Although the reliability of 14C dates of bone has increased greatly since AMS methods permitted better pretreatment on smaller samples, most old, badly contaminated or severely weathered bone still give serious problems. Several groups have recently proposed improvements to sample purification methods, often supported by a number of 14C measurements. We present here an overview of these improvements. The issue is complicated by the following:

1. 1. Different problems are presented depending on age, preservation and degree of contamination of bone.

2. 2. Methods may or may not be developed with routine application in mind.

3. 3. Determining the conditions for which any method can be regarded as reliable is not at all straightforward.

We describe the use of collagenase to digest ancient collagen samples to obtain a peptide mixture, and the use of reversed-phase chromatography to purify these peptides. The isolation of the tripeptides glycine-proline-hydroxyproline and glycine-proline-alanine is especially encouraging.

Osteocalcin, a non-collagen bone-matrix protein, has been examined as a possible source of autochthonous 14C data in fossil bones where collagen has been seriously degraded. Extraction procedures for osteocalcin yield a well-characterized product that can be clearly distinguished from collagen. The Gla content indicates that osteocalcin is present in the fossil bones at levels similar to the range present in modern bone. However, it appears to be extracted primarily as proteolytic polypeptide fragments rather than as an intact protein. Concordant 14C determinations are obtained on osteocalcin and gelatin extracts from the same bone when the collagen is relatively well preserved. However, increasing discordances in the 14C values of the osteocalcin and gelatin fractions are associated with reduced concentrations of the gelatin extract in the bone.

We describe the operation of a commercial system as modified for preparation of CO2 for the CO2 source. AMS samples are automatically combusted in a CHN analyzer, and stable isotope measurements are made on line. We describe the performance of this equipment, with reference to yield, sample contamination, memory effect, accuracy of isotope measurement, convenience and cost. We discuss the current status of dating using the CO2 source. This is the only source in operation at Oxford, and has been in routine dating since September 1989. We assess the practicalities of operation, including the latest measurements on background, memory, sample-size requirements and operating schedules. We also describe modifications to the sputter beam optics and to the gas handling systems.

The Radiological Dating Laboratory in Trondheim relatively often dates samples with ages >30 ka BP. Contaminated background materials are known to affect the accuracy of very old dates. We have found, by measurements of different materials, that such contamination is small when using our conventional gas proportional counting (GPC) system. We have also studied contamination levels of our target preparation for 14C accelerator mass spectrometry (AMS) dating in Uppsala. A significant lower background is obtained for Icelandic double spar than for marbles, probably due to a crystal structure of the double spar that is more insensitive to contaminating processes. The background for combusted samples is at the same level as for samples of double spar, indicating that additional 14C contamination due to combustion is negligible. Dates obtained on interstadial samples (T >30 ka bp) by both GPC and AMS agree well.

I describe a simple, rapid and inexpensive method for converting CO2 samples into AMS target graphite. The technique is applicable for both small- and large-scale production facilities. With some modification, the method is suitable for use with very small samples.

We conducted a study of relative gas composition changes of CO2, CO and CH4 during the formation of graphite targets using different temperatures, catalysts and methods. Reduction with H2 increases the reaction rate without compromising the quality of the AMS target produced. Methane is produced at virtually any temperature, and the amount produced is greater at very low temperatures. The reduction of CO to graphite is very slow when H2 is not included in the reaction.

Graphite is the most common type of target for 14C accelerator mass spectrometry (AMS). It is readily produced by catalytic reduction of CO2, but the presence of a small amount of impurities (e.g., sulfur compounds) may retard the reaction. We have tested some techniques to find a method that reduces the impurity content of CO2 produced by combustion of organic material. We found that using water during combustion reduces the average time for graphite conversion of CO2 from organic matter from >3 h to ca. 2 h. This is the time for graphite production from CO2 obtained by acid hydrolysis of calcite. Measurements of known-age and background samples show that this combustion method neither changes the isotopic ratios nor introduces any additional background.

Many factors influence the preparation and quality of graphite targets for 14C accelerator mass spectrometry (AMS). We identified four factors (sample size, H2 pressure, catalyst temperature and pretreatment time) as potentially critical, and investigated their effects on two particular characteristics: the integrated rates of CO2 reduction (to graphite) and methane production. We used a 2-level fractional factorial experimental design and determined chemical reduction yield rates through manometry and partial pressure monitoring of residual gases by mass spectrometry.

Chemical reduction yield rates ranged from 0.2% to 6.2% per hour. With respect to their influence on percent yield rate, the factors we studied were ordered as: sample size > level of hydrogen > pretreatment of the catalyst. The temperature of the catalyst, and the sample size × hydrogen (2-factor) interaction, were only marginally influential. Other interactions did not appear to be significantly important. We estimated uncertainty in the order of influence and magnitudes of the effects by the Monte Carlo method of error propagation.

We observed significant methane production in only one experiment, which suggests that methane originates from indigenous carbon in untreated iron catalyst only in the presence of hydrogen and only at thermodynamically favorable temperatures. This exploratory investigation indicates that factorial design techniques are a useful means to investigate multivariate effects on the preparation and quality of AMS graphite targets.

The application of AMS to the detection of 14C makes possible a new class of sensitive experiments in molecular biology. Such experiments inherently produce large numbers of samples for the determination of biological variability in molecular interactions. The samples vary in 14C concentration over many orders of magnitude. We added TiH2 to aid the reduction of CO2 by zinc in a sealed tube to reproducibly make graphite without sample cross-contamination. The CO2 is transferred from a combustion tube to the reaction tube through a disposable plastic manifold. The sealed tubes are heated to a single-reaction temperature in a muffle furnace. The process is complete within 5 h. Bulk isotopic fractionation in the finished graphite is less than 0.5%.

Calculations show the possibility of detecting 14C remaining after 10 or 11 decay half-lives in natural materials, such as wood, using commercially available liquid scintillation (LS) detectors. Assuming in-situ14C production has contributed insignificantly to the measured 14C, one can calculate finite ages approaching 70 ka. In practice, defensible finite age determinations involve careful considerations of several critical elements in the procedure. These critical elements are: 1) the integrity of the sample itself, in terms of younger contaminants and of in situ-produced 14C; 2) the availability of “dead” background material; 3) chemical blank in laboratory preparation of, in this case, benzene; and 4) stability of background and efficiency of the LS counting system. High-sensitivity 14C dating reveals a low level of memory effect 14C in benzene synthesized in the laboratory from anthracite or marble. This level of 14C activity is equivalent to that found in 53 ka old wood, and thus, is not distinguishable from petrochemical benzene used in routine dating. If careful control of laboratory conditions can maintain this 14C blank constant, reproducible dating beyond 53 ka would be possible. Although we have not completed a systematic analysis of the origins of memory effect, lithium reactors used in acetylene production and organic solvents in wood pretreatment are likely sources.

We consider the suitability of commonly used Poisson counting statistics applied to background count rates measured in the new generation of low-background Packard liquid scintillation spectrometers. We also investigate the stability of these systems over long time intervals. Undetected instability will result in an underestimation of the precision of any result (i.e., the calculated error will be too small), and, in the presence of a systematic source, could lead to inaccurate results. The work described here forms only a small part of a project to investigate the statistical criteria that should be applied to the performance of such counters. The procedures to be discussed here include the Poisson index of dispersion, x and s control charts and the MSSD test for detection of drift. These are illustrated on background count rates derived from the Packard 2260XL and 2000CA/LL.

A new, improved photomultiplier has been incorporated and a more efficient active guard installed in a single photomultiplier liquid scintillation counter. A special quartz vial was designed to improve the photoelectron cathode index (photoelectrons/keV), and thus, to reduce background. Four photomultipliers were installed so that four samples can be counted simultaneously. The measurements presented here cover background spectra and counting efficiency obtained under various conditions in the laboratory. These measurements and derived parameters, such as the figure of merit (E2/B) and the factor of merit (S/√B), are used to evaluate the system performance. Also presented are the results of measurements on the cathode index and further background spectra, obtained at an underground site.

In China, the development of 14C dating is closely related to that of liquid scintillation counting (LSC). New demands are continuously made on LSC by 14C dating, and at the same time, the development of LSC promotes 14C dating. Benzene synthesis for 14C sample preparation was first developed in China in 1974 by the 14C Laboratory of the Department of History, Peking University. This has laid the foundation for 14C dating by β counting with LS counters. Initially, we used a modified conventional LS counter. Despite its poor detection efficiency and high background, it did make a great contribution to 14C dating and LSC in China. Additional 14C laboratories were established for the fields of archaeology, geology, geography, vertebrate paleontology, marine geology and seismology, for which 14C dating is becoming an important research tool. At present, over 50 14C laboratories have been established in China; 90% of the labs use LS counters for β counting, most of which are manufactured in China. 14C dating in China has been primarily concerned with sample synthesis techniques. Since 1979, we have developed three types of LS counters. One is of conventional design; the others are with anticoincidence shielding. We describe here an anticoincidence-shielded LS counter.

Considerable research has been carried out in this laboratory on the use of Packard scintillation counters employing burst-counting circuitry. The nature of the pulse-shape discrimination circuit has led to redefining our sample vialing philosophy and scintillation cocktail optimization. Results presented here demonstrate: 1) the enhancements in efficiency that are achievable using a two-component cocktail compared with the use of a primary fluor only; 2) the development of a cocktail in which efficiency and quenching are relatively unaffected by moderate variations in fluor concentrations and ratio; 3) the rationale behind the use of screw-cap vials rather than sealable ampules; 4) the advantage of nickel/chromium-plated brass caps over standard plastic caps.

I have made an evaluation of 0.3-ml minivials for 14C dating of small samples by liquid scintillation counting of benzene. A calibrated wood standard was diluted by varying amounts of ancient CO2, with synthesized benzene counted in both conventional 3.0-ml vials and 0.3-ml minivials in a 1220 Quantulus. The accuracy and precision of results are compared for samples ranging in weight from 50 to 240 mg of carbon. I examined two significant potential problems associated with handling small samples, namely, memory effects within the vacuum system, and signal within the dilution gas. Although accurate radiocarbon dates can be obtained using either standard vials or minivials, minivials are more suitable for dating small samples because they are less influenced by these sources of error.

We evaluate high-purity synthetic silica vials in both delrin and active plastic holders (Pico adapters™) for 14C dating, using liquid scintillation counting of benzene. We have designed synthetic silica vials in the form of simple cylinder-cells based on the standard 7-ml vial. We have also designed a delrin holder for supporting 7-ml silica or glass vials. We compare the counting efficiency and background of the silica vials with Teflon, plastic and low-K glass vials for both delrin holders and Pico adapters in the 1220 Quantulus and Packard Tri-Carb 2660 XL, fitted with a plastic detector guard. In the 1220 Quantulus, synthetic silica vials in Pico adapters have the highest figure of merit (FM), closely followed by silica in delrin holders and Teflon. In the Packard Tri-Carb 2660 XL, plastic vials in Pico adapters give the highest FM.