Tải bản đầy đủ - 0 (trang)
G. X-Ray Absorption Near Edge Structure Spectroscopy

G. X-Ray Absorption Near Edge Structure Spectroscopy

Tải bản đầy đủ - 0trang

CHARACTERIZATION OF P IN ORGANIC WASTES



47



XAS experiment typically measures the x‐ray absorbance of a sample over a

certain energy range to generate an absorption spectrum. This absorption

spectrum can be divided into two regions: the near edge and the extended

region.

The XANES spectroscopy is the region before the absorption edge to

around 100 eV past the edge. The XANES spectra are typically very sensitive

to the oxidation state of the central atom and to changes in coordination

number from tetrahedral to octahedral. Past the edge, the XANES region

is typically dominated by resonance and multiple scattering contributions.

The XANES spectra of unknown and environmental samples are typically

analyzed by comparison to standards, a technique known as ‘‘fingerprinting.’’ However, newly developed computer packages can now model

XANES data by comparison with theoretical standards. In contrast, the

extended portion of the spectrum is dominated by oscillations due to longer‐range interactions between photoelectrons and other atoms. The XAS

studies, which utilize these oscillations to determine chemical information

about the sample, use a technique known as extended x‐ray absorption fine

structure (EXAFS) spectroscopy. The EXAFS data are analyzed by fitting

it to theoretical scattering paths and the spectra provide information on

coordination number, the identity of nearest neighbors, and bond distances.

X‐ray absorption spectroscopy provides a molecular scale method to

probe the local chemical environment of P in heterogeneous samples such

as soils and organic wastes. EXAFS spectroscopic studies at the P K‐edge

are possible (Rose et al., 1997) but are currently of limited use in soils

because P concentrations in soil samples are often too low for EXAFS and

many of the common cations associated with phosphate in soil (Al, C, Ca,

and Si) are extremely weak backscatterers. P K‐edge EXAFS studies will

likely become more common as beamline configurations and detectors are

upgraded and as the intensity of synchrotron light sources increases.

In contrast, XANES spectroscopy provides a method to probe the speciation of P in animal manures and biosolids with minimal sample modification

and much simpler data reduction. Unlike many other routine spectroscopic

tools, XANES can be employed with heterogeneous samples (e.g., soils and

manures) and one can analyze samples in situ (with normal amounts of water

at room temperature and pressure). Fingerprinting techniques work well with

P K‐edge XANES due to the fact that there are three diVerent features of

phosphate XANES spectra that are aVected by local chemical environment:

(i) intensity and position of a pre‐edge peak, (ii) position of the white line

energy, and (iii) resonance features past the edge.

The eVects of local chemical environment on phosphate XANES spectra,

pre‐edge intensity and white line position, have been studied using a variety

of mono‐, di‐, and tri‐cationic metal phosphates. Franke and Hormes (1995)

studied the eVects that metal substitution has on P K‐edge XANES spectra,



48



G. S. TOOR ET AL.



and found that changing the identity of the metal cation bound to the

phosphate tetrahedron aVected the P XANES spectra in a systematic manner. The intensity and position of a pre‐edge peak changed as metals were

varied, and the position of the white line peak was also aVected. Past the

edge, continuum resonances in the spectra were attributed to a combination

of multiple scattering eVects in the first coordination shell and to the interactions with atoms at higher coordination shells. In an especially insightful

study, Okude and coworkers (1999) used P K‐edge XANES to study the

first row transition series of metal‐phosphates and then correlated spectral

features with the 3d electron structure of the metal phosphates (Fig. 10). It

was determined that the intensity and position of a pre‐edge peak and the

position of the main peak could both be correlated with the number of 3d

electrons in the transition metal. The position of the white line (main

absorption peak) increases linearly with increasing number of d electrons

in the metal center. The position of the white line peak, therefore, is a clue to

the bonding environment of phosphate in XANES samples. Another spectral feature that changes systematically with metal electronic configuration is

the pre‐edge peak intensity and position. This pre‐edge peak is attributed to

hybridization/orbital mixing between 3p orbitals of phosphate and 3d and 4s

orbitals of the transition metals. As the number of d electrons is increased,

the intensity of this peak decreases because fewer d orbitals are available for

mixing with P orbitals.

The eVect of mineral structure upon resonance features in P XANES

spectroscopy has not been systematically evaluated, but various researchers

have provided reference spectra of many diVerent CaPO4, AlPO4, and

FePO4 phases. One complication is that some researchers have not collected

data very far past the absorption edge and so it may be diYcult to use

their results as standards. The XANES spectra of many common phosphate

minerals and adsorbed phosphate species are collected in Fig. 11. The

important features to note in the Ca phosphate minerals are that

the XANES spectra change considerably as structure of the Ca phosphate

phase changes. For monetite (a dicalcium phosphate phase), there is a single

peak $4 eV past the white line and an otherwise relatively featureless

spectrum. For the other Ca phosphates, a shoulder is present at 2 eV past

the edge that increases in intensity as the Ca phosphate phases become

more crystalline and less soluble. This shoulder is not observed in other

phosphate minerals and can be used to identify more crystalline Ca phosphates reliably in mixed samples. There is an additional peak that appears

between 6 and 8 eV past the edge due to resonance features in Ca phosphate

minerals for all phases other than monetite, and some shifts in the shape

of the large oxygen oscillation, but these XANES features are much

less diagnostic than the shoulder in apatite. For Al phosphate minerals

(variscite, wavelite, and amorphous aluminum phosphate) there is a peak



CHARACTERIZATION OF P IN ORGANIC WASTES

49



Figure 10 EVect of 3d metal electron configuration on phosphate pre‐edge peak intensity and white line peak position. From Okude et al. (1999).



50

G. S. TOOR ET AL.



Figure 11 XANES spectra of reference CaPO4, AlPO4, and FePO4 materials.



CHARACTERIZATION OF P IN ORGANIC WASTES



51



at approximately 5 eV past the white line energy that is present in all

minerals. This peak is absent in the adsorption samples, and therefore can

be used to distinguish adsorbed from precipitated phosphate. This feature is

also diVerent enough from the Ca phosphate features to distinguish Al

phosphates from Ca phosphates in mixed systems. For Fe phosphate materials, the oxidation state of Fe has a large role on the XANES spectral

features. The Fe(III) phosphate minerals have a strong pre‐edge feature

due to orbital mixing that can be seen very clearly in strengite, P‐siderite,

and weakly in the adsorption sample. This feature is absent in the Fe(II)

phosphate mineral vivianite.

Several researchers have successfully utilized these diagnostic XANES

features of phosphate minerals to study phosphate chemistry in soils and

manures. Hesterberg et al. (1999) used XANES spectroscopy to study

adsorbed and mineral forms of phosphate. They found XANES spectroscopy could distinguish phosphate adsorbed on Fe and Al oxides from Fe and

Al phosphate precipitates. They also collected XANES spectra for a variety

of Ca phosphate minerals and concluded that phosphate in a soil sample

from North Carolina was in the form of dicalcium phosphate. Peak et al.

(2002) used P, S, and Ca K‐edge XANES spectroscopy to perform speciation of unamended and alum‐amended poultry litter samples, as discussed in

the following Case Study 1. Both of these studies were qualitative in nature

and merely compared the spectral signature of standards to diVerent samples. More advanced P XANES spectroscopic analysis techniques have

(linear combination fitting or LC XANES) been used to quantitatively assess

P speciation in soils (Beauchemin et al., 2003) and in poultry litters and

turkey manures (Toor et al., 2005c). Both of these studies assessed the

composition of natural samples by fitting their spectra to a group of reference standards. The application of LC XANES to determine the relationship

between animal diet and P speciation in manures will be discussed in Case

Study 2. This form of speciation is complicated by the fact that fluorescence

spectra of concentrated samples are distorted by self‐absorption eVects. This

is a well‐known issue with XAS experiments in the 2000–4000 eV range, and

researchers using P K‐edge XANES should be well versed in the possible

pitfalls associated with these problems.

Self‐absorption occurs in an XAS sample when concentrations of the

element of study are high enough that the ejected core electron interacts

with other atoms in the sample matrix rather than being released into the

continuum. This has the eVect of distorting the XANES spectrum around

the white line energy and decreasing the main peak’s intensity (similar to

saturation). There are several strategies to deal with self‐absorption eVects.

In hard x‐ray experiments, the easiest method of dealing with self‐absorption

is to use thin film samples and collect data in transmission mode (log I/I0)

with a detector placed directly before (I0) and after (I) the sample. This



52



G. S. TOOR ET AL.



sampling environment requires that some of the x‐ray beam must pass

completely through the sample. At the P K‐edge, the depth of penetration

is only a few microns, and transmission mode data collection is not possible.

There are, however, two experimental approaches that one can take to

eliminating self‐absorption issues in their XANES data. First of all, one

can dilute all samples to the same P concentration (in mg kgÀ1) with a

nonabsorbing powder, such as BN3, and then collect data in fluorescence

mode. Another possibility is to conduct the analysis under ultra high vacuum (UHV) conditions and collect a total electron yield (TEY) spectrum of

the surface only. The sampling depth of TEY experiments will depend on the

energy range, but researchers previously determined that at the Si K‐edge

(1840 eV) TEY probes 60 nm deep versus several hundred nm for the

fluorescence yield (FY) measurements (Kasrai et al. 1996). As a result,

TEY spectra are typically free of self‐absorption eVects. Figure 12 shows a

comparison of phytic acid reference standard analyzed as an aqueous solution in fluorescence mode, and as a powder with TEY and FY measurements. The decreased absorbance at the white line of the fluorescence

samples is caused by self‐absorption. The aqueous sample is considerably

less distorted due to a lower P concentration in this sample compared to the

salt. If the aqueous and salt samples were further diluted, their spectra would

approach the intensity of the TEY spectrum. If one were to use the phytic

acid salt sample collected in fluorescence mode as a standard for linear

combination XANES analysis, the results would be much diVerent from

using the TEY or aqueous standard. The aqueous phosphate spectrum is

included in Fig. 12 to illustrate another point. There are no clear spectral

features of phytic acid that distinguish them from aqueous inorganic phosphate. Subtle diVerences in the intensity of the white line peak and the shape

of the oxygen oscillation are present in phytic acid, but there is nothing that

makes it possible to conclusively separate phytic acid from aqueous inorganic phosphate. This makes P K‐edge XANES spectroscopy poorly suited to

conclusive identification of organic phosphates.

From the above discussion, one might infer that TEY spectra are always

the preferred choice of analysis for powder samples. There is however

another complication that should be addressed. Since TEY only probes the

top $60 nm of the sample, it is possible for the surface structure of the

samples to diVer from the bulk structure of the material. Two examples of

this are shown in Fig. 13. In the case of vivianite (a natural ferrous phosphate mineral), the surface was oxidized due to exposure to the atmosphere

and the TEY spectrum resembles that of strengite (a ferric phosphate mineral) from Fig. 11. In the case of monetite (a naturally occurring dicalcium

phosphate mineral), there is no change in oxidation state, but the TEY

spectrum has much less structure than the FY spectrum. One explanation

for this is that the mineral is relatively ionic in nature and that the surface



CHARACTERIZATION OF P IN ORGANIC WASTES



53



Figure 12 XANES spectra of phytic acid collected in fluorescence yield (FY) on an aqueous

solution and a salt as well as a total electron yield spectrum of the phytic acid salt. The decreased

intensity of the white line peak in the fluorescence samples is caused by self‐absorption eVects.



layer of phosphate, when no longer coordinatively saturated with calcium, is

relatively weakly bound. This could also explain the high solubility of

dicalcium phosphate compared to other calcium phosphate minerals.

The above discussion should highlight the fact that no sampling method is

perfect and that care must be made in both sample preparation and data

analysis. Which method of data collection is most useful depends upon the

physical state of the samples (UHV beamlines require dried samples for TEY

measurements) and the amount of beam time available (dilution to suitably

low concentrations for fluorescence studies will demand much longer data

collection times to achieve the same signal‐to‐noise ratio).

Detection limits are dependent on the type of detectors and sample

chamber chosen, but can often make performing phosphate speciation in



54



G. S. TOOR ET AL.



Figure 13 Examples of phosphate samples where total electron yield data and fluorescence

yield (FY) data diVer significantly due to surface modification. In the vivianite Fe(II)PO4

sample, surface oxidation has occurred that results in a total electron yield spectrum consistent

with Fe(III)PO4. In the monetite sample, the total electron yield spectrum appears substantially

diVerent from the bulk fluorescence in structure. This may be evidence that the surface layers of

phosphate are very weakly bound in this mineral.



natural soils with XANES diYcult. Conversely, self‐absorption can significantly distort the XANES spectrum if concentrations are too high. Another

current limitation is the availability of a suitable beam lines for environmental samples that operate in the 2–3 keV energy range. For chemical speciation in soils, UHV beamlines are not generally recommended because

analysis of aqueous references, moist references, and samples is the primary

goal. However, manure, biosolids, and soil samples are typically air‐dried

and sieved prior to chemical analysis and so the limitations of UHV beamlines may not be as large of an issue as with typical environmental XAS



CHARACTERIZATION OF P IN ORGANIC WASTES



55



samples. And as mentioned previously, TEY mode data collection has some

advantages for quantitative XANES analysis.



Case Study 1: Natural and Alum‐Amended Poultry Litter

Peak and coworkers (2002) have shown that P K‐edge XANES can

provide important information about the chemical speciation of P in animal

manures. Beamline X‐19A of the National Synchrotron Light Source at

Brookhaven National Laboratory was utilized for all XANES experiments.

The XANES spectra of aqueous and organic phosphates as well as Ca

phosphates, Al phosphates, and phosphate adsorbed on Al oxides were all

collected as references for chemical fingerprinting. The XANES spectra of

many diVerent P reference materials were previously shown in Fig. 11 (Ca,

Al, and Fe phosphates) and Fig. 13 (phytic acid). The important spectral

features to keep in mind from those standards are: (i) All Al phosphate

precipitates have a slight peak in the 2163–2165 eV range, which can be used

to distinguish phosphate adsorbed on Al oxides from Al phosphate precipitate phases, and (ii) all Ca phosphates have a spectral feature near the white

line position that appears as a peak or a shoulder between 2160 and 2162 eV.

The XANES spectra of poultry litter samples were then compared to

the reference compounds to determine the chemical forms of P. Two primary

types of litter collected from Delaware poultry houses were analyzed:

unamended litter and aluminum‐sulfate amended litter (Sims and Luka‐

McCaVerty, 2002). Spectra of a representative unamended and alum‐

amended poultry litter are shown in Fig. 14. This figure also includes spectra

of potentially occurring P reference standards for comparison. In the unamended sample, the P XANES spectrum seems to contain a mixture of

aqueous phosphate, organic phosphate, and dicalcium phosphate. However,

in the alum‐amended sample, dicalcium phosphate peaks are no longer

present. Instead, the P XANES spectrum is consistent with a mixture of

phosphate adsorbed on Al oxide, organic phosphate, and aqueous phosphate. No evidence of Al phosphate precipitation could be seen in any alum‐

amended litter. There are several reasons that Al phosphate precipitation

may not occur in alum‐amended litter. Of primary importance is the way

that litters pH changes over time. When alum is first added, the litter pH

initially drops to approximately pH 5 and then gradually increases to around

pH 7 when the flock is removed from the house. Above pH 5, Al phosphate

stability decreases greatly, so Al phosphates may initially form and then are

dissolved as pH rises over time. Alternatively, the reactivity of Al and

phosphate could be kinetically controlled rather than dictated purely by

the solubility of minerals. If Al hydroxides precipitation rates are much

faster than Al phosphate precipitation rates at 20–30  C and in the presence



56



G. S. TOOR ET AL.



Figure 14 XANES spectra of natural and alum‐amended poultry litter taken at P K‐edge.

Adapted from Peak et al. (2002).



of large quantities of organic matter, then Al phosphate formation may be

kinetically inhibited. This theory is supported by the observation (Grossl

and Inskeep, 1991, 1992; Inskeep and Silvertooth, 1988) that inhibition of

Ca phosphate precipitation by organic ligands is responsible for stabilizing

thermodynamically instable Ca phosphates in natural systems.

While diVerences in the amended and unamended litters can be seen, the

fingerprinting technique is to some extent relying on the absence of peaks to

assign speciation in these litters. To gain confidence in peak assignment, S

and Ca XANES was conducted on the samples along with P XANES. Since

peaks consistent with dicalcium phosphate were seen in the unamended

P XANES spectrum, one would also expect these peaks to appear at the

Ca edge. Figure 15 shows the spectra of an unamended and alum‐amended

poultry litters taken at the S and Ca edge, and compared to various reference

compounds. The Ca XANES spectrum of the unamended poultry litter also



CHARACTERIZATION OF P IN ORGANIC WASTES

57



Figure 15 XANES spectra of natural and alum‐amended poultry litter sample taken at the S and Ca edge.



Tài liệu bạn tìm kiếm đã sẵn sàng tải về

G. X-Ray Absorption Near Edge Structure Spectroscopy

Tải bản đầy đủ ngay(0 tr)

×