The effect of surface coatings on the association of orthophosphate with natural colloids

The Science of the Total Environment 263 Ž2000. 23᎐35

The effect of surface coatings on the association of orthophosphate with natural colloids Bailin Chen, Janine Hulston1, Ronald Beckett U CRC for Freshwater Ecology, Water Studies Centre, Department of Chemistry, Monash Uni¨ ersity, Clayton, Victoria 3800, Australia Received 10 February 2000; accepted 29 May 2000

Abstract A new method has been utilised for the characterisation of natural particle surface coatings. The method involves the use of sedimentation field-flow fractionation ŽSdFFF., radiolabelling and inductively coupled plasma-high resolution mass spectrometry ŽICP-HR MS. techniques to study the effect of colloidal surface coatings on the adsorptive behaviour of orthophosphate. Colloidal river sediment and soil samples were chemically treated in an attempt to selectively remove metal hydroxyoxides and natural organic matter. The samples were then radiolabelled with 33 PO3y and analysed by SdFFF to determine the surface adsorption density ŽSAD. of orthophosphate as a 4 function of particle size. The SdFFF unit was directly coupled to an ICP-HR MS to determine the chemical composition of the colloidal samples as a function of particle size. Element concentrationrUV detector signal and element atomic molar ratios were plotted against particle size, and the trends used in the interpretation of SAD distribution ŽSADD. changes for the samples were studied. In general, non-constant trends in the orthophosphate SADDs were found, except for the river sediment treated with hydroxylamine hydrochloride. The results indicated that, in the soil sample studied, the Mn oxide coating was a dominant factor in determining phosphorus adsorption. This method could also be applicable to other industrial or similar samples. 䊚 2000 Elsevier Science B.V. All rights reserved. Keywords: Colloids; ICP-HR MS; Orthophosphate adsorption; Sedimentation field-flow fractionation; Surface coatings

U

Corresponding author. Tel.: q61-3-9905-4555; fax: q61-3-9905-4196. E-mail address: [email protected] ŽR. Beckett.. 1 Current address: School of Chemistry, The University of Melbourne, Parkville, Victoria, Australia. 0048-9697r00r$ - see front matter 䊚 2000 Elsevier Science B.V. All rights reserved. PII: S 0 0 4 8 - 9 6 9 7 Ž 0 0 . 0 0 6 0 7 - 0

24

B. Chen et al. r The Science of the Total En¨ ironment 263 (2000) 23᎐35

1. Introduction Natural colloids often play a dominant role in the transport, fate and environmental effects of pollutants ŽStumm, 1982; Allan, 1986; Rees, 1991.. If pollutants are associated with these natural colloids and are concentrated in locations such as reservoirs, estuaries, lakes or soils, they can become an ecological and human health hazard ŽBeckett et al., 1990.. One example is the occurrence of algal blooms, which can form in areas of high nutrient concentrations. It is thought that one of the factors responsible for algal blooms is the release of orthophosphate from suspended particles and bottom sediments under favourable conditions ŽManning, 1987; Bostrom ¨ et al., 1988; Gachter and Wehrli, 1998. ¨ A useful conceptual model for a typical soil or sediment particle consists of a mineral core, which is usually made up of silica, silicate or aluminosilicates, coated with hydrous metal oxides Že.g. iron, manganese, aluminium. and natural organic matter. The association of pollutants with natural colloidal particles will be dependent upon the surface area as well as the nature of the particle surface. Hydrous oxide and natural organic matter coatings would be expected to play an important role in controlling the surface characteristics of colloids ŽBeckett and Le, 1990; Stumm and Sulzberger, 1992; Hart et al., 1993.. Information on the nature and influence of particle surface coatings on pollutant adsorption can be obtained using a combination of sedimentation field-flow fractionation ŽSdFFF. and radiolabelling techniques ŽBeckett et al., 1990; van Berkel and Beckett, 1996, 1997.. The suspension of colloidal particles is treated with 33 P-labelled orthophosphate, then separated by SdFFF. Fractions are collected throughout the SdFFF run and the ␤ radiation activity is determined using a scintillation counter. The resulting adsorption data could then be used to calculate a surface adsorption density distribution ŽSADD., which is the amount of pollutant adsorbed per unit area of particle surface, plotted as a function of particle size. For a homogeneous sample, a constant SADD plot would be expected. A non-constant

SADD plot may be due to changes in particle shape, mineralogy or surface coatings across the particle size range. In previous work, van Berkel and Beckett Ž1997. found that changes in surface adsorption density distribution were influenced by changes in particle mineralogy or surface coatings across the size distribution. If particle shape changes occur in different size ranges, this may need to be compensated for. Changes in the chemical composition of the sample with particle size can be determined by directly combining SdFFF with inductively coupled plasma-high resolution mass spectrometry ŽICP-HR MS.. The distribution of element concentrations in the particles and element molar ratio distributions can be used to help interpret the reasons for obtaining non-constant orthophosphate SADD plots with some colloid samples. The purpose of this work was to investigate the effect of iron Žor other metal. hydroxyoxide and natural organic matter surface coatings on the orthophosphate surface adsorption density. Information such as this should lead to a better understanding of the behaviour of pollutants in the environment.

2. Theory

2.1. Sedimentation field-flow fractionation

Sedimentation field-flow fractionation ŽSdFFF. is a high resolution liquid chromatography-like elution method. The mechanism and theory of SdFFF have been detailed elsewhere ŽGiddings et al., 1980, 1983; Giddings, 1984; Janca, 1988; Beckett and Hart, 1993; Beckett and Giddings, 1997.. A schematic diagram of the SdFFF-ICP MS apparatus is presented in Fig. 1. The equivalent spherical diameter d of sample particles could be estimated from the measured elution volume Ž Vr .. In the case of normal mode and constant field SdFFF runs where retention is

B. Chen et al. r The Science of the Total En¨ ironment 263 (2000) 23᎐35

25

Fig. 1. Schematic representation of the SdFFF-ICP-HR MS apparatus.

high Žsay Vr ) 3Vo ., the approximate expression in Eq. Ž1. can be used ŽGiddings et al., 1983.: 3

ds

(

36 kTVr ␲ ␻2 rw⌬␳Vo

Ž1.

where k is Bolzman’s constant, T is the absolute temperature, ␻ is the centrifuge speed Žradians sy1 ., r is the centrifuge radius, w is the channel thickness, ⌬␳ is the density difference between the particle and the carrier solution and Vo is the channel void volume. The equation has been validated in many studies using well characterised standard samples ŽKirkland and Yau, 1982; Giddings et al., 1983; Caldwell, 1984..

2.3. Surface adsorption density distribution

The adsorptive behaviour of orthophosphate could be monitored using radiolabelled orthophosphate, which was adsorbed onto colloidal particles and separated by SdFFF. Fractions collected as a function of elution time were measured for their radioactivity using a scintillation counter and the distribution of orthophosphate adsorbed per unit mass of particle Ž dmcPrdmc . at any point i along the elution time or volume axis was calculated as follows ŽBeckett et al., 1990.: dmPc i s dmci

dmPc i dVr i

r

dmic dVr i

½ž / ž /5

A

DPMi UVi

Ž2.

2.2. Particle size and composition distributions The UV detector signal was collected during the elution run to monitor the particle concentration. These data enabled the particle size distribution of the sample to be computed. If some of the eluent from the UV detector was fed into the ICP-HR MS, element concentrations were also recorded. This information could be similarly processed to obtain elemental concentration distributions and element atomic ratio distributions.

where DPMi is the ␤ activity in disintegrationr minrml of eluent, and UVi is the UV detector signal at elution volume Vr i . Note that the superscript c in these quantities signifies that it is the cumulative amount eluted up to point i on the fractogram. These data are then used to calculate the surface adsorption density distribution ŽSADD., which is a plot of the amount of orthophosphate adsorbed per unit area of particle surface

B. Chen et al. r The Science of the Total En¨ ironment 263 (2000) 23᎐35

26

Ž dmcP irdAci , in arbitrary units. as a function of particle size. Assuming a constant spherical shape and density for the particles, the following equation can be used to determine the y-axis of SADD: dmPc i s dAic

dmPc i ␦m ci = ␦ Aci dmic

½ž / ž /5

A

DPMi = di UVi

Ž3.

3. Experimental 3.1. Colloid sample preparation A river sediment ŽYarra River, Melbourne, Australia. and soil sample ŽLilydale, Melbourne, Australia. were investigated in this paper. The Yarra River sediment was sampled at Dights Falls in Melbourne. This particular river system was chosen because its nutrient dynamics are currently being studied in our laboratories. A clay soil sample was collected from the B-horizon at Lilydale in Melbourne. Samples were suspended in ultra-pure Milli-Q water ŽMillipore. and screened through a 25-␮m mesh nylon sieve. The - 1-␮m diameter fraction was isolated by repeated centrifugation and stored at 4⬚C. 3.2. Chemical extraction procedures To selectively remove the oxyhydroxy coating as well as organic surface coatings, three separate extractants were used ŽTessier et al., 1979.. Hydroxylamine hydrochloride ŽNH2 OH ⭈ HCl. Ž0.25 M. in 0.25 M HCl was used to broadly release the Fe and Mn associated with reducible phases. The procedure was similar to that reported by Chao and Zhou Ž1983.. Sodium hypochlorite ŽNaOCl. Ž10% vrv. was used to extract pollutants held on organic sites. The method used to destroy organic matter was adapted from Lavkulich and Wiens Ž1970., who found that NaOCl extracted more organic matter with less destruction of Fe and Mn oxides than procedures employing hydrogen peroxide ŽH2 O2 .. Aqua regia Ž3:1 HCl and HNO3 . was used to release all metals in the samples

except those incorporated in the crystal lattice of the particle core ŽDavidson et al., 1994.. Extractions were performed on 5 ml of a 100grl colloid suspension. Samples were dispersed and homogenised by sonicating twice for 5 s each with a sonic horn ŽModel VCX-600, Sonics and Materials Inc., Danbury, CT, USA. at 650 W. Samples treated with 62.5 ml NH2 OH ⭈ HCl in HCl and 10 ml NaOCl were placed in an endover-end shaker for 16 h at 60 rev.rmin. Samples treated with 10 ml of aqua regia were gently digested overnight on a hot plate. When the volume of the mixture was reduced to approximately 1 ml, more aqua regia was added and the digestion repeated. Upon the completion of the extraction procedure, the samples were centrifuged for 30 min at 3750 rev.rmin. The extractable Fe and Mn in the supernatant were analysed by flame atomic absorption spectrometry ŽFAAS. ŽPerkin-Elmer model 1100.. The residue solids were washed three times with Milli-Q water. The suspension Ž1 ml. containing approximately 1 mg of residue solids in each sample was placed in a 6-ml scintillant vial and stored at 4⬚C for 33 P adsorption and ICP-HR MS element determination. Another sub-sample of the extracted residue solids was dried for natural organic matter determination using a carbon analyser ŽShimadzu, SSM-5000A.. 3.3. Orthophosphate adsorption 33

P-labelled orthophosphate in dilute hydrochloric acid was obtained from the Australian Nuclear Science and Technology Organisation. A Ž610 KBqrml. was sample Ž200 ␮l. of 33 PO3y 4 added to a 1-ml sub-sample of the colloid suspension Žapproximately 1 mgrml., shaken manually and allowed to stand for 48 h prior to SdFFF separation. During a SdFFF run, sample fractions were collected over 3-min intervals using a fraction collector ŽISCO Retriever 500.. A 1-ml aliquot was taken from each fraction and added to 3 ml of scintillant ŽUltima Gold. and thoroughly mixed. The ␤ activity of each sample was measured for 10 min using a scintillation counter ŽBeckmann LS6000TA .. The results were

B. Chen et al. r The Science of the Total En¨ ironment 263 (2000) 23᎐35

27

Table 1 The SdFFF run parameters Parameter

t1 ta Start RPM Hold RPM Relaxation time Žmin. Flow rate Žmlrmin. Sample concentration Žmgrml. Sample injection Ž␮l.

33

SdFFF system 1 for P adsorption experiment

SdFFF system 2 for ICP-HR MS

5.367 y42.93 1500 20 11 2.00 1.0 90

3.67 y29.37 1500 20 7 1.00 1.0 90

recorded in countsrminrml, from which the disintegrationrmin ŽDPMrml. was calculated. 3.4. SdFFF and ICP-HR MS instrumentations The SdFFF systems used here were as described in Ranville et al. Ž1999. and van Berkel and Beckett Ž1997.. The carrier solution consisted of 0.0005% Žwrv. sodium dodecyl sulphate ŽSDS. and 0.0002% Žwrv. sodium azide. The outlet stream from the SdFFF channel was passed through a LDC Milton Roy SpectroMonitor variable wavelength UV absorbance detector operating at 254 nm. There were two SdFFF systems used. System 1 was used for 33 P adsorption experiments and system 2 was coupled to the ICP-HR MS Žvan Berkel and Beckett, 1997.; the major difference being that the channel thickness w was 0.282 mm in system 1 and 0.147 mm in system 2. All samples were sonicated for 10 s using a sonicator probe prior to being injected into the SdFFF. The sample was introduced onto the channel through a rubber septum and was then relaxed under stop flow conditions. The run was then started by pumping the carrier through the channel. The initial field strength was held for a lag time t1 once the run started. A power program decay of the field reduced the centrifuge speed to a hold rev.rmin ŽWilliams and Giddings, 1987, 1991; Beckett and Hart, 1993.. The SdFFF run parameters are listed in Table 1. To obtain a high resolution element distribution as a function of particle size, the SdFFF system was directly coupled to an ICP-HR MS ŽFinnigan, MAT, ELEMENT.. The power sup-

porting the plasma was 1250 W, the reflected power was less than 5 W, and the carrier gas flow was 0.8 lrmin. The instrument operated in pulsecounting mode at medium resolution. A mass range of 23᎐133 a.m.u was scanned at a rate of 15 scan sweepsrmin. The measuring time was 0.01 s. The element masses used for analysis were: Mg, 23.9850; Al, 26.9815; Si, 27.9769; P, 30.9738; Mn, 54.9380; Fe, 55.9349; Co, 58.9332; and Cs, 132.9054. Calibration was achieved using standard solutions before and after each SdFFF run. Co and Cs were added to the SdFFF carrier and the standard solutions were used as internal standards to correct instrumental noise and drift, respectively. A blank sample was used for background correction. An in-house QBASIC software program was used to compute the element concentration based on ion currents obtained for the above standard solutions and to perform elementby-element drift corrections and noise reduction, as well as data smoothing.

4. Results and discussion 4.1. Efficiency of extraction techniques The effectiveness of the extractants in removing Fe, Mn and Natural Organic Matter ŽNOM. from the samples is summarised in Table 2. It should be noted that the percentage removal of Fe and Mn was calculated relative to that of the aqua regia extraction, which assumes that all of the Fe and Mn in the surface coatings had been removed. The percentage of NOM removed was

28

B. Chen et al. r The Science of the Total En¨ ironment 263 (2000) 23᎐35

Table 2 Extractable Fe, Mn and NOM in F 1.0 ␮m colloid fractions of Yarra River sediment and Lilydale soil Sample

Extractant

Fe extracted y1

Mn extracted y1

NOM extracted

mg⭈ kg

%

mg⭈ kg

% 100 81 0.2

Yarra River sediment

Aqua regia NH 2 OH ⭈ HCl NaOCl

49 000 12 800 15

100 26 0.03

360 290 0.8

Lilydale soil

Aqua regia NH 2 OH ⭈ HCl NaOCl

14 600 1620 10

100 11 0.07

19 2.9 0.8

computed relative to the organic carbon in the untreated samples. The Yarra River sediment sample was found to contain significantly higher Fe, Mn and NOM concentrations compared to the Lilydale soil sample ŽTable 2.. The values of these quantities are typical for sediments and soils in south-eastern Australia ŽEllaway et al., 1982; Chartres and Walker, 1988.. Acidic NH2 OH ⭈ HCl was found to remove 26% of the aqua regia extractable Fe and 81% of Mn, compared to 11 and 15%, respectively, for the Lilydale soil. This reduction in extraction efficiency may be due to different hydroxyoxide phases present in the samples. The oxidising agent NaOCl was found to remove very little Fe and Mn, but destroyed a high percentage of the organic matter Ž79 and 87% for the Yarra and Lilydale samples respectively.. These results demonstrated that the NH2 OH ⭈ HCl and NaOCl reagents showed a reasonable degree of selectivity in removing either the iron hydroxyoxides, manganese hydroxyoxides or NOM coatings from the particles. 4.2. Effect of surface coatings on orthophosphate adsorption Orthophosphate adsorption experiments were undertaken on untreated Yarra River sediment and Lilydale soil colloids, as well as the NH2 OH ⭈ HCl and NaOCl treated samples. Yarra Ri¨ er Sediment: The raw SdFFF fractograms obtained with adsorbed radiolabelled orthophosphate for Yarra River sediment colloids are shown in Fig. 2. The corresponding size dis-

100 15 4

g ⭈ kgy1

%

19 2.4 25

60 8 79

1.4 - 0.1 1.4

57 -4 57

tributions are shown in Fig. 3. The mass based fractograms and particle size distributions were found not to change significantly following NH2 OH ⭈ HCl and NaOCl treatments ŽFigs. 2 and 3.. This indicates that the particle mineral cores were not significantly attacked by the extraction procedures. The particle size distributions ranged from 0.05 to 0.5 ␮m. The corresponding surface adsorption density distributions ŽSADD. calculated from the fractograms and adsorbed orthophosphate analyses are shown in Fig. 4. The SADD plots for the samples were found to be non-constant, except for the NH2 OH ⭈ HCl-treated sample. The adsorption density values decreased significantly after extraction with NH2 OH ⭈ HCl, compared with the untreated sample. This finding was anticipated, because NH2 OH ⭈ HCl removes Fe and Mn oxide surface coatings, which we assumed contributes the majority of binding sites for orthophosphate adsorption. The opposite was observed for the sample treated with NaOCl, for which the SADD was found to increase markedly compared to the untreated sample. This observation conformed to the hypothesis that NOM competes with PO3y 4 for adsorption sites. Furthermore, the SADD trend for the sample treated with NaOCl was found to increase most significantly in the smaller particle size range, compared to the untreated sample. This finding may indicate that more NOM was bound onto the small particles than the large particles or that the adsorption binding was stronger. Lilydale Soil: The Lilydale colloidal soil sample was more complex. The raw SdFFF fractograms

B. Chen et al. r The Science of the Total En¨ ironment 263 (2000) 23᎐35

29

with adsorbed radiolabelled orthophosphate for the sample are shown in Fig. 5. The corresponding size distributions are shown in Fig. 6. The fractograms and particle size distributions remained unchanged after treatment with NH2 OH ⭈

Fig. 3. Particle mass-based and adsorbate-based size distributions of Yarra River sediment colloids for: Ža. untreated sediment; Žb. sediment treated with NH2 OH ⭈ HCl; and Žc. sediment treated with NaOCl. The y-axis records UV detector Ž dmcrdd, arbitrary units. and scintillation counter response Ž dmcPrdd, disintegrationrminr␮m.. Fig. 2. SdFFF fractograms of Yarra River sediment colloids with adsorbed radiolabelled orthophosphate for: Ža. untreated sediment; Žb. sediment treated with NH2 OH ⭈ HCl; and Žc. sediment treated with NaOCl. The y-axis records UV detector Ž dmcrdV, arbitrary units. and scintillation counter response Ž dmcP rdV, disintegrationrminrml..

HCl and NaOCl, indicating no significant attack of the mineral core by the reagents. The particle size was found to range between 0.05 and 0.50 ␮m with a high proportion of material present in

30

B. Chen et al. r The Science of the Total En¨ ironment 263 (2000) 23᎐35

the 0.05᎐0.20 ␮m size range. The orthophosphate adsorption experiments showed that a significant amount of orthophosphate was adsorbed in the smaller particle size region. This was consistent with the general trend that the smaller particles contained higher pollutant concentrations ŽBeckett et al., 1990.. One unusual feature observed was the presence of a distinct peak in orthophosphate adsorption at 0.20᎐0.35 ␮m for the untreated sample, which was eliminated by both chemical extractions. The corresponding SADD plots for the Lilydale soil sample are shown in Fig. 7. Again, a significant peak in the orthophosphate adsorption density was found in the 0.20᎐0.35-␮m size range for the untreated sample, which was not observed in samples treated with reagents. This result was unexpected for the sample treated with NaOCl, as NaOCl was expected to destroy natural organic matter but not to substantially affect Fe and Mn oxide surface coatings, resulting in an increase in PO3y adsorption. A possible explanation for this 4 phenomenon may be the presence of some easily extractable Fe and Mn surface coatings in this size range. This possibility was explored using SdFFF-ICP-HR MS element analysis of the colloids. 4.3. Analysis of chemical composition by SdFFFICP-HR MS The hyphenated technique of SdFFF-ICP-HR MS has excellent potential for determining changes in particle mineralogy and the composi-

Fig. 5. SdFFF fractograms of Lilydale soil colloids with adsorbed radiolabelled orthophosphate for: Ža. untreated soil; Žb. soil treated with NH2 OH ⭈ HCl; and Žc. soil treated with NaOCl. The y-axis records UV detector Ž dmcrdV, arbitrary units. and scintillation counter response Ž dmcP rdV, disintegrationrminrml..

Fig. 4. Orthophosphate surface adsorption density ŽSAD, dmcPrdAc, arbitrary units. distributions of Yarra River sediment colloids for: B, untreated sediment; `, sediment treated with NH2 OH ⭈ HCl; and ⌬, sediment treated with NaOCl.

tion of surface coatings across the size distribution ŽMurphy et al., 1993.. SdFFF-ICP-HR MS studies were undertaken on the untreated soil and after extraction with NH2 OH ⭈ HCl and

B. Chen et al. r The Science of the Total En¨ ironment 263 (2000) 23᎐35

31

Fig. 7. Orthophosphate surface adsorption density ŽSAD, dmcP rdAc, arbitrary units. distributions of Lilydale soil colloids for: B, untreated soil; `, soil treated with NH2 OH ⭈ HCl; and ⌬, soil treated with NaOCl.

8, w E xirUVi vs. particle diameter Ž d . plots for Si and Al are shown. The SirUV and AlrUV ratio plots were constant across the whole size range, probably indicating that the particle mineralogy remained unchanged. The plots also showed that the extraction processes employed in this experiment did not significantly attack the sample min-

Fig. 6. Particle mass based and adsorbate based size distributions of Lilydale soil colloids for: Ža. untreated soil; Žb. soil treated with NH2 OH ⭈ HCl; and Žc. soil treated with NaOCl. The y-axis records UV detector Ž dmcrdd, arbitrary units. and scintillation counter response Ž dmcP rdd, disintegrationr minr␮m..

NaOCl. The concentrations of Fe, Mn, Si, Al, Mg and P were determined as a function of particle size. The element composition trends were effectively monitored by computing the element concentrationrUV signal ratio distributions. In Fig.

Fig. 8. Element concentration distributions in Lilydale soil colloids plotted as element concentration from ICP-HR MSrUV detector response vs. particle diameter for: Ža. SirUV; and Žb. AlrUV. B, untreated soil; `, soil treated with NH2 OH ⭈ HCl; and ⌬, soil treated with NaOCl.

32

B. Chen et al. r The Science of the Total En¨ ironment 263 (2000) 23᎐35

Fig. 9. Atomic ratio distributions of Lilydale soil colloids for: Ža. SirAl; Žb. MgrAl; Žc. FerAl; Žd. MnrAl; and Že. PrAl. B, untreated soil; `, soil treated with NH2 OH ⭈ HCl; and ⌬, soil treated with NaOCl.

eral cores, since the sample treated with the extractants had almost the same composition trends as the untreated sample. Changes in mineralogy were more clearly seen by plotting element molar ratios against particle size. Since the Al and Si composition distributions are often relatively constant in aquatic sediment and soil samples, either Al or Si could be selected as an element to compare with other elements. Fig. 9a shows a constant SirAl ratio trend in this sample. In contrast, Fig. 9b shows that the MgrAl ratio rose significantly with increased particle size. The amount of Mg in 0.45-␮m sized particles was

nearly twice that in 0.08-␮m sized particles. Since Mg was probably present as either an isomorphous replacement ion in the clay lattice or in interlayer cation exchange positions, this trend suggested that the clay mineralogy changed with particle size. The fact that the extractants had not significantly affected the MgrAl distribution gave further support to the statement that these treatments did not attack the mineral core of the particles. Fig. 9c shows how the FerAl molar ratios changed with particle size. The sample extracted with NaOCl had the same molar ratio trend as

B. Chen et al. r The Science of the Total En¨ ironment 263 (2000) 23᎐35

33

the untreated sample. This supported the previous observation that the Fe oxide coatings were not significantly attacked by NaOCl extraction. The sample treated with NH2 OH ⭈ HCl had a distinct decrease in FerAl ratio over the entire size range. This would be anticipated if a significant amount of iron was in the form of surface coatings of hydrous iron oxides. The difference in FerAl molar ratio between the untreated sample and the sample treated with NH2 OH ⭈ HCl was greater in the small particles than the large particles. This result supported the hypothesis that a significant proportion of the Fe in these particles was associated with the surface layer coating; the small particles with large specific surface area could be coated with more hydrous Fe oxides per unit mass than the large particles.

sample was possible using SdFFF-ICP-HR MS. PrAl molar ratios are shown in Fig. 9e. The PrAl ratio decreased significantly for the NH2 OH ⭈ HCl-extracted sample, illustrating that most phosphorous was removed after treatment with this extractant. Probably both the Fe and Mn surface coatings were involved in the P association in the original particles. The Fe surface coating was likely to have been a factor in P adsorption across the whole particle size range. However, the Mn surface coating may have been a dominating factor in the 0.20᎐0.35 ␮m size range, where the peaks in both MnrAl and PrAl occurred.

4.4. Role of Mn hydroxyoxide coatings in orthophosphate adsorption

SdFFF separations, when combined with other analytical methods, provide a very powerful characterisation tool. In this paper, we illustrated that very detailed information can be obtained for soil and sediment colloids. Radiolabelled orthophosphate adsorption can be followed using fraction collection and offline scintillation counting, and element composition distributions can be obtained by the direct online coupling of SdFFF and ICP-MS. The results show that Fe and Mn surface coatings are important in orthophosphate uptake on soil and sediment particles. NOM coatings tend to inhibit PO3y adsorption to some 4 extent. The Lilydale soil sample contained elevated levels of P and Mn in the narrow size range 0.20᎐0.35 ␮m, which were removed by extraction with NH2 OH ⭈ HCl and NaOCl. This detailed analysis provided convincing evidence that Mn hydroxyoxide particle coatings can be important in orthophosphate binding in some cases.

It is obvious that the FerAl molar ratio trends for the Lilydale soil do not follow the SADD plots, especially for the untreated sample. However, it was found that the MnrAl molar ratio trend for the untreated sample ŽFig. 9d. followed the similar SADD trend of the untreated sample ŽFig. 7.. In particular, the double peak in both the MnrAl and SADD plots in the range 0.20᎐0.35 ␮m was eliminated by both NH2 OH ⭈ HCl and NaOCl extractions. This strongly indicated that the Mn oxide surface coating was involved in orthophosphate adsorption for this soil sample. The reduction of FerAl for the sample treated with the reductant NH2 OH ⭈ HCl was expected. However, the removal of Mn in the size range 0.20᎐0.35 ␮m for the sample treated with NaOCl was not expected. Flame AAS results showed that only 4.8% of Mn was removed, compared to the amount removed by aqua regia. This suggested that the Mn oxide surface coating within the narrow size range 0.20᎐0.35 ␮m was in a very easily extractable phase, most likely an amorphous hydroxide. The direct determination of phosphorous in the

5. Conclusions

Acknowledgements

This work was supported by the Australia Re-

34

B. Chen et al. r The Science of the Total En¨ ironment 263 (2000) 23᎐35

search Council and the CRC for Freshwater Ecology. Bailin Chen received a scholarship from Monash University. We thank Finlay Shanks for assistance with the SdFFF-ICP-HR MS work, Sandra Sdraulig for help with the FAAS analyses and Ashley Liang for help with the NOM determinations.

References

Allan RJ. The role of particulate matter in the transport and burial of contaminants in aquatic ecosystems. In: Hart BT, editor. Water Quality Management: The Role of Particulate Matter in the Transport and Fate of Pollutants, Water Studies Centre, Chisholm Institute of Technology, Melbourne, Australia, 1986:1᎐55. Beckett R, Giddings JC. Entropic contribution to the retention of non-spherical particles in field-flow fractionation and electron microscopy. J Colloid Interface Sci 1997; 186:53᎐59. Beckett R, Hart BT. Use of field-flow fractionation techniques to characterize aquatic particles, colloids and macromolecules. In: Buffle J, van Leeuwen HP, editors. Environmental Particles, 2. Boca Raton, FL: Lewis Publishers, 1993:165᎐205. Beckett R, Hotchin DM, Hart BT. Use of field-flow fractionation to study pollutant᎐colloid interactions. J Chromatogr 1990;517:435᎐447. Beckett R, Le NP. The role of organic matter and ionic composition determining the surface charge of suspended particles in natural waters. Colloid Surf 1990;44:35᎐49. Bostrom ¨ B, Andersen JM, Fleischer S, Jansson M. Exchange of phosphorus across the sediment᎐water interface. Hydrobiologia 1988;170:229᎐244. Caldwell KD. Field-flow fractionation of particles. In: Barth HG, editor. Modern Methods of Particle Size Analysis. New York: Wiley, 1984:211᎐250. Chao TT, Zhou LY. Extraction techniques for selective dissolution of amorphous iron oxides from soils and sediments. Soil Sci Soc Am J 1983;47:225᎐232. Chartres CJ, Walker PH. The effect of aeolian accessions on soil development on granitic rocks in south-eastern Australia. III. Micromorphological and geochemical evidence of weathering and soil development. Aust J Soil Res 1988;26:33᎐53. Davidson CM, Thomas RP, McVey SE, Perala R, Littlejohn D, Ure AM. Evaluation of a sequential extraction procedure for the speciation of heavy metals in sediments. Anal Chim Acta 1994;291:277᎐286. Ellaway M, Hart BT, Beckett R. Trace metals in sediments from Yarra River. Aust J Mar Freshwater Res 1982;33: 761᎐778.

Gachter R, Wehrli B. Ten years of artificial mixing and ¨ oxygenation: no effect on the internal phosphorus loading of two eutrophic lakes. Environ Sci Technol 1998;32: 3659᎐3665. Giddings JC. Field-flow fractionation. Sep Sci Technol 1984;19:831᎐847. Giddings JC, Karaiskakis G, Caldwell KD, Myers MN. Colloid characterization by sedimentation field-flow fractionation. I. Monodisperse populations. J Colloid Interface Sci 1983;92:66᎐80. Giddings JC, Myers MN, Caldwell KD, Fisher SR. Analysis of biological macromolecules and particles by field-flow fractionation. In: Glick D, editor. Methods of Biochemical Analysis 26. New York: Wiley, 1980:79᎐136. Hart BT, Beckett R, Murphy D, Ranville J. Role of colloids in cycling contaminants in rivers. In: Adriano DC, editor. Advances in Environmental Science: Biogeochemistry of Trace Metals. Water Studies Centre, Monash University, Melbourne, Australia, 1993:193᎐216. Janca J. Field-flow fractionation: analysis of macromolecules and particles. Chromatographic Science Series 39. New York: Marcel Dekker, 1988:336. Kirkland JJ, Yau WW. Sedimentation field-flow fractionation: applications. Science 1982;218:121᎐127. Lavkulich LM, Wiens JH. Comparison of organic matter destruction by hydrogen peroxide and sodium hypochlorite and its effects on selected mineral constituents. Soil Sci Soc Am Proc 1970;34:755᎐758. Manning PG. Phosphate ion interactions at the sediment᎐water interface in Lake Ontario: Relationship to sediment adsorption capacities. Can J Fish Aquat Sci 1987; 44:2204᎐2211. Murphy DM, Garbarino JR, Taylor HE, Hart BT, Beckett R. Determination of size and element composition distributions of complex colloids by sedimentation field-flow fractionation᎐inductively coupled plasma-mass spectrometry. J Chromatogr 1993;642:459᎐467. Ranville JF, Chittleborough DJ, Shanks F et al. Development of sedimentation field-flow fractionation᎐inductively coupled plasma mass-spectrometry for the characterisation of environmental colloids. Anal Chim Acta 1999;381:315᎐329. Rees TF. Transport of contaminants by colloid-mediated processes. In: Hutzinger O, editor. The Handbook of Environmental Chemistry, 2. Heidelberg: Springer᎐Verlag, 1991:165᎐184. Stumm W. Surface chemical theory as an aid to predict the distribution and the fate of trace constituents and pollutants in the aquatic environment. Water Sci Technol 1982; 14:481᎐491. Stumm W, Sulzberger B. The cycling of iron in natural environments, consideration based on laboratory studies of heterogeneous redox processes. Geochim Cosmochim Acta 1992;56:3233᎐3257. Tessier A, Campbell PGC, Bisson M. Sequential extraction procedure for the speciation of particulate trace metals. Anal Chem 1979;51:844᎐851.

B. Chen et al. r The Science of the Total En¨ ironment 263 (2000) 23᎐35 van Berkel J, Beckett R. Determination of adsorption characteristics of the nutrient orthophosphate to natural colloids by sedimentation field-flow fractionation. J Chromatogr A 1996;733:105᎐117. van Berkel J, Beckett R. Estimating the effect of particle surface coatings on the adsorption of orthophosphate using sedimentation field-flow fractionation. J Liq Chromatogr Rel Tech 1997;20:2647᎐2667.

35

Williams PS, Giddings JC. Power programmed field-flow fractionation: a new program form for improved uniformity of fractionating power. Anal Chem 1987;59:2038᎐2044. Williams PS, Giddings JC. Comparison of power and exponential field programming in field-flow fractionation. J Chromatogr 1991;550:787᎐797.