- Email: [email protected]
water extractant mixtures
Analytica Chimica Acta 585 (2007) 24–31
A systematic study on the extractability of arsenic species from algal certified reference material IAEA-140/TM (Fucus sp., Sea Plant Homogenate) using methanol/water extractant mixtures b , Markus Kahn c , Walter Goessler c ˇ Johannes Teun van Elteren a,∗ , Zdenka Slejkovec a
National Institute of Chemistry, Hajdrihova 19, SI-1001 Ljubljana, Slovenia b Joˇzef Stefan Institute, Jamova 39, SI-1000 Ljubljana, Slovenia c Karl-Franzens Universit¨ at Graz, Institut f¨ur Chemie, Analytische Chemie, Universit¨atsplatz 1, A-8010 Graz, Austria Received 20 September 2006; received in revised form 12 December 2006; accepted 13 December 2006 Available online 19 December 2006
Abstract Using methanol/water mixtures (from pure water to pure methanol), with different desorption and solubility parameters, and varying extractant volume to algal mass (V/m) ratios, the extractability of arsenic species from CRM IAEA-140/TM was investigated. A linear sorption isotherm-based model was developed to process the data obtained with variable volume extraction, allowing the unambiguous deduction of the maximal extractable species concentrations under the specific extraction conditions, even for more stable species. The maximal extractable arsenic fraction ranged from 41 to 68% of the total arsenic concentration in CRM IAEA-140/TM, depending on the extractant composition, with pure methanol giving the lowest extraction yield and pure water giving erratic extractability (probably due to bad wettability). The main arsenic species quantified in the methanol/water extracts were arsenosugars, with arsenosugars 1 (glycerol arsenosugar), 3 (sulfonate arsenosugar) and 4 (sulfate arsenosugar) making up ca. 90% of the maximal extractable arsenic. The rest accounts for DMA (dimethylarsinate), arsenosugar 2 (phosphate arsenosugar) and As(V). There is no clear extraction pattern emerging from the data although it may be seen that extraction of more polar species (e.g. arsenosugar 1) is favoured in pure methanol and less polar more ionic species (e.g. arsenosugar 2 and As(V)) in methanol extractants with a higher water percentage. The precise and highly accurate data may be used for quality control purposes under strictly followed extraction conditions since the extraction is operationally defined. Additionally, the variable volume extraction methodology presented may be applied to other elemental species in other matrices using other extractants. Although this approach does not maximise the absolute extractability but only that which is extractant-specific, experimentators are forewarned that in most cases only a fingerprint of the extractant-specific species is produced unless a quantitative extraction of all species is obtained. © 2006 Elsevier B.V. All rights reserved. Keywords: Desorption; Solubility; Speciation; Algae; High performance liquid chromatography; Inductively coupled plasma mass spectrometry
1. Introduction One of the critical aspects in solid material characterization is the extractability of analytes from the sample matrix since most instrumental techniques are unsuitable for direct chemical speciation on the low concentration levels generally encountered in solid material. To accomplish this extraction, the extractant must interact with the target analytes thereby solvating or hydrat-
∗
Corresponding author. Tel.: +386 1 4760288; fax: +386 1 4760300. E-mail address: [email protected] (J.T. van Elteren).
0003-2670/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.aca.2006.12.019
ing them in the respective organic or aqueous extractant. The solid–liquid interfacial phenomena describing this process may be given by the following physico-chemical steps [1]: (i) desorption of the target analytes from the matrix surface, (ii) diffusion into the extractant layer covering the matrix surface and (iii) solubilization in the extractant. Mass transfer of target analytes to the extractant continues until equilibrium is reached, i.e. until the target analytes have been distributed so that their equilibrium concentrations in solid and liquid fulfill the distribution criterion (partitioning coefficient, D). The extraction efficiency may be limited, mostly through the interactions between target analytes and sample matrix. One
J.T. van Elteren et al. / Analytica Chimica Acta 585 (2007) 24–31
must be aware that analytes in cells must be released from cells first by disrupting cell membranes although this does not garantee the release of analytes bound to cellular components [2]. Then, when the extractant is properly chosen desorption and not the solubility is the most important factor to consider [3], which may be seen from the extractability of arsenic species from different matrices. Mild extractants, such as 0.05 mol L−1 (NH4 )2 SO4 are incapable of extracting large amounts (<2%) of As(V), the predominant arsenic species in most soils due to strong association of As(V) with hydrous oxides of Fe [4]. However, assessing the arsenic extractability from, e.g. certain algae, the release of As(V) in water may be quantitative, suggesting that As(V) is “free” (in cytoplasm and organelles) although this depends on the algal type [2]. Thus, in speciation analysis of solid material it is important to consider sorption isotherms when an extraction step is involved, especially when the recovery within the extract is incomplete. In this way it is possible to distinguish between solubility and desorption issues. In their excellent review paper Francesconi and Kuehnelt [5] discussed some misconceptions regarding the extraction of arsenic species, based on the general assumption that all organoarsenic species are organic and will prefer methanol to water as an extractant. However, they mention that since most naturally-occurring arsenic species are polar or ionic and very water-soluble they favour water over methanol as an extractant (see, e.g. Ref. [6]). Although methanol may still extract some non-polar arsenic, the fact that commonly the methanol is removed and the residue taken up in an aqueous mobile phase prior to chromatographic separation means that non-polar arsenic species are “lost” in this stage. Extraction and identification of non-polar arsenic species is still in its infancy and only recently work on arsenolipids has shown significant progress with the advance of HPLC–ICPMS (high performance liquid chromatography–inductively coupled plasma mass spectrometry) analytical procedures for direct and quantitative measurement of arsenolipids in hexane extracts [7]. To date, shakers and rotators are still the most commonly used devices in the extraction of arsenic species from biological tissues. The extraction efficiency may be enhanced by using classical devices, such as the Soxhlet apparatus but less elaborate protocols using sonication, accelerated solvent extraction and microwave-assisted extraction have taken over [8–10]. These procedures are meant to increase the desorption characteristics of arsenic species by weakening the bond between arsenic species and sample matrix and are thus unable to probe, e.g. As(III) complexation due to inherent speciation alteration. Speciation techniques in use are mostly ion-exchange chromatographic separation procedures combined with elementselective detectors such as ICPMS, HGAFS (hydride generation atomic fluorescence spectrometry), HGAAS (hydride generation atomic absorption spectrometry), etc. [5]. In this work the extractability of arsenic species from the algal CRM IAEA-140/TM material (Fucus sp., Sea Plant Homogenate) was studied in a systematic way using a shaker with methanol/water extractant mixtures of different ratios, from pure methanol to pure water, therefore defining varying solubilization and desorption parameters. To study desorption
25
characteristics the extractant volume to algal mass (V/m) ratios were varied in the range 25–250 mL g−1 , thereby allowing the deduction of the initial arsenic species concentrations in algal material under the conditions applied using a simple model based on the assumption of linear sorption isotherm behaviour. For speciation analysis water/methanol extracts were directly brought onto an HPLC column interfaced with an ICPMS. The data will give us more insight into solubility and desorption phenomena of arsenic species with regards to the specific algal material in this study. 2. Experimental 2.1. Sample, standards and reagents The certified reference material IAEA-140/TM material (Fucus sp., Sea Plant Homogenate) was purchased from the International Atomic Energy Association (Vienna, Austria). The certified arsenic concentration is 44.3 mg kg−1 , with an associated uncertainty of 5%. All reagents were of analytical-reagent grade unless mentioned otherwise. Standard solutions of 1000 mg As L−1 were prepared in Milli-Q water (18.2 M cm) for the following species: arsenite and arsenate prepared from NaAsO2 and Na2 HAsO4 ·7H2 O, respectively (Merck, Darmstadt, Germany); methylarsonate (MMA) prepared from methylarsonic acid synthesised from As2 O3 and CH3 I; dimethylarsinate (DMA) prepared from sodium dimethylarsinate (Fluka, Buchs, Switzerland); and the arsenosugars 1–4, obtained from natural sources [11], often referred to as glycerol, phosphate, sulfonate and sulfate arsenosugar, respectively (see Fig. 1, for structures). Ammonium dihydrogen phosphate (p.a.), methanol (p.a.) and aqueous ammonia solution (25%, Suprapur) were purchased from Merck. 2.2. Extraction procedure Varying amounts of algal certified reference material (0.1–1.0 g dw) were accurately weighed into 50 mL “Falcon” tubes and 25 mL of extractant solution was added to the tubes, yielding V/m ratios in the range of 25–250 L kg−1 . Extractant solutions were prepared from mixtures of methanol and water with (v/v) ratios of 0, 10, 25, 50 and 100% methanol (in water).
Fig. 1. Structures of arsenosugars.
26
J.T. van Elteren et al. / Analytica Chimica Acta 585 (2007) 24–31
The tubes were shaken in a reciprocal shaking water bath (Julabo SW 22, Julabo Labortechniek, Germany) with a stroke of 15 mm and a shaking speed of 200 strokes min−1 for 16 h, thermostatted at 20 ◦ C. After the required shaking time the samples were immediately centrifuged at 3200 rpm for 15 min. The solutions were carefully removed with a poly(propylene) Pasteur pipette and kept frozen (−20 ◦ C) until analysis. Extracts were analysed within 1 week of preparation to prevent changes in species. For determination of the moisture content samples were weighed in a glass vial and dried (85 ◦ C) to a constant weight (ca. 4 h).
where m is the mass of the solid material (kg), V the volume of extractant (L), ci the equilibrium concentration of species i in extract (mg L−1 ) and ai is the remaining equilibrium concentration of species i in solid material (mg kg−1 ). Under the assumption of a linear sorption isotherm, i.e. ai = Di × ci , with Di denoting the partitioning coefficient (L kg−1 ) of species i between solid and liquid, Eq. (2) may be rewritten to:
2.3. Arsenic species by HPLC–ICPMS
Note that Di is a function of the extraction parameters such as extractant composition, temperature, shaking time, shaking frequency, shaking amplitude, etc. The extraction yield Yi (% of ai ,0) of species i in the extract is given by: ci × V/m V/m Yi = × 100 (4) × 100 = ai , 0 V/m + Di
HPLC was performed with a Hewlett-Packard 1100 Series system (Hewlett-Packard, Waldbronn, Germany) equipped with quarternary pump, vacuum degasser, column oven set to 40 ◦ C, and thermostatted autosampler with a variable injection loop set at 20 L. Separation was carried out on an anion-exchange column (PRP-X100; 250 mm × 4.1 mm i.d.; Hamilton Company, Reno, NV, USA) with a mobile phase of 20 mmol L−1 NH4 H2 PO4 at pH 5.6 (adjusted with 25% aqueous NH3 ) and a flow rate of 1.5 mL min−1 . The outlet of the HPLC column was directly connected to an ICPMS (Agilent 7500c, Waldbronn, Germany), equipped with a Babington-type nebulizer with PEEK (polyetheretherketone) capillary tubing (0.125 mm i.d.). The ion intensities at m/z, 75 and 77 were monitored. The ICPMS signal was optimised with a solution of the mobile phase containing 10 g As L−1 to give maximum response on the As signal (m/z, 75). Arsenic species were quantified with an external calibration against standard solutions of the relevant species.
The total arsenic concentration in the methanol/water extracts was determined by ICPMS (Agilent 7500ce, Agilent Technologies, Palo Alto, CA, USA), equipped with a concentric nebulizer. The ion intensity at m/z, 75 was monitored. Methanol/water extracts were diluted in such a way that they all contained 10% (v/v) methanol (in water). Arsenic species were quantified with an external calibration against a standard solution of inorganic arsenic in 10% (v/v) methanol (in water). 3. Extraction theory When an element is present in a solid material, the total concentration at (mg kg−1 ) may be subdivided into n subconcentrations for n species containing the element of interest: n
ai , 0
(1)
i=1
with ai ,0 (mg kg−1 ) the initial concentration of species i in the solid material. For each species i extracted from this solid material the following mass balance is valid: m × ai , 0 = ci × V + m × ai
ai , 0 V/m + Di
(2)
(3)
The total extraction yield Yt (% of at ) of the element of interest, comprising n species, is given by: n
Yt =
Yi × a i , 0
i=1
(5)
at
From Eq. (4), it follows that the highest yields for the individual species i may be expected at high V/m ratios and/or low Di values, thereby giving the highest total extraction yield Yt ,max: n
Yt , max =
2.4. Total arsenic by ICPMS
at =
ci =
i=1
ai , 0
at
× 100
(6)
To illustrate the above assume that an element in a solid comprises three species with identical initial concentrations: a1 ,0 = a2 , 0 = a3 ,0 = 10 mg kg−1 . Furthermore, assume that the species have different extractabilities: species 1, completely extractable (D1 = 0); species 2, moderately extractable (D2 = 25 L kg−1 ); species 3, non-extractable (D3 = ∞). In Fig. 2, the extraction fate of the species under varying V/m ratios is given; it can be observed that: the concentration ci in extract decreases hyperbolically to 0, except for species 3 which is 0 for all V/m ratios due to their non-extractability (A); the extractability visualised by ci × V/m increases hyperbolically to ai ,0, except for species 1 and 3 which are a1 ,0 and 0 for all V/m ratios due to complete extractability and non-extractability, respectively (B); the remaining concentration ai in solid decreases hyperbolically to 0, except for species 1 which is 0 for all V/m ratios due to complete extractability and species 3 which stays on the initial level a3 ,0 due to non-extractability (C); the extraction yield Yi for the species increases hyperbolically, except for species 1 which is 100% due to complete extractability and species 3 which is 0 due to non-extractability (D); the total extraction yield Yt increases hyperbolically to Yt ,max = 66.67% (E). From Fig. 2D and E, it can be remarked that although a total extraction yield Yt may show little overall variation with V/m ratio this does not necessarily imply that this is also the case for the extraction
J.T. van Elteren et al. / Analytica Chimica Acta 585 (2007) 24–31
27
Fig. 2. Extraction fate of an element comprising of three species (a1 ,0 = a2 ,0 = a3 ,0 = 10 mg kg−1 ; D1 = 0; D2 = 25 L kg−1 ; D3 = ∞) as a function of V/m ratio.
yield Yi of the individual species in case ai ,0 is small compared to the bulk at and the Di value is high. Also, the unextractable fraction becomes immediately apparent when a plateau is reached which is less than 100% in the Yt versus V/m graph. From the above it is obvious that obtaining the initial species concentrations ai ,0 is not straightforward when the extractability of the individual species is in the range of approximately 5–95% (outside this range we may regard species as nonextractable or completely extractable); this translates into Di ’s in the range of 13.2–4750 L kg−1 in the case of a high V/m ratio, such as 250 L kg−1 (according to Eq. (4)). To this end we need to find ai ,0 (or Di ) indirectly which may be achieved by
rewriting Eq. (2): V/m Di 1 = + ci ai , 0 ai , 0
(7)
From a plot of 1/ci versus V/m the concentration ai ,0 of species i can be derived from the slope and the partitioning coefficient Di of the species from the intercept; when the intercept is zero the species is completely extractable. This approach has been applied for determination of methylmercury in mushrooms [12], organotin compounds in sediments [13] and heavy metals in soil [14,15]. This is illustrated in Fig. 2F for the above example of three species. The following can be observed from Fig. 2F: (i)
28
J.T. van Elteren et al. / Analytica Chimica Acta 585 (2007) 24–31
species 1 has a slope of 0.1 kg mg−1 and an intercept of 0, denoting an a1 ,0 of 10 mg kg−1 and a D1 of 0, respectively; (ii) species 2 has a slope of 0.1 kg mg−1 and an intercept of 2.5 L mg−1 , denoting an a2 ,0 of 10 mg L−1 and a D2 of 25 L kg−1 , respectively; (iii) species 3 can not be plotted since its unextractability and thus infinite D3 value leads to an infinitely low concentration of species 3 in the extract and therefore infinitely high reciprocal concentrations. This shows us that the initial concentration ai ,0 in the solid, of moderately to completely extractable species, may be derived from this variable volume approach. The above theory for deduction of the extraction yield of elemental species from solid material may be restricted when significant losses in the subsequent measurement procedure occur. Since the extracted species cannot be determined directly but only indirectly via chromatographic separation and elementselective detection, losses may occur by (partial) binding of certain species on the chromatographic column. Strong retention of, e.g. thio-arsenosugars on an anion exchange column (PRPX100) has been reported a problem recently [16] resulting in the misinterpretation of the extraction data when not identified. To circumvent such misinterpretations it is essential to compare the total concentration of the element of interest before and after the chromatographic column. As the total concentration after the column the sum of the initial arsenic species concentrations ( ai ,0, mg kg−1 ) may be taken. As the total concentration before the column the total arsenic concentration ct in the extract (for each extractant) as a function of V/m ratio may be used to calculate the maximal extractable arsenic concentration (at ,ext, −1 ), using a linearisation similar to Eqs. (6) and (7). In mg kg case ai ,0 is significantly lower than at ,ext, and thus (partial) binding on the column occurs, conclusions on extraction yields for individual species may not be drawn. 4. Results and discussion Algae, especially the ones from the Fucus family, are interesting in that they often contain high concentrations of arsenosugars (up to 100 mg kg−1 wet mass) [17]. As far as we know the specific algal material used in this study (CRM IAEA-140/TM, Fucus sp., Sea Plant Homogenate) has only been certified for total arsenic and the only arsenic speciation data we are aware of ˇ are reported by Slejkovec et al. [2]. The following arsenic species were found in this particular algal material (in chromatographical order): arsenosugar 1, DMA, arsenosugar 2, arsenosugar 3, As(V) and arsenosugar 4. The presence of high amounts of mostly arsenosugars is in accordance with findings reported for related algae such as Fucus serratus [18] (arsenosugars 1–4), Fucus spiralis [19] (arsenosugar 4), Fucus distichus [20] (arsenosugars 1–4) and Fucus virsoides [2] (arsenosugars 1, 3 and 4). Initial concentrations ai ,0 in algal material for various methanol/water extractant mixtures may be estimated from ci × V/m versus V/m graphs by extrapolation to high V/m ratios (Fig. 3). Although some “noise” in the graphs may be expected for higher V/m ratios since the sample gets proportionally diluted (e.g. at a V/m ratio of 250 mL g−1 the total arsenic concentration
in extract is only ca. 0.18 mg L−1 , assuming extraction yields of 100% for all individual species), this does not explain the wavy curves obtained for the 0% methanol (in water) extractant. This aspect may have to do with irreproducible wettability (due to too high surface tension) of the sample with the extractant since data from a later extraction, where the sample was sonicated for ca. 5 min prior to extraction, show much less of this “wavyness” (Fig. 4). From Fig. 4, it is obvious that in particular arsenosugar 4 is still badly affected by the wettability phenomenon, whereas for all other species this phenomenon is much less pronounced; only at higher V/m ratios some abnormalities still show. The other extractants (see Fig. 3) used for extraction of arsenic species from the algal material show no irregularities. It can be observed from Fig. 3, that arsenosugars 1, 3 and 4 are the predominant arsenic species extracted, with concentrations in the order of 5–10 mg kg−1 , whereas arsenosugar 2, DMA and As(V) are present at (much) lower levels (0–2 mg kg−1 ). It should be noted that the 100% methanol (in water) extract, in contrast to the other extracts, contained unmeasurably low As(V) concentrations (<0.0005 mg L−1 , DL). This implies that As(V) is badly soluble or non-soluble in pure methanol as supported by literature data [21]. Furthermore, this extractant was the only one where the arsenic species extractabilities were (slightly) dependent on the V/m ratio as may be best visualised by the extraction yields Yi (Fig. 5). In particular for arsenosugar 2 a high dependency of Yi on V/m ratio was found which may be explained by the functionality of the groups on the arsenosugar. Although the dimethylarsinoyl group (Me2 As–) is likely to be protonated below pH 3, thereby giving polar behaviour to arsenosugar 1 in methanol, the basic property of the dimethylarsinoyl moiety is outweighed by the acidic groups in the aglycone for the other three arsenosugars [18]. The phosphate, sulfonate and sulfate groups in the respective arsenosugars 2–4 show increasing acid strength, implying that the arsenosugar 2 is relatively most prone to be present in ionised form, thus yielding the least polar arsenosugar with an inherent worst extraction efficiency in methanol. By linearising the data according to Eqs. (6) and (7), plotting 1/ci versus V/m, the initial concentration ai ,0 and partitioning coefficient Di can be derived with a very high degree of precision and accuracy, even for moderately extractable species, such as the above mentioned arsenosugar 2 in methanol. An initial arsenosugar 2 concentration in CRM IAEA140/TM of 0.32 ± 0.01 mg kg−1 and a partitioning coefficient of 52 ± 9 L kg−1 (n = 10) may be derived from the slope and intercept, respectively. The partitioning coefficient is so high that even three repetitive extractions with pure methanol do not increase the total (calculated) extraction efficiency for this species to more than 69.3% at a V/m ratio of 25 L kg−1 . For the other arsenic species hold-up in the algal material is much less of a problem in pure methanol and even less for the water/methanol extractants. However, when we tabulate all initial arsenic species concentrations, derived from linearisation of the data according to Eqs. (6) and (7), it is obvious that some arsenic is “missing” (see Table 1). In Table 1, the data for extraction with pure water are discarded due to problems with erratic extraction,
J.T. van Elteren et al. / Analytica Chimica Acta 585 (2007) 24–31
29
Fig. 3. Extractability of arsenic species from CRM IAEA-140/TM material as a function of V/m ratio for various methanol (in water) extractants.
probably as a result of wettability (see Fig. 4). Only 40–70% (depending on the extractant composition) of the total arsenic at is accounted for; this is in-line with extraction efficiencies of 50–60% reported for related algae [2]. This recalcitrant arsenic concentration of 13–25 mg kg−1 has necessarily a high Di value, of at least of 26,000–50,000 L kg−1 (for a V/m ratio of 25 L kg−1 ), to be undetectable (ci < 0.0005 mg L−1 , DL). However, to exclude trivial artifacts such as irreversible binding on the chromatographic column it is essential to compare the sum of arsenic species after the column with the total arsenic con-
centration in the extract prior to chromatographic separation (see theoretical section). In Fig. 6, such a comparison is made and it can be concluded that no irreversible binding on the column occurs for methanol/water extractants with 10, 25 and 50% methanol (in water), at least not at the 5% significance level (18 d.f.). With 100% methanol (in water) there is a slight discrepancy, showing statistically less (84%) arsenic after the column than before the column. This may be attributed to the binding of certain species on the column under these conditions which may have
30
J.T. van Elteren et al. / Analytica Chimica Acta 585 (2007) 24–31
Fig. 4. Extractability of arsenic species from CRM IAEA-140/TM material using pure water (0% methanol) after initial sonication (5 min).
Fig. 5. Extraction yield of arsenic species from CRM IAEA-140/TM material using pure methanol (100% methanol).
Table 1 Initial arsenic species concentrations ai ,0 in CRM IAEA-140/TM for varying methanol/water extractant mixtures Species, i
Assug1 DMA Assug2 Assug3 As(V) Assug4
ai ,0
8.36 1.85 0.82 7.85 0.41 9.53
25 ± ± ± ± ± ±
0.06 0.02 0.02 0.07 0.01 0.07
28.82 ± 0.12
8.86 1.90 0.75 8.20 0.40 10.13
50 ± ± ± ± ± ±
0.04 0.02 0.02 0.09 0.02 0.08
30.24 ± 0.13
been caused by release of specific non-ionic components upon extraction, responsible for holdup behaviour on the column. From Fig. 6, it can be seen that there is a certain extraction pattern, with 25% methanol extracting the highest amount of arsenic. However, this extraction pattern is not exactly the same for the individual arsenic species (see Table 1) where extraction of arsenosugar 1 seems to be enhanced, in contrast to the other arsenic species, at higher (>25%) methanol concentrations, probably due to arsenosugar 1 being the most polar arsenic species present. At lower (<25%) methanol concentrations arsenosugar 2 and As(V) extraction seem to be favoured over the other arsenic species due to their inherent ionic character (see above). The remaining, unextractable, fraction stays either physico-chemically bound to cellular components (e.g. protein bound arsenic), requiring more severe extraction conditions for desorption, or is in a form (e.g. lipid arsenic) where the extractant is unable to solubilize the associated arsenic species. Evidence for the former species has been found in Fucus vesiculosus, where As(III)-sequestering has been demonstrated to occur through the binding of five As(III) atoms to methallothionein [22]. The latter arsenolipids have been found, but not identified in fish oil and cod liver [7]. 5. Conclusions
% (v/v) methanol (in water) 10
Fig. 6. Total arsenic concentration in algae derived in different stages of the extraction-measurement procedure: before the column (at ,ext) and after the col umn ( ai ,0); as reference value the total arsenic concentration in algae is given (at = 44.3 mg kg−1 ).
8.94 1.62 0.56 7.59 0.19 8.31
100 ± ± ± ± ± ±
0.04 0.01 0.01 0.14 0.02 0.17
27.21 ± 0.23
6.97 ± 0.07 1.01 ± 0.02 0.32 ± 0.01 4.00 ± 0.07 ND 5.85 ± 0.13 18.15 ± 0.17
Standard deviations (σ i ) in ai ,0 (n = 10) were calculated using Microcal (TM) Origin® , version 7.5 (Microcal Software Inc., Northampton, USA) by fitting the data in Eqs. (6) and (7). The total arsenic concentration at certified in this material is 44.3 mg kg−1 .
The processed data from the variable volume extraction approach are precise and highly accurate, showing that varying methanol/water mixtures yield varying maximal extractabilities for a range of arsenic species in CRM IAEA-140/TM. Since this topology corrects for solubility issues of these arsenic species, it is obvious that the extraction is desorption-controlled. Therefore, part of these arsenic species are unaccounted for as a result of the desorption conditions being too mild, whereas a fraction of unknown arsenic species may still be “forgotten” due to solubility/desorption issues under the extraction conditions used. As a consequence it is not possible to definitely quantify the
J.T. van Elteren et al. / Analytica Chimica Acta 585 (2007) 24–31
arsenic species present in CRM IAEA-140/TM but only give extractant-specific concentrations. For algal CRM IAEA-140/TM material a large (unidentified) fraction of the arsenic is unextractable (Di is very high) in methanol/water extractant mixtures as a result of strong binding to cellular components and/or inability for solubilization by the extractants used. A more severe desorption procedure, using, e.g. microwave assisted extraction, may release (part) of this recalcitrant arsenic fraction and either the less polar or nonpolar extractant may reach the hydrophobic arsenic species. The complementary, mostly arsenosugar fraction, is easily solubilized (Di is 0) in all extractant mixtures except pure methanol which shows a particularly bad affinity for arsenosugar 2, comparatively the most ionic arsenosugar. The fact that varying maximal extractabilities are found for different methanol/water mixtures can not be attributed to differences in the solubilities of arsenic species, at least not when the same species are extracted, since an extrapolation to infinite extractant volume to algal mass ratios has been carried through. However, assuming that (some) arsenic species are sorbed onto algal cell components it is likely that the desorption is extractantdependent. Thus, we are not able to definitely quantify the arsenic species present in CRM IAEA-140/TM since they vary with extractant composition. Even the large unextractable fraction may ultimately contain no other species than the ones found except on much higher levels, although it is also possible that different types of physico-chemical species, e.g. hydrophobic ones, are present which are unextractable in the extractants used. The general conclusion of this work is that it is impossible to give defined species distribution concentrations when the very species are not completely extractable. As a consequence this implies that species in materials can generally not be certified, unless a given extraction condition is specified. So, specific extraction conditions will only take us as far as an elemental species fingerprint of a material under the very extraction conditions. Furthermore, it seems that we have to redefine our notion
31
of extractability for speciation purposes as striving for maximal extractability might be conflicting with species integrity. References [1] J. Pawliszyn, J. Chromatogr. Sci. 31 (1993) 31. ˇ [2] Z. Slejkovec, E. K´apolna, I. Ipolyi, J.T. van Elteren, Chemosphere 63 (2006) 1098. [3] M. Letellier, H. Budzinski, Analusis 27 (1999) 259. ˇ [4] J.T. van Elteren, Z. Slejkovec, I. Arˇcon, H.J. Glass, Environ. Pollut. 139 (2006) 477. [5] K.A. Francesconi, D. Kuehnelt, Analyst 129 (2004) 373. [6] J. Szakova, P. Tlustos, W. Goessler, D. Pavlikova, J. Balik, Appl. Organomet. Chem. 19 (2005) 308. [7] E. Schmeisser, W. Goessler, N. Kienzl, K.A. Francesconi, Analyst 130 (2005) 948. [8] J.L. Gomez-Ariza, D. Sanchez-Rodas, I. Giraldez, E. Morales, Analyst 125 (2000) 401. [9] C.G. Yuan, G.B. Jiang, B. He, J. Anal. Atom. Spectrom. 20 (2005) 103. [10] R. Wahlen, T. Catterick, Rapid Commun. Mass Spectrom. 18 (2004) 211. [11] K.A. Francesconi, J.S. Edmonds, R.V. Stick, B.W. Skelton, A.H. White, J. Chem. Soc., Perkin Trans. 1 (1991) 2707. [12] R.G. Fischer, S. Rapsomanikis, M.O. Andreae, F. Baldi, Environ. Sci. Technol. 29 (1995) 993. [13] Y. Cai, S. Rapsomanikis, M.O. Andreae, J. Anal. Atom. Spectrom. 8 (1993) 119. [14] J.T. van Elteren, B. Budiˇc, Anal. Chim. Acta 514 (2004) 137. ˇ [15] J.T. van Elteren, Z. Slejkovec, R. Milaˇciˇc, Int. J. Environ. Anal. Chem. 83 (2003) 389. [16] E. Schmeisser, R. Raml, K.A. Francesconi, D. Kuehnelt, A.L. Lindberg, C. S¨or¨os, W. Goessler, Chem. Commun. 16 (2004) 1824. [17] W.R. Cullen, K.J. Reimer, Chem. Rev. 89 (1989) 713. [18] A.D. Madsen, W. Goessler, S.N. Pedersen, K.A. Francesconi, J. Anal. Atom. Spectrom. 15 (2000) 657. [19] G. Raber, K.A. Francesconi, K.J. Irgolic, W. Goessler, W. Fresenius, J. Anal. Chem. 367 (2000) 181. [20] C.F. Harrington, A.A. Ojo, V.W.M. Lai, K.J. Reimer, W.R. Cullen, Appl. Organomet. Chem. 11 (1997) 931. ˇ [21] A.R. Byrne, Z. Slejkovec, T. Stijve, L. Fay, W. Goessler, J. Gailer, K.J. Irgolic, Appl. Organomet. Chem. 9 (1995) 305. [22] M.E. Merrifield, T. Ngu, M.J. Stillman, Biochem. Biophys. Res. Commun. 324 (2004) 127.
A systematic study on the extractability of arsenic species from algal certified reference material IAEA-140/TM (Fucus sp., Sea Plant Homogenate) using methanol/water extractant mixtures b , Markus Kahn c , Walter Goessler c ˇ Johannes Teun van Elteren a,∗ , Zdenka Slejkovec a
National Institute of Chemistry, Hajdrihova 19, SI-1001 Ljubljana, Slovenia b Joˇzef Stefan Institute, Jamova 39, SI-1000 Ljubljana, Slovenia c Karl-Franzens Universit¨ at Graz, Institut f¨ur Chemie, Analytische Chemie, Universit¨atsplatz 1, A-8010 Graz, Austria Received 20 September 2006; received in revised form 12 December 2006; accepted 13 December 2006 Available online 19 December 2006
Abstract Using methanol/water mixtures (from pure water to pure methanol), with different desorption and solubility parameters, and varying extractant volume to algal mass (V/m) ratios, the extractability of arsenic species from CRM IAEA-140/TM was investigated. A linear sorption isotherm-based model was developed to process the data obtained with variable volume extraction, allowing the unambiguous deduction of the maximal extractable species concentrations under the specific extraction conditions, even for more stable species. The maximal extractable arsenic fraction ranged from 41 to 68% of the total arsenic concentration in CRM IAEA-140/TM, depending on the extractant composition, with pure methanol giving the lowest extraction yield and pure water giving erratic extractability (probably due to bad wettability). The main arsenic species quantified in the methanol/water extracts were arsenosugars, with arsenosugars 1 (glycerol arsenosugar), 3 (sulfonate arsenosugar) and 4 (sulfate arsenosugar) making up ca. 90% of the maximal extractable arsenic. The rest accounts for DMA (dimethylarsinate), arsenosugar 2 (phosphate arsenosugar) and As(V). There is no clear extraction pattern emerging from the data although it may be seen that extraction of more polar species (e.g. arsenosugar 1) is favoured in pure methanol and less polar more ionic species (e.g. arsenosugar 2 and As(V)) in methanol extractants with a higher water percentage. The precise and highly accurate data may be used for quality control purposes under strictly followed extraction conditions since the extraction is operationally defined. Additionally, the variable volume extraction methodology presented may be applied to other elemental species in other matrices using other extractants. Although this approach does not maximise the absolute extractability but only that which is extractant-specific, experimentators are forewarned that in most cases only a fingerprint of the extractant-specific species is produced unless a quantitative extraction of all species is obtained. © 2006 Elsevier B.V. All rights reserved. Keywords: Desorption; Solubility; Speciation; Algae; High performance liquid chromatography; Inductively coupled plasma mass spectrometry
1. Introduction One of the critical aspects in solid material characterization is the extractability of analytes from the sample matrix since most instrumental techniques are unsuitable for direct chemical speciation on the low concentration levels generally encountered in solid material. To accomplish this extraction, the extractant must interact with the target analytes thereby solvating or hydrat-
∗
Corresponding author. Tel.: +386 1 4760288; fax: +386 1 4760300. E-mail address: [email protected] (J.T. van Elteren).
0003-2670/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.aca.2006.12.019
ing them in the respective organic or aqueous extractant. The solid–liquid interfacial phenomena describing this process may be given by the following physico-chemical steps [1]: (i) desorption of the target analytes from the matrix surface, (ii) diffusion into the extractant layer covering the matrix surface and (iii) solubilization in the extractant. Mass transfer of target analytes to the extractant continues until equilibrium is reached, i.e. until the target analytes have been distributed so that their equilibrium concentrations in solid and liquid fulfill the distribution criterion (partitioning coefficient, D). The extraction efficiency may be limited, mostly through the interactions between target analytes and sample matrix. One
J.T. van Elteren et al. / Analytica Chimica Acta 585 (2007) 24–31
must be aware that analytes in cells must be released from cells first by disrupting cell membranes although this does not garantee the release of analytes bound to cellular components [2]. Then, when the extractant is properly chosen desorption and not the solubility is the most important factor to consider [3], which may be seen from the extractability of arsenic species from different matrices. Mild extractants, such as 0.05 mol L−1 (NH4 )2 SO4 are incapable of extracting large amounts (<2%) of As(V), the predominant arsenic species in most soils due to strong association of As(V) with hydrous oxides of Fe [4]. However, assessing the arsenic extractability from, e.g. certain algae, the release of As(V) in water may be quantitative, suggesting that As(V) is “free” (in cytoplasm and organelles) although this depends on the algal type [2]. Thus, in speciation analysis of solid material it is important to consider sorption isotherms when an extraction step is involved, especially when the recovery within the extract is incomplete. In this way it is possible to distinguish between solubility and desorption issues. In their excellent review paper Francesconi and Kuehnelt [5] discussed some misconceptions regarding the extraction of arsenic species, based on the general assumption that all organoarsenic species are organic and will prefer methanol to water as an extractant. However, they mention that since most naturally-occurring arsenic species are polar or ionic and very water-soluble they favour water over methanol as an extractant (see, e.g. Ref. [6]). Although methanol may still extract some non-polar arsenic, the fact that commonly the methanol is removed and the residue taken up in an aqueous mobile phase prior to chromatographic separation means that non-polar arsenic species are “lost” in this stage. Extraction and identification of non-polar arsenic species is still in its infancy and only recently work on arsenolipids has shown significant progress with the advance of HPLC–ICPMS (high performance liquid chromatography–inductively coupled plasma mass spectrometry) analytical procedures for direct and quantitative measurement of arsenolipids in hexane extracts [7]. To date, shakers and rotators are still the most commonly used devices in the extraction of arsenic species from biological tissues. The extraction efficiency may be enhanced by using classical devices, such as the Soxhlet apparatus but less elaborate protocols using sonication, accelerated solvent extraction and microwave-assisted extraction have taken over [8–10]. These procedures are meant to increase the desorption characteristics of arsenic species by weakening the bond between arsenic species and sample matrix and are thus unable to probe, e.g. As(III) complexation due to inherent speciation alteration. Speciation techniques in use are mostly ion-exchange chromatographic separation procedures combined with elementselective detectors such as ICPMS, HGAFS (hydride generation atomic fluorescence spectrometry), HGAAS (hydride generation atomic absorption spectrometry), etc. [5]. In this work the extractability of arsenic species from the algal CRM IAEA-140/TM material (Fucus sp., Sea Plant Homogenate) was studied in a systematic way using a shaker with methanol/water extractant mixtures of different ratios, from pure methanol to pure water, therefore defining varying solubilization and desorption parameters. To study desorption
25
characteristics the extractant volume to algal mass (V/m) ratios were varied in the range 25–250 mL g−1 , thereby allowing the deduction of the initial arsenic species concentrations in algal material under the conditions applied using a simple model based on the assumption of linear sorption isotherm behaviour. For speciation analysis water/methanol extracts were directly brought onto an HPLC column interfaced with an ICPMS. The data will give us more insight into solubility and desorption phenomena of arsenic species with regards to the specific algal material in this study. 2. Experimental 2.1. Sample, standards and reagents The certified reference material IAEA-140/TM material (Fucus sp., Sea Plant Homogenate) was purchased from the International Atomic Energy Association (Vienna, Austria). The certified arsenic concentration is 44.3 mg kg−1 , with an associated uncertainty of 5%. All reagents were of analytical-reagent grade unless mentioned otherwise. Standard solutions of 1000 mg As L−1 were prepared in Milli-Q water (18.2 M cm) for the following species: arsenite and arsenate prepared from NaAsO2 and Na2 HAsO4 ·7H2 O, respectively (Merck, Darmstadt, Germany); methylarsonate (MMA) prepared from methylarsonic acid synthesised from As2 O3 and CH3 I; dimethylarsinate (DMA) prepared from sodium dimethylarsinate (Fluka, Buchs, Switzerland); and the arsenosugars 1–4, obtained from natural sources [11], often referred to as glycerol, phosphate, sulfonate and sulfate arsenosugar, respectively (see Fig. 1, for structures). Ammonium dihydrogen phosphate (p.a.), methanol (p.a.) and aqueous ammonia solution (25%, Suprapur) were purchased from Merck. 2.2. Extraction procedure Varying amounts of algal certified reference material (0.1–1.0 g dw) were accurately weighed into 50 mL “Falcon” tubes and 25 mL of extractant solution was added to the tubes, yielding V/m ratios in the range of 25–250 L kg−1 . Extractant solutions were prepared from mixtures of methanol and water with (v/v) ratios of 0, 10, 25, 50 and 100% methanol (in water).
Fig. 1. Structures of arsenosugars.
26
J.T. van Elteren et al. / Analytica Chimica Acta 585 (2007) 24–31
The tubes were shaken in a reciprocal shaking water bath (Julabo SW 22, Julabo Labortechniek, Germany) with a stroke of 15 mm and a shaking speed of 200 strokes min−1 for 16 h, thermostatted at 20 ◦ C. After the required shaking time the samples were immediately centrifuged at 3200 rpm for 15 min. The solutions were carefully removed with a poly(propylene) Pasteur pipette and kept frozen (−20 ◦ C) until analysis. Extracts were analysed within 1 week of preparation to prevent changes in species. For determination of the moisture content samples were weighed in a glass vial and dried (85 ◦ C) to a constant weight (ca. 4 h).
where m is the mass of the solid material (kg), V the volume of extractant (L), ci the equilibrium concentration of species i in extract (mg L−1 ) and ai is the remaining equilibrium concentration of species i in solid material (mg kg−1 ). Under the assumption of a linear sorption isotherm, i.e. ai = Di × ci , with Di denoting the partitioning coefficient (L kg−1 ) of species i between solid and liquid, Eq. (2) may be rewritten to:
2.3. Arsenic species by HPLC–ICPMS
Note that Di is a function of the extraction parameters such as extractant composition, temperature, shaking time, shaking frequency, shaking amplitude, etc. The extraction yield Yi (% of ai ,0) of species i in the extract is given by: ci × V/m V/m Yi = × 100 (4) × 100 = ai , 0 V/m + Di
HPLC was performed with a Hewlett-Packard 1100 Series system (Hewlett-Packard, Waldbronn, Germany) equipped with quarternary pump, vacuum degasser, column oven set to 40 ◦ C, and thermostatted autosampler with a variable injection loop set at 20 L. Separation was carried out on an anion-exchange column (PRP-X100; 250 mm × 4.1 mm i.d.; Hamilton Company, Reno, NV, USA) with a mobile phase of 20 mmol L−1 NH4 H2 PO4 at pH 5.6 (adjusted with 25% aqueous NH3 ) and a flow rate of 1.5 mL min−1 . The outlet of the HPLC column was directly connected to an ICPMS (Agilent 7500c, Waldbronn, Germany), equipped with a Babington-type nebulizer with PEEK (polyetheretherketone) capillary tubing (0.125 mm i.d.). The ion intensities at m/z, 75 and 77 were monitored. The ICPMS signal was optimised with a solution of the mobile phase containing 10 g As L−1 to give maximum response on the As signal (m/z, 75). Arsenic species were quantified with an external calibration against standard solutions of the relevant species.
The total arsenic concentration in the methanol/water extracts was determined by ICPMS (Agilent 7500ce, Agilent Technologies, Palo Alto, CA, USA), equipped with a concentric nebulizer. The ion intensity at m/z, 75 was monitored. Methanol/water extracts were diluted in such a way that they all contained 10% (v/v) methanol (in water). Arsenic species were quantified with an external calibration against a standard solution of inorganic arsenic in 10% (v/v) methanol (in water). 3. Extraction theory When an element is present in a solid material, the total concentration at (mg kg−1 ) may be subdivided into n subconcentrations for n species containing the element of interest: n
ai , 0
(1)
i=1
with ai ,0 (mg kg−1 ) the initial concentration of species i in the solid material. For each species i extracted from this solid material the following mass balance is valid: m × ai , 0 = ci × V + m × ai
ai , 0 V/m + Di
(2)
(3)
The total extraction yield Yt (% of at ) of the element of interest, comprising n species, is given by: n
Yt =
Yi × a i , 0
i=1
(5)
at
From Eq. (4), it follows that the highest yields for the individual species i may be expected at high V/m ratios and/or low Di values, thereby giving the highest total extraction yield Yt ,max: n
Yt , max =
2.4. Total arsenic by ICPMS
at =
ci =
i=1
ai , 0
at
× 100
(6)
To illustrate the above assume that an element in a solid comprises three species with identical initial concentrations: a1 ,0 = a2 , 0 = a3 ,0 = 10 mg kg−1 . Furthermore, assume that the species have different extractabilities: species 1, completely extractable (D1 = 0); species 2, moderately extractable (D2 = 25 L kg−1 ); species 3, non-extractable (D3 = ∞). In Fig. 2, the extraction fate of the species under varying V/m ratios is given; it can be observed that: the concentration ci in extract decreases hyperbolically to 0, except for species 3 which is 0 for all V/m ratios due to their non-extractability (A); the extractability visualised by ci × V/m increases hyperbolically to ai ,0, except for species 1 and 3 which are a1 ,0 and 0 for all V/m ratios due to complete extractability and non-extractability, respectively (B); the remaining concentration ai in solid decreases hyperbolically to 0, except for species 1 which is 0 for all V/m ratios due to complete extractability and species 3 which stays on the initial level a3 ,0 due to non-extractability (C); the extraction yield Yi for the species increases hyperbolically, except for species 1 which is 100% due to complete extractability and species 3 which is 0 due to non-extractability (D); the total extraction yield Yt increases hyperbolically to Yt ,max = 66.67% (E). From Fig. 2D and E, it can be remarked that although a total extraction yield Yt may show little overall variation with V/m ratio this does not necessarily imply that this is also the case for the extraction
J.T. van Elteren et al. / Analytica Chimica Acta 585 (2007) 24–31
27
Fig. 2. Extraction fate of an element comprising of three species (a1 ,0 = a2 ,0 = a3 ,0 = 10 mg kg−1 ; D1 = 0; D2 = 25 L kg−1 ; D3 = ∞) as a function of V/m ratio.
yield Yi of the individual species in case ai ,0 is small compared to the bulk at and the Di value is high. Also, the unextractable fraction becomes immediately apparent when a plateau is reached which is less than 100% in the Yt versus V/m graph. From the above it is obvious that obtaining the initial species concentrations ai ,0 is not straightforward when the extractability of the individual species is in the range of approximately 5–95% (outside this range we may regard species as nonextractable or completely extractable); this translates into Di ’s in the range of 13.2–4750 L kg−1 in the case of a high V/m ratio, such as 250 L kg−1 (according to Eq. (4)). To this end we need to find ai ,0 (or Di ) indirectly which may be achieved by
rewriting Eq. (2): V/m Di 1 = + ci ai , 0 ai , 0
(7)
From a plot of 1/ci versus V/m the concentration ai ,0 of species i can be derived from the slope and the partitioning coefficient Di of the species from the intercept; when the intercept is zero the species is completely extractable. This approach has been applied for determination of methylmercury in mushrooms [12], organotin compounds in sediments [13] and heavy metals in soil [14,15]. This is illustrated in Fig. 2F for the above example of three species. The following can be observed from Fig. 2F: (i)
28
J.T. van Elteren et al. / Analytica Chimica Acta 585 (2007) 24–31
species 1 has a slope of 0.1 kg mg−1 and an intercept of 0, denoting an a1 ,0 of 10 mg kg−1 and a D1 of 0, respectively; (ii) species 2 has a slope of 0.1 kg mg−1 and an intercept of 2.5 L mg−1 , denoting an a2 ,0 of 10 mg L−1 and a D2 of 25 L kg−1 , respectively; (iii) species 3 can not be plotted since its unextractability and thus infinite D3 value leads to an infinitely low concentration of species 3 in the extract and therefore infinitely high reciprocal concentrations. This shows us that the initial concentration ai ,0 in the solid, of moderately to completely extractable species, may be derived from this variable volume approach. The above theory for deduction of the extraction yield of elemental species from solid material may be restricted when significant losses in the subsequent measurement procedure occur. Since the extracted species cannot be determined directly but only indirectly via chromatographic separation and elementselective detection, losses may occur by (partial) binding of certain species on the chromatographic column. Strong retention of, e.g. thio-arsenosugars on an anion exchange column (PRPX100) has been reported a problem recently [16] resulting in the misinterpretation of the extraction data when not identified. To circumvent such misinterpretations it is essential to compare the total concentration of the element of interest before and after the chromatographic column. As the total concentration after the column the sum of the initial arsenic species concentrations ( ai ,0, mg kg−1 ) may be taken. As the total concentration before the column the total arsenic concentration ct in the extract (for each extractant) as a function of V/m ratio may be used to calculate the maximal extractable arsenic concentration (at ,ext, −1 ), using a linearisation similar to Eqs. (6) and (7). In mg kg case ai ,0 is significantly lower than at ,ext, and thus (partial) binding on the column occurs, conclusions on extraction yields for individual species may not be drawn. 4. Results and discussion Algae, especially the ones from the Fucus family, are interesting in that they often contain high concentrations of arsenosugars (up to 100 mg kg−1 wet mass) [17]. As far as we know the specific algal material used in this study (CRM IAEA-140/TM, Fucus sp., Sea Plant Homogenate) has only been certified for total arsenic and the only arsenic speciation data we are aware of ˇ are reported by Slejkovec et al. [2]. The following arsenic species were found in this particular algal material (in chromatographical order): arsenosugar 1, DMA, arsenosugar 2, arsenosugar 3, As(V) and arsenosugar 4. The presence of high amounts of mostly arsenosugars is in accordance with findings reported for related algae such as Fucus serratus [18] (arsenosugars 1–4), Fucus spiralis [19] (arsenosugar 4), Fucus distichus [20] (arsenosugars 1–4) and Fucus virsoides [2] (arsenosugars 1, 3 and 4). Initial concentrations ai ,0 in algal material for various methanol/water extractant mixtures may be estimated from ci × V/m versus V/m graphs by extrapolation to high V/m ratios (Fig. 3). Although some “noise” in the graphs may be expected for higher V/m ratios since the sample gets proportionally diluted (e.g. at a V/m ratio of 250 mL g−1 the total arsenic concentration
in extract is only ca. 0.18 mg L−1 , assuming extraction yields of 100% for all individual species), this does not explain the wavy curves obtained for the 0% methanol (in water) extractant. This aspect may have to do with irreproducible wettability (due to too high surface tension) of the sample with the extractant since data from a later extraction, where the sample was sonicated for ca. 5 min prior to extraction, show much less of this “wavyness” (Fig. 4). From Fig. 4, it is obvious that in particular arsenosugar 4 is still badly affected by the wettability phenomenon, whereas for all other species this phenomenon is much less pronounced; only at higher V/m ratios some abnormalities still show. The other extractants (see Fig. 3) used for extraction of arsenic species from the algal material show no irregularities. It can be observed from Fig. 3, that arsenosugars 1, 3 and 4 are the predominant arsenic species extracted, with concentrations in the order of 5–10 mg kg−1 , whereas arsenosugar 2, DMA and As(V) are present at (much) lower levels (0–2 mg kg−1 ). It should be noted that the 100% methanol (in water) extract, in contrast to the other extracts, contained unmeasurably low As(V) concentrations (<0.0005 mg L−1 , DL). This implies that As(V) is badly soluble or non-soluble in pure methanol as supported by literature data [21]. Furthermore, this extractant was the only one where the arsenic species extractabilities were (slightly) dependent on the V/m ratio as may be best visualised by the extraction yields Yi (Fig. 5). In particular for arsenosugar 2 a high dependency of Yi on V/m ratio was found which may be explained by the functionality of the groups on the arsenosugar. Although the dimethylarsinoyl group (Me2 As–) is likely to be protonated below pH 3, thereby giving polar behaviour to arsenosugar 1 in methanol, the basic property of the dimethylarsinoyl moiety is outweighed by the acidic groups in the aglycone for the other three arsenosugars [18]. The phosphate, sulfonate and sulfate groups in the respective arsenosugars 2–4 show increasing acid strength, implying that the arsenosugar 2 is relatively most prone to be present in ionised form, thus yielding the least polar arsenosugar with an inherent worst extraction efficiency in methanol. By linearising the data according to Eqs. (6) and (7), plotting 1/ci versus V/m, the initial concentration ai ,0 and partitioning coefficient Di can be derived with a very high degree of precision and accuracy, even for moderately extractable species, such as the above mentioned arsenosugar 2 in methanol. An initial arsenosugar 2 concentration in CRM IAEA140/TM of 0.32 ± 0.01 mg kg−1 and a partitioning coefficient of 52 ± 9 L kg−1 (n = 10) may be derived from the slope and intercept, respectively. The partitioning coefficient is so high that even three repetitive extractions with pure methanol do not increase the total (calculated) extraction efficiency for this species to more than 69.3% at a V/m ratio of 25 L kg−1 . For the other arsenic species hold-up in the algal material is much less of a problem in pure methanol and even less for the water/methanol extractants. However, when we tabulate all initial arsenic species concentrations, derived from linearisation of the data according to Eqs. (6) and (7), it is obvious that some arsenic is “missing” (see Table 1). In Table 1, the data for extraction with pure water are discarded due to problems with erratic extraction,
J.T. van Elteren et al. / Analytica Chimica Acta 585 (2007) 24–31
29
Fig. 3. Extractability of arsenic species from CRM IAEA-140/TM material as a function of V/m ratio for various methanol (in water) extractants.
probably as a result of wettability (see Fig. 4). Only 40–70% (depending on the extractant composition) of the total arsenic at is accounted for; this is in-line with extraction efficiencies of 50–60% reported for related algae [2]. This recalcitrant arsenic concentration of 13–25 mg kg−1 has necessarily a high Di value, of at least of 26,000–50,000 L kg−1 (for a V/m ratio of 25 L kg−1 ), to be undetectable (ci < 0.0005 mg L−1 , DL). However, to exclude trivial artifacts such as irreversible binding on the chromatographic column it is essential to compare the sum of arsenic species after the column with the total arsenic con-
centration in the extract prior to chromatographic separation (see theoretical section). In Fig. 6, such a comparison is made and it can be concluded that no irreversible binding on the column occurs for methanol/water extractants with 10, 25 and 50% methanol (in water), at least not at the 5% significance level (18 d.f.). With 100% methanol (in water) there is a slight discrepancy, showing statistically less (84%) arsenic after the column than before the column. This may be attributed to the binding of certain species on the column under these conditions which may have
30
J.T. van Elteren et al. / Analytica Chimica Acta 585 (2007) 24–31
Fig. 4. Extractability of arsenic species from CRM IAEA-140/TM material using pure water (0% methanol) after initial sonication (5 min).
Fig. 5. Extraction yield of arsenic species from CRM IAEA-140/TM material using pure methanol (100% methanol).
Table 1 Initial arsenic species concentrations ai ,0 in CRM IAEA-140/TM for varying methanol/water extractant mixtures Species, i
Assug1 DMA Assug2 Assug3 As(V) Assug4
ai ,0
8.36 1.85 0.82 7.85 0.41 9.53
25 ± ± ± ± ± ±
0.06 0.02 0.02 0.07 0.01 0.07
28.82 ± 0.12
8.86 1.90 0.75 8.20 0.40 10.13
50 ± ± ± ± ± ±
0.04 0.02 0.02 0.09 0.02 0.08
30.24 ± 0.13
been caused by release of specific non-ionic components upon extraction, responsible for holdup behaviour on the column. From Fig. 6, it can be seen that there is a certain extraction pattern, with 25% methanol extracting the highest amount of arsenic. However, this extraction pattern is not exactly the same for the individual arsenic species (see Table 1) where extraction of arsenosugar 1 seems to be enhanced, in contrast to the other arsenic species, at higher (>25%) methanol concentrations, probably due to arsenosugar 1 being the most polar arsenic species present. At lower (<25%) methanol concentrations arsenosugar 2 and As(V) extraction seem to be favoured over the other arsenic species due to their inherent ionic character (see above). The remaining, unextractable, fraction stays either physico-chemically bound to cellular components (e.g. protein bound arsenic), requiring more severe extraction conditions for desorption, or is in a form (e.g. lipid arsenic) where the extractant is unable to solubilize the associated arsenic species. Evidence for the former species has been found in Fucus vesiculosus, where As(III)-sequestering has been demonstrated to occur through the binding of five As(III) atoms to methallothionein [22]. The latter arsenolipids have been found, but not identified in fish oil and cod liver [7]. 5. Conclusions
% (v/v) methanol (in water) 10
Fig. 6. Total arsenic concentration in algae derived in different stages of the extraction-measurement procedure: before the column (at ,ext) and after the col umn ( ai ,0); as reference value the total arsenic concentration in algae is given (at = 44.3 mg kg−1 ).
8.94 1.62 0.56 7.59 0.19 8.31
100 ± ± ± ± ± ±
0.04 0.01 0.01 0.14 0.02 0.17
27.21 ± 0.23
6.97 ± 0.07 1.01 ± 0.02 0.32 ± 0.01 4.00 ± 0.07 ND 5.85 ± 0.13 18.15 ± 0.17
Standard deviations (σ i ) in ai ,0 (n = 10) were calculated using Microcal (TM) Origin® , version 7.5 (Microcal Software Inc., Northampton, USA) by fitting the data in Eqs. (6) and (7). The total arsenic concentration at certified in this material is 44.3 mg kg−1 .
The processed data from the variable volume extraction approach are precise and highly accurate, showing that varying methanol/water mixtures yield varying maximal extractabilities for a range of arsenic species in CRM IAEA-140/TM. Since this topology corrects for solubility issues of these arsenic species, it is obvious that the extraction is desorption-controlled. Therefore, part of these arsenic species are unaccounted for as a result of the desorption conditions being too mild, whereas a fraction of unknown arsenic species may still be “forgotten” due to solubility/desorption issues under the extraction conditions used. As a consequence it is not possible to definitely quantify the
J.T. van Elteren et al. / Analytica Chimica Acta 585 (2007) 24–31
arsenic species present in CRM IAEA-140/TM but only give extractant-specific concentrations. For algal CRM IAEA-140/TM material a large (unidentified) fraction of the arsenic is unextractable (Di is very high) in methanol/water extractant mixtures as a result of strong binding to cellular components and/or inability for solubilization by the extractants used. A more severe desorption procedure, using, e.g. microwave assisted extraction, may release (part) of this recalcitrant arsenic fraction and either the less polar or nonpolar extractant may reach the hydrophobic arsenic species. The complementary, mostly arsenosugar fraction, is easily solubilized (Di is 0) in all extractant mixtures except pure methanol which shows a particularly bad affinity for arsenosugar 2, comparatively the most ionic arsenosugar. The fact that varying maximal extractabilities are found for different methanol/water mixtures can not be attributed to differences in the solubilities of arsenic species, at least not when the same species are extracted, since an extrapolation to infinite extractant volume to algal mass ratios has been carried through. However, assuming that (some) arsenic species are sorbed onto algal cell components it is likely that the desorption is extractantdependent. Thus, we are not able to definitely quantify the arsenic species present in CRM IAEA-140/TM since they vary with extractant composition. Even the large unextractable fraction may ultimately contain no other species than the ones found except on much higher levels, although it is also possible that different types of physico-chemical species, e.g. hydrophobic ones, are present which are unextractable in the extractants used. The general conclusion of this work is that it is impossible to give defined species distribution concentrations when the very species are not completely extractable. As a consequence this implies that species in materials can generally not be certified, unless a given extraction condition is specified. So, specific extraction conditions will only take us as far as an elemental species fingerprint of a material under the very extraction conditions. Furthermore, it seems that we have to redefine our notion
31
of extractability for speciation purposes as striving for maximal extractability might be conflicting with species integrity. References [1] J. Pawliszyn, J. Chromatogr. Sci. 31 (1993) 31. ˇ [2] Z. Slejkovec, E. K´apolna, I. Ipolyi, J.T. van Elteren, Chemosphere 63 (2006) 1098. [3] M. Letellier, H. Budzinski, Analusis 27 (1999) 259. ˇ [4] J.T. van Elteren, Z. Slejkovec, I. Arˇcon, H.J. Glass, Environ. Pollut. 139 (2006) 477. [5] K.A. Francesconi, D. Kuehnelt, Analyst 129 (2004) 373. [6] J. Szakova, P. Tlustos, W. Goessler, D. Pavlikova, J. Balik, Appl. Organomet. Chem. 19 (2005) 308. [7] E. Schmeisser, W. Goessler, N. Kienzl, K.A. Francesconi, Analyst 130 (2005) 948. [8] J.L. Gomez-Ariza, D. Sanchez-Rodas, I. Giraldez, E. Morales, Analyst 125 (2000) 401. [9] C.G. Yuan, G.B. Jiang, B. He, J. Anal. Atom. Spectrom. 20 (2005) 103. [10] R. Wahlen, T. Catterick, Rapid Commun. Mass Spectrom. 18 (2004) 211. [11] K.A. Francesconi, J.S. Edmonds, R.V. Stick, B.W. Skelton, A.H. White, J. Chem. Soc., Perkin Trans. 1 (1991) 2707. [12] R.G. Fischer, S. Rapsomanikis, M.O. Andreae, F. Baldi, Environ. Sci. Technol. 29 (1995) 993. [13] Y. Cai, S. Rapsomanikis, M.O. Andreae, J. Anal. Atom. Spectrom. 8 (1993) 119. [14] J.T. van Elteren, B. Budiˇc, Anal. Chim. Acta 514 (2004) 137. ˇ [15] J.T. van Elteren, Z. Slejkovec, R. Milaˇciˇc, Int. J. Environ. Anal. Chem. 83 (2003) 389. [16] E. Schmeisser, R. Raml, K.A. Francesconi, D. Kuehnelt, A.L. Lindberg, C. S¨or¨os, W. Goessler, Chem. Commun. 16 (2004) 1824. [17] W.R. Cullen, K.J. Reimer, Chem. Rev. 89 (1989) 713. [18] A.D. Madsen, W. Goessler, S.N. Pedersen, K.A. Francesconi, J. Anal. Atom. Spectrom. 15 (2000) 657. [19] G. Raber, K.A. Francesconi, K.J. Irgolic, W. Goessler, W. Fresenius, J. Anal. Chem. 367 (2000) 181. [20] C.F. Harrington, A.A. Ojo, V.W.M. Lai, K.J. Reimer, W.R. Cullen, Appl. Organomet. Chem. 11 (1997) 931. ˇ [21] A.R. Byrne, Z. Slejkovec, T. Stijve, L. Fay, W. Goessler, J. Gailer, K.J. Irgolic, Appl. Organomet. Chem. 9 (1995) 305. [22] M.E. Merrifield, T. Ngu, M.J. Stillman, Biochem. Biophys. Res. Commun. 324 (2004) 127.