Dispersive liquid–liquid micro extraction of boron as tetrafluoroborate ion (BF₄⁻) from natural waters, wastewater and seawater samples and determination using a micro-flow nebulizer in inductively coupled plasma-quadrupole mass spectrometry

Abstract

Boron, present in groundwater and seawater, is extracted as tetrafluoroborate anion by dispersive liquid–liquid microextraction (DLLME) and determined by inductively coupled plasma quadrupole mass spectrometry (ICP-QMS). A low flow rate (200 μL min⁻¹) SeaSpray™ micro-flow nebulizer was used for the sample introduction. In this method, the tetrafluoroborate anion formed in the presence of 0.9 mol L⁻¹ H₂SO₄ and 0.1 mol L⁻¹ F⁻ was extracted into chloroform in the presence of Aliquat® 336 (tricaprylmethylammonium chloride) at room temperature. The bulky cationic surfactant, Aliquat® 336, acts as a phase transfer agent, which not only forms an ion-pair complex with tetrafluoroborate anion but also helps in the rapid conversion of boric acid to BF₄⁻ ion. The tetrafluoroborate anion was back-extracted from the chloroform layer with nitric acid for determination by ICP-QMS. Effective parameters for the complex formation and its extraction, such as volume of extractant/disperser solvent, extraction time and concentration of the surfactant have been optimized. Under optimum conditions, an average preconcentration factor of 18 was obtained for 8 mL of water sample for determination by ICP-QMS. The calibration graph was linear in the range of 1–50 μg L⁻¹ for boron, with a limit of detection of 0.3 μg L⁻¹, calculated based on 3 s of blank (n = 6). The precision was close to 3% R.S.D. (n = 3), when processing 8 mL aliquots of sample. The method has been applied to determine boron in bottled mineral water, groundwater, wastewater and seawater samples. The recoveries obtained for the boron spiked to 30 μg L⁻¹ levels in these water samples were 97–102%.

---

Introduction

Boron (B) is a ubiquitous element in rocks, soil and water, forming compounds with oxygen, hydrogen, the halogens, nitrogen, phosphorus and carbon. It is most commonly used in its compound forms, especially borax (Na₂B₄O₇·10H₂O) and boric acid (H₃BO₃). Boric acid, a very weak acid with a pK_a of 9.15, exists predominantly as undissociated boric acid B(OH)₃ in dilute aqueous solutions at pH < 7, whereas at pH > 10 the metaborate anion B(OH)₄⁻ becomes the major species.

Concentrations of boron in surface water range widely, from <0.001 mg L⁻¹ to as much as 360 mg L⁻¹.¹ Concentrations of boron in groundwater throughout the world have a more limited range, <0.3 to >100 mg L⁻¹.² Seawater contains approximately 5 mg L⁻¹ of B in saline estuaries in association with other salts.³ The geographic source of water is the major determinant of these widely variable levels of boron in drinking waters. Drinking water rarely contains more than 1 mg L⁻¹ and generally less than 0.1 mg L⁻¹, concentrations that are considered innocuous for human consumption.

The potential sources of boron contamination in water are either anthropogenic (sewage effluents, boron enriched fertilizers, landfill leachates and discharge from soap and detergent manufacturing) or natural (e.g., water–rock interactions, sea water encroachment, mixing with fossil brines or hydrothermal fluids).⁴

Due to its interaction with the environment, the boron concentration in both drinking water and wastewaters, should be limited according to the WHO and European regulations. The maximum permissible limit of boron in drinking water is set at 0.5 mg L⁻¹ and at 1 mg L⁻¹ in the case of wastewaters discarded to the environment.⁵ In many countries seawater and groundwater are desalinated and levels of boron are reduced before it is used for drinking or irrigation purposes. Thus the determination of boron content in different water resources is important.

Spectrophotometric methods using chromogenic reagents such as azomethine-H, curcumin, carminic acid and methylene blue are commonly used for the determination of boron in water samples after separation of the matrix by distillation of boron as methyl borate.^6,7 These methods, in general, suffer from numerous interferences and have low sensitivity and precision. Consequently, robust instrumental techniques and analytical methods for monitoring and quantifying boron with significant sensitivity and precision become essential. Among the various other techniques reported for boron determination, inductively coupled plasma mass spectrometry (ICP-QMS) provides higher sensitivity, lower detection limits and measurement of ¹⁰B to ¹¹B isotopic ratio in a sample. Boron being a light element (atomic mass 10.8) with relatively low ionization in the argon plasma (approximately 58% at 7500 K), is expected to have serious non-spectroscopic interferences due to the presence of very high salt content in groundwater and seawater.⁸ A sample with high salt content, if introduced directly, can clog the nebulizer and deposit on the sampler and skimmer cones. In addition, it can also change the plasma characteristics. Hence a suitable separation and preconcentration method is needed to separate the matrix prior to determination of boron by ICP-QMS.

Extraction of boron from aqueous samples has generally been reported to be carried out by methyl borate distillation⁶ or by ion exchange using anion, cation or a boron-specific resin such as Amberlite IRA 743.⁹ These methods do present some limitations such as being tedious and time consuming. Modern trends in analytical chemistry are towards the simplification and miniaturization of sample preparation procedures as they lead inherently to a minimum reagent consumption and drastic reduction of laboratory wastes.¹⁰ Recently, a microextraction technique—dispersive liquid–liquid microextraction (DLLME),¹¹ based on a ternary solvent system has been reported, wherein an appropriate mixture of an extraction solvent and a disperser solvent is rapidly injected into an aqueous sample so that a cloudy solution is formed. The analyte in the sample transfers into the fine droplets of the extraction solvent and subsequent phase separation is achieved by centrifugation. The advantages of this method include its simplicity of operation, rapidity, low cost, low consumption of organic solvents and high preconcentration factors. The technique has been applied to the determination of trace organic pollutants¹¹ and metal ions in environmental samples.^12–14

Recent reviews on DLLME^15,16 show that ETAAS has been the technique of choice for analysis, as the volume of sample required is a few micro-litres, but it cannot be applied to boron due to the formation of refractory boron carbide in the graphite furnace. Flow-Injection-ICP-QMS^17,18 has been used with a micro-litre sample loop for sample introduction. We have reported recently, for the first time, a direct analysis of all the 14 lanthanides using the micro-litre volume of the preconcentrated phase using a micro-flow nebulizer with ICP-QMS for DLLME procedure.¹⁹

In the present procedure a micro-flow nebulizer (SeaSpray™), with low sample uptake rates (200 μL min⁻¹) was used for sample introduction, for the determination of boron at a few parts per billion levels. The determination of boron by ICP-QMS however gives rise to serious memory effects.^20,21 We have eliminated the memory effect successfully in the present study by coating the inner walls of the spray chamber employing a siliconizing fluid.

Rusnáková et al.²² have recently reported the application of DLLME method for the determination of boron in water by spectrophotometric detection. In their procedure, after addition of reagents, ultrasound was used to accelerate the conversion of boron to tetrafluoroborate ion and its subsequent extraction as an ion-pair complex. In the present study we have employed a bulky cationic surfactant acting as a phase transfer agent that helps in the rapid conversion of boric acid to BF₄⁻ ion and also forms an ion-pair complex with tetrafluoroborate anion. The method has been applied to different sources of water (mineral water, groundwater, wastewater and seawater) for determining boron concentration.

Experimental

Instrumentation

A VG Plasmaquad 3 Inductively Coupled Plasma-Quadrupole Mass Spectrometer (VG Elemental, Winsford, Cheshire, UK) situated in a class 100 laboratory was used in this study. A micro-flow nebuliser—SeaSpray™ [AR35-1-USS0.4]—was used for sample introduction through a water cooled (1 °C) Scott type double-pass spray chamber. Parameters such as plasma rf power, nebulizer gas flow and lens voltages were optimized daily by aspirating a 10 μg L⁻¹ boron solution prepared from boric acid in 5% HNO₃ passed at a flow rate of 200 μL min⁻¹. The reagents were introduced using a peristaltic pump (REGLO Digital MS-4/12, ISMATEC, Switzerland). The operating parameters of ICP-QMS are listed in Table 1.

Table 1 Instrumental conditions and measurement parameters for ICP-QMS

Model	VG Plasmaquad 3
Rf forward power	1350 W
Reflected power	2 W
Coolant gas flow rate	13.5 L min^−1
Auxiliary gas flow rate	0.9 L min^−1
Nebulizer gas flow rate	0.8 L min^−1; optimized daily
Nebulizer	SeaSpray™ (AR35-1-USS0.4)
Spray chamber	Double pass Scott-type, water cooled at 1 °C
Ni sampler cone	1.0 mm dia
Ni skimmer cone	0.7 mm dia
Dwell time	10 ms
Isotopes measured	m/z^9Be, ^10B, ^11B
Data acquisition mode	Peak jump
Points per peak	3
Sample flow rate	200 μL min^−1

---

Reagents and solutions

All chemicals were of analytical grade unless stated otherwise. Sub-boiled HNO₃ was prepared in-house in class 10 clean bench, by sub-boiling in quartz stills using reagent-grade feedstock and used wherever necessary. Ultra-pure water with >18.2 MΩ cm resistivity, obtained using a Milli-Q Element water system (Millipore, Bedford, MA, USA), located in class 100 area, was used for dilution of standards, for preparing samples and for final rinsing of the acid cleaned vessels. All sample preparations were carried out in class 10 clean bench.

The stock standard solution of boron (1000 mg L⁻¹) was prepared from NIST 951 boric acid by weighing appropriate amounts and dissolving in ultrapure water. Working standard solutions were prepared by serial dilutions prior to analysis. A solution of 5.0 mol L⁻¹ potassium fluoride (AR Grade, S.D. Fine-Chem limited, Mumbai) and 5.0 mol L⁻¹ sulphuric acid (GR grade, Merck, India) was prepared in ultra pure water. Suitable aliquots of these solutions were added for the required final concentration. Methanol (GR grade, Merck, India) was used as the disperser solvent. A stock solution of 40% (v/v) Aliquat® 336 (tricaprylmethylammonium chloride) (Sigma Aldrich, USA) was prepared in chloroform (GR grade, Merck, India) and used as the extraction solvent after appropriate dilution to the required concentration with chloroform. Beryllium stock standard (1000 mg L⁻¹) (Merck, Germany) was diluted and used as internal standard because of its comparable mass and ionization potential with boron. A multi element stock standard solution (100 mg L⁻¹) containing 26 elements (Ag, Al, Ba, Bi, Ca, Cd, Co, Cr, Cu, Fe, K, Mg, Mn, Na, Ni, Pb, Sr, Tl, Zn, Mo, W, Se, Hg, As, Cs, V) was prepared in 5% HNO₃ solution and used for interference study.

Coating the spray chamber

The interior walls of the (Scott type double pass, water cooled) quartz spray chamber used in the sample introduction system in ICP-QMS, was coated with a hydrocarbon soluble siliconizing fluid −5% (v/v) dimethyldichlorosilane in carbon tetrachloride (Princeton Applied Research, Oak Ridge, USA; used for siliconizing glass capillaries). About 10 mL of the solution was poured into the spray chamber, slowly rotated to completely wet the whole inner surface and this was continued for 15 min. Any traces of liquid were drained and the spray chamber was air dried and then heated using an IR lamp for 30 min.

Sample collection

A groundwater sample was collected, drawn by a borewell. A sea-water sample collected from the Bay of Bengal was used as a representative sample for seawater matrix. The samples were filtered through a 0.45 μm membrane filter, collected in a 100 mL pre-cleaned polypropylene screw capped vessel, which was previously thoroughly rinsed with the respective water. Three bottled mineral waters samples, drawn from different manufacturers was bought from the local supermarket, and the BCR® IRMM reference materials of different matrices viz., groundwater containing low and high content of trace elements, effluent wastewater, influent wastewater, industrial effluent wastewater and seawater were analyzed after appropriate dilution with ultra-pure water. These samples were stored in a refrigerator at 4 °C, until use.

DLLME procedure

8 mL of ultra-pure water was taken in a 15 mL screw capped polypropylene centrifuge vial and spiked with 10 μg L⁻¹ of boron. The conversion of boron to tetrafluoroborate was carried out maintaining acidity (0.9 mol L⁻¹) and concentration of fluoride ion (0.1 mol L⁻¹) as reported by Škrliková et al.²⁴ Potassium fluoride was used instead of sodium fluoride due to its higher solubility (102 g/100 mL at 25 °C), which helped in the preparation of a higher concentration of stock solution for fluoride. 1.8 mL of 5 mol L⁻¹ sulphuric acid and 0.2 mL of 5 mol L⁻¹ potassium fluoride were added and shaken well. A mixture of 500 μL of methanol (disperser solvent) and 300 μL of 10% (v/v) Aliquat® 336 in CHCl₃ (extraction solvent) was prepared in a glass vial. The mixture was injected rapidly into the sample solution with a disposable syringe. A cloudy solution was formed due the ternary mixture of water, methanol and chloroform. The cloudy mixture was shaken vigorously and centrifuged at 5000 rpm for 10 min. After centrifugation, the chloroform layer settled at the bottom of the vial (∼160 μL). The aqueous layer at the top was removed with the help of a peristaltic pump. To the organic layer, 200 μL of concentrated HNO₃ was added (to back extract boron), shaken well and then centrifuged at 5000 rpm for 5 min. The aqueous layer was then pipetted out carefully into a 2 mL graduated centrifuge vial, 50 μL of 500 μg L⁻¹ Be was added as internal standard and made up to 0.5 mL with ultra pure water. The mixture was shaken well and analyzed in replicates (n = 3) by ICP-QMS. A process blank was prepared in a similar way without spiking boron. The signal obtained for process blank was subtracted from the signal obtained for samples and standards.

Results and discussion

The preconcentration of boron as tetrafluoroborate ion by DLLME involved the prior formation of an ion-association complex with sufficient hydrophobicity to enable extraction into the small volume of the sediment phase. To obtain good preconcentration factors, various parameters associated with the DLLME such as the volumes of extraction and disperser solvents, concentration of the surfactant and extraction time were optimized. The possible interference on the extraction of boron due to the presence of major concomitant ions in groundwater and seawater was studied.

The preconcentration factor (PF), was calculated as

PF = C_f/C_o

where C_f and C_o are the concentrations of the analyte in the back extracted solution from the sedimented organic phase and the aqueous sample, respectively.

Conversion and extraction of boron as tetrafluoroborate anion

The determination of boron after converting it to tetrafluoroborate ion has been reported by many authors. It is reported that the conversion to the tetrafluoroborate anion is slow and requires heating²³ or ultrasonication^22,24 to speed up the process. It has also been stated that it is necessary to maintain acidic pH conditions and a high concentration of fluoride for complete conversion. In the present method the concentration of fluoride ion (0.1 mol L⁻¹) and the acidity using H₂SO₄ (0.9 mol L⁻¹) was maintained similar to that reported by Škrliková et al.²⁴ for conversion to BF₄⁻. The BF₄⁻ formed was extracted as an ion-pair complex with Aliquat® 336 into the organic solvent chloroform. The present method is rapid and does away with any heating or ultrasonication for the conversion of boric acid to tetrafluoroborate anion. Katagiri et al.²³ have reported that the formation reaction of BF₄⁻ from boric acid reached an equilibrium state within 20 min regardless of reaction temperature, at higher concentration of total boron (66.7 × 10⁻³ mol L⁻¹) and total fluoride (1.0 mol L⁻¹).

H₃BO_3(aq) + 4F⁻_(aq) + 3H⁺_(aq) = BF₄⁻_(aq) + H₂O_(l) (1)

Although the above reaction proceeds slowly, the addition of Aliquat® 336 is found to speed up the reaction at room temperature, which is due to the extraction of the tetrafluoroborate anion into the organic phase based on the phase transfer catalyst principle. Phase-transfer methods employ a system consisting of two mutually insoluble phases, either liquid–liquid or solid–liquid, in which inorganic ions are transported into the organic phase by formation of a complex soluble in the organic phase. They have been widely adopted in industrial processes, since they often result in faster and cleaner reactions and can be carried out at reduced temperatures.

The tetrafluoroborate ions are poorly extracted into the chloroform layer. It is well known that formation of ion-association compounds with bulky organic groups can increase the solubility of the metal complexes in organic solvents due to increase of their hydrophobicity. In the present method Aliquat® 336, a quaternary ammonium cation acts as the phase transfer agent:

H⁺BF₄⁻_(aq) + C₂₅H₅₄N⁺Cl⁻_(org) = C₂₅H₅₄N⁺BF₄⁻_(org) + H⁺Cl⁻_(aq) (2)

The Aliquat® 336 in chloroform dispersed into the aqueous medium as fine micro-droplets forms an ion-pair complex with BF₄⁻ and gets extracted rapidly into the chloroform layer. As the BF₄⁻ is extracted into the chloroform layer, the product being continuously removed, reaction (1) proceeds in the forward direction and the conversion of boric acid to tetrafluoroborate anion is nearly complete. Thus in this method, Aliquat® 336 not only acts as a phase transfer agent, but also indirectly helps in the rapid and complete conversion of boric acid to tetrafluoroborate anion. Other parameters affecting the extraction process have been discussed in the following sections. The concentration of boron in water was maintained at 10 μg L⁻¹ during the optimization of various parameters.

Effect of volumes of the disperser and extractant solvent

In the present procedure methanol was used as the disperser solvent. Before the start of the experiment, the formation of the cloudy solution by the addition of different volume combinations of extraction and disperser solvent was tested. A mixture of 500 μL of methanol and 200 μL of chloroform gave a stable cloudy solution. This was chosen as the base value for further optimization. The concentration of Aliquat® 336 in chloroform was maintained at 5% throughout this optimization. The effect of volume of the disperser solvent was examined by changing the volume of methanol (200, 300, 500, 700, 800 μL), keeping the volume of chloroform constant (200 μL). The extraction of tetrafluoroborate ion-pair complex was found to increase up to 500 μL. Thereafter, further increase in volume lowered the preconcentration factor of the analyte and also resulted in an unstable cloudy solution.

Most of the ion-pair complexes are extracted quantitatively into chloroform.²⁵ Chloroform, an extractant having higher density than water, capable of extracting the analyte-complex of interest and having low solubility in water, well suited for DLLME. In the developed procedure, Aliquat® 336 used as an ion-association reagent to complex the anionic BF₄⁻ ions, is highly soluble in chloroform. The effect of volume of the extractant was examined by changing the volume of chloroform and carrying out the DLLME procedure. 500 μL of disperser solvent-complexing agent medium was added along with different volumes of chloroform (100, 200, 300, 400, 500 μL). When less than 200 μL of chloroform was added, the phase separation was not clearly marked at the bottom of the centrifuge vial. The volume of the sedimented phase increased from 100 to 200 μL, as the volume of chloroform was increased from 200 to 500 μL. At 300 μL, the volume of the sedimented phase was about 160 μL (±10 μL) and the extraction of tetrafluoroborate ion-pair complex was found to be maximum. Since boron was back extracted from the sedimented phase into acid, further increase in the volume of sediment phase was not desirable. Hence 300 μL of chloroform was used in all further experiments. The concentration of Aliquat® 336 (AQ) in the extraction solvent was optimized as given in detail in the next section.

Effect of the concentration of Aliquat® 336

The concentration of Aliquat® 336 (AQ) in chloroform was optimized for complete recovery of boron as tetrafluoroborate. A set of solutions with different amounts of AQ [1 to 10% (v/v)] in chloroform was prepared from the 40% (v/v) stock solution of AQ. 300 μL of each of these solutions was investigated, for optimization, during the DLLME procedure. As shown in Fig. 1, the addition of AQ improved the extraction of tetrafluoroborate ion. It also improved the cloud formation during the DLLME procedure. The preconcentration factor for boron reached a maximum, when 10% (v/v) AQ was used for extraction. Therefore a solution of 10% (v/v) AQ in chloroform as extraction solvent was used throughout the subsequent experiments.

Fig. 1 Effect of concentration of Aliquat® 336 in % (v/v) in CHCl₃ on the preconcentration factor of boron obtained from DLLME. Extraction conditions: water sample volume = 8 mL; disperser solvent (MeOH) volume = 500 μL; extractant solvent (CHCl₃) volume = 300 μL; [H₂SO₄] = 0.9 mol L⁻¹; [F⁻] = 0.1 mol L⁻¹; concentration of boron = 10 μg L⁻¹ (error bars correspond to n = 3 replicates).

Effect of extraction time

In DLLME, extraction time is defined as the interval between injecting the disperser into the extraction mixture and centrifugation. The influence of the extraction time was evaluated in the range of 0–15 min under constant experimental conditions. It was observed that the extraction time has no significant effect on the extraction efficiency. The area of contact between the extraction solvent and the aqueous phase being infinitely large due to fine cloud formation, the extraction of tetrafluoroborate–AQ ion association complex was nearly completed instantly. The mixture was centrifuged for about 10 min at 5000 rpm.

Interference effect of other concomitant ions

Under the present extraction conditions (0.1 mol L⁻¹ F⁻ and 0.9 mol L⁻¹ H₂SO₄), the effect of other concomitant ions commonly present in groundwater, on the percentage recovery of 10 μg L⁻¹ boron was studied. A multi element standard solution containing 26 elements (Ag, Al, Ba, Bi, Ca, Cd, Co, Cr, Cu, Fe, K, Mg, Mn, Na, Ni, Pb, Sr, Tl, Zn, Mo, W, Se, Hg, As, Cs, V) was added to 10 μg L⁻¹ boron standard solution, to check for their interference on the extraction of boron. 8 mL of the 10 μg L⁻¹ of boron standard solution and known aliquots of the multi-element stock standard at different concentration levels (10 μg L⁻¹, 100 μg L⁻¹, 1000 μg L⁻¹, 5000 μg L⁻¹ and 10 000 μg L⁻¹) were treated according to the proposed DLLME procedure. The % recovery of boron decreases as the concentration of multi-element standard increases beyond 5000 μg L⁻¹. It follows that the recovery of boron up to an interferent to analyte ratio of 500 remained quantitative (98–106%) (Fig. 2).

Fig. 2 Effect of interferents on the recovery of 10 μg L⁻¹ boron in water sample using DLLME-ICP-QMS. Extraction conditions: water sample volume = 8 mL; disperser solvent (MeOH) volume = 500 μL; extractant solvent (CHCl₃) volume = 300 μL; [Aliquat® 336] = 10%; [H₂SO₄] = 0.9 mol L⁻¹; [F⁻] = 0.1 mol L⁻¹.

A simulated seawater containing the ions with concentration as given in Table 2, was prepared to study the recovery of boron spiked at 5 mg kg⁻¹ level by the present DLLME procedure. When 8 mL of the simulated seawater was directly subjected to DLLME process, the % recovery of boron was found to be close to 60%. This shows that the high total dissolved solid content in seawater affects the micro-extraction process. But when the simulated seawater was diluted to reduce the matrix—sea water was thus diluted by 100 times—and the DLLME procedure was applied, the % recovery was quantitative (>98%). Hence this study showed that the present method can be used in the extraction of boron from seawater and groundwater, after appropriate dilution.

Table 2 Concentrations of the major nonvolatile constituents of 35‰ salinity seawater

Constituent	Concentration (mg kg^−1)	Constituent	Concentration (mg kg^−1)
Chloride	19 350	Bicarbonate	142
Sodium	10 760	Bromide	67
Sulphate	2 710	Strontium	8
Magnesium	1 290	Boron	5
Calcium	413	Fluoride	1
Potassium	387

---

Memory effect of boron

Boron is known to be one of the difficult elements to determine due to a significant memory effect and issues with prolonged rinsing times. This affects both the analysis time and the accuracy of the results. Al-Ammar et al.²⁰ and Sun et al.²¹ have reported that a primary source of the memory effect was the volatilization of boric acid droplets in the spray chamber. In order to eliminate the memory effect, a small amount of ammonia gas was also introduced with the nebulizer gas flow.²⁰ Sun et al.²¹ used a combination of ammonia with mannitol, as both diluent and rinse solution for the determination of boron in biological samples, to prevent boron from binding to the spray chamber walls.

In this work, a siliconizing fluid viz., 5% (v/v) dimethyldichlorosilane in carbon tetrachloride was used to coat the inner walls of the spray chamber, as explained in the Experimental section. Surface treatment with siliconizing fluids helps to reduce adsorption of polar compounds, proteins and trace metals onto glass surfaces.²⁶ The dimethyldichlorosilane deactivates the silanol (Si–OH) groups on the glass surface, resulting in a hydrophobic surface that resists nonspecific binding.

The effectiveness of the coating was evaluated by analyzing a set of preconcentrated samples (SAM1–SAM9 in Fig. 3) with concentration ranges from 140–600 μg L⁻¹. A rinse solution of 5% (v/v) HNO₃ was used for washing between two successive samples. The rinse solution was passed at the maximum speed of the peri-pump during washing. After a wash for 30–40 s, the boron signal of the rinse solution equals the value measured at the beginning of the experiment (Fig. 3), whereas in the absence of coating the spray chamber with siliconizing fluid, the time taken for eliminating the memory effect of boron with the same rinse solution is >200 s.

Fig. 3 Profile showing the effect of coating the spray chamber on the memory between two successive samples with sample concentration of boron varying between 140 and 600 μg L⁻¹ in different water samples.

The samples and the rinse solution were analyzed alternately for boron content, every 3 min. Results of this experiment are tabulated in Table 3. The concentration of boron in the rinse solution was initially determined to be 9.0 ± 0.4 μg L⁻¹. As may be seen from the table, at the end of 1 hour of sample analysis, the concentration of boron in the rinse solution has increased by only 2 μg L⁻¹ and had stabilized close to 11 μg L⁻¹. This demonstrates that the contribution due to memory effect is not significant in the coated spray chamber despite the wide variation in the concentration of boron in the preconcentrated samples.

Table 3 Elimination of boron memory effect by coating the inner walls of the spray chamber with siliconizing fluid

Time of measurement (min)	Concentration of boron in rinse solution^a (μg L^−1) (n = 3)	Concentration of boron in sample solution (μg L^−1) (n = 3)
a The rinse solution is 5% (v/v) HNO3.
0	9.0 ± 0.4
3		169 ± 0.7
6	9.2 ± 0.2
9		355 ± 2.7
12	8.9 ± 0.1
15		352 ± 2.6
18	11.1 ± 0.2
21		564 ± 2.8
24	10.7 ± 0.5
27		310 ± 3.1
30	10.6 ± 0.7
33		203 ± 2.5
36	11.2 ± 0.3
39		143 ± 1.9
42	11.6 ± 0.8
45		204 ± 1.2
48	10.9 ± 0.9
51		288 ± 2.0
54	11.3 ± 0.5

---

The use ammonia²⁰ and mannitol²¹ as diluents have resulted in the loss of sensitivity of boron. In addition, the use of mannitol by Sun et al.,²¹ gave rise to huge interference on the most abundant isotope of boron, ¹¹B⁺ due to ¹²C⁺, and therefore ¹⁰B⁺ was used for quantification. In the present study the combined use of nitric acid and coated spray chamber, did not show any spectral interference on both the isotopes of boron (¹⁰B⁺ and ¹¹B⁺), that might arise due to the leaching of the coating material. This shows that the spray chamber coating was stable throughout the experiment. The ¹⁰B/¹¹B isotope ratio recorded throughout the experiment showed natural abundance ratio. Hence both the isotopes can be used for quantification of boron in various types of water. In the present study both coating of spray chamber with siliconizing fluid and the use of low flow-rate nebulizer had helped in minimizing/eliminating memory effect of boron.

Analytical characteristics of the method

The calibration standard solutions for boron in the range 1–50 μg L⁻¹ was prepared in 0.1 mol L⁻¹ F⁻ and 0.9 mol L⁻¹ H₂SO₄. The standard solutions were processed by the optimized DLLME procedure. The calibration curves in the above ranges exhibited good linearity with correlation coefficients >0.999. The preconcentration factor (PF) calculated as the ratio of the concentration of the analyte in the final aqueous phase (back extracted from the sedimented organic phase) to the initial concentration of the analyte in the aqueous sample was 18 for boron determined by ICP-QMS in the given range. The PF value of 18 obtained in the final back extracted aqueous phase (volume = 0.5 mL) shows that the extraction of boron is nearly quantitative in the sedimented organic phase volume of 160 μL. This means that the concentration of boron in sedimented organic phase may be approximately 3 times more than the back extracted aqueous phase and the actual PF based on the sedimented organic phase may be close to 55. However to avoid the direct introduction of sedimented organic phase into ICP-QMS, the boron were back-extracted into nitric acid medium, which limits the PF to 18 due to dilution. The limit of detection (LOD), calculated based on three times of the standard deviation of the process blank (n = 6), was 0.3 μg L⁻¹ of boron. The relative standard deviations (R.S.D.) of 6 replicate measurements of 5 μg L⁻¹ pure aqueous standard of boron by ICP-QMS was calculated to be close to 3%. The above analytical characteristics are summarized in Table 4.

Table 4 Analytical characteristics of DLLME-ICPQMS for the determination of boron

Parameter	DLLME-ICPQMS
a Concentration of boron at which R.S.D. was calculated – 5 μg L^−1.
Linear range (μg L^−1)	1–50
R^2	>0.999
Limit of detection (μg L^−1) (3σ, n = 6)	0.3
Reproducibility (% R.S.D, n = 6)	3^a
Enrichment factor	18
Sample volume (mL)	8
Extraction time (min)	5

---

Validation of the method

Due to non-availability of certified reference material of boron in water, the present method was validated using a reference method (ISO 9390:1990; azomethine-H method)²⁷ and the values were found to be in agreement. Applying the t-test (paired-comparison) with multiple samples the results obtained by the two methods showed agreement at the 95% confidence level. The results are shown in Table 5. Further, a known concentration of boron was spiked into three different water matrices—bottled mineral water (MW-1), groundwater (GW) and seawater (SW-1)—and its % recovery was calculated after the DLLME procedure. The recovery of the spiked boron ranged from 97–102%, as shown in Table 6. In addition, an ICP Multi Element Standard Solution IV CertiPUR® (Merck, Germany) containing 23 elements (Ag, Al, B, Ba, Bi, Ca, Cd, Co, Cr, Cu, Fe, Ga, In, K, Li, Mg, Mn, Na, Ni, Pb, Sr, Tl, Zn) including boron, was diluted to 10 μg L⁻¹ with ultra pure water. This standard solution was subjected to the proposed DLLME method and the % recovery of boron was also found to be quantitative (101 ± 2%).

Table 5 Comparison of the results by present method with those by an independent spectrophotometric reference method (azomethine-H method)

Sample	Present method, mg L^−1 mean ± S.D. (n = 3)	Reference method, mg L^−1 mean ± S.D. (n = 3)
Effluent wastewater (BCR® 713)	2.33 ± 0.02	2.1 ± 0.1
Influent wastewater (BCR® 714)	1.75 ± 0.01	1.8 ± 0.1
Bay of Bengal seawater (SW-1)	5.35 ± 0.15	5.1 ± 0.2
North seawater (SW-2)	6.58 ± 0.15	6.2 ± 0.3

---

Table 6 Percentage recovery of boron spiked in real water samples by DLLME-ICPQMS

Sample	Concentration, mean ± S.D. (μg L^−1), (n = 3)	Added, (μg L^−1)	Found, mean ± S.D. (μg L^−1)	Recovery (%)
a MW-1 = bottled mineral water.b GW = groundwater, diluted 10 times before spiking.c SW = seawater, collected from Bay of Bengal (SW-1), diluted 100 times before spiking.
MW-1^a	25.4 ± 0.3	30	54.5 ± 0.5	97
GW^b	29.4 ± 1.0	30	60.1 ± 1.0	102
SW-1^c	53.5 ± 1.5	30	82.9 ± 2.0	98

---

Comparison with other methods

In comparison to the methods reported by Rusnáková et al.²² and Škrliková et al.²⁴ on the conversion of boric acid to tetrafluoroborate anion, the present procedure is simpler and more rapid as it does not require any heating or ultrasonication. Quantitative conversion followed by extraction of tetrafluoroborate occurs in the presence of Aliquat® 336, which acts as phase transfer agent, as explained in an earlier section. The LOD for the present method (0.3 μg L⁻¹) is 50 times better than that reported by Rusnáková et al.²² In addition the present DLLME method does not require any auxiliary reagent for extraction. The advantage of this method lies in its applicability to a wide range of water matrices; from a relatively simple matrix like bottled mineral water to seawater. The combined use of spray chamber coated with a siliconizing fluid and nitric acid as rinse solution has minimized/eliminated the memory effect of boron. In addition the major isotope of boron (¹¹B⁺) can be used for quantification, which was not possible when mannitol and/or ammonia were reported to be used as rinse solution for removing the memory effect, due to spectral interference.

Analysis of real samples

The developed DLLME procedure was applied to the determination of boron in bottled mineral waters (MW-1, MW-2, MW-3), groundwater samples (GW, BCR® 609, BCR® 610), effluent wastewater (BCR® 713), influent wastewater (BCR® 714), industrial effluent wastewater (BCR® 715), sea water collected from Bay of Bengal (SW-1) and North Sea Water-BCR® 403 (SW-2). The groundwater, wastewater and seawater samples were diluted 10, 100 and 100 times, respectively, before subjecting them to the DLLME procedure, to reduce the total dissolved solid content. The concentration of boron in these samples was determined based on the calibration curve obtained for pure aqueous standards subjected to the same DLLME procedure. Although these BCR® reference materials are not certified for their boron contents, the concentration obtained by the developed method can be used as “information value” and would be useful for future certification of these reference materials. The analytical results for the above samples are presented in Table 7.

Table 7 Determination of boron in real water samples by DLLME-ICPQMS

Sample	Concentration, mean ± S.D. (n = 3)	Unit
Mineral water (MW-1)	25.4 ± 0.3	μg L^−1
Mineral water (MW-2)	36.0 ± 0.3	μg L^−1
Mineral water (MW-3)	54.0 ± 1.5	μg L^−1
Groundwater (GW)	0.29 ± 0.01	mg L^−1
Low level groundwater (BCR® 609)	0.24 ± 0.01	mg L^−1
High level groundwater (BCR® 610)	0.34 ± 0.01	mg L^−1
Effluent wastewater (BCR® 713)	2.33 ± 0.02	mg L^−1
Influent wastewater (BCR® 714)	1.75 ± 0.01	mg L^−1
Industrial effluent wastewater (BCR® 715)	2.01 ± 0.01	mg L^−1
Bay of Bengal seawater (SW-1)	5.35 ± 0.15	mg L^−1
North seawater (SW-2)	6.58 ± 0.15	mg L^−1

---

Conclusion

Dispersive liquid–liquid microextraction has been shown to offer a simple way to separate and preconcentrate boron from various sources of water using green chemistry principles. Only micro-litre volumes of organic solvents are employed for the extraction of boron from seawater and wastewaters. Low limits of detection were obtained with 8 mL of the sample. The use of micro-flow nebulizer for sample introduction into ICP-QMS is very useful for the direct analysis using only a few micro-litres of the extracted phase obtained in the DLLME procedure. The coating of the spray chamber with a siliconizing fluid helps in fast wash-out of boron memory between two successive samples. The present method could also be very useful for the determination of boron in drinking water and groundwater post reverse osmosis treatment using various membranes.

Acknowledgements

The authors are grateful to Dr T. Mukherjee, Director, Chemistry Group, B.A.R.C, for his constant support and encouragement.

References

1. http://www.greenfacts.org/en/boron/l-2/boron-3.htm.
2. WHO, Guidelines for Drinking-water Quality, WHO/SDE/WSH/03.04/54, 2003.
3. N. Öztürk, D. Kavak and T. E. Köse, Desalination, 2008, 223, 1–9.
4. Agency for Toxic Substances and Disease Registry, Division of Toxicology/Toxicology Information Branch (ATSDR), Toxicological Profile for Boron, ATSDR, Atlanta, USA, 2007, pp. 1–249.
5. P. Dydo, M. Türek, J. Ciba, J. Trojanowska and J. Kluazka, Desalination, 2005, 185, 131–137.
6. A. A. Nemodruk and Z. K. Karalova, Analytical Chemistry of Boron, Ann Arbor-Humphrey Science Publishers, Ann Arbor, London, 1969.
7. R. Ruggieri, Anal. Chim. Acta, 1961, 25, 145–149.
8. R. N. Sah and P. H. Brown, Microchem. J., 1997, 56, 285–304.
9. D. Lemarchand, J. Gaillardet, C. Göpel and G. Manhès, Chem. Geol., 2002, 182, 323–334.
10. A. N. Anthemidis and K. G. Ioannou, Talanta, 2009, 80, 413–421.
11. M. Rezaee, Y. Yamini and M. Faraji, J. Chromatogr., A, 2010, 1217, 2342–2357.
12. R. E. Rivas, I. López-García and M. Hernández-Córdoba, Spectrochim. Acta, Part B, 2009, 64, 329–333.
13. H. M. Jiang, Y. C. Qin and B. Hu, Talanta, 2008, 74, 1160–1165.
14. E. Z. Jahromi, A. Bidari, Y. Assadi, M. R. M. Hosseini and M. R. Jamali, Anal. Chim. Acta, 2007, 585, 305–311.
15. S. Dadfarnia and A. M. H. Shabani, Anal. Chim. Acta, 2010, 658, 107–119.
16. A. Z. Grześkowiak and T. Grześkowiak, TrAC, Trends Anal. Chem., 2011, 30, 1382–1399.
17. X. Jia, Y. Han, X. Liu, T. Duan and H. Chen, Microchim. Acta, 2010, 171, 49–56.
18. K. Chandrasekaran, D. Karunasagar and J. Arunachalam, Anal. Methods, 2011, 3, 2140–2147.
19. K. Chandrasekaran, D. Karunasagar and J. Arunachalam, J. Anal. At. Spectrom., 2012, 27, 1024–1031.
20. A. Al-Ammar, E. Reitznerová and R. M. Barnes, Spectrochim. Acta, Part B, 2000, 55, 1861–1867.
21. D. Sun, R. Ma, C. W. McLeod, X. Wang and A. G. Cox, J. Anal. At. Spectrom., 2000, 15, 257–261.
22. L. Rusnáková, V. Andruch, I. S. Balogh and J. Škrliková, Talanta, 2011, 85, 541–545.
23. J. Katagiri, T. Yoshioka and T. Mizoguchi, Anal. Chim. Acta, 2006, 570, 65–72.
24. J. Škrliková, V. Andruch, I. S. Balogh, H. Sklenářová and P. Solich, Anal. Methods, 2010, 2, 1275–1279.
25. E. B. Sandell, Photometric Determination of Traces of Metals, John Wiley and Sons, 1978, pp. 513–535.
26. www.thermo.com/eThermo/CMA/PDFs/Various/File_52881.pdf, Thermo Fisher Scientific Inc., 2008, 21–22.
27. ISO 9390, Water Quality—Determination of Borate—Spectrometric Method Using Azomethine-H, International Organization for Standardization, Geneva, 1990.

---