Pergamon Geochimica et Cosmochimica Acta, Vol. 61, No. 1 I. pp. 2253-2263. 1997 Copyright 0 1997 Elsevier Science Ltd Printed in the USA. All rights reserved 0016-7037/97 $17.00 + .OO PI1 SOO16-7037( 97)00057-4 Negative thermal ion mass spectrometry of oxygen in phosphates C. Holmden,* D. A. Papanastassiou, and G. J. Wasserburg The Lunatic Asylum of the Charles Arms Laboratory, Division of Geological and Planetary Sciences. California Institute of Technology, Mail Code 170-25, Pasadena, California 91 125, USA (Received November 7, 1996; accepted in revised ,form January 28, 1997) Abstract-A novel technique for the precise measurement of oxygen isotopes by negative thermal ion mass spectrometry (NTIMS) is presented. The technique is ideally suited to the analysis of oxygen isotopes in phosphates which form intense PO; ion beams. Since P is monoisotopic, the mass spectrum for PO, at 79, 80, and 81 corresponds to 160, “0, and l8O. Natural and synthetic phosphates are converted and loaded on the mass spectrometer filament as Ag,P04 precipitated directly from ammoniacal solution. To lower the work function of the filament, BaCl, is added in a 1: 1 molar ratio of P04:Ba. Using these procedures, Br- mass interference (at 79 and 81 amu) is eliminated for typical analyses. Experiments with ‘*O-enriched water show less than 1% O-exchange between sample PO, and adsorbed water, and there is no O-exchange with trace O2 present in the mass spectrometer source chamber. The ionization efficiency of PO, as PO; is >lO% compared to 0.01% for both conventional dual inlet Gas Isotope Ratio Mass Spectrometry (GIRMS) and secondary ion mass spectrometry (SIMS). Therefore, NTIMS offers exceptional sensitivity enabling routine and precise oxygen isotope analysis of sub- microgram samples of PO, ( <2 1 nmoles equivalent CO* gas) without need for lengthy chemical pretreat- ment of the sample. Overall external precision is ? 1%0 (2g) for ‘*O/“ O and ‘7O/‘6O with reproducibility of instrumental isotope fractionation (calculated from ‘8O/‘6O) of 20.5%~ amu-‘. Small phosphate samples including single mineral grains from meteorites, or apatite microfossils, can be analyzed by this technique. Cop&ight 0 1997 Ekevier Science Ltd 1. INTRODUCTION We report on a Negative Thermal Ion Mass Spectrometry (NTIMS) technique for the isotopic analysis of all oxygen isotopes ( 160, 170, ‘*O) measured as POT. This contribution builds on earlier work of Heumann et al. ( 1989) and Wachs- mann and Heumann ( 1991)) who showed that ion beams of PO, and PO; can be obtained from phosphate compounds. Since P is monoisotopic (“ P), the mass spectra of POX- and PO; reflect the isotopic composition of oxygen. We show that it is possible to measure oxygen isotope abun- dances in sedimentary, igneous, and biogenic phosphates by a modified direct loading technique, thus eliminating the need to convert phosphate oxygen to CO,, which is neces- sary for conventional dual inlet Gas Isotope Ratio Mass Spectrometry (GIRMS ) Although oxygen isotope analyses by GIRMS are highly precise (t0.2%~, 2a), relatively large volumes of sample gas (>O.l pmole) are required to achieve source pressures high enough to maintain viscous how through the capillary inlet and to compensate for the low Ionization Efficiency (IE) of the electron impact source (0.01%) (Brenna, 1994). We have obtained ionization efficiencies for PO, >lO% ( PO_: ions detected / PO, molecules loaded on the sample filament), readily permitting measurement of 70 ng of oxy- gen ( 100 ng Pod). The precision and accuracy of the oxygen isotope measurements depend on: ( 1) eliminating isobaric interferences with 79Br and “Br ~, (2) evaluating the poten- tial for high temperature O-exchange between PO, oxygen *Present address: Department of Geological Sciences, University of Saskatchewan, Saskatoon, Saskatchewan S7N 5E2, Canada. and extraneous oxygen on the filament or within the mass spectrometer source chamber, and (3) controlling the effects of instrumental isotope fractionation. 2. ANALYTICAL PROCEDURES 2. 1. Standard and Normal Solutions We have used several apatite standards including two igneous apatites UMS- 1 (University of Michigan) and LA- 1 (Laramie Anor- thosite), and Florida rock phosphate (SRM- 12Oc), the last distrib- uted by the National Institute of Standards and Technology. These apatites were dissolved in 1.0 N HNO?. Calcium was removed by passing the solution through a cation exchange column containing Dowex 5OW-X8, 200-400 mesh. The pH of the resulting H3POJ solution was adjusted to 28.0 with clean NH,OH. and approximately 1.5 times the stoichiometric amount of AgNO, was added. In an open container, the slow evaporation of NH, (overnight) lowered the solution pH, causing large yellow crystals of Ag,PO, to precipitate (Firsching, 1961; cf. also O’Neil et al., 1994). This precipitate was rinsed four times in clean water. No attempt was made to eliminate or characterize potentially coprecipitating AgCI, AgBr, Ag,S, and AgZS04. In addition to the apatite standards, Johnson Matthey Ag,P04 and NaJP04 salts were used as normals. The Ag,P04 normal was prepared by dissolution in 10% NHIOH. The NaTPOd normal was used to prepare Ag3P0, by first dissolving NaiPOl in water, followed by conversion to Ag,PO,, as above. Solutions for loading a sample on the mass spectrometer filament were prepared by dis- solving a known amount of AglPO, in 10% NH,OH. 2.2. Sample Loading Procedure We have performed extensive tests of sample loading techniques and have adopted the following procedure. A thin layer of colloidal Pt powder (Pt-black) is loaded as an aqueous slurry on a Pt filament (0.020” wide X 0.0012” thick) using a microsyringe fitted with polyethylene microtubing (PE-10); the colloid is dried with a small current through the filament. The deposit of Pt powder covers the width of the filament and is about 3 mm in length. Next, an aliquot 2253