Sample Preparation for Accelerator-based Radiocarbon Dating

J. 0. Wand, R. Gillespie and R. E. M. Hedges

Although several efficient sample preparation methods have been devised for conventional radiocarbon dating, the advent of the accelerator mass spectrometer (AMS) dating technique with its much smaller sample requirement and different counting method has meant that new preparation routes have had to be developed. Techniques used for the Oxford AMS systemare described in this paper.
Kr?~words:

RADIOCARBON DATING, METER, SAMPLE PREPARATION.

ACCELERATOR

MASS SPECTRO-

Introduction

In conventional radiocarbon dating, only the small fraction of 14C atoms decaying during the counting period are measured, whereas with the AMS method the majority of 14C atoms present in a sample can be detected without waiting for radioactive decays to take place. Consequently, much smaller amounts of carbon are required, typically l-5 mg rather than lL5g. The negligible background of the AMS system means that the available age range extends beyond 50,000 years. Thus the sample preparation technique must be capable of efficiently handling milligramme sized samples without introducing carbon contamination or allowing cross-contamination between samples. The essence of the AMS method is the production of carbon ion beams which are analysed by means of magnetic and electric fields into separate beams for each of the three carbon isotopes (12C, r3C and r4C) before measurement. For the production of large, stable beams of negative carbon ions, a caesium beam is directed at a carbon target in a sputter source. The best performance of the sputter source is obtained with graphite samples, which yield good carbon ion beams with minimal cross-contamination between different samples mounted in the source. General experience with carbon ion beams in AMS systems is described in Radiocarbon (1980) and a detailed description of the approach used at Oxford is given in Hedges ct al. (1980). For convenience, the graphite should be confined to a small area and be easy to handle and store after preparation. Thus, all samples, after decontamination and pretreatment as
Research Laboratory for Archaeolog> Oxford OX 1 305. England. and the Histor) of Art. 6 Kchlc Road.

1.5)
0305 4403 X4 020159 05 503.00 0 ( 1984 Academic Press Inc. (London) Lmutcd

160

J. 0.

WAND

ET AL.

described in Gillespie or al. (1984). are converted to graphite using a combination of conventional and novel techniques :

on tantalum

wire.

(1) carbon dioxide preparation-organic materials arc burned in oxygen, or in vacuum with copper (II) oxide as catalyst and a source of oxygen; carbonates are treated with acid; (2) acetylene preparation-carbon dioxide is reacted with molten lithium metal to form lithium carbide, this reacts with water to form acetylene; (3) graphite preparation --acetylene is pyrolized by a hot tantalum wire. graphite being deposited as a bead on the wire.

- Tantalum

wre Sample tcuo

lV,30A

H,O trap

CzH,

trap

Hz0 system.

trap

tigure

1. Ciraphitc

preparation

Sample preparation The methods for preparation of carbon dioxide follow standard practice. Organic samples are burned in a slow flow of pure oxygen. complete combustion being assured by an afterburner of supported platinum at 450 C or copper (II) oxide at 750 C. Combustion is most satisfactory for the majority of samples with a flow of oxygen rather than a static system, but the pressure must be kept below 1Ombar in the liquid-nitrogen-cooled carbon dioxide traps to avoid some oxygen being trapped. We have observed impurities in the oxygen cylinders which could be serious contaminants with very old samples; these impurities include carbon dioxide, carbon monoxide and hydrocarbons. For this reason, elimination of oxygen may be preferred. In this case. samples are mixed with wire-form copper (II) oxide in a silica tube, evacuated and heated to about 800 C for 5-10min. For carbonates, carbon dioxide is released from the sample by treatment with acid; 88,6 phosphoric acid is convenient because of the low vapour pressure. Water is removed from the carbon dioxide produced in all systems by dry-ice-cooled traps.

SAMPLE

PREPARATION

IN AMS DATING

101

The acetylene preparation is similar to that described by Tamers (1975) but scaled down for use with I--5mg samples. An all stainless steel system is used, with PTFE ferrules in compression fittings for case of demounting and cleaning. Because of the small volumes involved, it has been found convenient to freeze carbon dioxide directly into the lithium reactor and heat with about 50mg of lithium (a large excess). This also keeps the pressure higher than the 50mbar which is apparently essential for high yields of carbide. The reasons for this, and a proposed reaction mechanism, arc given in Wand ( 1980). The overall reaction may be written : 1OLi + 2C02 -3 4Li,O -4 Li,C, which proceeds at a temperature of 60&700 C. Afterheating to 900 C advocated by Polach et ul. (1972) has not been found to increase the yield, and the radon removal effected by this step is not required for the AMS. Water is slowly added to the cooled reactor, the reactions giving off acetylene. which is collected in liquid-nitrogencooled traps, and hydrogen (from the excess lithium), which is pumped away. Acetylene is routinely produced in better than 900/, yields. Pressures of carbon dioxide and acetylene are measured with stainless steel diaphragm transducers. The vacuum conditions are monitored with Pirani and Penning gauges. and we have found a residual gas analyser to be helpful in the development of thu system (for leak detection and gas purity analysis). .A cryopump is used to minimize contamination from vacuum pump oils. The final stage of graphite. achieved by pyrolysis of acetylene. is new to radiocarbon dating. A tantalum wire 0.5 mm in diameter is resistively heated between stainless steel electrodes. a current of about 30A at 1.5 V AC being used to raise the temperature of the wire to 2000 C. Maximum yield of about 607 graphite deposited on the wire is obtained after 15-20 min heating, using an acetylene pressure of 20-50 mbar. Yields are independent of the metal used : tungsten. molybdenum and hafnium all work satisfactorily, but tantalum is superior since it does not display embrittlemcnt under the conditions employed. The yield is also independent ofsurface area, so short wires can be prepared which are well suited to the sputter source and are easily handled and stored. No deterioration has been observed in the condition of the graphite after storage for 2 years, any surface contamination present being removed by the ion source sputtering in less than 1 min. This apparently simple cracking of acetylene in fact proceeds by a series of gasphase reactions forming polyacetylenic and polycyclic species which are deposited on the wire and there dehydrogenated (Homan & Wagner, 1967; Tanzawa & Gardincr, 1978; Lahaye & Prudo. 1978). Side reactions lead to the formation of unreactive polycyclic species which are deposited on the walls of the cracking vessel. Rigorous clean-up procedures are essential after each cracking reaction -we find that rinsing with chloroform followed by ultrasonic agitation in 1 M HCI, distilled water and methanol. then drying at 100 C is satisfactory. If a build-up of insoluble deposit is observed, sand blasting is an effective final solution. It is also possible to thermally crack benzene to graphite. We have used petroleumderived benzene as a background standard and the Harwell benzene standards (Otlet (I al., 1980) as known age samples for calibration. The benzene can be introduced into the cracking vessel either by syringe through a septum or via a micro needle valve. To obtain a full system background, benzene may be burned on hot copper (II) oxide and processed to graphite normally. Using these techniques. the directly cracked petroleum benzene yields graphite with a C/12C ratio <O.lo/,; of that of the modern reference standard; when the benzene is burned and converted via lithium carbide and

162

J. 0. WAND

/T AL

acetylene to graphite, a ratio about 0.4% of the modern standard is obtained. This is equivalent to a maximum measurable age of about 45,000 years, which we consider to be satisfactory for our initial work. Tracer experiments were used to determine whether memory or cross-contamination effects were present in the various stages of the reaction sequence. Carbon highly enriched in the mass 13 isotope was burned and converted to graphite according to the scheme described above, and samples of carbon dioxide ,after combustion and from burning the prepared graphite were analysed for i3C/12C ratio on a stable isotope mass spectrometer. A memory effect was observed when using the platinum combustion catalyst, equivalent to about 0.5% of the enriched sample carbon retained on the catalyst and which was released as carbon dioxide during the next combustion of normal carbon. This was not observed when using the copper (II) oxide catalyst, which was sulTicient reason for our choice of the latter method. The lithium reaction also showed a memory effect, confirming the work of Radnell & Muller (1980). Up to 1.5% of the enriched carbon was retained in the lithium reactor, and was not completely removed by four subsequent reactions with normal carbon. This problem can be reduced by using disposable crucibles inside the lithium reactor, but even in this situation a memory effect is possible since lithium can vaporize onto the outer vessel walls during the outgassing procedure. The solution adopted is to use disposable lithium reactors; as these are only 75 mm long by 12.5 mm diameter and can be made from an easily machined stainless steel the cost is not prohibitive. Auto-production of acetylene from lithium, suggested by Geyh (1969), has also been observed. The amount produced from our standard reaction was always less than 0.2% of the sample acetylene, which we do not consider to be significant at the current level of accuracy. Acetylene was also produced, at about 0.5% of sample size, from used lithium reactors, even when no new lithium was added. Results from the AMS on 0.1 mg of graphite from such blank acetylene gave a 14C/12C ratio about 70% of the modern standard. This is an indication of a memory effect averaged over many reactions. Further experiments using materials of known age indicated that as much as 3% contamination could be introduced from the lithium reaction, obviously a serious problem for routine dating of very old samples. These results all reinforce the decision to use disposable lithium reactors. Since the yield of graphite from acetylene using the pyrolysis method as described above is never greater than 60%, the possibility exists for isotopic fractionation to occur. Measurements of the i3C/i2C ratio on carbon dioxide from the combustion of samples and from the combustion of the prepared graphite indicate that fractionation does occur at the level of -0.8 + 0.1%. This is constant for all sample materials, and while it is a significant amount it is probably less than that taking place in the AMS system. During a dating run both i4C/12C and i3C/12C ratios are measured, so that these fractionation effects should be correctable by the on-line computing, White (1981). Acknowledgements The authors would like to thank R. Burleigh of the British Museum Research Laboratory for his help with stable isotope measurements on samples for this work. A. Bowles and M. Humm provided technical assistance.

References
Geyh, M. A. (1969). Problems in radiocarbon dating of small samples by means of acetylene. ethane or benzene. Intrrnational Journal of Applkd Radiation and Isotopc~ 20, 46Z-466.

SAMPLE

PREPARATION

IN AMS DATING

163

Gillespie, R.. Hedges, R. E. M. & Wand, J. 0. (1984). Radiocarbon dating of bone by accelerator mass spcctrometry. JournalqfArr~hac~ological Science 11, 165170. Hedges. R. E. M., Wand, J. 0. & White, N. R. (1980). The production of C - beams for radiocarbon dating with accelerators. Nuclear Instruments and Methods 173,40%421. Homann, K. H. & Wagner, H. G. (1967). Some new aspects of the mechanism of carbon formation in premixed flames. Proceedings qf 11th Symposium on Combustion, pp. 371-379. Lahaye, J. & Prado. G. (1978). In (P. L. Walker & P. A. Thrower, Eds) The Chemistry and Physics of Carbon 14, 167. Otlet, R. L., Walker. A. J., Hewson, A. D. & Burleigh, R. (1980). 14C interlaboratory comparison in the UK : Experiment design, preparation and preliminary results. Radiocarbon 22, 936946. Polach, H. A., Cower, J. & Fraser, 1. (1972). Synthesis of high purity benzene for radiocarbon dating. Proceedings qf 8th International Conference on Radiocarbon Dating, Wellington. B36 49. Radnell, C. J. & Muller, A. B. (1980). Memory effects in the production of benzene for radiocarbon dating. Radiocarbon 22,479-486. Rudiocurhon ( 1980). 22 Tamers, M. A. (1975). Chemical yield optimization of the benzene synthesis for radiocarbon dating. International Journal of Applied Radiation and Isotopes 26,67&682. Tanzawa, T. & Gardiner, W. C. (1978). Thermal decomposition of acetylene. Proceedings qfl7th Symposium on Combustion, 563-573. Wand, J. 0. (1980). Microsample preparation for radiocarbon dating. Unpublished D.Phil thesis. University of Oxford. White. N. R. (1981). Isotopic fractionation in accelerator based radiocarbon dating. Proceedings of 2nd Symposium on Accelerator Muss Spectrometr.v, Argonne. 359-372.