| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
School of Earth and Environmental Sciences, The Univ. of Adelaide, South Australia, 5005
* Corresponding author (warwick.dougherty{at}adelaide.edu.au)
 |
ABSTRACT |
|---|
 TOPNOTES ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Abbreviations: CP, cross polarization • CSA, chemical shift anisotropy • DP, direct polarization or Bloch Decay • EC, electrical conductivity • ICP–AES, inductively coupled plasma–atomic emission spectrometry • MAS, magic angle spinning • OC, organic carbon • Pobs, percentage of sample phosphorus observable by spin counting using 31P NMR • SSB, spinning side band
|
INTRODUCTION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Nuclear magnetic resonance spectroscopy is an ideal analytical technique for differentiating and quantifying elements in different chemical environments. It is particularly well suited for P characterization because the 31P nucleus is NMR-sensitive (more so than 13C or 15N), has 100% natural isotopic abundance, and has a nuclear spin of 1/2 that ensures relatively easy detection and spectral interpretation. Two different modes of 31P NMR spectroscopy—solution and solid-state—have been applied to P characterization in soils and related materials. Each mode has inherent advantages and disadvantages.
Studies of soil P using NMR have mainly utilized solution NMR. Solution NMR spectra provide much better resolution than solid-state NMR spectra, enabling clearer differentiation of chemically similar species. Solution 31P NMR spectroscopy has been widely used to differentiate and quantify classes of organic P molecules found in soils (see, for example, Cade-Menun and Preston, 1996; Mahieu et al., 2000; Makarov et al., 2004; Turner and Richardson, 2004). Impressive resolution is possible using this technique, for example, a range of chemically similar inositol phosphates can be differentiated (Turner and Richardson, 2004). However, in most studies, differentiation of only broad classes of organic P compounds, such as orthophosphate monoesters and diesters, phospholipids, and phosphonates, has been achieved (Cade-Menun and Preston, 1996; Mahieu et al., 2000; Makarov et al., 2004). The obvious drawback to solution NMR is that only soluble species can be detected. These soluble species represent a highly variable proportion of total soil P (e.g., 16.1– 99.6%) (Cade-Menun et al., 2000) and are typically a mixture of organic and inorganic species. Furthermore, the high pH required to solubilize most of the organic matter may cause hydrolysis of sensitive phosphate esters (Makarov et al., 2002; Turner et al., 2003).
The obvious attraction of solid-state NMR spectroscopy for the analysis of soil is that soil is (for the most part) a solid, and a solid-state technique avoids the problems inherent with insoluble species and aggressive extractants. Solid-state 31P NMR spectroscopy has found application mainly in the characterization of inorganic P in soil and other materials (Hinedi et al., 1989; Frossard et al., 1994; McDowell et al., 2002, 2003), although some researchers have used it to characterize organic P (Newman and Condron, 1995; Condron et al., 1997). The main disadvantage of solid-state 31P analysis is that the spectra lack the resolution of solution spectra (Shand et al., 1999). Furthermore, paramagnetic species, especially Fe, can limit the quantitative potential of solid-state NMR analysis of soils. This limitation has been studied in depth for solid-state 13C NMR analysis of soils (see, for example, Kinchesh et al., 1995; Dai and Johnson, 1999; Smernik and Oades, 2000a, 2000c). Although the potential for paramagnetic species to compromise quantitation in solid-state 31P NMR analyses has been recognized (Hinedi et al., 1989; Frossard et al., 1994; Condron et al., 1997; McDowell et al., 2002, 2003; Hunger et al., 2004), the issue of quantitation has not been considered in depth in previous 31P NMR studies.
In this paper, we describe solid-state 31P NMR analyses of eight whole soils, sampled from an area used for grazing of dairy cows, that had a wide range of P contents but otherwise similar chemistry. The aims were to (i) investigate the observability of P using spin counting and (ii) determine the forms of P identifiable in the CP and DP 31P NMR spectra using a combination of NMR spectroscopy and wet chemistry. Although spin counting has been widely used in solid-state 13C (Smernik and Oades, 2000a, 2000b) and 15N (Smernik and Baldock, 2005) analyses of soils, it has not previously been applied in 31P NMR analyses of soil.
|
MATERIALS AND METHODS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Soil samples (0–1 cm) were collected from the upper, mid-, and lower slope positions of the landscape (designated U, M, and L, respectively). At each of these positions in the landscape, samples of low, medium, and high P fertility were selected (designated L, M, and H, respectively), except for the upper part of the landscape where no low-fertility surface horizon was available. Samples are identified by their slope position followed by their fertility level, for example, U-H refers to the soil from the upper part of the landscape with a high P status. Sampling was undertaken at the end of summer when there was minimal pasture growth and no pasture mat, precluding the possibility that the soil samples were contaminated with large amounts of undecomposed plant material. Samples were dried and passed through a 2-mm sieve. A subsample of each of the soils was ground to pass a 200-µm sieve for use in all analyses except for pH, electrical conductivity (EC), and Olsen P, for which the <2-mm fraction was used.
Soil Analysis
Soils were digested in a HNO3–HClO4 mixture (6:1) at 160°C (Olsen and Sommers, 1982) and concentrations of elements, including P, measured by inductively coupled plasma–atomic emission spectrometry (ICP–AES). The organic P content of the soils was determined using two different methods, the ignition method of Saunders and Williams (1955) and a sequential extraction using concentrated H2SO4 and NaOH (Bowman, 1989). The ignition method involved extracting both a sample of soil ashed at 500°C and an un-ashed sample with 0.5 M H2SO4, the difference being defined as organic P. As is the case for most chemical extraction methods, the ignition method of Saunders and Williams (1955) is subject to a number of possible sources of error. Consequently, for the purposes of comparison, organic P was also measured by the method of Bowman (1989) by extracting the soil with concentrated H2SO4 followed by 0.5 M NaOH.
Olsen P was determined by shaking soil with 0.5 M NaHCO3 (pH 8.5, 20:1 solution/soil) for 30 min (Olsen et al., 1954). Organic carbon (OC) was determined by wet oxidation with chromic and sulfuric acids followed by colorimetric titration with FeSO4 (Rayment and Higginson, 1992). Electrical conductivity and pH were measured after shaking 5 g of soil with 25 mL of deionized water on an end-over-end shaker for 1 h. Oxalate extractable Al, Fe, and P were separated by shaking 1 g of soil in 100 mL acid ammonium oxalate for 4 h in the dark (Ross and Wang 1993) and measuring the concentrations in the filtered supernatant by ICP–AES. The degree of P saturation was calculated as Pox/(Feox + Alox) (Hooda et al., 2000). Organically complexed Fe was determined by extraction with Na4P2O7 (Ross and Wang, 1993). This involved shaking 0.3 g of soil with 30 mL of 0.1 M Na4P2O7 for 16 h and centrifuging (15000 g) and analysis of the supernatant by ICP–AES.
The M-H soil was subjected to more detailed analysis to assist in our interpretation of spectrum features. A subsample of soil M-H was extracted with NaOH and EDTA using the method of Cade-Menun and Preston (1996). Five g of soil was shaken with 100 mL of 0.25 M NaOH and 0.05 M Na2EDTA for 16 h, centrifuged at 1500 g, and the supernatant filtered, frozen (–20°C), and then freeze dried. The soil residue was dried at 40°C, ground, and passed through a 2-mm sieve. In a separate treatment, 2% HF was used to isolate organic matter by removing inorganic soil components (Skjemstad et al., 1994). Briefly, 2.5 g of soil was shaken with 50 mL of HF for periods of 1 h (five times), 16 h (three times), and 64 h (once). Between treatments, samples were centrifuged and the supernatant discarded and replaced with fresh HF. Following the final treatment, the residue was rinsed three times with deionized water and then freeze dried.
Nuclear Magnetic Resonance Spectroscopy
Solid-state 31P NMR spectra were acquired with magic angle spinning (MAS) and high-power 1H decoupling on a Varian Unity INOVA 400 spectrometer (Varian, Palo Alto, CA) with a Doty Scientific supersonic MAS probe (Doty Scientific, Columbia, SC) at a 31P frequency of 161.9 MHz. Samples were packed into 7-mm-diam. cylindrical zirconia rotors with Kel-F end-caps (Doty Scientific, Columbia, SC) and spun at 5 kHz at the magic angle. Free induction decays were acquired with a sweep width of 50 kHz. A total of 1216 data points was collected for all spectra, representing an acquisition time of 12 ms. All spectra were zero-filled to 131072 data points and processed with a 100-Hz Lorentzian line broadening and a 0.010-s Gaussian broadening. Chemical shifts were externally referenced to NH4H2PO4 at 0.72 ppm (Frossard et al., 2002).
Phosphorus-31 CP NMR spectra of the whole soils and soil fractions represent the accumulation of 20000 scans, except for the HF-treated residue of soil M-H (10000 scans), and were acquired using a 1-ms contact time and a 1-s recycle delay. Preliminary variable contact time experiments (data not shown) on soil L-H showed that signal intensity was maximized at a contact time of 1 ms. Preliminary inversion–recovery experiments (data not shown) on soil L-H showed that this recycle delay was sufficient to avoid saturation. The total acquisition time for 31P CP NMR spectra was around 6 h (3 h for the HF-treated residue of soil M-H).
Phosphorus-31 DP NMR spectra of the whole soils and soil fractions represent the accumulation of 2700 to 5000 scans and were acquired using a 20-s recycle delay. The total acquisition time for 31P DP NMR spectra was around 15 to 28 h. The 31P DP NMR spectra were corrected for a broad background signal by subtracting the 31P DP NMR spectrum acquired for an empty rotor (there was no such background signal for 31P CP NMR spectra). The effect of the recycle delay on signal intensity for 31P DP NMR spectra of soil L-H and fractions of soil M-H is shown in Fig. 1 . In all cases, the signal intensity at a recycle delay of 20 s was 87 to 97% of the signal intensity at a recycle delay of 100 s. Thus, while a recycle delay of 20 s is insufficient to completely avoid signal loss through saturation for these whole soils and soil fractions, saturation is unlikely to cause signal loss of more than 15%. Longer recycle delays were considered impractical as they would require either very long run times, or the acquisition of fewer scans that would adversely affect signal-to-noise ratios. Previously published 31P DP NMR spectra of soils and related materials have employed recycle delays of 0.1 to 20 s (Hinedi et al., 1989), 0.5 s (McDowell et al., 2002; McDowell et al., 2003), 20 s (Frossard et al., 1994; Frossard et al., 2002), and 60 s (Hunger et al., 2004).
|
H (Smernik et al., 2002). The T1H relaxation rate for NH4H2PO4 was found to be 120 ms. The T1H relaxation rate for soil L-H was found to be 3.33 ms; this value was used in the T1H correction for all of the other soils and soil fractions. Uncertainty in the precision of Pobs values is estimated to be ±10% CP and ±15% in DP (Smernik and Oades, 2000a). |
RESULTS AND DISCUSSION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
|
|
In contrast, the 31P DP NMR spectrum of whole soil M-H (Fig. 2) contains both sharp and broad resonances. The broad resonances appear to correspond with those detected by CP, both in terms of the broad lineshape and the extent and shape of the SSBs. The central peak of the sharp resonance is centered at 2.7 ppm, 4.3 ppm downfield of the broad resonance detected by CP. This resonance has a width at half-height of approximately 1.8 ppm (i.e., it is nearly five times narrower than the resonances detected by CP). Spinning side-bands for the sharp peak of the central resonance are less prominent, indicating that the P species that give rise to it have much lower CSA.
It would thus appear that two different types of P are detected in the 31P DP NMR spectrum, and that only one of these two types of P is detected by CP. To assign these two distinct P types we applied solid-state 31P NMR analysis to soil fractions generated during the wet chemical determination of organic and inorganic P and the residue after HF treatment of soil M-H.
The procedure of Saunders and Williams (1955) used to determine the organic P content involved first ashing the soil at 500°C to convert all P to inorganic forms. Because ashing removes all organic carbon, the residual P must be inorganic. The 31P DP NMR spectrum of soil M-H after ashing (but before H2SO4 extraction) contains only the sharp resonances with small SSBs (Fig. 2). No signal could be detected in the 31P CP NMR spectrum of soil M-H after ashing. Therefore, we assigned the sharp resonance present in the 31P DP spectra of both the ashed residue and the whole soil to inorganic P. Furthermore, we propose that for our soils, inorganic P is only detected by 31P DP NMR.
Shaking the soil with 0.5 M H2SO4 for 16 h (Saunders and Williams, 1955) nominally removes inorganic P and leaves organic P intact. The 31P NMR spectra (both CP and DP) of the residue of soil M-H after H2SO4 extraction contain only the broad resonances with prominent SSBs (Fig. 2). We propose that these broad resonances are due to organic P. This is consistent with the findings of Shand et al. (1999), who found that organic P gave rise to broad 31P NMR resonances in a large range of soils. Furthermore, inositol phosphates, the dominant forms of organic P in soil, give a complex series of signals (Turner et al., 2003) that may contribute to the broad 31P NMR resonances in our spectra. However, the effect of extraction with 0.5 M H2SO4 on organic and inorganic P is not entirely straightforward. First, the extraction may hydrolyze some organic P. This may result in the extracted soil containing less organic P than the whole soil. Second, some native P minerals may be resistant to this treatment, such that inorganic P may still be present in the residue resulting from the H2SO4 extraction. However, the similarity of the organic P contents estimated by the H2SO4 extraction, the sequential H2SO4–NaOH extraction (see Table 2), and NaOH–EDTA extraction (data not shown) suggest that these potential deficiencies of the H2SO4 extraction are not major sources of error in the estimation of organic P content. Furthermore, some imprecision in the isolation of organic and inorganic P in the method of Saunders and Williams does not invalidate our interpretation of the 31P CP and DP NMR spectra and the subsequent assignment of organic and inorganic P.
Further evidence for the assignment of the broad resonances in the whole soil to organic P is provided by examination of the HF-treated residue of soil M-H. De-ashing with HF is commonly used to concentrate organic matter and remove paramagnetic impurities before 13C NMR analysis of organic matter (Skjemstad et al., 1994). Soil M-H was treated with HF (see Materials and Methods) and subsequent wet chemical analysis of the HF-treated material confirmed that the majority of the P in the residue was organic P (98%) as defined by the method of Saunders and Williams (1955). The 31P CP and DP NMR spectra of the HF-treated residue of soil M-H are very similar to those of the H2SO4 residue, apart from the improvement in signal-to-noise ratio. As the HF treatment dissolved almost all of the inorganic soil components (96% of Fe and 99% of Al was removed), the P remaining in the HF residue must be organic in nature. Thus, the broad resonances in the CP and DP spectra of the HF residue must be from organic P.
Therefore, based on the evidence provided by spectra from the ashed sample, the H2SO4 extracted residue, and the HF-treated residue, we conclude that the broad resonances observed in the CP and DP spectra of the whole soils are indeed due to organic P and that inorganic P is only detected by DP. This differentiation of organic and inorganic P found here has rarely been reported in solid-state 31P NMR analyses of whole soils and related materials. Indeed, many solid-state 31P NMR studies do not assign any NMR signal to organic P (Hinedi et al., 1989; Frossard et al., 1994; McDowell et al., 2002), or mention organic P only briefly (Frossard et al., 2002; McDowell et al., 2003; Hunger et al., 2004). Newman and Condron (1995) could not differentiate organic and inorganic P in 31P CP NMR spectra of dairy pond sludge, although circumstantial evidence pointed toward the majority of the resonance coming from organic P. Condron et al. (1997) reported similar results for soils and soil humic acids. Benitez-Nelson et al. (2004) used spectral deconvolution to apportion the solid-state 31P NMR spectra of marine particulates and sediments to broad P groups including organic and inorganic P. They made broad assignments of orthophosphate, P esters, and phosphonates based on chemical shift derived by spiking of their samples with known compounds.
Precise identification of the mineral phase detected in our 31P DP NMR spectrum is difficult given the dependence of 31P chemical shift on variables such as pH and the purity of the mineral (Hunger et al., 2004). More detailed identification of the organic P is restricted by the broadness of the organic P resonance. This is in contrast to the detailed characterization of organic P possible in solution 31P NMR spectra of soil extracts, where line-widths are at least two orders of magnitude smaller (Cade-Menun and Preston, 1996; Mahieu et al., 2000; Makarov et al., 2004; Turner and Richardson, 2004).
It should be noted that the finding that inorganic P gave rise to sharp resonances with relatively small SSBs in the 31P DP NMR spectrum may be particular to these soils, and may not be true in general. On the other hand, our results appear to confirm the contention of Shand et al. (1999) that organic P always gives rise to broad resonances with prominent SSBs. The potential to differentiate inorganic and organic P by solid-state 31P NMR as we have proposed adds to the use of NMR for the study of soil P. However, in the case of the soils we examined, little more information was garnered from complex solid-state 31P NMR experiments than could be gathered from simple wet chemistry techniques.
Spin Counting
Given the differentiation of organic and inorganic P in the 31P DP NMR spectra of the whole soil we have proposed, it would be tempting to use spectral deconvolution to quantify the proportion of each. However, this would only be a measure of the relative amounts of NMR signal produced by each of the components. Either or both of these P types may be under-represented in the NMR spectra, perhaps to differing degrees, and there may be other types of P that are essentially undetected by either technique. Therefore, an essential prerequisite for the use of spectral deconvolution is confirmation that the NMR spectrum is quantitative (i.e., the majority of the P in the sample being analyzed must be detected by 31P NMR). This requires a way to gauge how much of the total P in the sample is actually detected in each 31P NMR spectrum. Spin counting provides such a gauge.
The degree to which solid-state NMR spectroscopy can provide quantitative information on soil chemistry has been widely debated, especially with regard to soil C chemistry (Fründ and Lüdemann, 1989; Kinchesh et al., 1995; Preston, 1996; Mao et al., 2000; Smernik and Oades, 2000a, 2000b; Keeler and Maciel, 2003). Two potential causes of quantitation problems in solid-state 31P NMR spectra of soils are widely recognized—close association with paramagnetic ions, especially Fe (Hinedi et al., 1989; Frossard et al., 1994; Condron et al., 1997; McDowell et al., 2002, 2003; Hunger et al., 2004), and remoteness of 31P nuclei from nearest 1H neighbors, which affects only CP spectra (Frossard et al., 1994; Condron et al., 1997; Hunger et al., 2004).
That quantitation can be achieved at all from NMR spectra is a consequence of a fundamental NMR property—that, under suitable conditions, each nuclear spin of a given isotope (e.g., 31P, 13C, or 15N) produces the same quantum of NMR signal, regardless of its chemical environment. This is not true for other spectroscopic techniques, such as infrared or ultraviolet spectroscopy, for which different chromophores (the chemical structures that give rise to the signal) have different sensitivities. However, the potential for NMR to provide quantitative characterization of soil chemistry is often limited because the properties of the soils themselves are not consistent with the "suitable conditions" required for producing quantitative NMR spectra. The two main problems are interference by paramagnetic species present in the soil and low sensitivity, which can make it impossible to obtain a quantitative NMR spectrum on a practical timescale even when it is theoretically possible to do so given unlimited spectrometer time.
Spin counting is a convenient way to gauge the potential "quantitativeness" of an NMR spectrum. The causes of NMR quantitation problems have a common feature—they all decrease the amount of NMR signal produced by the affected nuclei. Spin counting simply calibrates the amount of NMR signal produced by a sample against that of a standard, with the result, the "NMR observability" or Pobs, expressed as the amount of NMR signal per unit mass of the nucleus in the sample divided by the amount of NMR signal per unit mass of the nucleus in a reference material.
The results of spin counting on the 31P CP and DP spectra of the whole and treated soil fractions are shown in Table 3. The Pobs(CP) values for the whole soil were in the range 7 to 12%, indicating that only around 10% of potential NMR signal for these soils was detected using this technique. The Pobs(DP) values were higher by a factor of 2.0 to 2.7, ranging between 16 and 28%. These values may be slightly affected by signal saturation, which would result in an underestimation of Pobs(DP) by at most 15% (see Materials and Methods). In any case, it is clear that the majority of P in all of the soils was "undetected" even in the DP spectra.
|
Another advantage of spin counting is that it facilitates a more meaningful comparison of corresponding CP and DP spectra. The 31P CP and DP spectra of soil M-H shown in Fig. 2 are scaled so that the highest peak of the central resonance in each spectrum is the same height. The vertical scales in Fig. 2 are thus unrelated, and although the broad resonances present in each of the spectra have very similar chemical shifts, line-shapes, and patterns of SSBs, their relative intensities (i.e., heights), cannot be compared meaningfully across the two techniques in this representation. This is the way corresponding CP and DP spectra are usually presented (see, for example, Frossard et al., 1994, 2002; Condron et al., 1997; McDowell et al., 2002). In Fig. 3 to 5
, corresponding CP and DP spectra are presented in a way that allows direct comparison between the two techniques (although not among soils as each of these have been scaled differently for clarity). This was achieved by representing the vertical scale of the CP spectra to take account of the greater sensitivity of the CP technique. The ratio of signal detected by the two techniques (CP/DP) for the spin counting standard was 1.82:1. Allowing for this and the larger signal losses in the soil CP spectra caused by its more rapid T1H relaxation rate (see Materials and Methods section), the soil 31P CP spectra were scaled down by a factor of 1.37 relative to the 31P DP spectra to facilitate direct comparison of the spectra on the vertical scale.
|
|
|
The Effect of Paramagnetic Iron on Nuclear Magnetic Resonance Observability
The presence of paramagnetic species, especially Fe, is the most likely cause of the low NMR observability of 31P nuclei in these soils. Paramagnetic species can affect NMR observability it two ways: (i) they can cause nonselective signal loss by affecting magnetic field homogeneity, and (ii) they can cause selective loss of signal for nuclei in close contact with the paramagnetic centers. Smernik and Oades (2000a) provided an example of the first case, in which the NMR observabilities of physical mixtures of cellulose and paramagnetic salts were determined. Addition of 3.3 and 8.1% Mn decreased both CP and DP observability of 13C nuclei by around 30 and 60%, respectively. The decrease in NMR observability affected all resonances equally. The most abundant paramagnetic element present in the soil studied here is Fe (0.86–1.64%, Table 1). The concentration of Mn in these soils is around two orders of magnitude lower (data not shown), and at these concentrations Mn is unlikely to have a substantial effect on NMR Pobs via this mechanism.
To test whether the paramagnetic species present in the soil affect the NMR observability of remote 31P nuclei, a physical mixture of soil L-H (357 mg) and NH4H2PO4 (15 mg) was prepared. The Pobs(CP) and Pobs(DP) values for the added phosphate salt were 96% and 84%, respectively, showing that the bulk magnetic properties of the soil did not substantially affect the observability of 31P nuclei in the added phosphate salt.
Much lower concentrations of paramagnetic species can affect NMR observability when there is contact between the paramagnetic species and the observed nuclei at the molecular level. For example, the NMR observability of 13C nuclei in pectin was decreased by 80% when complexed paramagnetic Cu2+ was present at a concentration of 11.9% (Smernik and Oades, 2000c), whereas the observability of 13C nuclei in cellulose was unaffected when 9.8% Cu2+ was added in a physical mixture (Smernik and Oades, 2000a). Hinedi et al. (1989) reported that a range of Fe and Mn phosphate minerals produced no signal at all in 31P DP NMR spectra. McDowell et al. (2002)( 2003) also assumed that Fe phosphates were not visible in 31P DP NMR spectra and thus used wet chemical techniques to calculate Fe phosphate contents to supplement the Ca and Al phosphate contents determined by NMR.
To test whether closely associated Fe is responsible for the low NMR observability of P in the whole soils, extraction with NaOH and EDTA (traditionally used for solution 31P NMR analysis) was undertaken. Treatment with NaOH solubilizes organic P while the EDTA, a strong chelating ligand, increases the efficiency of P extraction from soils by chelating the cations that may bind P, including paramagnetic Fe and Mn (Cade-Menun and Preston, 1996; Turner and Richardson, 2004). Treatment of soil M-H with NaOH and EDTA resulted in a residue containing 255 mg kg–1 inorganic P, 249 mg kg–1 organic P, and 14012 mg kg–1 Fe. The freeze-dried NaOH–EDTA extract contained 2110 mg kg–1 inorganic P, 555 mg kg–1 organic P, and 792 mg kg–1 Fe.
The solid-state 31P CP and DP NMR spectra of the residue and extract fractions from NaOH–EDTA extraction of soil M-H are shown in Fig. 6 . The 31P CP NMR spectrum of the NaOH–EDTA residue is very similar to the corresponding CP spectrum of the whole soil and the H2SO4 residue (Fig. 2). This similarity is consistent with the results of the wet chemical analysis that shows that there is still a substantial proportion of organic P in the residue. The 31P DP NMR spectrum of the residue is also similar to the corresponding DP spectrum of the whole soil (Fig. 2), except that the broad organic P resonances are much less intense. This reduced intensity can be attributed to the substantially reduced organic P concentration in the residue relative to the whole soil.
|
The Pobs(CP) value for the NaOH–EDTA residue was 14% (Table 3), slightly higher than the Pobs(CP) values for the corresponding whole soil (9%) and the H2SO4 residue (10%). The Pobs(DP) value for the NaOH–EDTA residue was 43% (Table 3), approximately double the Pobs(DP) values for the corresponding whole soil (21%) and the H2SO4 residue (20%). This indicates that either the P not extracted by NaOH and EDTA had a higher NMR observability than the fraction that was extracted, or that NaOH–EDTA extraction reduced the effect of paramagnetic species in the residue fraction (by complexation of and/or the loss of paramagnetics) that were compromising NMR observability.
The Pobs (CP) value for the freeze-dried NaOH–EDTA extract was 14% (Table 3), slightly higher that the Pobs (CP) value for the corresponding whole soil (9%). The Pobs(DP) value for the NaOH–EDTA extract was 78% (Table 3), approximately four times the Pobs(DP) values for the corresponding whole soil (21%). The high Pobs (DP) value is likely to be the result of the much lower Fe content of the freeze-dried extract (792 mg kg–1) relative to that of the whole soil (14761 mg kg–1) and complexation of some of the paramagnetics by EDTA in the freeze-dried extract.
To determine whether the high pH of the NaOH–EDTA extract affected its chemical shift and SSBs, a subsample of the freeze-dried extract was neutralized by first dissolving it in deionized water and then adding HCl until the pH was reduced to 5.9. The neutralized sample was subsequently freeze dried. The neutralized, freeze-dried NaOH–EDTA extract contained 1944 mg kg–1 inorganic P and 386 mg kg–1 organic P. The neutralization of the extract resulted in a small change in the proportions of inorganic P (83%) and organic P (17%) relative to the untreated NaOH–EDTA extract, and a small decrease in the total P concentration, attributable to the addition of chloride ions. There was no removal or addition of material other than the HCl added in the pH adjustment.
Both the 31P CP and DP NMR spectra of the neutralized, freeze-dried NaOH–EDTA extract are very different in appearance to the corresponding spectra of the untreated freeze-dried NaOH–EDTA extract. The chemical shift of the central resonance was slightly downfield of that of the unamended freeze-dried NaOH–EDTA extract, this shift most likely being the result of the change in pH (Hunger et al., 2004). In both spectra of the neutralized extract, the SSBs are much more prominent. The 31P CP NMR spectrum for the neutralized freeze-dried NaOH–EDTA extract is similar to that of the whole soil. However, the 31P DP NMR spectrum for the neutralized freeze-dried NaOH–EDTA extract contains much more prominent SSBs than does the DP spectrum of the whole soil. Both the whole soil and the neutralized freeze-dried extracts have similar total P contents (2098 and 2330 mg kg–1, respectively) and in both cases are dominated by inorganic P. The most likely explanation for the differences in the appearance of the 31P DP spectra of the whole soil and neutralized NaOH–EDTA extract is that they contain different inorganic P minerals; the P minerals present in the latter having resulted from dissolution of the original P minerals, followed by rapid precipitation on neutralization or during freeze drying.
The Pobs(CP) of the neutralized extract was 24%, which is considerably higher than that for the untreated extract (see Table 3). At least part of the increase in Pobs(CP) can be attributed to increased protonation of the phosphate groups in the soils with the decreased pH. Cross polarization relies of transfer of magnetization from 1H to 31P nuclei. The increase in the degree of protonation of the weakly acidic phosphate ions and esters brought about by a decrease in pH increases the concentration of 1H nuclei in close proximity to 31P nuclei and hence increases Pobs(CP) (Hunger et al., 2004). The Pobs(DP) was 72%, which is similar to that of the untreated NaOH–EDTA extract.
Both the soil residue and freeze-dried extract resulting from treatment with NaOH and EDTA have substantially higher Pobs (both CP and DP) than the corresponding whole soil. This can be attributed to the strong affinity of EDTA for Fe and other cations and the subsequent displacement of P. Consequently, less of the P is in close proximity to Fe and there is a substantial increase in the observability of P (particularly inorganic P).
The Pobs(CP) and Pobs(DP) values for the HF-treated residue of 73 and 77%, respectively (Table 3), were the highest of any of the treated soil fractions, providing further evidence that closely associated paramagnetic Fe is the main cause of poor NMR quantitation for the organic P in these soils. As previously noted, 
17% of the soil Fe is associated with organic matter. However, it should be noted that only 12% of P present in the whole soil was recovered in the HF-treated residue. This accounts for only 41% of the organic P in the whole soil. By contrast, the recovery of C on HF treatment for this soil was 91%. The most likely cause of the low recovery of organic P is that majority of organic P may have been in small, soluble molecules held in the soil by association with clay minerals and Fe oxides and was released into solution during HF treatment. In other words, it may be that the organic matter lost during HF treatment, which represented only 9% of soil carbon, may have been disproportionately rich in P.
|
CONCLUSIONS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
We recommend that spin counting be routinely used in solid-state 31P NMR analyses of soils, and that Pobs values be reported. Furthermore, in the further development of solid state 31P NMR, methods should be utilized that allow the Pobs of the various forms of P identified to be assessed. A combination of both CP and DP 31P NMR methods and a range of wet chemical methods aided in the assignment of peaks in the spectra to the broad classes of soil P—organic and inorganic. We recommend that these be considered in the assignment of peaks in future solid state studies in combination with the examination of model compounds. We believe that significant improvements in the information garnered from solid-state 31P NMR analysis of soil will come not from improving resolution—there are fundamental limitations here—but in using information contained in nonfrequency parameters, such as observability, chemical shift anisotropy, and relaxation rates.
|
NOTES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Received for publication January 17, 2005.
|
REFERENCES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
H relaxation rates in charred and uncharred wood and consequences for NMR quantitation. Solid State Nucl. Magn. Reson. 22:50–70.[Medline]This article has been cited by other articles:
|
|
 |
|
  W. J. Dougherty, R. J. Smernik, E. K. Bunemann, and D. J. Chittleborough On the Use of Hydrofluoric Acid Pretreatment of Soils for Phosphorus-31 Nuclear Magnetic Resonance Analyses Soil Sci. Soc. Am. J., June 8, 2007; 71(4): 1111 - 1118. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. M. McBeath, R. J. Smernik, E. Lombi, and M. J. McLaughlin Hydrolysis of Pyrophosphate in a Highly Calcareous Soil: A Solid-State Phosphorus-31 NMR Study Soil Sci. Soc. Am. J., March 29, 2006; 70(3): 856 - 862. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||
