Quantifying hydrogen peroxide in iron-containing solutions using leuco crystal violet

  • Corey A Cohn1Email author,

    Affiliated with

    • Aimee Pak2,

      Affiliated with

      • Daniel Strongin3, 5 and

        Affiliated with

        • Martin A Schoonen4

          Affiliated with

          Geochemical Transactions20056:47

          DOI: 10.1186/1467-4866-6-47

          Received: 24 February 2005

          Accepted: 26 April 2005

          Published: 14 June 2005

          Abstract

          Hydrogen peroxide is present in many natural waters and wastewaters. In the presence of Fe(II), this species decomposes to form hydroxyl radicals, that are extremely reactive. Hence, in the presence of Fe(II), hydrogen peroxide is difficult to detect because of its short lifetime. Here, we show an expanded use of a hydrogen peroxide quantification technique using leuco crystal violet (LCV) for solutions of varying pH and iron concentration. In the presence of the biocatalyst peroxidase, LCV is oxidized by hydrogen peroxide, forming a colored crystal violet ion (CV+), which is stable for days. The LCV method uses standard equipment and allows for detection at the low microM concentration level. Results show strong pH dependence with maximum LCV oxidation at pH 4.23. By chelating dissolved Fe(II) with EDTA, hydrogen peroxide can be stabilized for analysis. Results are presented for hydrogen peroxide quantification in pyrite–water slurries. Pyrite–water slurries show surface area dependent generation of hydrogen peroxide only in the presence of EDTA, which chelates dissolved Fe(II). Given the stability of CV+, this method is particularly useful for field work that involves the detection of hydrogen peroxide.

          Introduction

          Hydrogen peroxide is an important reactant in natural aquatic systems, environmental remediation technologies, and in biological systems. Reactive oxygen species (ROS), which include hydrogen peroxide (H2O2) and hydroxyl radicals (·OH) occur in rain[1, 2] and surface waters. [3, 4] ROS play an important role in natural processes in aquatic systems, including, radiolysis,[5, 6] pyrite oxidation, [79] and photochemical oxidation.[10, 11] Hydrogen peroxide also is intentionally added to wastewaters to promote in situ oxidation processes[12, 13] by leveraging the extreme reactivity of radicals toward organic pollutants. The generation of ·OH, via the decomposition of H2O2, degrades contaminants when methods involving microbiological degradation are ineffective. In organisms, including humans, ROS are produced during metabolic and immune system function.[14] When ROS concentrations are above normal for prolonged periods of time, however, their presence can lead to oxidative stress.[15] Oxidative stress is now recognized to be an important factor in the development or enhancement of many diseases.[16]

          Among ROS, H2O2 is relatively stable, in the presence of ferrous iron, H2O2 forms ·OH via the Fenton reaction

          Fe(II) + H2O2·OH + OH- + Fe(III)     (1)

          ·OH is far more reactive than H2O2 and its reaction with aqueous species is diffusion limited.[17] Therefore, H2O2, in the presence of Fe(II), represents great potential for reactivity.

          In natural systems, ·OH is a transient species and its steady-state concentration may only be in the nM range. The inherent reactivity of ·OH precludes its detection via a direct measurement, and hence, methods generally rely on the detection of a stable reaction product resulting from the reaction of ·OH and a target molecule. One strategy is to add a reactant as a target species that will oxidize in the presence of ·OH and form a product that can be analyzed by UV–Vis spectroscopy[2] or fluorescence[18]. Verifying ·OH involvement in oxidation of the target species is difficult. Addition of competing scavengers that react with ·OH may inhibit oxidation of the target molecule; however, there is little conformity when using different scavengers and target molecules.[19] Another strategy that is often employed is to "trap" the unpaired electron of the radical into a compound that is less reactive. By using electron paramagnetic resonance (EPR), the EPR intensity of the spin-trapped molecule can be directly related to the concentration of ·OH in the solution. EPR spin-trapping is a very sensitive technique and has been widely used in chemistry, environmental sciences, and bio-(medical) sciences; however, it does require a significant capital investment and spin-trapping cannot be conducted in the field. The spin-trapping technique has also been used for ·OH detection from aqueous mineral slurries, however reaction between Fe(III) and the spin-trap complicates interpretation of the data.[20]

          Where detection of ·OH is impracticable, identifying H2O2 may elucidate a reaction mechanism involving ·OH formation. In iron-containing systems, both H2O2 and ·OH are short-lived. However, chelation of iron [i.e., Fe(II)] avoids its interaction and decomposition of H2O2. Hence, we argue that hydrogen peroxide detected in iron-containing systems with an iron-chelator added is a proxy for the capacity of the system to generate ·OH in the absence of an iron chelator. This strategy may prove useful as an alternative method for radical detection in natural systems.

          Several methods are available for quantifying H2O2. UV–Vis absorbance (240 nm, ε = 43.6 M-1 cm-1)[21] is convenient for H2O2 solutions containing no other UV absorbing chromophores. Another spectrophotometric method uses copper(II) and 2,9-dimethyl-l, 10-phenanthroline (DMP) for μM H2O2 detection in wastewater.[22] The colored complex does not change in the presence of humic acid; however, metal chelators affect the copper reactivity, complicating its use when iron chelation is necessary. For higher sensitivity detection, fluorometric techniques are available. The scopoletin/horseradish peroxidase method[23] allows for detection of low nM concentrations, but the need for standard additions and quantification via a decrease in fluorescence prove to be time consuming. Another fluorometric technique involves the oxidation of non-fluorescent 2',7'-dichloro-fluorescin (DCFH) to fluorescent 2',7'-dichlorofluorescein (DCF) in the presence of H2O2 and peroxidase.[18] This technique is often used to detect the formation of H2O2 in cells. The technique makes use of the non-fluorescent (DCFH-diacetate) crossing of cell membranes, which is enzymatically deacetylated to non-fluorescent DCFH. Cellular production of H2O2 produces the fluorescent DCF. The deacetylation process can be achieved chemically, but H2O2 may be produced in the process, [24] resulting in background levels of H2O2. Another disadvantage for using the DCFH technique is the necessity for daily preparation of the DCFH reagent.

          Here, we present results on an improved method for H2O2 detection in the μM to several hundred nM range, which has several advantages over preexisting methods. The leuco crystal violet (LCV) method[25] involves oxidation of 4,4'4"-methylidynetris (N,N-dimethylaniline) (C25H31N3) (LCV) in the presence of H2O2 and horseradish peroxidase (HRP) to form the crystal violet ion, CV+, which absorbs at 590 nm. CV+ remains stable for several days, which makes it possible to treat samples upon collection and perform the analysis at a later time. This is of value in field studies or shipboard analysis. [26] Daily preparation of new reagents is not required and iron chelators do not affect the analysis.

          In this report, the LCV method was employed for analysis of H2O2 in pyrite/aqueous slurries. Recently,[9] pyrite (FeS2), the most abundant metal sulfide mineral on Earth, has been shown to produce H2O2 in aqueous solutions. Several adjustments were made to the original LCV method so that it is now possible to determine H2O2 concentration in aqueous systems that contain iron-bearing minerals. Here, we determined effects of iron, pH, and addition of EDTA. To verify that our technique was specific to the presence of H2O2, and that other species were not responsible for the oxidation of LCV in the presence of HRP, we carried out experiments in the presence of catalase. This enzyme selectively decomposes H2O2, and its addition to the solution prior to the addition of LCV and HRP eliminates the production of the CV+ species.

          Experiment

          Leuco crystal violet (Spectrum) in the presence of the enzyme horseradish peroxidase (HRP) type II (Aldrich) forms a crystal violet cation, which has an absorbance maximum at 590 nm. The absorbance of CV+ was measured with a Hach DR4000 spectrometer interfaced with a laptop for data storage. Calibration curves were used to quantify the effects of pH, iron (Fisher brand ferrous ammonium sulfate), and EDTA (Sigma) on the absorbance of CV+. Catalase (Sigma 40 000 to 60 000 units/ mg bovine liver) was used at a concentration around 100 000 units per sample vial to verify the presence of H2O2. Pyrite (Wards brand from Huanzala, Peru) was crushed and sieved under an ambient atmosphere, acid washed (0.1 M HCl), and then thoroughly rinsed with N2-purged water (Easy Pure 18.3 MΩ cm, UV-irradiated, and ultrafiltered) under N2 atmosphere to remove surface oxides. The acid-washed and cleaned pyrite was subsequently dried in a vacuum desiccator and stored under vacuum. The size fraction used in these experiments was between 10 and 90 μm with a five point N2 adsorption BET surface area of roughly 1.25 m2/g. This BET surface area is an approximation due to the inherent variations for low surface area measurements.[27] Varying amounts of the pyrite were mixed with water in the absence or presence of 1 – l0 mM EDTA, then immediately filtered (Millipore 0.45 μm). Reagents [all stored at 4°C and brought up to room temperature (22 ± 1°C) before analyses] were added to the aqueous filtrate in the following order for a total volume of 2 ml (also shown in Table 1): 100 mM KH2PO4 (Aldrich) pH 4 buffer, 41 μM leuco crystal violet (dissolved with HC1), and 4 mg/50 ml HRP [containing 1.5 mM azide (Sigma) to prevent bacterial growth]. Samples were kept in the dark at room temperature (22 ± 1°C) for 30 min, upon which absorbance stabilized. Absorbance measurements were taken in 1 cm path-length cuvettes.
          Table 1

          LCV method details.a

          Reagent

          Reagent preparation

          Stock concentration

          Volume added (μl)

          Final concentration

          Sampleb

          ...

          ...

          1700

          ...

          KH2PO4

          136.07 g + 0.5 L H2O

          1 M

          200

          100 mM

          pH buffer

          pH adjusted to 4.2 with H3PO4

             

          LCV

          31 mg LCV + 30 ml H2O + 19.2 ml of 0.25 N HCl

          1.65 mM

          50

          41 μM

          HRP

          4.0 mg HRP + 50 ml H2O + 92 μl of 1 M sodium azide

          0.08 mg/ ml (14.4 units/ml)

          50

          1 μg (0.18 units)

          aReagent preparations, concentrations, volume added, and final concentrations of the solutions added to a 4 ml cuvette and kept in the dark before analysis. Reagents listed in order of addition. H2O2 calibration solutions were prepared by dilutions of a stock.

          bSamples containing pyrite were filtered.

          Results and Discussion

          The reagent preparations and volumes used are shown in Table 1. It is useful to present our conditions in the context of prior LCV studies that were carried out by Zhang et al. [26] Compared to the concentrations used by Zhang et al. for pH buffer, LCV, and HRP, we used 1.25, 4, and 10 times those previously used, respectively, to increase (1) the buffer concentration, (2) the higher range of detection (more LCV), and (3) the reaction rate (more HRP). By increasing the concentration of our reagents, we were able to limit the volume of reagent addition so that a maximum volume would be available for the sample.

          Figure 1 shows color development over time after the last reagent, HRP, is added to hydrogen peroxide and LCV. Absorbance levels plateau in 5–10 min and remain stable for several days. Zhang et al. [26]waited 5 min before taking absorbance measurements; we waited 30 min just as a cautionary measure to ensure stability of the absorbance. The inset is a plot of wavelength scans showing maximum absorbance at 590 nm and an overlapping of scans after 2 min. Absorbance measurements at 590 nm (A590) were taken at 30 min for all other samples.
          http://static-content.springer.com/image/art%3A10.1186%2F1467-4866-6-47/MediaObjects/12932_2005_Article_52_Fig1_HTML.jpg
          Figure 1

          Oxidation of LCV in the presence of 20 μM H2O2 and HRP as a function of time after addition of all reagents. Insert shows wavelength scans of the solutions showing maximum absorbance at 590 nm.

          Addition of catalase is important for verifying the presence of H2O2, since other ROS or other reactions may also oxidize LCV. When catalase is added, no H2O2 is detected, which limits the possibility of false positives.

          The LCV method is strongly affected by pH, and Fig. 2 shows A590 as a function of pH. The optimal pH range is 3.6–4.2, where the peak intensity is the highest, but the intensity decreases by half as the pH increases to 4.37 (Fig. 2). Zhang et al. [26] have shown a similar increase in absorbance up to around pH 4.5. They report CV+ precipitation at higher pH, and here, the effect at pH greater than 4.5 is shown. Although A590 is significantly reduced and the solutions are slightly turbid, calibration curves can still be produced and H2O2 quantified.
          http://static-content.springer.com/image/art%3A10.1186%2F1467-4866-6-47/MediaObjects/12932_2005_Article_52_Fig2_HTML.jpg
          Figure 2

          Absorbance of the buffered LCV and HRP solution as a function of pH in the presence of 20 μM H2O2.

          Figure 3 shows calibration curves made between pH 2.99 and 3.75 with 0–24 μM H2O2. At pH values of 3.20 and less, the method is not valid. In the pH range of 3.38–3.55, the calibration curves are only useful at low H2O2 concentrations. At pH 3.64 and 3.75, the curves are nearly linear. These results are consistent with those presented in Fig. 1, which showed the optimal pH range of around 3.6–4.2; here, the calibration curves are nearly linear at pH 3.64 and 3.75. Above pH 3.75, the slopes are expected to be linear (results from higher-pH experiments are presented in the following). In separate experiments, which are not shown, calibration curves were extended into the nM region indicating a lower limit around 0.5 μM.
          http://static-content.springer.com/image/art%3A10.1186%2F1467-4866-6-47/MediaObjects/12932_2005_Article_52_Fig3_HTML.jpg
          Figure 3

          Calibration curves as a function of pH. H2O2 quantification using Beer's law is only applicable at H2O2 ≲12 μM for pH = 3.55 and ≲6 μM for pH = 3.38.

          In the presence of Fe(II), EDTA is necessary for stabilizing H2O2. As solution pH is increased, EDTA is deprotonated, therefore having a higher capacity for chelation (the EDTA stock solution is pH 8). In our experiments, 1 mM EDTA with LCV reagents resulted in a pH around 3.67. At this pH, calibration curves with EDTA are equivalent to those without EDTA. Figure 4 shows the effect of iron on A590 at several H2O2 concentrations in the presence of 1 mM EDTA. Even without H2O2, EDTA and iron oxidize LCV in the presence of HRP. This may be due to reaction of LCV with either the electron-deficient chelated iron or iron-EDTA complex. When ferrous iron is added at low concentrations (0–15 μM) there is an increase in absorbance but the A590 is not affected as the iron concentration is increased above 15 μM iron. At low iron concentrations (0 – 15 μM), the slopes of the calibration curves are nearly identical to those calculated from experiments without EDTA and iron. As iron is increased, H2O2 decreases. This experimental observation is due to a higher proportion of nonchelated iron, which reacts with H2O2. Although some of the H2O2 was decomposed in the presence of iron, linear calibration curves can still be generated. In these experiments, the concentration of EDTA is much greater than that for iron. EDTA chelation capacity decreases with decreasing pH. The optimal pH for the LCV technique is around 4, which necessitates the addition of high concentrations of EDTA.
          http://static-content.springer.com/image/art%3A10.1186%2F1467-4866-6-47/MediaObjects/12932_2005_Article_52_Fig4_HTML.jpg
          Figure 4

          Effect of dissolved ferrous iron on A590 at several H2O2 concentrations (given in μm) in the presence of 1 mM EDTA at pH = 3.67.

          Figure 5 shows calibration curves with 10 mM EDTA solutions at varying ferrous iron concentrations. The pH of these solutions is between 5.9 and 5.6. The A590 does not vary by much as a function of pH in this region. Therefore, differences in the curves are due to iron. Compared to the high EDTA concentration, relatively low iron concentrations decompose H2O2. This is probably due to the low pH where EDTA has a lower chelation capacity. At higher iron concentrations, around 150 – 200 μM, only the higher H2O2 concentrations are detected. The pH for EDTA-containing solutions was buffered but the pH was not forced to the optimal pH of 4.23 because addition of highly acidic buffers alter EDTA chelation capacity and, in the field, it is easier to quantify H2O2 by using a pH-specific calibration curve than trying to force the pH to 4.23.
          http://static-content.springer.com/image/art%3A10.1186%2F1467-4866-6-47/MediaObjects/12932_2005_Article_52_Fig5_HTML.jpg
          Figure 5

          Calibration curves as a function of ferrous iron (added as ferrous ammonium sulfate, given in μm) in 10 mM EDTA.

          Using the DMP method[28] and an ultrahigh vacuum mass spectroscopy technique,[9] pyrite has recently been shown to produce H2O2. Here, we used the LCV method and employed EDTA to verify H2O2 at several particle loadings (Fig. 6). These results show a surface area dependence on H2O2 generation. The H2O2 concentrations reported here are about 200 times lower than those obtained previously (34 μM at a 4 g/L loading).[28] The higher readings in the prior study may have been due to the presence of dissolved Fe(II), which enhances H2O2 readings with the DMP method. The DMP method uses Cu, so EDTA would interfere with the analysis. Hence the values reported in our earlier work based on the DMP method are overestimated.
          http://static-content.springer.com/image/art%3A10.1186%2F1467-4866-6-47/MediaObjects/12932_2005_Article_52_Fig6_HTML.jpg
          Figure 6

          Pyrite-generated H2O2 as a function of particle loadings. 1 mM EDTA was used for the 20 and 40 g/L loadings and 10 mM EDTA was used for the 80 and 160 g/L loadings. The EDTA solutions were quickly mixed with pyrite particles for about 3 s and filtered. LCV and HRP were then added to the filtrate. When catalase is added prior to LCV, H2O2 is not detected.

          Addition of catalase to the solution prior to the addition of LCV results in a colorless solution, suggesting that hydrogen peroxide was not present upon addition of LCV and HRP (Fig. 7). Catalase specifically reacts with H2O2, so that it will remove any H2O2 from the EDTA-treated solution. By performing two measurements, one with EDTA and one with EDTA plus catalase, we rule out false positives due to other reactions that could conceivably lead to the formation of CV+. Without EDTA, there is no LCV oxidation, since in this case ferrous iron, either on the pyrite surface or in solution, is not chelated and it can react with hydrogen peroxide to form ·OH. In a separate study,[29] we have shown that without EDTA, pyrite readily degrades biomolecules including RNA, presumably due to hydroxyl radical generation from the reaction of ferrous iron with hydrogen peroxide. This study also showed that addition of EDTA protects the biomolecules, which are only degraded if ·OH is formed.
          http://static-content.springer.com/image/art%3A10.1186%2F1467-4866-6-47/MediaObjects/12932_2005_Article_52_Fig7_HTML.jpg
          Figure 7

          (Color) Photograph of the cuvettes containing pyrite slurry filtrate and LCV reagents. Both contain 1 mM EDTA. The left cuvette is purple due to CV+. On the right, catalase was added before addition of LCV. Catalase reacts rapidly with H2O2, preventing LCV oxidation and resulting in a colorless solution.

          For H2O2 detection in natural waters, the LCV technique may be one of the more suitable methods. It has already been shown that the method can be used for seawater analyses [26] and here we show that it can be used for a pH range of around 3.5 – 6.0 and in the presence of EDTA. Compared to techniques involving fluorescence, where reagents are prepared on the day of analyses, our solutions of LCV and HRP remained stable for months at 4°C in opaque centrifuge vials. The stability of CV+ upon reaction of H2O2 and the LCV reagents makes it possible to quantify H2O2 several days after sampling. This can be exploited in field studies. For example, the LCV technique could be used to study the spatial and temporal distribution of H2O2 in hot spring waters in Yellowstone National Park. It has been shown that steady-state levels of photochemically produced H2O2 in the surface geothermal waters at Yellowstone National Park reach 200 – 600 nM by late afternoon and decrease to less than 50 nM during the night.[30] With the LCV technique a large number of water samples can be collected and prepared for later analysis. Temperature is expected to affect the rate of the HRP-mediated reaction and it could possibly also affect the stability of the LCV or CV+. Probably the best strategy working with hydrothermal waters is to rapidly cool the sample down to a temperature between 20 and 30°C before adding the reagents. Cooling the samples to much lower temperature may impede the enzyme reaction. Further experimental work would be needed to resolve this temperature dependence. The LCV technique may also prove to be useful to evaluate the performance of environmental remediation projects involving the injection of H2O2 into contaminated waters. Many groundwaters contain dissolved iron, which could make it difficult to determine the residual H2O2 concentration. With the LCV techniques samples can be treated with ETDA and preserved for latter analysis.

          Conclusion

          This study demonstrates a reliable and efficient method for quantifying H2O2 from iron-containing mineral slurries and waste by use of separate calibration curves to account for pH and iron concentrations. The stability of the colored CV+ makes this method suitable for the field or when immediate access to a spectrophotometer is not possible. Relative to the concentration of H2O2 consumed, the reported[25] high molar absorptivity of 75 000 M-1 cm-1 for CV+ makes it possible to determine H2O2 at sub-μM concentration levels. In iron-containing systems at low pH, hydrogen peroxide reacts to form ·OH. Under those conditions, the presence of H2O2 as an intermediate to ·OH formation would be extremely difficult to detect. The LCV technique as outlined provides a relatively simple method to demonstrate the involvement of H2O2 under those conditions.

          Declarations

          Acknowledgements

          This work was funded by the Department of Energy through grants to D.R.S. and M.A.S., Basic Energy Sciences Grant Nos. DEFG029ER14644 and DEFG0296ER14633, respectively. The Center for Environmental Molecular Science (NSF CHE 0221934) facilitated the contribution of A.P. to this project. C.C. would like to acknowledge support from a National Defence Science and Engineering Fellowship.

          Authors’ Affiliations

          (1)
          Department of Geosciences and Center for Environmental Molecular Science, Stony Brook University
          (2)
          Department of Chemistry and Center for Environmental Molecular Science, Stony Brook University
          (3)
          Department of Chemistry, Temple University
          (4)
          Department of Geosciences and Center for Environmental Molecular Science, Stony Brook University
          (5)
          Center for Environmental Molecular Science, Stony Brook University

          References

          1. Willey JD, Kieber RJ, Lancaster RD, Atmos J: Chem. 1996, 25: 149.
          2. Faust BC, Allen JM: Environ Sci Technol. 1993, 27: 1221-10.1021/es00043a024.View Article
          3. Hakkinen PJ, Anesio AM, Graneli W: Can J Fish Aquat Sci. 2004, 61: 1520-10.1139/f04-098.View Article
          4. Gerringa LJA, Rijkenberg MJA, Timmermans KR, Buma AGJ: Netherlands J Sea Res. 2004, 51: 3.View Article
          5. Amme M, et al: Environ Sci Technol. 2005, 39: 221.View Article
          6. Clarens F, et al: Environ Sci Technol. 2004, 38: 6656-10.1021/es0492891.View Article
          7. Ahlberg E, Broo AE: Int J Min Process. 1996, 47: 49-10.1016/0301-7516(95)00100-X.View Article
          8. Allen JM, Lucas S, Allen SK: Envir Toxicol Chem. 1996, 15: 107-10.1897/1551-5028(1996)015<0107:FOHROI>2.3.CO;2.
          9. Borda MJ, Elsetinow AR, Strongin DR, Schoonen MA: Croat Chem Acta. 2003, 67: 935.
          10. Stumm W, Morgan JJ: Aquatic Chemistry: Chemical Equilibria and Rates in Natural Waters. 1995, Wiley-Interscience, New York, 1022-3
          11. Vaughan P, Blough N: Environ Sci Technol. 1998, 32: 2947-10.1021/es9710417.View Article
          12. Gogate PR, Pandit AB: Adv Environ Res. 2004, 8: 501-10.1016/S1093-0191(03)00032-7.View Article
          13. Arienzo M: Chemosphere. 1999, 39: 1629-10.1016/S0045-6535(99)00061-2.View Article
          14. Dröge W: Physiol Rev. 2002, 82: 47.View Article
          15. Nordberg J, Arnér ESJ: Free Radic Biol Med. 2001, 31: 1287-10.1016/S0891-5849(01)00724-9.View Article
          16. Finkel T, Holbrook NJ: 2000, Nature (London), 408: 239.
          17. Pryor WA: Annu Rev Physiol. 1986, 48: 657-10.1146/annurev.ph.48.030186.003301.View Article
          18. Lebel CP, Ischiropoulos H, Bondy SC: Chem Res Toxicol. 1992, 5: 227-10.1021/tx00026a012.View Article
          19. Winterbourn CC: Free Radic Biol Med. 1987, 3: 33-10.1016/0891-5849(87)90037-2.View Article
          20. Makino K, Hagiwara T, Hagi A, Nishi M, Murakami A: Biochem Biophys Res Commun. 1990, 172: 1073-10.1016/0006-291X(90)91556-8.View Article
          21. Hildebrandt AG, Roots I: Arch Biochem Biophys. 1975, 171: 385-10.1016/0003-9861(75)90047-8.View Article
          22. Kosaka K, Yamada H, Matsui S, Echigo S, Shishida K: Environ Sci Technol. 1998, 32: 3821-10.1021/es9800784.View Article
          23. Holm T, George G, Barcelona M: Anal Chem. 1987, 59: 582-10.1021/ac00131a010.View Article
          24. Rota C, Chignell CF, Mason RP: Free Radic Biol Med. 1999, 27: 873-10.1016/S0891-5849(99)00137-9.View Article
          25. Mottola HA, Simpson BE, Gorin G: Anal Chem. 1970, 42: 410-10.1021/ac60285a017.View Article
          26. Zhang LS, Wong GTF: Talanta. 1994, 41: 2137-10.1016/0039-9140(94)00199-5.View Article
          27. Gregg SJ, Sing KSW: Adsorption Surface Area and Porosity. 1982, Academic, London
          28. Borda M, Elsetinow A, M Schoonen, Strongin D: Astrobiology. 2001, 1: 283-10.1089/15311070152757474.View Article
          29. Cohn CA, Borda MJ, Schoonen MA: Earth Planet Sci Lett. 2004, 225: 271-10.1016/j.epsl.2004.07.007.View Article
          30. Wilson CL, et al: Environ Sci Technol. 2000, 34: 2655-10.1021/es9906397.View Article

          Copyright

          © American Institute of Physics 2005