Environ. Sci. Technol. 2010, 44, 2555–2561
Improvement of 31P NMR Spectral Resolution by 8-Hydroxyquinoline Precipitation of Paramagnetic Fe and Mn in Environmental Samples S H I M I N G D I N G , * ,† D I X U , † B I N L I , † C H E N G X I N F A N , †,‡ A N D CHAOSHENG ZHANG§ State Key Laboratory of Lake Science and Environment, Nanjing Institute of Geography and Limnology, Chinese Academy of Sciences, Nanjing 210008, China, College of Forest Resources and Environment, Nanjing Forestry University, Nanjing 210037, China, and School of Geography and Archaeology, National University of Ireland, Galway, Ireland
Received November 23, 2009. Revised manuscript received February 12, 2010. Accepted February 24, 2010.
Solution 31P nuclear magnetic resonance (NMR) spectroscopy is currently the main method for the characterization of phosphorus (P) forms in environment samples. However, identification and quantification of P compounds may be hampered by poor resolution of spectra caused by paramagnetic Fe and Mn. In this study, a novel technique was developed to improve spectral resolution by removing paramagnetic Fe and Mn from alkaline extracts via 8-hydroxyquinoline (8-HOQ) precipitation. Batch experiments showed that both Fe and Mn were effectively removed by the precipitation at pH 9.0, with the removal efficiencies of 83-91% for Fe and 67-78% for Mn from the extracts of five different environmental samples, while little effect was found on concentration of total P. The 31P NMR analysis of a model P solution showed that addition of 8-HOQ and its precipitation with metal ions did not alter P forms. Further analyses of the five extracts with 31P NMR spectroscopy demonstrated that the 8-HOQ precipitation was an ideal method compared with the present postextraction techniques,suchasbicarbonatedithionate(BD),EDTAandChelex100 treatments, by improving spectral resolution to a large extent with no detrimental effects on P forms.
Introduction Solution 31P nuclear magnetic resonance (NMR) spectroscopy is a technique using the magnetic resonance of 31P nucleus to identify its chemical forms in solution samples. This technique was first used by Newman and Tate (1) to investigate P species in extracts of New Zealand grass-land soils. Since then it has been applied in analyses of a wide variety of environmental samples and shown to be an effective tool for determining P composition (2). A major limitation for solution 31P NMR spectroscopy is the overlap of peaks, causing uncertainty in peak assignment and quantification. This is primarily due to the excessive paramagnetic ions that are extracted during extraction for * Corresponding author phone: 86-25-86882198; fax: 86-2586882198; e-mail:
[email protected]. † Chinese Academy of Sciences. ‡ Nanjing Forestry University. § National University of Ireland. 10.1021/es903558g
2010 American Chemical Society
Published on Web 03/04/2010
P (3). The paramagnetic ions, such as Fe(III), Mn(II), Co(II), Ni(II) and Cu(II), can substantially improve the relaxation rate of 31P nuclei. However, they can also increase line broadening and thus cause peak overlap (4). This drawback is particularly evident when Fe(III) and Mn(II), which are generally the dominant paramagnetic ions in alkaline extracts, present at excess concentrations (3, 5, 6). Poor resolution most likely caused by Fe(III) and Mn(II) has been reported in a large number of 31P NMR studies on different environmental samples using various extraction solutions such as H2O (7), NaHCO3 (8), BD (9), H2SO4 (8), NaOH (10, 11), NaOHNaF (12-14), and NaOH-EDTA (15-17). Most of the distorted spectra have been found with overlapping peaks between orthophosphate and orthophosphate monoesters (7-13, 15-18), making it hard to accurately quantify orthophosphorus monoesters such as myo-inositol hexakisphosphate (phytic acid) (19). Some spectra with poor resolution also contained other overlapping peaks such as between DNA and phospholipids (8, 11-15) and between pyrophosphate and the terminal group of polyphosphate (9). Attempts have been made to reduce the interference of paramagnetic ions. Pretreatments of samples with chemicals, such as HCl (20, 21), Na-EDTA (21), citrate-dithionitebicarbonate (CBD) (22), dithionite-bicarbonate (BD) (23), Ca-EDTA-dithionite (6), and Na-EDTA-dithionite (24), have been used to remove excessive paramagnetic Fe and Mn from the samples prior to P extraction. However, pretreatments have a risk of removing organic P compounds or hydrolyzing acid-labile organic P compounds. The level of improvement on spectral quality is also limited and varies with sample properties. For example, poor resolution was still found in some spectra after pretreatment with HCl on zonal steppe soils (12), CBD on lacustrine sediments (22), and BD on lake settling particles (23). Addition of cation exchange resins such as Chelex (Bio-Rad Laboratories) in extraction solutions (6, 25) or treatment of the extracts with cation exchange resins (26, 27) has also been used to remove Fe and Mn during or after P extraction. However, there is also a risk of removing P by binding it to the resins possibly via cation linkages (25, 27). Addition of BD (11) or EDTA (18, 28) in concentrated sample solution provides a safe alternative, which can maintain Fe and Mn ions in the reduced or complexed state during the subsequent 31P NMR analysis. However, their effects may be limited under strong alkaline conditions (e.g., pH > 13). For example, serious overlaps of peaks were observed in some spectra from a lake sediment profile following BD addition (11). In this study, a novel technique was developed by using 8-hydroxyquinoline (8-HOQ), an organic precipitant that has been commonly used in batch applications since 1926 (29), to precipitate paramagnetic Fe and Mn from solutions and enhance spectral resolution. The pH values and amount of 8-HOQ to be added were first determined to achieve an optimal removal efficiency for both Fe and Mn from NaOHEDTA extracts of five different environmental samples. A model P solution associated with 31P NMR analysis was used to test whether addition of 8-HOQ and its precipitation altered P forms. This technique associated with 31P NMR analysis was then applied to the extracts, and the results were compared with other postextraction techniques.
Experimental Section Reagents and Solutions. The following ten model P compounds were used for preparation of a model P solution. Eight model phosphate compounds were purchased from Sigma-Aldrich Chemicals (Shanghai, China), including polyVOL. 44, NO. 7, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2555
phosphate (sodium salt, polyP), myo-inositol hexakisphosphate (myo-IHP), adenosine 5′ monophosphate (AMP), D-glucose-6-phosphate (g6P), L-a phosphatidyl ethanolamine (PE), deoxyribonucleic acid (DNA), R-(-)-1′-1-binaphthyl2,2′-diyl hydrogen phosphate (RbdP), and 2-aminoethyl phosphonic acid (AEP). The other two model P compounds, pyrophosphate (tetrasodium salt, pyroP) and disodiumhydrogen phosphate (orthoP), were provided by Sinopharm Chemical Reagent Co., Ltd. The NaOH-EDTA solution used for P extraction was composed of 0.25 M NaOH and 50 mM EDTA, which has been widely applied for solution 31P NMR measurement of various environmental samples (15-17). The model P solution was prepared by combining 3 to 12 mg of each model P compound and dissolving the mixture in 100 mL of the NaOH-EDTA solution. Concentration of total P in this solution was 86 mg P L-1. BD solution was composed of 0.11 M NaHCO3 and 0.11 M NaS2O4. The 8-HOQ solution (3%, w/v) was prepared by dissolving 3 g of 8-HOQ with HCl and diluting with deionized water, and the solution was made up to 100 mL. Sampling and P Extraction. Environmental samples used in this study included two lake sediments, one soil, one pig manure, and one settling seston. The two sediments were collected from two shallow and eutrophic lakes, Lakes Taihu and Dianchi, respectively. Surface sediments (1 cm) were collected in Meiliang Bay of Lake Taihu (120.22° E, 31.54° N) in November 2006 and Fubao Bay of the Dianchi Lake (112.67° E, 40.62° N) in September 2006 using a gravity core sampler. The settling seston was collected at 1 m depth of water column in Meiliang Bay of Lake Taihu (120.24° E, 31.52° N) in April 2008 and represented a 1-month settling period. The pig manure was collected from a commercial farm located in the south of Nanjing City, China, in July 2007, which had been stored in a lagoon over 3 months. The soil was collected from a surface layer (0-20 cm) in a Yingtan field experimental site in Jiangxi Province, China, in September 2008. The soil belonged to Ultisols in the Soil Taxonomy System of the USA. All the samples were frozen at -80 °C, lyophilized, and ground to pass a 0.5 mm sieve. P extraction was performed by mixing each dried sample with the NaOH-EDTA solution in a 1: 20 w/v ratio and shaking for 16 h at 20 °C. The extract was collected after centrifugation at 10,000 rpm for 30 min. An aliquot was taken for determination of total P, Ca, Al, Fe, and Mn by ICP-AES. The remaining solution was used for the following experiments. Optimization of Precipitation Condition. The sediment extract from Lake Dianchi was used to determine the pH condition and amount of 8-HOQ to be added for an optimal removal of both Fe and Mn. Concentrations of total P and various metals in this extract are listed in Table S1. A total of 12 subsamples (each with 10 mL of the extract) were prepared, and an amount of 1.1 mL 8-HOQ solution was added to each sample (10%, v/v). Yellow precipitates were immediately observed in all the solutions. The pH value was adjusted to between 3 and 12, respectively, with the addition of NaOH (0.5-3.0 M) or HCl (0.5-3.0 M) dropwise. Each sample was allowed for standing for 0.5 h, and the solution pH value was then determined. All the solutions were centrifuged at 10,000 rpm for 10 min, followed by filtration through a 0.45 µm filter (cellulose acetate membrane, Whatman). The precipitates were rinsed three times with 1 mL of deionized water each time. The rinsed water samples were combined with the filtrate and diluted to 25 mL for determination of total P, Ca, Al, Fe, and Mn by ICP-AES. Meanwhile, for the other 7 subsamples (10 mL each) of the sediment extract from Lake Dianchi, a volume of 0.05, 0.09, 0.18, 0.36, 0.55, 0.73, and 1.09 mL of the 8-HOQ solution was added, respectively. The amount of the added 8-HOQ was accordingly 0.25, 0.5, 1.0, 2.0, 3.0, 4.0, and 6.0 times of the 2556
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 7, 2010
sum of mole concentrations of Fe and Al in the subsamples, accounting for 0.5, 0.9, 1.8, 3.6, 5.5, 7.3, and 11% of the total volume (v/v). Solution pH values for all the subsamples were adjusted to 9.0 ( 0.1. After standing for 0.5 h, the pH value of each subsample was rechecked and adjusted if it shifted (0.1 pH units. Solution was then collected through centrifugation and filtration according to the above procedure, and the total concentrations for P, Ca, Al, Fe, and Mn were determined by ICP-AES. The extracts from the sediment of Lake Taihu, pig manure, soil, and settling seston were further used to test the removal efficiency for Fe and Mn by 8-HOQ precipitation under the optimal conditions determined above. A total of 10 mL of each extract was prepared and an amount of 1.1 mL of 8-HOQ solution (10%, v/v) was added to each sample. Solution pH was adjusted to 9.0 ( 0.1. After standing for 0.5 h, each solution was collected through centrifugation and filtration according to the above procedure, and total concentrations for P, Ca, Al, Fe, and Mn were determined by ICP-AES. Test of Model P Compounds with 31P NMR Spectroscopy. The model P solution was used to examine if 8-HOQ precipitation altered P forms. The solution was divided into 5 subsamples of 10 mL. The pH value for the first subsample was adjusted to 9.0 ( 0.1, frozen at -80 °C, lyophilized, and then redissolved with 0.9 mL of 1 M NaOH for 31P NMR analysis. For the second subsample, a volume of 2.5 mL of 8-HOQ solution was added (20%, v/v), and the pH value was adjusted to 9.0 ( 0.1. The solution was then frozen at -80 °C, lyophilized, and redissolved with 0.9 mL of 1 M NaOH for 31P NMR analysis. For the other three subsamples, the following solutions were added respective: 5 mL of 8-HOQ and 10 mL of 500 mg Al L-1 (AlCl3), 3.3 mL of 8-HOQ solution and 3.3 mL of 500 mg Fe L-1 (FeCl3), and 2.8 mL of 8-HOQ and 1.4 mL of 500 mg Mn L-1 (MnCl2), resulting in 20% (v/v) of 8-HOQ with 200 mg L-1 Al(III), 100 mg L-1 Fe(III), and 50 mg L-1 Mn(II) in the solutions, respectively. The concentrations of the three metal ions were higher than most of those reported in NaOH-EDTA extracts (6, 21). Each subsample was then adjusted to pH 9.0 ( 0.1 and standing for 0.5 h. Solution samples were collected through centrifugation and filtration as described above, followed by being frozen at -80 °C, lyophilized, and redissolved with 0.9 mL of 1 M NaOH for NMR analysis. Test of Environmental Samples with 31P NMR Spectroscopy. The two sediment extracts were used for detailed investigations of the 8-HOQ precipitation effect on improving spectral resolution in comparison with the present postextraction techniques. Each extract was divided into five subsamples of 20 mL. The first subsample was frozen at -80 °C and lyophilized, followed by redissolving with 0.9 mL of 1 M NaOH for 31P NMR analysis. The second and third subsamples were frozen at -80 °C and lyophilized. The residues for the second subsample were redissolved with 0.8 mL of 1 M NaOH and 0.1 mL of BD (15). The other lyophilized residues were redissolved with a solution composed of 1 M NaOH and 0.1 M EDTA (28). An amount of 5 g Chelex-100 (sodium form, Bio-Rad Laboratories) was added to the fourth subsample and subsequently shaken overnight at 20 °C. The mixture was filtered through a 0.45 µm membrane. The filtrate was frozen at -80 °C, lyophilized, and redissolved with 0.9 mL of 1 M NaOH solution for 31P NMR analysis. For the last subsample, a volume of 2.2 mL of 8-HOQ solution (10%, v/v) was added, and the mixture was adjusted to pH 9.0 ( 0.1. After standing for 0.5 h, the solution was centrifuged at 10,000 rpm for 10 min followed by filtration through a 0.45 µm filter. The filtrate was frozen at -80 °C, lyophilized, and redissolved with 0.9 mL of 1 M NaOH solution for 31P NMR analysis. The extracts from pig manure, soil, and settling seston were only treated with and without 8-HOQ, according to the above procedures for the first and last subsamples.
31P NMR Analysis. Prior to 31P NMR analysis, all the NaOH solutions were centrifuged at 10,000 rpm for 10 min to remove any possible particles. A volume of 0.1 mL of heavy water (D2O) was added into each solution for signal lock. The 31P NMR spectra were measured at 161.98 MHz on a Bruker AV400 spectrometer equipped with a 5-mm broadband probe, using a 45° pulse, relaxation delay 4.5 s, and acquisition time 0.5 s, with around 400 transients acquired for the model P solution and 20000 transients for the environmental samples. Chemical shifts were recorded relative to 85% H3PO4 via the signal lock, and the orthophosphate peak for each sample was standardized to 6 ppm in all spectra to simplify the comparison among samples (30). There were slight differences in chemical shifts for other peaks, possibly due to the differences of salt concentrations and pH values in the final solution samples. Peak area was quantified through manual integration for obvious peaks. The P compounds were identified based on model P spectra in this study and literature reports (2, 31). All spectral processing was carried out using NMR Utility Transform Software (NUTS) for Windows (2000 edition; Acorn NMR, Livermore, CA). To assess the analytical errors associated with sample preparation and 31P NMR analysis, three duplicate analyses of the model P solution and the two sediment extracts from Lakes Dianchi and Taihu were performed. Each solution was divided into three subsamples. Each subsample was frozen, lyophilized, redissolved with 0.9 mL of 1 M NaOH, and then analyzed using 31P NMR spectroscopy as described above.
Results and Discussion Optimization of Conditions for 8-HOQ Precipitation. The removal efficiencies of various metal ions and P from the extract of Lake Dianchi were investigated by 8-HOQ precipitation under different pH conditions. Obvious increasing trends of removal efficiencies for Al, Fe, Mn, and Mg were observed with the increase of pH value from 3 to 9, and the peak values of the removal efficiencies were found at the pH value of around 9 (Figure 1). Approximately 90% of Al and Fe were removed in the pH range of 7.5 to 9.6, and 80% of Mn and 70% of Mg were removed in the pH range of 8.8 to 10. Therefore, an optimal removal efficiency for both Fe and Mn could be determined at pH 9.0. The 8-HOQ precipitation had relatively little effect on Ca, with the removal efficiency varying slightly around 4.8 ( 3% throughout the pH range investigated. The removal efficiency for total P was extremely low (1.2 ( 2%), and thus the 8-HOQ precipitation had little effect on concentration of total P. The removal efficiency for the above ions was further investigated under pH 9.0, with the addition of different doses of 8-HOQ. The removal efficiencies for Al, Fe, Mn, and Mg increased significantly with the increasing addition of 8-HOQ (Figure 1). When the addition of 8-HOQ reached three times of the sum of mole concentrations of Al and Fe originally contained in the extract (accounting for 5.5% of the total volume, v/v), the removal efficiencies for Al and Fe reached their maxima and then remained stable after that. The removal efficienciess for Mn and Mg were also high (around 70% and 60%, respectively) at this level, followed by small increases until the added 8-HOQ was four times of the sum of mole concentrations of Al and Fe in the extract (accounting for 7.3% of the total volume, v/v). Concentrations of Ca and P showed no notable changes with the addition of 8-HOQ. Precipitation reactions of 8-HOQ with metal ions can be demonstrated as follows:
Mn+ ) Al(III), Fe(III), Mn(II), Mg(II) The solubility product constants of 8-HOQ with Al(III) and Fe(III) are much lower than those with Mn(II) and Mg(II) (29), leading to preferential precipitation of Al(III) and Fe(III) by 8-HOQ. As the sum of mole concentrations of Al(III) and Fe(III) were much higher (20 times more) than those of Mn(II) and Mg(II) in this extract, the precipitation reaction was dominated by Al(III) and Fe(III), resulting in a nearly stoichiometric precipitation of 8-HOQ with them. Removal efficiencies of Fe, Mn, and P were also measured with the extracts from another sediment (Lake Taihu), pig manure, soil, and settling seston at pH 9.0. The added amount of 8-HOQ (10%, v/v) was calculated to be high enough to stoichiometrically precipitate Al(III), Fe(III), Mn(II), and Mg(II) contained in these extracts. About 83-91% of Fe and 67-78% of Mn were removed from these extracts, while total P showed no significant changes (Table S2). It could be concluded that 8-HOQ precipitation had a high and stable efficiency of removing Fe and Mn, without causing detrimental effects on concentration of total P. Effects of 8-HOQ Precipitation on Model P Compounds. Prior to the analyses of environmental samples, an experiment was conducted on the model P solution to investigate whether 8-HOQ addition and precipitation affected P forms, either by degrading specific P compounds or causing loss of some P compounds. The 31P NMR spectra were found to be similar among the different treatments, except that the peak for DNA originally appeared in the spectrum of the untreated sample was divided into two separate peaks in those of Fe and Al precipitation treatments (Figure S1). Each P compound had very small changes in proportion after addition of 8-HOQ and its subsequent precipitation with Fe, Al, or Mn (Table S3). The coefficient variations (CVs) for most of the P compounds were
