3.1 Introduction

The main difference between the protocol by Tamburini et al. (2010) and other methods is the precipitation of cerium phosphate to purify extracts (McLaughlin et al. 2004). An issue with the precipitation of cerium phosphate is that it precipitates around pH 5.5, which is the same range where iron oxides flocculate. While flocculating iron oxides can strip out other components like dissolved organic matter (DOM) from the solution (Hiradate et al. 2006). This combination of iron oxides and OM stays until the end and the obtained silver phosphate will be contaminated (Tamburini et al. 2010).

The following purification protocol is based on the original protocol published by Tamburini et al. (2010) and the protocol described in Chap. 2 of the IAEA-TECDOC-1783 (IAEA 2016). The original protocol was developed for 1 M HCl extracts of soils and fertilizers but has been adapted since its publication to a range of other extracts including NaOH-EDTA from soils (Tamburini et al. 2018) and trichloroacetic acid (TCA) extracts from plants (Pfahler et al. 2013). Those modifications are described in Chap. 4.

3.2 Reagents

  1. (1)

    Ammonium nitrate solution, 35% (w/v) (= 4.2 M)

    Weigh out 538.5 g of ammonium nitrate salt (NH4NO3; molecular weight (MW) 80.052 g/mol). Add 1000 mL of ultrapure water and dissolve the salt, stirring well. Store at room temperature.

  2. (2)

    Ammonium nitrate solution, 5% (w/v) (= 0.6 M)

    Weigh out 105.3 g of NH4NO3 salt. Add 2000 mL of ultrapure water. Stir well to dissolve salt. Store at room temperature.

  3. (3)

    Ammonium heptamolybdate solution, 10% (w/v)

    Weigh out 53.3 g of ammonium molybdate salt (NH4Mo x7H2O; tetrahydrate form; MW 1235.86 g/mol). Add 480 mL of ultrapure water. Dissolve well. Enough for 12 samples. Prepare fresh before each use.

  4. (4)

    Ammonium-citrate solution

    Weigh out 10 g of citric acid. Add 300 mL of ultrapure water and 140 mL of concentrated ammonia solution (NH4OH). Prepare and use under hood. Stable at room temperature.

  5. (5)

    Magnesia solution

    Weigh out 50 g of magnesium chloride (MgCl2 × 6H2O; hexa-hydrate salt; MW 203.3 g/mol) and 100 g of ammonium chloride (NH4Cl; MW 53.49 g/mol). Dissolve them in 500 ml of ultrapure water. Acidify to pH 1 with conc. HCl. Volume is then adjusted to 1 L with ultrapure water. Stable indefinitely at room temperature. Caution: MgCl2 can be contaminated with P.

  6. (6)

    1:1 and 1:20 ammonia solutions

    Measure in a volumetric cylinder the concentrated ammonia (NH4OH; 50 mL for the 1:1 and 100 mL for the 1:20). Pour into an appropriate glass bottle and dilute with ultrapure water (50 mL for the 1:1 and 1900 mL for the 1:20). Store at room temperature.

  7. (7)

    1 M HCl

    In a volumetric flask add 82.7 mL concentrated hydrochloric acid (HCl; 37%) in 900 mL ultrapure water and make up to 1 L.

  8. (8)

    0.5 M HNO 3

    Measure 967 mL of ultrapure water and add 33 mL of concentrated nitric acid (HNO3).

  9. (9)

    1 M HNO 3

    In a volumetric flask add 66 mL concentrated HNO3 to 800 mL ultrapure water and make up to 1 L.

  10. (10)

    7 M HNO 3

    Add 463 mL concentrated HNO3 to 537 mL ultrapure water.

  11. (11)

    AG 50 × 8 cation resin

    Take the equivalent of 6 mL of resin per sample and place it in a plastic bottle. Add 1.5 BV of 7 M HNO3 and shake well and let rest overnight. The next day, discard the acid and wash it thoroughly using ultrapure water till pH is close to neutrality. After use, collect the resin and store it in 1 M HNO3 (66 mL concentrated HNO3 + 934 mL ultrapure water).

  12. (12)

    Silver ammine solution

    Weigh out 10.2 g of silver nitrate (AgNO3; MW 169.87 g/mol) and 9.6 g of NH4NO3. Dissolve in 81.5 mL of ultrapure water and add 18.5 mL of concentrated NH4OH. Store in amber bottle in the dark.

Some of the above-mentioned chemicals are quite hazardous. According to the Globally Harmonized System of Classification and Labelling of Chemicals (GHS), concentrated HCl, for example, has the following hazardous statements: H290 may be corrosive to metals, H314 causes severe skin burns and eye damage and H335 may cause respiratory irritation. It is therefore advised to check the GHS hazardous and precautionary statements and the safety data sheets for the chemicals before using the protocol. It is also recommended to follow the good laboratory practice which includes wearing personal protective equipment (PPE) like lab coats, safety glasses and gloves.

3.3 Equipment and Consumables

  1. (1)

    Standard lab glassware and equipment.

  2. (2)

    50 mL polypropylene tubes.

  3. (3)

    GF/F filters.

  4. (4)

    Cellulose acetate filters (pores 0.2 µm).

  5. (5)

    Cellulose nitrate filters (pores 0.2 µm).

  6. (6)

    Items 4 and 5 could be substituted by the GPWP filters by Millipore or equivalent, resistant to a larger pH range.

  7. (7)

    Polycarbonate filters (pores 0.2 µm).

  8. (8)

    Parafilm.

  9. (9)

    Silver capsules, pressed, 4 × 3.2 mm.

  10. (10)

    Lamp black (Gasruß; glassy carbon), conditioned, 10 mL.

  11. (11)

    Fume hood.

  12. (12)

    Fridge (max. +8 °C).

  13. (13)

    Water bath shaker set at 50 °C.

  14. (14)

    Multiplate magnetic stirrer.

  15. (15)

    Drying oven (no ventilation).

  16. (16)

    IAEA-601 and 602 benzoic acid standards.

  17. (17)

    Silver phosphate standards.

3.4 Procedure

The following describes the ideal case, i.e. 1 M HCl extract with around 20 µmol P, low in organic matter and no other compounds which might interfere. How to deal with other extracts and potential problems during the purification is described in Chap. 5. The extractants are first purified in four consecutive steps (A1–A4; Fig. 3.1), before the precipitation of the final analyte in Step A5 (Fig. 3.1).

Fig. 3.1
figure 1

Stepwise purification of extracts and final precipitation of silver phosphate

Full size image

The first two steps (A1 and A2) are removing, for example, organic matter and are also concentrating the phosphate in the extract by reducing the volume. Certain cations could interfere with the precipitation of silver phosphate and are removed in Step A3. Silver chloride, which, if not removed, could co-precipitate with silver phosphate is removed in Step A4. The final analyte is then precipitated in Step A5. The filtration steps can be quite tedious, and using vacuum filtration equipment is therefore recommended.

Step A1: Ammonium phosphor-molybdate (APM) mineral precipitation and dissolution

  • Pour supernatant (about 100–150 mL) into 200 mL Erlenmeyer flasks and place the flasks into the water bath set at 50 °C.

  • Add 25 mL of 35% ammonium nitrate solution and 40 mL of the 10% NH4-Mo solution. Shake gently overnight in the warm water bath.

  • The following day filter the formed crystals (generally bright yellow, but variations might happen) by using cellulose acetate filters (or filters resistant to low pH). Wash thoroughly with the 5% ammonium nitrate solution (about 200–300 mL). Collect crystals and discard supernatant.

  • Place the filter with the crystals into 100 mL Erlenmeyer flasks and add about 50 mL of the NH4-citrate solution (work under the fume hood!). Gently swirl to dissolve the crystals. Remove the filter.

Step A2 : Magnesium ammonium phosphate (MAP) mineral precipitation and dissolution

  • Place the Erlenmeyer flasks onto the multiplate magnetic stirrer. Add 25 mL of the Mg solution, while stirring. Then add slowly about 7 mL of the 1:1 ammonia solution. Cover with parafilm and make small holes to vent. Leave overnight.

  • The following day filter the white crystals by using the cellulose nitrate filters (or filters resistant to high pH). Wash thoroughly with the 1:20 ammonia solution. Discard supernatant.

  • Collect the filter and white, fine crystals into 50 mL polypropylene tubes. Add about 20 mL of 0.5 M HNO3 and shake to dissolve the crystals.

Step A3 : Cation removal

  • Add about 6 mL of the cation resin slurry which was brought to pH 7. Seal with parafilm, shake overnight.

  • The following day filter the samples by using the 0.2 µm polycarbonate filters. Rinse the resin with 1–2 mL ultrapure water.

  • Collect supernatant and place the resin to be reconditioned.

Step A4 : Silver chloride (AgCl) removal

  • Check if Cl− is still present in solution by adding a little amount of AgNO3. If a white cloud forms (AgCl crystals), wait 10 min and filter again.

Step A5: Ag 3 PO 4 precipitation

  • Once that the supernatant is Cl free, add 5 mL of the Ag-ammine solution. Place the tubes open into the oven set at 50 °C. Add ultrapure water to keep volume as constant as possible.

  • If yellow crystals are not formed in the next 24–48 h, check the pH of the solution and bring it to 7, using either HNO3 or NH4OH (absolutely no HCl or NaOH). When NH4OH is used to adjust the pH value, the pH value of the solution changes sharply, so NH4OH should be added slowly drop by drop.

  • Once that the crystals are formed, filter them using the 0.2 µm polycarbonate filters. Wash thoroughly with ultrapure water. Collect filters and crystals, and discard supernatant.

  • Place the filters on Petri dishes and put them into the oven set at 50 °C for at least 1 day. Cover the Petri dishes to prevent filters (and crystals) from flying away!

  • By gently scraping the filter, collect the dried crystals and put them into little vials. Store possibly inside a desiccator. Homogenize well before weighing, for example, by using a small pestle.

3.5 Preparation for TC/EA-IRMS and Data Analysis

Following Step A5, the silver phosphate samples need to be weighed in for the measurement with a thermal conversion elemental analyser (TC/EA) coupled to a continuous flow isotope ratio mass spectrometer (IRMS). In the TC/EA the silver capsules containing the samples and standards are pyrolysed at 1450 °C and the oxygen is converted into CO. The CO is then, via a helium stream, transported through a GC column which is separating CO from any nitrogen (N2) that might be present to the IRMS. In the IRMS the isotopic composition of CO is then measured. Before each run, the linearity of the IRMS should be checked and this will inform us about the range of weight that should be used for the silver phosphate standards and samples. In general, most instruments are linear between 250 and 350 µg of silver phosphate; however, linearity should still be checked (Carter and Barwick 2011). Likewise, a stability check of the IRMS should be done before each run (Carter and Barwick 2011). Further information about normalization and selection of reference materials can also be found in Skrzypek (2013) and Skrzypek and Sadler (2011).

  • Store all isotope reference standards, laboratory standards, samples and the capsule in desiccator with a drying material to avoid condensation of water. After weighing in, transfer the samples and standards back to the desiccator and dry for a minimum of 24 h.

  • Weigh the silver phosphate crystals into silver capsules in triplicates in the previously determined linearity range. Depending on the instrument preconditioned addition of glassy carbon powder can be added to improve the pyrolysis. Close tightly to avoid trapping air by forming little balls using tweezers.

  • Weigh 100 µg of the two benzoic acid standards IAEA-601 and IAEA-602 into silver capsules. Together with the silver phosphate standards these have to cover the range of the δ18OP values of the samples. There is no need to add glassy carbon even if it was added to the silver phosphate samples and standards. Close tightly by forming little balls using tweezers and gloves (Werner and Brand 2001).

  • Weigh in the silver phosphate standard, which was calibrated against international certified reference material (Halas et al. 2011; Watzinger et al. n.d.) into silver capsules in triplicates. If glassy carbon was added to the samples, it should also be added to the standards (Werner and Brand 2001). Close tightly by forming little balls using tweezers.

  • Weigh different amounts of pure silver phosphate obtained from a chemical company or of the internal standards (Lécuyer et al. 2007) into silver capsules. If glassy carbon was added to the samples, it should also be added to the standards (Werner and Brand 2001). They should cover the range of weights used for the samples. Close tightly by forming little balls using tweezers and gloves.

  • Prepare 5–6 blanks (adding glassy carbon if added to silver phosphate samples and standards) to empty capsules and closing them tightly. These are placed at the beginning of the run. The blanks aid preconditioning the column and are also necessary if the IRMS does a blank correction.

  • Transfer closed capsules to coded racks. Most commonly 96-well microtiter plates are used as racks.

During the whole preparation process neither the crystals nor the silver capsules should be touched with bare hands in order to avoid contamination with any O-containing compounds.

A sample sequence might look as follows (see the Annex for an example):

  • Five blanks.

  • Silver phosphate standard from certified source (×5 at different weights).

  • Additional standards (for example, internal silver phosphate standard, IAEA-601, IAEA-602) are used for normalization, at least four of each, distributed along the run.

  • Samples can be put in groups of 12, weighing each sample in triplicates.

  • After each group, standards should be added to obtain a sequence like

    • 12 samples,

    • 1 silver phosphate standard,

    • 1 IAEA-601 and

    • 12 samples.

  • Repeat until the run is almost full and finish with

    • 4× certified silver phosphate standard and

    • Additional standard.

Once the δ18OP values are analysed with the TC/EA-IRMS, a quality check needs to be done. The first three steps of the data analysis therefore are

  1. 1.

    Checking the oxygen yield; 15.3% for pure silver phosphate, between 14 and 17%, is acceptable.

  2. 2.

    Checking for nitrogen (N2) contamination.

  3. 3.

    Drift correction and normalization to international standards.

  4. 4.

    Calculating average δ18OP value and standard deviation.

Oxygen yield above or below expectation implicates that the silver phosphate sample was not pure silver phosphate and hence the obtained δ18OP value might be erroneous. To check for the purity of the Ag3PO4, we control the O% given by the TC/EA. The silver phosphate standards weighed at different amounts provide a calibration curve used to determine the O content of all samples. If the analysed Ag3PO4 samples are pure, their O content should be in the range of the Ag3PO4 standard.

Similarly, a contamination with N2 could also lead to erroneous δ18OP values. Either due to interference during the mass spectrometer measurements or because of N compounds containing oxygen (Pederzani et al. 2020). Sometimes, if the final wash of the Ag3PO4 is not done properly, NO3 could remain on the crystals. The presence of N2 is checked by looking at the chromatogram from the TCEA and from the IRMS. The N2 peak should be identified before the CO peak. Only when the first two checks have been passed, an average δ18OP value can be calculated.

The isotopic values of the certified standards should remain stable along the run. If this is not the case, a drift correction needs to be done (Carter and Barwick 2011). First the drift needs to be calculated:

$${\text{Drift}} = \left( {{\text{Average}}_{{{\text{STD1}}0 - {12}}} - {\text{Average}}_{{{\text{STD1}} - {4}}} } \right)/\left( {{\text{Position}}_{{{\text{STD1}}0}} - {\text{Position}}_{{{\text{STD1}}}} } \right),$$
(3.1)

where AverageSTD10–12 is the average δ18OP value of the last three replicates of the certified silver phosphate standard in a run, AverageSTD1–4 is the average δ18OP value of the first four replicates of the certified silver phosphate standard in a run, PositionSTD10 is the position of the 10th certified silver phosphate standard in a run and PositionSTD1 is the position on the 1st certified silver phosphate standard in a run.

The drift can then be used to perform a drift correction of the δ18O values:

$${\updelta }^{{{18}}} {\text{O}}_{{{\text{D}}\_{\text{corr}}}} = {\updelta }^{{{18}}} {\text{O}}_{{{\text{meas}}}} - {\text{drift}} \cdot \left( {{\text{Position}}_{{{\text{sample}}}} {-}{\text{ Position}}_{{{\text{STD1}}}} } \right),$$
(3.2)

where δ18Omeas is the measured δ18O value of a sample/standard, drift is the previously calculated drift value, Positionsample is the position of a sample/standard in a run and PositionSTD1 is the position on the 1st certified silver phosphate standard in a run.

After drift correction of samples and standards, average measured and drift corrected values of the standards should be correlated to the reported values (Brand et al. 2009). Using the slope and intercept of this correlation, calculate the real value of the samples.

A good practice is to compare the standard deviation of each sample to the standard deviations of the silver phosphate and benzoic acid standards within the same run. Standard deviations are around ±0.3‰ (Brand et al. 2009; Halas et al. 2011; Watzinger et al. 2021). Only if the standard deviation of a sample is equal or smaller than the standard deviations of the standards, a δ18OP value can be trusted. In case oxygen yield is outside the acceptable range and/or N2 contamination was found, the silver phosphate sample can be treated with H2O2 (removing organic compounds), thoroughly rinsed again with ultrapure water (removing N-containing compounds) and/or homogenized better (in case of crystals with different sizes). If the oxygen yield and N2 are ok, but the standard deviation (SD) of the replicates is still high (higher than SD of certified standard), then repeat analysis and possibly homogenize sample better. Vacuum roasting is also sometimes used to remove contamination with other O-bearing compounds; however, this procedure also has some issues including potential reoxidation when samples are stored too long (Mine et al. 2017).