shown in Table 4. F-value is shown for every model term and is sufficiently large in case of statistically significant variables. As shown in the table as well, variables with a significance level (*p*-value) less than 0.05 are statistically significant, and the opposite is true. Table 4 also shows that lack-of-fit has a *p*-value of 0.633 (statistically not significant) inferring goodness-of-fit.

λ α

′ <sup>λ</sup>− λ

%R.ହ = 12.46 − 1.442 pH + 1134 AD − 0.450 DC − 0.1495 CT − 7465 AD × AD

+ 0.000810 CT × CT − 42.21 pH × AD + 0.0468 pH × DC + 0.00668 pH × CT − 0.448 AD × CT − 0.001133 DC × CT,

**Figure 1.** Pareto chart of standardized effects following response transformation.


**Table 4.** Analysis of variance (ANOVA) for the transformed response.

\* DF is degrees of freedom SS is sum of squares and MS is mean of squares.

It is noteworthy to mention that response surface regression was performed versus blocks, pH, AD, DC as well as the CT employing Box-Cox transformation [37] where the transformation factor, λ = 0.75 and backward elimination of terms (α to remove = 0.1) was used, Equation (3):

$$\text{(Transformed response)}\,\text{Y} = (\text{Y}\_{\lambda} - 1) \lambda \text{ (transformation factor)}\tag{3}$$

The outcome of the response surface regression is the following mathematical paradigm shown in Equation (4):

$$\begin{aligned} \text{\%} \text{R}^{0.75} &= 12.46 - 1.442 \text{ pH} + 1134 \text{ AD} - 0.450 \text{ DC} - 0.1495 \text{ CT} - 7465 \text{ AD} \times \text{AD} \\ &+ 0.000810 \text{ CT} \times \text{CT} - 42.21 \text{ pH} \times \text{AD} + 0.0468 \text{ pH} \times \text{DC} \\ &+ 0.00668 \text{ pH} \times \text{CT} - 0.448 \text{ AD} \times \text{CT} - 0.001133 \text{ DC} \times \text{CT}, \end{aligned} \tag{4}$$

Equation (4) shows that increasing the pH value would reduce the %R. Conversely, increasing the dose of RPAL would enhance the removal of RB. Model summary shows that the value of R<sup>2</sup> was relatively high (R<sup>2</sup> = 96.95%) and close to the value of R2—adjusted (R<sup>2</sup> (adj) = 94.27%), indicating the linearity of the proposed model. The value of R2—predicted was also high (R<sup>2</sup> (pred) = 89.99%), implying that the proposed model is significantly capable of detecting new observations. This finding could be further confirmed by referring to Table 3 where both experimental and predicted values are revealed together with the difference between the experimental and actual values, relative to the actual values expressed as the relative error (RE). The shown error is relatively small reflecting a close match between observed and predicted responses.

#### *2.4. Contour Plots of %R and Surface Optimization*

Figure 2 illustrates the two–dimensional (2D) plots for the measured fitted response surface. Each of the shown panels reveals the effect of two factors on %R. As shown in the attached legend, the dark red color implies a lower %R, while dark grey color means higher %R. Having the upper left panel as an example (AD × pH), having an AD of 0.042–0.048 g/50 mL and a pH level of 5.00–5.20, the %R is in the range of 50%–60%. Similar conclusions can be obtained from the rest of panels for each factorial combination.

**Figure 2.** Response contour plots for the %R. Dark grey regions represent regions where maximum %R could be obtained using the factorial combination in each panel.

A typical strategy to deal with how a mixture of factorial settings satisfies the destinations that they were setup for is the use of "optimization plot"—a tool offered by Minitab to optimize the measured response. As shown in Figure 3, the objective was set to attain a 100% removal of RB, and the variable settings were fluctuated to achieve the objective. As shown, a blend of the tested variables at the level denoted as 'Cur' would produce a response value of 92.53%. The desirability value (*d*) was high enough, entailing the favorability of the mentioned blend. Figure 3 also shows that increasing the dose of RPAL enhances the adsorption process. This can be attributed to the increase in the number of adsorption sites available for the uptake of RB. The figure also shows that increasing the pH would decrease the %R, and similarly the DC. Impact of CT and as shown in the figure has a varying effect, where increasing the CT from 5 to 92.5 min. has resulted in reduced %R, while with increasing the time from 92.5 to 180 min., removal was improved. These findings are similar to the conclusions obtained from the mathematical paradigm described in Equation (4). Yet, explanation of these findings will be considered in lights of surface chemistry and nature of the dye throughout the next subsections.

−

−

**Figure 3.** Optimization plot. A factorial combination of pH = 5.00 ± 0.20, AD = 0.050 g/50 mL, DC = 10 ppm and CT of 5 min. would achieve %R = 92.53%.

#### *2.5. Adsorbent Characterization*

#### 2.5.1. Thermogravimetric Analysis (TGA)

Thermogravimetric analysis of RPAL was done under N<sup>2</sup> with a heating rate of 10 ◦C/min. The data represented in Figure 4 shows that the weight loss for the RPAL sample occurs over three steps as follows:


**Figure 4.** TGA graph of air—dried pineapple leaves (RPAL).

− As shown from the TGA findings, thermal treatment of the RPAL might have resulted in the evaporation of the small molecules such as H2O, CO, and CO2. Absence of these functionalities in the

− −

−

−

− −

−

−

−

−

− −

thermally treated samples would explain their diminished adsorption capabilities compared to the RPAL sample.

#### 2.5.2. Fourier Transform Infrared Spectroscopic Analysis (FT-IR)

FT-IR spectra of RPAL and TTPAL250 are given in Figure 5. As previously indicated, PAL are mainly composed of lignocellulosic material [28,29,32]. The obtained spectra show the existence of almost the same peaks in the two samples but with lower intensity in the thermally treated one due to the decomposition of lignocellulosic material, a finding that explains the subordinate adsorption capability of the later compared to the former [33]. The obtained data show a broad absorption band centered at 3325 cm−<sup>1</sup> for the RPAL and 3318.4 cm−<sup>1</sup> for TTPAL250. This peak could be assigned to the hydrogen–bonded—OH vibration of the cellulosic structure of the RPAL. In addition, it could be attributed to N-H group which is confirmed later by the presence of a high concentration of nitrogen in both raw and thermally treated samples in the CHN analysis. The spectra also show the presence of the absorption band at 2913–2920 cm−<sup>1</sup> in both samples, which could be ascribed to the C-H stretching of aliphatic—CH groups. The absorption bands at 1595–1585.8 cm−<sup>1</sup> confirm the presence of bending N-H of amines. The two bands at 1365 and 1375 cm−<sup>1</sup> can be assigned to bending—OH. The absorption band at 1034.3 cm−<sup>1</sup> for the RPAL and 1033.5 cm−<sup>1</sup> for TTPAL250 can be ascribed to the presence of C–O stretching. The FT-IR results confirm the presence of surface functional groups that should have played an important role in the adsorption of RB onto PAL.

− **Figure 5.** FT-IR spectra of RPAL and TTPAL250.

− − − <sup>−</sup> − <sup>−</sup> − − − − − − − By combining the FT-IR data together with the FCCCD analysis findings, it can be recognized that the pH has a substantial influence on RB sorption process. Measurements were made at three pH values 5.00, 8.00, and 11.00. These values were carefully selected, where RB had the same absorption maxima in the three solutions. Moreover, the color of RB disappeared at pH less than 4.00. As per the design analysis, biosorption of RB onto RPAL was maximum at the acidic side (pH = 5.00 ± 0.20) and further elevation in the pH has resulted in a diminutive removal, Figure 3. RB and as previously reported, is an anionic dye with a pKa value of 4.50 [38,39]. Therefore, at pH > pKa, RB will start to be ionized (deprotonated, negatively charged). On the other hand, the surface of RPAL at the acidic side and as per the FT-IR analysis might have some positively charged functionalities. The existence of negatively charged RB on the positively charged RPAL surface would encourage electrostatic interaction. Conversely, at pH = 11.00 surface of RPAL will be negatively charged, therefore, less interaction between RB and RPAL surface. Similar results for better sorption of RB in acidic media has been

previously reported using different adsorbents such as Fe (III)–montmorillonite [16], chitosan–TiO<sup>2</sup> nanocomposite [17], and bottom ash [40]. −

#### 2.5.3. Raman Analysis

Raman spectra of raw and thermally treated pineapples are shown in Figure 6. The obtained spectra show the absence of any peaks in the range between 1000 to 2000 cm−<sup>1</sup> in the raw sample. This could be explicated taking in consideration that carbon in the raw sample exists in the form of organic matter. Contrariwise, the Raman spectra of the burnt samples (TTPAL250 and TTPAL500) show two peaks which could be ascribed to the D–and G–bands at approximately 1351 cm−<sup>1</sup> (D–band) and 1585 cm−<sup>1</sup> (G–band). It is imperative to mention that these two bands are characteristic peaks for carbon materials. In addition, the resulted D–, and G–bands pattern is close to the bands present in graphene oxide [41]. Besides, the D–band reflects the carbon lattice properties including defects and sizes, but the G–band shows the stretching of C-C in sp<sup>2</sup> system [42]. Furthermore, the ration between intensity of D–band to G–band was calculated (ID/IG) and compared for the two thermally treated samples. Interestingly, the ID/I<sup>G</sup> for TTPAL250 was 0.90 compared to 1.07 for TTPAL500. This finding confirms the fact that the number of defects has increased by increasing the burning temperature. Yet, it can be also observed that the burning process (carbonization) might have resulted in the elimination of some essential functional groups, which in turn might have an important role in the diminished removal efficiency of the TTPAL250 and TTPAL500 compared to RPAL sample. − − − <sup>−</sup> − <sup>−</sup> − − − − − − −

**Figure 6.** Raman spectra of the raw pineapple leaves (RPAL) and the thermally treated samples (TTPAL250 and TTPAL500).

#### 2.5.4. Scanning Electron Microscopy Analysis (SEM)

The surface structure of the raw and the thermally treated PAL was explored using the scanning electron microscope (SEM). The SEM micrographs presented in Figure 7 showed that the RPAL (Figure 7A) has plain surface without any pores and the same was also observed following the burning process at 250 ◦C (Figure 7B). On the other hand, the surface has completely changed after burning at 500 ◦C. Figure 7C shows the presence of high porous surface compared to the raw material, confirming the formation of carbonaceous material with advanced pore structure and the loss of organic matter after burning at 500 ◦C. These findings are in a good match with the obtained data by FT-IR and TGA analyses. Furthermore, EDX analysis shows the effect of the burning process on the concentration of carbon and oxygen. Results show that carbon content has increased from 75.79% in the RPAL to 82.90% in the burnt sample (Figure 7D,E). In addition, the oxygen content has decreased from 22.91% in the RPAL to 10.27% in RPAL500. This decrease might be attributed to the loss of water oxygen during the burning process, an issue that might have a negative impact on the removal efficiency of the thermally treated samples and as was confirmed by the FT-IR and Raman analyses.

−

−

**Figure 7.** The upper panel is the SEM micrographs of RPAL (**A**), TTPAL250 (**B**), and TTPAL500 (**C**). The lower panel is the EDX analysis of RPAL (**D**), and TTPAL500 (**E**).

#### 2.5.5. Carbon, Hydrogen, and Nitrogen Analysis (CHN)

Data shown in Table 5 represent a comparison between three samples RPAL, TTPAL250, and TTPAL500 in terms of the percentage Carbon, Hydrogen, and Nitrogen. The collected data show that the %C and %N has increased following the thermal treatment in contrast to the %H. These findings indicate that the burning process might cause the loss of hydrogen in crystalline and physical water in contrast to the carbon concentration, which has increased because of the conversion of the biomass into carbon during the burning process.


**Table 5.** CHN Elemental analysis of the prepared adsorbents.

#### − − 2.5.6. Brunauer–Emmett–Teller (BET) Surface Area Analysis

− Table 6 shows the measured BET surface area and the total pore volume of the three adsorbents using N<sup>2</sup> adsorption–desorption measurements. The obtained data show that the surface area of RPAL is 4.59 m<sup>2</sup> /g and this area has increased (almost doubled) following thermal treatment to 9.81 m<sup>2</sup> /g for TTPAL500 with no much difference between TTPAL250 and TTPAL500. On the other hand, the total pore volume has increased from 0.016 to 0.041 cm<sup>3</sup> /g for RPAL and TTPAL500, respectively. This increase in the pore volume is confirmed by the SEM micrographs. Conversely, the pore radius has decreased in the thermally treated samples compared to the raw one. These findings together with the FT-IR and Raman, and FCCCD analyses might explain the superiority of RPAL as adsorbent compared to the TTPAL250 and TTPAL500 samples, and confirm that the adsorption process is controlled by the chemical structure of the adsorbent surface, which in turn is affected by the adsorption conditions. Figure 8 displays that the three adsorbents show a type III adsorption isotherm with H3—hysteresis loop, indicating the unrestricted multilayer formation and that lateral interactions between the adsorbate molecules are stronger than the interactions between adsorbent and the adsorbate. The H3—hysteresis indicates the aggregation of plate–like particles to form slit–like pores in loose assemblies. Furthermore, it also shows the presence of two types of pores including mesopores (2–50 nm diameter) and macropores

−

(>50 nm diameter, according to the IUPAC classification), in alignment with the analysis of SEM micrographs, Figure 7 [43]. − −

−

**Table 6.** Brunauer–Emmett–Teller (BET) analysis of RAPL and thermally treated samples.

**Figure 8.** BET analysis of (**A**) RPAL, (**B**) TTPAL250, and (**C**) TTPAL500.

#### *2.6. Equilibrium and Kinetics Studies of the Adsorption of RB onto PAL*

The data displayed in Table 2 prove that RPAL has higher adsorption efficiency compared to the thermally treated samples, hence, the equilibrium isotherms and kinetics studies were carried out using the RPAL sample. Important information, such as the maximum quantity adsorbed, the type of interaction (chemi—or physisorption) between the adsorbate and the adsorbent surface, are by and large obtained using adsorption isotherms. Kinetics studies, on the other hand, are used to find the different factors affecting the adsorption process including adsorption rate, type of the layer formed on the surface of the adsorbent (mono or multilayer), and the type of the adsorption mechanisms. The data given below will show the kinetics and adsorption isotherms of the adsorption of RB dye onto the RPAL sample.

#### 2.6.1. Equilibrium Isotherms

The biosorption of RB dye onto the RPAL was studied using four isotherms: (1) Langmuir, (2) Freundlich, (3) Temkin, and (4) Dubinin–Radushkevich (DR) paradigms [44–47]. Single–layer homogeneous adsorption on the surface of the adsorbent was explained by Langmuir isotherm as shown in Figure 9A and Table 7. The Langmuir equation is shown below:

$$q\_{\varepsilon} = \frac{q\_m \text{ K}\_L \text{ \textdegree C}\_{\varepsilon}}{1 - \text{K}\_L \text{ \textdegree C}\_{\varepsilon}} \tag{5}$$

**Figure 9.** Adsorption isotherms of RB on RPAL including (**A**) Langmuir, (**B**) Freundlich, (**C**) Temkin, and (**D**) Dubinin–Radushkevich (DR).

− **Table 7.** General and linearized equation of Langmuir, Freundlich, Temkin and Dubinin–Radushkevich isotherms, beside their parameters for the adsorption of RB on ADPP.


߳ = ܴܶ(1 + ܥ ) ௦ݍ 1 = ܧ <sup>൘</sup> ߚ2ඥ In Equation (5), *q<sup>m</sup>* and *K<sup>L</sup>* stand for the maximum adsorption capacity and the Langmuir equilibrium coefficient, respectively. Langmuir equation can be expressed using the following formula:

1

$$R\_L = \frac{1}{1 - K\_L \, \text{C}\_0} \tag{6}$$

ܧ

− − − − − where *R<sup>L</sup>* and *C*<sup>0</sup> represent the separation factor and the initial concentration (mg/L) respectively. The *R<sup>L</sup>* value reflects the feasibility of the sorption process. Therefore, if *R<sup>L</sup>* is higher than 1, the adsorption process is counted as unfavorable and if *R<sup>L</sup>* is equal to 1, the adsorption isotherm is linear. In cases where the *R<sup>L</sup>* value is in the range between 0 and 1, then the adsorption process is favorable, and it occurs spontaneously, while if *R<sup>L</sup>* is equal to 0, the adsorption is expressed as irreversible process [47]. Based on the obtained data for the current work, the *R<sup>L</sup>* value was found to be less than 1 and higher than 0, indicating that the biosorption of RB onto RPAL was spontaneous and the monolayer maximum adsorption capacity (*qmax*) = 58.80 mg/g.

ሽܮܣܴܲ − ܤܴሼ→

ܮܣܴܲ + ܤܴ

The heterogeneous adsorption is usually portrayed using the Freundlich isotherm described by the following equation:

$$q\_{\varepsilon} = \mathcal{K}\_{\mathcal{F}} \mathcal{C}\_{\varepsilon}^{\frac{1}{n}} \tag{7}$$

where *C<sup>e</sup>* is the equilibrium concentration of RB (mg L−<sup>1</sup> ); *q<sup>e</sup>* is the amount of RB adsorbed/unit mass (mg·g −1 ), while K<sup>F</sup> (mole·g −1 ) (L·mole−<sup>1</sup> ) <sup>1</sup>/*<sup>n</sup>* and 1/*n*, are the Freundlich coefficients. This model presumes neither homogenous adsorption nor restricted level of biosorption. According to the obtained data shown in Figure 9B and Table 7, the highest R<sup>2</sup> value (0.943)–among the four studied models—was obtained using the Freundlich isotherm, implying that this model holds for the RB—RPAL system. Freundlich coefficient 1/*n* = 0.609 and *n* = 1.642, signifying that the biosorption of RB is favorable where the value of 1/*n* < 1. This isotherm also designates that the adsorption might not be monolayer and that adsorption sites with higher affinity might be inhabited first. This finding also explains why the removal efficiency (%R) has decreased with increasing [RB].

The adsorbate–adsorbent interaction was also studied using the Temkin isotherm as shown in Figure 9C and Table 7. Temkin isotherm, however, cannot be used to explain the adsorption of RB onto RPAL since the R<sup>2</sup> value = 0.881. The DR isotherm, (Figure 9D and Table 7) was used to detect the type of adsorption on a heterogeneous surface [47]. Based on the reported information on the relation between the free energy value and the adsorption mechanism, where if the free energy value is <8.0 kJ/mol, the adsorption process is physisorption while if the free energy is >8.0 kJ/mol then the adsorption process will be chemisorption. According to the data revealed in Table 7, the free energy for adsorption of RB onto RPAL is physisorption where the amount of free energy equals 7.07 kJ/mole. Yet, this type of isotherm might not be applicable in the current investigation where data did not show an excellent goodness–of–fit with R<sup>2</sup> = 0.858. These findings together with the characterization outcomes show that and though free energy implies physisorption, occurrence of chemisorption cannot be ruled out [48].

#### 2.6.2. Biosorption Kinetics

In this study, four models were tested; pseudo–first order (PFO), pseudo–second order (PSO), Elovich and Weber–Morris (W–M) to explain the kinetics of the adsorption process of RB onto RPAL. The data shown in Figure 10A,B represent the plots of [ln(*qe*–qt) vs. time] and [time/*q<sup>t</sup>* vs. time] for the two tested kinetic models; PFO and PSO, respectively. Other parameters together with their values are listed in Table 8. By comparing the linearity and the calculated adsorption capacity at equilibrium for these two models, it can be detected that the PSO model is more applicable in explaining the adsorption of RB onto RPAL [49–51]. Therefore, the reaction of RB with RPAL can be expressed as:

$$\text{RB} + \text{RPAL} \stackrel{k}{\rightarrow} \{\text{RB} - \text{RPAL}\} \tag{8}$$

**Table 8.** The kinetics study results corresponding to Figure 10.


α <sup>−</sup> <sup>−</sup> β <sup>−</sup>

−

− −

− − **Figure 10.** Kinetic models for the adsorption of RB on RPAL including (**A**). Pseudo first order, (**B**). Pseudo second order, (**C**). Elovich and (**D**). intra–particle diffusion (Weber–Morris) curves.

− ln(qe − qt) = ln(qe) − kଵt − − ୲ ୯ୣ = ଵ ୩మ୯ <sup>మ</sup> + ଵ ୯ *− −* − − q<sup>୲</sup> = β ln(αβ) + βln (t) *α* Α Β Therefore, the rate of the reaction can be expressed as: *k*[RB][RPAL], implying that the adsorption rate depends mainly on both RB and RPAL concentrations. Weber–Morris intra–particle diffusion model, Figure 10C, indicates that the diffusion rate is very fast with the value of K<sup>1</sup> = 1.262. The mechanism of adsorption process using this model involves the formation of a layer of RB around the particles of RPAL, which will prevent any penetration of more RB and form a boundary layer (53.66 mg/g). This value is close to the *qmax* obtained from the Langmuir isotherm. Finally, the Elovich model, Figure 10D, shows a low R<sup>2</sup> value (0.953) compared to PSO model. This model shows that the initial adsorption rate (α = 3.79 × 10<sup>12</sup> mg·g −1 ·min−<sup>1</sup> ) is higher than the desorption rate (β = 1.817 g·mg−<sup>1</sup> ). Therefore, the adsorption of RB onto RPAL involves a second–order uptake rate vs. the existing surface sites.

#### − − ff **3. Materials and Methods**

*<sup>−</sup> <sup>−</sup> β*

q<sup>୲</sup> = K୍ t .ହ + C

#### *− ff 3.1. Materials and Reagents*

The chemicals used were of the analytical grade and were used as acquired with no additional purification. Sodium hydroxide, sodium tetraborate–10–hydrate and hydrochloric acid were purchased from Sigma–Aldrich (Eschenstrasse, Taufkirchen, Germany). Rose bengal (RB) was a product of BDH Laboratory Supplies (Poole, UK). Values of pH were adjusted as previously mentioned [10]. Pure water was used for diluting the RB dye solutions to 1000 ppm. Pineapple leaves (PAL) were used after drying as will be described in their method of preparation.

*− −*

#### *3.2. Instrumentation and Software*

A Jenway pH meter was used for the preparation of different pH dye solution. An ST8 Benchtop Centrifuge (Thermo Scientific, Waltham, MA, USA) was used for separating the components of each sample mixture. The absorbance was measured using an UV–Vis spectrophotometer (Agilent DAD, Agilent, Santa Clara, CA, USA). The surface morphology of the prepared pineapple leaves was identified using a scanning electron microscope (SEM– Quanta 200, Thermo Scientific, Waltham, MA, USA) and energy–dispersive X–ray spectroscopy (EDX, Thermo Scientific, Waltham, MA, USA). Fourier transform infrared radiation (FT-IR, Bruker Alpha, MA, USA) was used to determine the functional groups on the surface of pineapple leaf. The Raman spectrum was recorded in the range from 50–3500 cm−<sup>1</sup> using a Raman microscope (DXR Raman Microscope, Thermo Scientific, Waltham, MA, USA), with a laser beam at 532 nm as excitation source. Furthermore, a thermal gravimetric

analyzer (TGA400, PerkinElmer, Waltham, MA, USA was utilized to inspect the thermal stability of the pineapple leaf. Finally, Minitab®19 software (Minitab Inc., Chicago, IL, USA) was used to construct the face–centered central composite design (FCCCD).

#### *3.3. Face—Centered Central Composite Design (FCCCD)*

The design of experiment chosen to conduct the current study is FCCCD. The percentage removal (%R) as a single response was optimized as a function for four independent variables, pH, DC, AD, and CT (Table 3). The design matrix involved conducting 30 basic runs in one replicate over two blocks with α = one. Design points involved 16 cube points, eight axial points, and total of six center points. The full design matrix as shown in Table 3.

#### *3.4. Preparation of RB*

Ultra–pure water was artificially contaminated with RB dye to have a stock solution of 1000 ppm. Serial dilutions of the RB solution were prepared by adjusting the desired pH value using the previously prepared pH adjusting solutions. Three calibration curves were prepared, therefore, at three pH values, Table 3, and measured at 548 nm.

#### *3.5. Adsorbent Preparation*

#### 3.5.1. Air–Dried Raw Pineapple Leaves (RPAL)

Pineapples were purchased from a local market in Doha–Qatar. The leaves of the pineapple were separated from the bottom of the pineapple fruit using a metal blade. The crown base was detached, then the pineapple leaves were cut into small pieces approximately 1 × 1 cm. These pieces were rinsed with tap water followed by distilled water to remove any impurities or pollutants present on their surface. The cut leaves were then dried and exposed to the sunlight directly for three consecutive days until they are completely dry. Dry leaves were allotted as three portions. The first portion was further dried in air and labeled as raw pineapple leaves (RPAL).

#### 3.5.2. Thermal Treatment of Pineapple Leaves

Portions 2 and 3 were activated in the oven at 250 ◦C and 500 ◦C for 1 h, and labeled as, thermally treated pineapple leaves; (TTPAL250), and (TTPAL500), respectively. The three portions and after the previous treatment were chopped well with electrical grinder until it becomes fine powder.

#### *3.6. Evaluation of the Adsorption Perfomance of the Prepared Adsorbents*

Two batches of 15 mL centrifuge tubes were prepared. The first set was the sample and the second set was for the blanks. In each tube for both sets, 30–150 mg of RPAL was added. The pH value of the RB solutions was adjusted to the desired figure (Table 3). Next, the two sets of samples and blanks were centrifuged at 4200 rpm for the time specified in Table 3 to facilitate obtaining the supernatant. UV–Vis spectrophotometer was used to measure the absorbance of the supernatant.

#### **4. Conclusions**

The present work has emphasized that economic PAL adsorbents could be efficiently used for the adsorption of rose bengal (RB) from wastewater. Three types of adsorbents were developed for that purpose, raw (RPAL) and thermally treated PAL at 250 ◦C and 500 ◦C. Results showed that RPAL is more efficient for the removal of RB. A smart and ecofriendly platform has been proposed to engineer the removal process. In this context, a response surface methodological approach (face–centered central composite design, FCCCD) was used to optimize the variables influencing the adsorption process. The response (%R) was measured as a function of four factors (pH, AD, DC, and CT). As per the response surface regression model, increasing the dose of RPAL improves the adsorption of the dye, in contrast to pH and DC. FT-IR and Raman spectra were used to examine the prepared adsorbents. FT-IR data

showed the presence of–OH, N–H, C–H, and C–O function groups in RPAL as well as in the thermally treated sample but with a lower intensity. Raman spectra showed the formation of carbonaceous material after the burning process as confirmed by the presence of D– and G–bands. The equilibrium studies revealed that the biosorption of RB on RPAL could be represented by the Freundlich isotherm. The maximum monolayer adsorption capacity was 58.80 mg/g as determined by the Langmuir isotherm. Furthermore, the adsorption of RB onto RPAL is physisorption with free energy equals 7.07 kJ/mol as calculated by the Dubinin–Radushkevich (DR) isotherm. However, and considering the SEM and BET analyses together with the FT-IR findings, occurrence of chemisorption cannot be ruled out. The kinetic studies showed that the adsorption process was a second–order reaction and adsorption rate depends mainly on both RB and RPAL concentrations.

**Author Contributions:** Conceptualization, M.E.-A. and S.S.H.; methodology, A.S.E.-S. and N.Z.; software, M.E.-A.; validation, M.E.-A., A.S.E.-S. and S.S.H.; formal analysis, M.E.-A., S.S.H., A.S.E.-S.; investigation, M.E.-A., S.S.H., A.S.E.-S., and N.Z.; resources, M.E.-A.; data curation, M.E.-A., N.Z., A.S.E.-S. and S.S.H.; writing—original draft preparation, M.E.-A. and N.Z.; writing—review and editing, M.E.-A., S.S.H., A.S.E.-S. and N.Z.; visualization, M.E.-A., and S.S.H.; supervision, M.E.-A.; project administration, M.E.-A. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** The project members would like to extend their special thanks to the Central Lab Unit (CLU) at Qatar University.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Evidence for Phytoremediation and Phytoexcretion of NTO from Industrial Wastewater by Vetiver Grass**

**Abhishek RoyChowdhury <sup>1</sup> , Pallabi Mukherjee <sup>2</sup> , Saumik Panja <sup>2</sup> , Rupali Datta <sup>3</sup> , Christos Christodoulatos <sup>4</sup> and Dibyendu Sarkar 2,\***


**Abstract:** The use of insensitive munitions such as 3-nitro-1,2,4-triazol-5-one (NTO) is rapidly increasing and is expected to replace conventional munitions in the near future. Various NTO treatment technologies are being developed for the treatment of wastewater from industrial munition facilities. This is the first study to explore the potential phytoremediation of industrial NTO-wastewater using vetiver grass (*Chrysopogon zizanioides* L.). Here, we present evidence that vetiver can effectively remove NTO from wastewater, and also translocated NTO from root to shoot. NTO was phytotoxic and resulted in a loss of plant biomass and chlorophyll. The metabolomic analysis showed significant differences between treated and control samples, with the upregulation of specific pathways such as glycerophosphate metabolism and amino acid metabolism, providing a glimpse into the stress alleviation strategy of vetiver. One of the mechanisms of NTO stress reduction was the excretion of solid crystals. Scanning electron microscopy (SEM), electrospray ionization mass spectrometry (ESI-MS), and Fourier-transform infrared spectroscopy (FTIR) analysis confirmed the presence of NTO crystals in the plant exudates. Further characterization of the exudates is in progress to ascertain the purity of these crystals, and if vetiver could be used for phytomining NTO from industrial wastewater.

**Keywords:** insensitive munitions; 3-nitro-1,2,4-triazol-5-one (NTO); industrial wastewater; vetiver grass; phytoremediation; phytoextraction

#### **1. Introduction**

Conventional explosives such as 2,4,6-trinitrotoluene (TNT) and 1,3,5-trinitro-1,3,5 triazine (RDX) have been used in weapons for decades. Since the 1990s, however, the focus has shifted to developing formulations of insensitive munitions (IMs), which are safer to handle but remain as effective as conventional explosives [1–3]. 3-nitro-1,2,4-triazol-5-one (NTO) is one of the main constituents of IM formulations [4]. It is less sensitive to heat and shock and is safer to handle and transport. Detailed chemical properties of NTO are presented in Supplementary Information (Table S1). NTO is more water-soluble than conventional explosives such as RDX and TNT. The solubility of NTO in water increases from 9.97 to 1989.67 g/L when the temperature increases from 11 to 33 ◦C [5]. Due to its high solubility, wastewater generated in munition plants containing NTO requires physical, chemical, and/or biological treatment according to regulatory standards before being released into the environment.

As wastewaters produced in industrial munition facilities have the potential to contain residues of explosive compounds and their transformation products, they are subjected to

**Citation:** RoyChowdhury, A.; Mukherjee, P.; Panja, S.; Datta, R.; Christodoulatos, C.; Sarkar, D. Evidence for Phytoremediation and Phytoexcretion of NTO from Industrial Wastewater by Vetiver Grass. *Molecules* **2021**, *26*, 74. https://dx.doi.org/10.3390/ molecules26010074

Academic Editors: Chiara Bisio and Monica Pica Received: 26 November 2020 Accepted: 22 December 2020 Published: 26 December 2020

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/ licenses/by/4.0/).

extensive industrial waste treatment processes under regulatory discharge guidelines [6–9]. These processes can be expensive and inefficient. Aerobic and anaerobic biological treatment processes have been explored for the remediation of NTO in wastewater. Under anaerobic conditions, NTO is biotransformed into ATO (3-amino-1,2,4-triazol-5-one), which requires further treatment based on regulatory standards. In a sequential anaerobic-aerobic biodegradation study, while NTO biotransformed into ATO under anaerobic conditions, ATO later mineralized under aerobic conditions [10,11]. Sorbents such as granular activated carbon (GAC) are ineffective, as NTO carries an electrostatic charge in aqueous solutions and sorbs very poorly to GAC [1,12]. Other processes such as reverse osmosis (RO) and electrochemical degradation for NTO removal either produce concentrated waste streams or additional regulated byproducts [1]. A Fe/Cu bimetal system was used to remove NTO from an aquatic medium, and a pH and a solid-to-liquid ratio-based removal of NTO was reported [4,13]. The phototransformation of NTO in an aqueous medium was also tested [14]. The fate of NTO during biological wastewater treatment was also studied, and the ability of wastewater sludges to promote the biotransformation of NTO to ATO was documented [15].

It is important to develop more effective treatment technologies for wastewater streams containing NTO, since the processes tested so far are expensive, ineffective, or produce harmful byproducts. The objective of this study was to evaluate the potential use of vetiver grass (*Chrysopogon zizanioides* L.) to remove NTO from wastewater. Vetiver is high biomass, fast-growing, perennial grass. It has an extensive root system that can penetrate deeply (3–4 m). Vetiver's ability to remove various environmental contaminants including various metals and antibiotics is well studied [16–20]. Vetiver was also shown to be effective in the remediation of various explosive compounds from the environment [21,22]. Studies showed that vetiver grass has the potential to remove TNT, RDX, HMX (1,3,5,7-tetranitro-1,3,5,7-tetrazocane), DNAN (2,4-dinitroanisole), and NQ (nitroguanidine) from hydroponic media [23–25]. Vetiver's ability to uptake TNT from the soil in the presence of urea has also been reported [26,27]. Studies showed that plant– microbe interaction plays a significant role in vetiver's ability to remove TNT from soil [28]. While vetiver has been recognized as an effective candidate for the phytoremediation of several explosive compounds, its potential for removing NTO from water or soil has not yet been tested.

#### **2. Results and Discussion**

#### *2.1. Uptake of NTO by Vetiver Grass*

The untreated NTO-wastewater used in this study was alkaline, with a pH of 10.2 ± 0.1 (±standard deviation, SD). NTO concentration in the wastewater was measured as 23,161 ± 135 mg/L (± SD). Nitrate, nitrite, and ammonium-nitrogen concentrations in the wastewater were 1680 ± 185 (± SD), 1.07 ± 0.05 (± SD), and 19.8 ± 2.3 mg/L (± SD), respectively. The wastewater also contained 1.5 ± 0.3 mg/L (± SD) of Na and 44 ± 3 mg/L (±SD) of Ca.

Two different treatments were used for this study: (1) vetiver grown in NTO-wastewater (NV), and (2) NTO-wastewater without plants (negative control, NC). NV and NC were set up in triplicate, resulting in six containers in total. The duration of the entire experiment was 100 days. Vetiver batches were replaced every 20 days in NV treatment resulting in five successive batch studies. Figure 1 and Table S2 present the trend of average NTO reduction in both NV and NC treatments. At the end of the 100-day study, the average NTO concentration was reduced by 83.68 ± 0.43% (± SD) of its initial concentration in NV treatments whereas only 5.0 ± 1.3% (± SD). NTO reduction was estimated in NC treatments. For NV treatments, the NTO concentration reduced steadily from the first to the fourth successive batches and eventually stayed steady from days 80–100 for the 5th batch of vetiver. Varying amounts of NTO were removed in the NV treatments by successive batches of vetiver. While the first batch removed 22.73 ± 1.07% (± SD) of the initial NTO, later batches removed 51.97 ± 0.56% (± SD), 68.9 ± 0.9% (± SD),

83.22 ± 1.06% (± SD), and 83.68 ± 0.43% (± SD). NTO reduction was measured at the end of the second, third, fourth, and fifth batches, respectively. No significant NTO reduction in NC treatments indicated that NTO reduction was caused by the vetiver plants. Microbiological transformation of NTO to ATO to urea, CO2, and N2 has been reported earlier [2,10,11]. For this study, no transformation products of NTO were detected in the NTO-wastewater in NV treatments, indicating that NTO had been taken up by vetiver over time. Previous studies showed that plants such as big bluestem (*Andropogon gerardii*), Indiangrass (*Sorghastrum nutans*), and switchgrass (*Panicum virgatum*) uptake DNAN and NQ as nitrogen sources and store them in their roots and shoots [29]. RDX bioaccumulates in the edible parts of plants such as lettuce, tomatoes, and corn [30–38]. We previously reported that vetiver is capable of taking up TNT from soil and water and degrading it within its tissue [22,26]. A decrease in NTO concentration in NV treatments indicate NTO uptake by vetiver. However, since the NTO-wastewater also contained high levels of nitrate, it is not clear if the vetiver used NTO as a nitrogen source. –

**Figure 1.** Change in 3-nitro-1,2,4-triazol-5-one (NTO) concentration in NTO-wastewater during the 100 d experiment. New batches of vetiver plants were introduced in NTO with plant (NV) treatments every 20 days. In control experiments (NC), containers of NTO were maintained without vetiver plants. At the end of the 100-day study, the average NTO concentration was reduced by 83.68 ± 0.43% (±SD) of its initial concentration in NV treatments, whereas only 5.0 ± 1.3% (±SD) NTO reduction was estimated in NC treatments.

The presence of NTO in vetiver roots and shoots was detected in plants collected from NV treatments. It was estimated that on average, vetiver shoots and roots contained 830.9 ± 95.1 (±SD) and 747.8 ± 73.5 mg/kg (±SD) NTO, respectively. The calculated translocation factor (TF) for NTO in vetiver was 1.11, which indicated that vetiver translocated NTO from root to shoot. The presence of NTO inside vetiver roots and shoots indicated that NTO had been taken up by vetiver, and a translocation factor above 1 demonstrates that vetiver translocated NTO from its roots to its shoots. Many studies have previously reported the translocation of explosives such as DNAN, NQ, TNT, and RDX

by plants to their above-ground biomass [22,23,27,29,31]. This is the first study, to our knowledge, to report the translocation of NTO in a plant.

Figure 2 presents the change in nitrate concentration in NTO-wastewater over time for NV and NC treatments. For NV treatments, 37.26 ± 4.52% (±SD) reduction in nitrate in the NTO-wastewater in comparison to its initial value was observed during this study. Nitrogen is an essential nutrient for plant growth, and the reduction in nitrate in NV treatments is due to its uptake by vetiver plants. In contrast, for the NC treatments, nitrate concentration increased by 14.88 ± 3.34% (±SD) on average from its initial concentration during the study. Nitrate efflux from plants is common and occurs in both stressed and non-stressed plants. Efflux increased in plants that are subjected to mechanical or transplantation stress or changes in the pH of the media [32].

**Figure 2.** Change in nitrate concentration in NTO-wastewater during the 100 d experiment. New batc **Figure 2.** Change in nitrate concentration in NTO-wastewater during the 100 d experiment. New batches of vetiver plants were introduced in NTO with plant (NV) treatments every 20 days. In control experiments (NC), containers of NTO were maintained without vetiver plants. For NV treatments, 37.26 ± 4.52% (±SD) reduction in nitrate in the NTO-wastewater in comparison to its initial value was observed during this study. In contrast, for the NC treatments, nitrate concentration increased by 14.88 ± 3.34% (±SD) on average from its initial concentration during the study.

#### *2.2. NTO Phytotoxicity Analysis*

– At the beginning of each successive batch study, vetiver plants were individually weighed before introducing them to the experimental treatment, NV (vetiver grown in NTO-wastewater). Each successive batch of plants introduced was of approximately the same size and weight (21.1 ± 0.7 g (±SD)) as the previous batch. Figure 3A presents the change in plant biomass over time. At the end of each successive batch, on average, vetiver lost 16.07 ± 6.83% (±SD) of its initial biomass in NV treatments. It was observed that in the first three successive batches, plant biomass loss was higher (22.01 ± 0.20% (±SD), 18.2 ± 0.6% (±SD), and 22.5 ± 0.5% (±SD), respectively). However, for the fourth and fifth batches, the loss in biomass was much lower (8.59 ± 0.30% (±SD), and 9.03 ± 0.40% (±SD) respectively). NTO concentration was reduced by 69–83% of its initial value when the last two batches of plants were introduced, which might be within the tolerance range of NTO for vetiver. From the loss in vetiver biomass in the NV samples, it is clear that NTO is toxic to vetiver plants. Our results are similar to other studies on the effect of explosive compounds on plants. Studies showed that the biomass of *L. sativa* was significantly reduced by TNT at a concentration higher than 32 mg/kg [33]. Plant biomass was reduced

by 40% and 70% at a TNT concentration of 100 and 1000 mg/kg in comparison to the control, respectively. It was also reported that the growth of *Morella cerifera* was impacted significantly at 30 and 100 mg/L TNT and RDX concentrations [34]. Other studies also reported significant growth inhibition in smooth bromegrass (*Bromus* sp.), switchgrass (*Panicum virgatum*), big bluestem (*Andropogon geraldii*), and blue grama (*Bouteloua gracilis*), due to TNT toxicity [34–37]. –

**Figure 3.** Phytotoxicity of NTO on vetiver grown in NTO-wastewater. (**A**) Plant biomass, (**B**) chlorophyll content. Successive batches of vetiver plants were introduced every 20 days. At the end of each successive batch, on average, vetiver lost 16.07 ± 6.83% (±SD) of its initial biomass in NV treatments. A significant loss in chlorophyll content was observed in vetiver in NV treatments for all batches. An average chlorophyll reduction of 60.18 ± 18.79% (±SD) was noted for the four successive batches.

Figure 3B presents the change in chlorophyll content of vetiver during the successive batches for NV treatment (vetiver grown in NTO-wastewater). A significant loss in chlorophyll content was observed in vetiver in NV treatments for all batches. Signs of chlorosis were visible in all the NV treatments. An average chlorophyll reduction of 60.18 ± 18.79% (±SD) was noted for the four successive batches (data for the fourth batch were not analyzed). These results show that NTO is toxic to vetiver and impacts its chlorophyll content. Other studies also reported significant chlorophyll loss in various plants due to TNT and RDX toxicity [33,38].

The metabolic profiles of vetiver shoot and root tissues exposed to NTO-wastewater (NV treatments) were compared to control tissues of healthy vetiver grown in hydroponic plant growth media (Figure 4). Control tissues showed statistically significant differences in response when compared to treated shoot and root tissue in the PLS-DA model (Figure 4A,B). The major pathways showing upregulation in shoot include (1) glycerophospholipid (GLP) metabolism, (2) galactose metabolism, (3) linoleic acid metabolism, and (4) sphingolipid metabolism (Figure 4A). In the root, the major pathways affected include (1) pyrimidine and purine metabolism, (2) amino acid metabolism (cysteine and methionine) (3) glycerophospholipid metabolism, and (4) linoleic acid metabolism. Figure 4A,B show the significance of the major upregulated metabolites ranked using the variable importance in projection (VIP) score (>1) from the PLS-DA model. The overall metabolic response resembles the osmotic stress response generated by metal or salt stress in plants. Enhancement of galactose and amino acid metabolism could serve to provide osmoprotectants. An increase in lipid peroxidation and membrane damage is indicated by the presence of high levels of phospholipids and linoleic acid. Glycerophospholipids are generated as a result of osmotic stress caused by salt or dehydration [39–41]. They act as signaling molecules that trigger several downstream effects that help plants respond to stress. Large increases in the levels of various amino acids have been reported to combat salt and metal stress in vetiver [32,33].

#### *2.3. Plant Exudates Analysis*

During the five successive batch studies, all NTO-wastewater-treated plants exuded an unknown solid material at the junction of their root and shoot (Figure S1). The amounts of plant exudates varied for the individual batches. The amounts of exudates showed a decreasing trend from the first to the fifth batches. While the highest amount was exuded in the second batch, very little exudation was seen in the fourth and fifth batches. This result indicates that the exudation correlated with the level of NTO in the wastewater. Halophytes excrete salts as well as metals from their salt glands or trichomes on leaves when exposed to high salt or metal-containing media [34,41–43]. It was reported that as much as 30%–50% of toxic compounds the plants take up are excreted as a detoxification mechanism to protect sensitive photosynthetic tissue from damage [34,40]. Plant exudates were collected and analyzed by SEM, electrospray ionization mass spectrometry (ESI-MS), and FTIR, and the results were compared with pure NTO to decipher any similarities in structure and composition between them.

#### *2.4. Optical Microscopy and Scanning Electron Microscope (SEM) Analysis*

Before performing the scanning electron microscope (SEM) imaging, plant exudates were initially inspected under an optical microscope. An AmScope digital microscope imaging camera was used to capture pictures under the optical microscope. Figure S2 presents the optical microscope image of plant exudates. Clear crystalline structures can be seen in this picture. NTO is known to form an agglomeration of rod-like large crystals once exposed to air [44], and our findings are in agreement with the earlier report. Figure 5 represents the SEM image of pure solid NTO. Distinctive block-like structures were visible when NTO particles were examined. To our knowledge, no earlier study has reported an SEM image of pure NTO; hence, it was not possible to compare our result with any other study. Figure 5 also presents the SEM image of plant exudates (dried and ground). It is clear from the picture that the plant exudates were a mixture of many different substances. No specific distinctive structural feature was found under the SEM to identify the composition of this material. The presence of block-like structures was seen under 1000× magnification, which showed that NTO is a part of the exudates.

–

**Figure 4.** Metabolic profile of vetiver (**A**) shoot and (**B**) root grown in NTO-wastewater treatments (NV) compared to control plants grown in a nutrient medium. GLP metabolism—glycerophospholipid metabolism, Gal—galactose metabolism; Linoleic Acid—linoleic acid metabolism; SL metabolism sphingolipid metabolism.

#### *2.5. Electrospray Ionization Mass Spectrometry (ESI-MS) Analysis*

The ESI-MS analysis of NTO-containing wastewater and exudates was done both in negative and positive mode. The mass to charge ratio (m/z) was calculated to identify the peaks obtained from the samples. The literature showed that in the negative mode, NTO appears at an *m*/*z* ratio of 129 Da [2]. Figure 6A presents the ESI-MS results of both NTO-wastewater and plant exudates (dissolved in DI water) in the negative mode. The NTO peak was recorded at an *m*/*z* ratio of 129 Da in both NTO-wastewater and plant exudates under negative mode. This result shows that exudates contain NTO particles. As ESI-MS is a qualitative tool, no measurement could be done to quantify the NTO. Several peaks were found in both NTO-wastewater and plant exudate samples. As our NTO-wastewater was an industrial sample, the presence of many other impurities was recorded by ESI-MS spectra. Most of the peaks were found to be adducts of sodium salts (sodium nitrate and sodium carbonate). Figure 6B shows the ESI-MS results of NTO-wastewater and plant exudates in the positive mode. In the positive mode, peaks

(**A**)

at m/z ratio of 23 Da and 39 Da position represent sodium and potassium, respectively, which were present in both NTO-wastewater and plant exudates. Our results show that NTO-wastewater contained 1.5 mg/L of total sodium (Na) throughout the study. So, it was clear that sodium, present in the plant exudates, came from the wastewater media.

**Figure 5.** Scanning electron microscope (SEM) image of (**A**) pure NTO solid (5000×) and (**B**) plant exudates (1000×). Distinctive block-like structures were visible when pure NTO particles were examined. The presence of block-like structures was seen in plant exudates, which showed that NTO is a part of the exudates.

**Figure 6.** Electrospray ionization mass spectrometry (ESI-MS) analysis of NTO-wastewater and plant exudates at (**A**) negative mode and (**B**) positive mode. Both figures contain a blank (top), NTO-wastewater (middle), and plant exudates (bottom). NTO peak was recorded at an *m/z* ratio of 129 Da in both NTO-wastewater and plant exudates under negative mode.

#### *2.6. Fourier-Transform Infrared Spectroscopy (FTIR) Analysis*

– − – <sup>−</sup> – − Figure 7A shows the FTIR spectra of pure NTO solids. Based on the chemical structure of NTO, three distinct peaks can be expected from an NTO molecule: 1800–1600 cm−<sup>1</sup> for C=O, 1550–1500 cm−<sup>1</sup> , and 1372–1290 cm−<sup>1</sup> for N-O bonds. These three distinct peaks were observed in the pure NTO solids (Figure 7). Figure 7B shows a comparison of FTIR spectra between pure NTO solids and plant exudates. The figure shows these three peaks in plant exudates with a slight shift in position. Several studies have shown that the FTIR peak shift can occur for various reasons, including specific molecular interactions, such as hydrogen bonding, presence of water molecule in the chemical structure, and dipole–dipole interactions [45,46]. Our analysis showed the presence of Na and Ca ions in the plant exudates. The interaction of these co-existing ions with the original NTO molecules could have attributed to the observed peak shift in the FTIR spectra. In addition, vetiver was

grown in NTO wastewater and the interaction between NTO molecules and other ions occurred in the hydroponic media, which might have resulted in the introduction of the water molecule(s) in the structure, which might have resulted in the peak shift. FTIR analysis also confirmed similarities in chemical structure between pure NTO and plant exudates, which establishes the presence of NTO in plant exudates.

–

– − – <sup>−</sup> – − **Figure 7.** FTIR spectra of (**A**) pure NTO solid, and (**B**) both solid NTO and plant exudates. Peaks at 1800–1600 cm−<sup>1</sup> (for C=O bonds), 1550–1500 cm−<sup>1</sup> , and 1372–1290 cm−<sup>1</sup> (for N-O bonds) are distinct for NTO. All three distinct peaks were also visible for plant exudates.

derstanding this process would help in 'mining' the exuded metals, which would be an Salt-tolerant plant species have been reported to detoxify metals in their tissue by phytoexcreting toxic metals through salt glands or trichomes on their leaves [42,43]. Understanding this process would help in 'mining' the exuded metals, which would be an added benefit for phytoremediation applications. We report for the first time the extrusion of a munition compound. Further studies are required to find out if the vetiver system

could be used to recover and reuse NTO discarded in the waste stream of industrial munition facilities.

#### **3. Materials and Methods**

#### *3.1. Wastewater Characterization*

NTO-wastewater and pure NTO solids were obtained from an industrial munition facility in the US. The detailed characterization of NTO-wastewater was performed before the study. The pH of the wastewater was measured using an PC 700 pH meter, Oakton, Vernon Hills, IL, USA. HACH test kits, HACH Company, Loveland, CO, USA were used to measure the total nitrogen (TN), ammonia-nitrogen (N-NH<sup>4</sup> + ), and total phosphorus (TP) concentrations of the wastewater sample. Nitrate (NO<sup>3</sup> <sup>−</sup>) and nitrite (NO<sup>2</sup> <sup>−</sup>) concentrations of the wastewater were measured using Dionex ion chromatography (IC) with IonPac AS16 (4 mm × 250 mm, Dionex, Thermo Fisher Scientific, Sunnyvale, CA, USA), equipped with a guard column IonPac AG16 (4 mm×50 mm, Dionex, Thermo Fisher Scientific, Sunnyvale, CA, USA). The total organic carbon (TOC) concentration of the wastewater sampled was measured using a UV-Persulfate TOC Analyzer Phoenix 8000 (Teledyne Tekmar, Mason, OH, USA). In addition, NTO-containing wastewater sample was analyzed for Na, Ca, K, and Mg using an inductively coupled plasma optical emission spectrometry (ICP-OES, 5100 SVDV, Agilent Technologies, Santa Clara, CA, USA). NTO concentration in wastewater samples was measured using a high-performance liquid chromatography (HPLC, Agilent Technologies, Santa Clara, CA, USA, Infinity Series 1260, equipped with a ProStar 410 Auto-sampler and a DAD detector and coupled with a porous graphite column Hypercarb 7 ram, 100 × 4.6 mm). The flow rate of the mobile phase was at 1 mL/min with an isocratic mixture of water: acetonitrile + 0.1% trifluoroacetic acid of 70:30 (*v/v*). The sample injection volume was 35 µL. The analytical wavelengths were 215 nm. Under these conditions, NTO elutes at 4.2 min. A calibration range from 1 to 50 mg/L was used for the analysis and the wastewater samples were diluted as required. Dilution factors were considered while calculating the final amount. A known quality check (QC) standard was inserted after every 10 samples to validate the efficiency of the analytical procedure. All analyses were done in triplicate.

#### *3.2. Experimental Setup and Analyses*

Vetiver grass (*Chrysopogon zizanioides* L.) was purchased from Agriflora Tropical, Puerto Rico, USA. Plants were initially potted in garden soil and grown there for 30 days. The plants were then placed in a hydroponic system in half-strength Hoagland's solution for 14 days for acclimatization. After 14 days, the plants were removed from the Hoagland's solution, dried completely using paper towels, weighed, and used for the experiment. The experiment was conducted in 1 L plastic bottles with a working volume of 500 mL. Two different treatments were used: vetiver in NTO-wastewater (NV) (triplicates), and NTO-wastewater without any vetiver plant (negative control, NC) (triplicates). A 4% plant to solution ratio was maintained for each treatment. All vetiver plants were trimmed from their shoots and roots in such a way that all of them were of approximately the same size and weight. No plant growth nutrients were provided for NV treatments. The 100-day-long experiment was conducted in five successive batch studies of 20 days each. After every 20 days, the old batches of vetiver plants were replaced by new batches of plants. Wastewater samples including all replicates were collected at the same time from each of the bottles periodically. Before collecting the samples, each bottle was mixed thoroughly by swirling so that a homogenous solution can be obtained inside the bottles. Samples were collected by submerging the pipette in the liquid part of the bottles. Samples were analyzed for their NTO concentration, and nitrate concentration. For each measurement (NTO and nitrate concentration), samples were analyzed in triplicates, and analyzed concentrations were compared in Microsoft Excel (version 2007) by calculating mean and standard deviation values.

Vetiver plants collected from NV treatments were also tested for NTO translocation inside the vetiver's body. Both vetiver roots and shoots were collected (in triplicates) at the end of each successive batch and were analyzed separately to determine NTO concentration in them. A 0.5 g sample (root and shoot) was initially ground to a powder with liquid nitrogen. The powdered tissue was transferred to a tube and 5 mL of acetonitrile was added to each sample. The tubes were kept on a tube rotator for 24 h. Subsequently, the samples were filtered using a 0.45 µm syringe filter and were analyzed for NTO using HPLC. NTO translocation inside the vetiver was measured by calculating the translocation factor (TF) following the standard protocol [29,47]. The analyzed data were compared in Microsoft Excel (version 2007) and mean and standard deviation values were calculated.

#### *3.3. Phytotoxicity Analysis*

At the end of every successive batch, phytotoxicity analysis was performed on the plant samples (collected from NTO treatment, NV) by conducting a total chlorophyll study and a plant biomass study. Total chlorophyll (as a combination of chlorophyll a and b) extraction from the vetiver samples was performed following standard protocols [30,48], and the absorbance was measured at 645 and 663 nm using a Cytation 3 microplate reader, Biotek Instruments, Winooski, VT, USA. The weight of each plant was recorded before and after each successive batch study. Before weighing the plants, the roots were dried thoroughly with paper towels.

#### *3.4. Plant Metabolomics Study*

Vetiver samples were frozen in liquid nitrogen and were stored at −80◦C until the metabolomics studies were performed. Metabolites were extracted according to a standard protocol [40] with a few modifications adopted by the earlier published literature [48,49]. Ampicillin (0.5 mg/mL) was added as an internal standard before extraction. Methanol: acetonitrile (50:50) with 0.125% formic acid was used as an extraction buffer. LC-MS/MS analysis was performed on the extracted samples using an ABSciex Qtrap 5500 mass spectrophotometer (Sciex, Framingham, MA, USA) equipped with a Turbo V electrospray ionization (ESI) source, a Shimadzu LC-20A system, and a PAL CTC autosampler following a standard protocol [33,49]. A total of 325 metabolites were targeted in multiple reaction monitoring (MRM) mode. Two injections, one for negative mode (ESI−) and one for positive mode (ESI+), were performed. The dwell time was set at 5 ms. Purified standards were used to optimize the compound-specific MS/MS parameters. Peaks were manually reviewed, and the peak area of each metabolite was intergraded through Multiquant v3.0 (Sciex). All data processing was done following standard protocol 3450. MetaboAnalyst 2.0 (http://www.metaboanalyst.ca) was used for all statistical analyses. Partial least-squares discriminant analysis (PLS-DA) was chosen for multivariate analysis. A VIP score >1.5 was considered as significant.

#### *3.5. Plant Exudates Analysis*

In all the successive batches, it was observed that vetivers grown in NTO-wastewater treatments exuded substances from their shoots, which eventually deposited at the junction of the plant's roots and shoots (Figure S1). During the first successive batch study, the plants exuded the material starting from the fifth day of the experiment. For different batches, the amount of total exuded material varied. Plants exuded the highest amount of material in the second batch of the study. The rate of exuded material subsequently decreased, and a very small amount was collected during the fourth and fifth batches of the study.

At the end of every successive batch study, plant exudates were carefully scraped off the plant surface and properly stored. The weight of the collected solids from every treatment was measured and noted. Special attention was given during the collection process so that no plant shoot part was scraped off with the exudates.

Plant exudates were completely dried and used for further analysis. Initially, plant exudates were inspected under an optical microscope. An AmScope digital microscope

imaging camera (AmScope, Irvine, CA, USA) was used to capture pictures under the optical microscope. As many researchers reported that electron microscopy is a good tool to check the purity and morphology of the energetic compounds, the microscopy scans were performed on both exuded solids and pure NTO solids using a field-emission scanning electron microscope Auriga 40 (ZEISS) (SEM, LEO DSM 982, LEO Electron Microscopy, Thornwood, NY, USA).

Electrospray ionization mass spectrometry (ESI-MS) analysis was also performed on both plant exudates (collected from the first two batches of successive batch studies) and NTO-wastewater (obtained from the industrial facility) using a Micromass Quattro Ultima mass spectrometer (Waters Micromass, Manchester, UK) equipped with an electrospray ion source. Many researchers reported that ESI-MS is a reliable qualitative tool to reflect solution-phase structures [50,51]. The ESI-MS comparison of both plant exudates and NTO-containing wastewater was performed to find similarities in chemical structures between these two samples. All analyses were done in triplicates.

Fourier-transform infrared spectroscopy (FTIR) of both plant exudates and pure NTO solids was performed using a Nicolet iS50 FT-IR (Thermo Scientific, Waltham, MA, USA). Both solids were ground to prepare finer particles using a mortar pestle. As plant exudates were moist initially, they were dried in an oven (60 ◦C for 2 h) before use for FTIR analysis. All analyses were done in triplicates.

#### **4. Conclusions**

We evaluated the potential of using vetiver grass to remove NTO from wastewater collected from a munition manufacturing facility. In addition to a high concentration of NTO, the wastewater also contained a high concentration of nitrate. The wastewater was treated with five successive batches of vetiver hydroponically, and the batches were replaced every 20 days. The average NTO concentration decreased by 84% in 100 days. In control tanks without vetiver, the reduction was about 5% during the same period. NTO was detected in root and shoot tissues of vetiver, and high translocation from root to shoot was observed. The vetiver plants showed toxicity symptoms such as a reduction in biomass and a decline in chlorophyll when exposed to NTO. Metabolomic studies indicated an increase in lipid peroxidation, membrane damage, and osmotic stress in vetiver exposed to NTO. During the batch studies, NTO-treated plants produced an exudate at the junction of root and shoot. The amounts of exudates showed a decreasing trend from the first to the fifth batches. While the highest amount was exuded in the second batch, very little exudation was seen in the fourth and fifth batches, as NTO levels declined. SEM, ESI-MS, and FTIR spectroscopic analysis confirmed the presence of NTO crystals in the plant exudates, indicating vetiver exudation of NTO as a mechanism to relieve stress in vetiver. Further studies are needed to understand whether any plant or microorganismmediated biotransformation or degradation of NTO occurs in vetiver. Further studies are also needed to test the feasibility of this technology in large-scale applications under controlled greenhouse environments. If proven feasible in scaled-up settings, existing NTO wastewater holding tanks can be retrofitted with floating treatment platforms of vetiver. At regular intervals, vetiver biomass can be removed and incinerated under controlled conditions. A significant reduction in the total amount of energetics waste is possible by applying this technology at an expense that is much lower than conventional hazardous waste treatment technologies.

**Supplementary Materials:** The following are available online, Table S1: Relevant chemical properties of NTO. Table S2: Change in NTO concentration in NTO-wastewater during the 100-d experiment in vetiver grown in NTO-wastewater. Figure S1: Plant exudates deposited at the junction of vetiver root and shoot. Figure S2: Plant exudates under optical microscope. Photos show the presence of crystalline structures that correspond to NTO.

**Author Contributions:** Conceptualization and research design, D.S. and R.D.; investigation—batch analysis, A.R., P.M., S.P., investigation—SEM, ESI-MS and FTIR analysis, P.M., S.P. A.R; investigation metabolomics, R.D.; data interpretation, A.R., D.S., and R.D.; writing—original draft, A.R.; writing review and editing, D.S., R.D., and C.C.; supervision and project management, funding acquisition, D.S., C.C. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Consortium for Energy, Environment, and Demilitarization (CEED) contract number SINIT-15-0013. The APC was funded by MDPI.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author.

**Acknowledgments:** The authors would like to thank Tsan-Liang Su, and Andrew Mai for their analytical help; Zhaoyu Zheng, and Athula Attygalle and the Center for Mass Spectrometry (Stevens Institute of Technology) for their help with ESI-MS analysis; and Tseng-Ming Chou for his help with SEM analysis (the Laboratory for Multiscale Imaging, Stevens Institute of Technology).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


*Article*
