We describe the design, construction, and testing of a new, simplified in situ radiocarbon extraction system at the University of Arizona. Blank levels for the new system are low ((234 ± 11) x 103 atoms (1 σ; n = 7)) and stable. The precision of a given measurement depends on the concentration of 14C, but is typically <5% for concentrations of 100 x 103 atoms g–1 or more. The new system is relatively small and easy to construct, costs significantly less than the original in situ14C extraction system at Arizona, and lends itself to future automation.

Chemical pretreatment and extraction techniques have been developed previously to extract in situ cosmogenic radiocarbon (in situ14C) from quartz and carbonate. These minerals can be found in most environments on Earth, but are usually absent from mafic terrains. To fill this gap, we conducted numerous experiments aimed at extracting in situ14C from olivine ((Fe,Mg)2SiO4). We were able to extract a stable and reproducible in situ14C component from olivine using stepped heating and a lithium metaborate (LiBO2) flux, following treatment with dilute HNO3 over a variety of experimental conditions. However, measured concentrations for samples from the Tabernacle Hill basalt flow (17.3 ± 0.3 ka4) in central Utah and the McCarty's basalt flow (3.0 ± 0.2 ka) in western New Mexico were significantly lower than expected based on exposure of olivine in our samples to cosmic rays at each site. The source of the discrepancy is not clear. We speculate that in situ14C atoms may not have been released from Mg-rich crystal lattices (the olivine composition at both sites was ∼Fo65Fa35). Alternatively, a portion of the 14C atoms released from the olivine grains may have become trapped in synthetic spinel-like minerals that were created in the olivine-flux mixture during the extraction process, or were simply retained in the mixture itself. Regardless, the magnitude of the discrepancy appears to be inversely proportional to the Fe/(Fe+Mg) ratio of the olivine separates. If we apply a simple correction factor based on the chemical composition of the separates, then corrected in situ14C concentrations are similar to theoretical values at both sites. At this time, we do not know if this agreement is fortuitous or real. Future research should include measurement of in situ14C concentrations in olivine from known-age basalt flows with different chemical compositions (i.e. more Fe-rich) to determine if this correction is robust for all olivine-bearing rocks.

The radiocarbon content of atmospheric CO2 (Δ14CO2) has long been of interest to atmospheric and Earth system researchers. Recent improvements in 14C measurement precision and reduction in sample size requirements have now made it possible to measure Δ14CO2 within existing trace gas sampling networks, most notably as a method to quantify recently added fossil-fuel-derived CO2 in the atmosphere. At INSTAAR, in collaboration with NOAA/ESRL, ∼600 atmospheric samples from around the globe are prepared each year, and that number is anticipated to grow in connection with various monitoring and data assimilation efforts. To accommodate the growing demand and reduce per sample costs, we developed an automated extraction system to quantitatively isolate CO2 from whole air for AMS 14C analysis. Twenty samples can be extracted in 1 fully automated run, taking 10–12 hr to complete and requiring only about 1 hr of operator time, a substantial improvement over the manual extraction system. CO2 is extracted cryogenically by flowing the whole air over a liquid nitrogen trap, after first removing water in a trap at –85 °C. Large volume vacuum lines are used to extract ∼30 μmol of CO2 in less than 10 min, keeping contamination from leaks to a minimum and allowing rapid processing and greater throughput. δ13C measurements on the resultant CO2 demonstrate that extraction is quantitative, and extractions of 14C-free air show that no significant modern contamination occurs. Replicate analyses of standard materials indicate that both mean values and precision are comparable to those for the manual extraction system.

To be able to successfully measure radiocarbon with accelerator mass spectrometry (AMS) in atmospheric aerosol samples, graphitization of small sample sizes (<50 μg carbon) must provide reproducible results. At Lund University, a graphitization line optimized for small samples has been constructed. Attention has been given to minimize the reduction reactor volume and each reactor is equipped with a very small pressure transducer that enables constant monitoring of the reaction. Samples as small as 25 μg of carbon have been successfully analyzed, and the mass detection limit of the system has probably not been reached.

Many previous studies on the sample preparation of various kinds of radiocarbon dating samples by accelerator mass spectrometry (AMS) have been examined at KIGAM (Korea Institute of Geoscience and Mineral Resources) and our own procedures have been established. Furthermore, an automated reduction system has been developed. The volume of the reduction region was minimized to improve the reduction yield, and air-actuated pneumatic valves and solenoid arrays were used for computer control of the system. Operation of all the valves and vacuum pumps and signals from the temperature sensors and pressure gauges were interfaced to a personal computer with an A/D board. A computer program was also developed to perform automatic operation of the reduction system. This system consistently shows a higher reduction yield than 90%. The reduction time of the system is currently 140 min.

In this paper, we describe improvements to the in situ cosmogenic radiocarbon extraction system at SUERC made since 2004, highlighting the factors that potentially control the reduction of analytical variability. We also present new results on system blanks and of measurements of in situ14C in shielded quartz and a surface quartz sample used at the University of Arizona as an in situ14C standard (PP-4). The SUERC in situ14C extraction system was built in 2001 and is based on a combustion technique following the design of the extraction system at the University of Arizona. Our preliminary results suggest that the continuous running of the extraction system and the monitoring of gas collecting time and of the temperature of the cryogenic traps used in the gas cleaning steps are key to maintaining low and stable system blanks. Our latest average system blank is 2.02 ± 0.23 x 10514C atoms. This is consistent with those recently published by the University of Arizona and ETH in situ14C labs. Measurements of in situ14C concentrations in sample PP-4 yield an average of 3.82 ± 0.23 x 105 atoms g–1 quartz, again consistent with published values.

Since iron artifacts generally contain trace amounts of carbon, an iron sample needs to be relatively large, as compared to other materials, and a specially designed combustion system is required. An elemental analyzer (EA) was used for the combustion of iron without any special chemical treatment. CO2 gas with 1 mg of carbon was obtained from the combustion of an iron artifact by using an EA and reduced to graphite for accelerator mass spectrometry (AMS) measurement. In this work, AMS dating results done at the Korea Institute of Geoscience and Mineral Resources (KIGAM) for several ancient iron artifacts are presented and compared with independently estimated ages. This method was found to be useful for the pretreatment of iron artifacts that contained >0.1% carbon. A simple pretreatment method using an EA was also applied to calcium carbonate (CaCO3) samples. Samples were preheated overnight at 100–300 °C, without any special chemical treatment. This removed modern CO2 contamination and the background level decreased to a comparable value measured in samples treated with phosphoric acid under vacuum.

Coastal archaeological sites that lack organic remains for radiocarbon dating are often abundant sources of molluscan shells. As a substitute for materials such as bone and charcoal, shells can be analyzed with 14C dating to determine a site's age. Despite their being convenient, non-mobile archaeological artifacts, molluscan shells are plagued by several issues, including carbonate remodeling, in which aragonite in shells is converted to calcite as predicted by thermodynamics. We present here a carbonate density separation technique that addresses the issue of carbonate remodeling. Using a density fractionation with bromoform, aragonite concentrations are enriched in shells that have undergone significant remodeling. The technique has been applied to archaeological shells and has returned dates that are younger than those previously determined for the same shells.

A method of thermal and acid treatments was developed at the Archaeo-metallurgy Laboratory of Hongik University in Korea to extract carbon from cast iron, and carbon objects thus prepared from cast iron artifacts of ancient Korea were dated at the University of Arizona's AMS Facility. The thermal treatments consist of heating a specimen to ∼1000 °C in a controlled environment with reduced oxygen potential, then cooling it rapidly to room temperature. The heating causes the cementite phase in white cast iron to be graphitized and the quenching suppresses pearlite formation. The specimen then consists of flakes of graphite embedded in a matrix of martensite. The next stage of the treatment is to dissolve the martensite matrix in a solution of nitric and hydrochloric acids to release the graphite as a powder. This material is then cleaned, dried, and pressed into target holders for accelerator mass spectrometry (AMS) analysis. The method was applied to a collection of artifacts from the Korean Three Kingdoms period (about AD 300–668) and the AMS results were compared with chronological estimates from other means.

The Keck Carbon Cycle AMS facility at the University of California, Irvine (KCCAMS/UCI) has developed protocols for analyzing radiocarbon in samples as small as ∼0.001 mg of carbon (C). Mass-balance background corrections for modern and 14C-dead carbon contamination (MC and DC, respectively) can be assessed by measuring 14C-free and modern standards, respectively, using the same sample processing techniques that are applied to unknown samples. This approach can be validated by measuring secondary standards of similar size and 14C composition to the unknown samples. Ordinary sample processing (such as ABA or leaching pretreatment, combustion/graphitization, and handling) introduces MC contamination of ∼0.6 ± 0.3 μg C, while DC is ∼0.3 ± 0.15 μg C. Today, the laboratory routinely analyzes graphite samples as small as 0.015 mg C for external submissions and ≅0.001 mg C for internal research activities with a precision of ∼1% for ∼0.010 mg C. However, when analyzing ultra-small samples isolated by a series of complex chemical and chromatographic methods (such as individual compounds), integrated procedural blanks may be far larger and more variable than those associated with combustion/graphitization alone. In some instances, the mass ratio of these blanks to the compounds of interest may be so high that the reported 14C results are meaningless. Thus, the abundance and variability of both MC and DC contamination encountered during ultra-small sample analysis must be carefully and thoroughly evaluated. Four case studies are presented to illustrate how extraction chemistry blanks are determined.

Charcoal is the result of natural and anthropogenic burning events, when biomass is exposed to elevated temperatures under conditions of restricted oxygen. This process produces a range of materials, collectively known as pyrogenic carbon, the most inert fraction of which is known as black carbon (BC). BC degrades extremely slowly and is resistant to diagenetic alteration involving the addition of exogenous carbon, making it a useful target substance for radiocarbon dating particularly of more ancient samples, where contamination issues are critical. We present results of tests using a new method for the quantification and isolation of BC, known as hydropyrolysis (hypy). Results show controlled reductive removal of non-BC organic components in charcoal samples, including lignocellulosic and humic material. The process is reproducible and rapid, making hypy a promising new approach not only for isolation of purified BC for 14C measurement but also in quantification of different labile and resistant sample C fractions.

Accurate accelerator mass spectrometry (AMS) measurements rely on standards with well-known isotopic ratios. For radiocarbon measurements, a number of standards with different properties are commercially available, of which the IAEA-C6 sucrose standard with a 14C value of 150.61 pMC is the most active. When analyzing biological samples resulting from studies using 14C-labeled substances, the activity content can be up to 100 times this value. Thus, there is a need for a standard material with higher activity content than IAEA-C6 for making accurate AMS measurements on this type of sample. This paper describes the attempts of producing a standard with an activity content of about 10 times modern carbon. The material chosen has to be chemically inert, preferably non-toxic, commercially available in 14C-labeled form, and the activity must be homogeneously distributed within the material. Two different standard materials were considered: urea and sucrose. Sucrose was chosen for the new standard, since it is non-toxic, inexpensive, and organic and on combustion, forms only carbon dioxide (CO2) and water (H2O). In this paper, we discuss our experience in the production and homogeneity analysis of this material, from the crystallization of the sucrose solution to the graphitization of the samples. When using an online combustion method and a septa-sealed vial reduction method, the AMS measurements indicated that the activity was not homogeneously distributed throughout the material. Contrary to this, measurements of the sucrose solution prior to recrystallization indicated that the activity was more homogeneously distributed before than after the recrystallization. In order to determine whether the inhomogeneity depended on the graphitization method (i.e. the combustion or the reduction method) or on the material itself, 3 different graphitization methods and 2 different methods of recrystallization were tested.

The main methods applied to clean plant material for radiocarbon dating are not compound-specific and generally remove only the easily exchangeable components by an acid-base-acid sequence and additional optional steps like Soxhlet extraction to remove resins and oxidative bleaching with NaClO2. The products are normally clean enough for standard 14C measurement, but in some cases it is desirable to have pure cellulose, which remains unchanged and immobile over longer time ranges, better representing the original plant material. In this work, 2 more compound-specific but still simple methods were tested to separate the cellulose from wood. The viscose method is based on the xanthification process used in the textile industry, where the alkali-cellulose with CS2 forms a soluble cellulose xanthate, which is then extracted and cellulose is recovered. The second procedure is based on the wood/cellulose dissolution in ionic liquid 1-butyl-3-methylimidazolium chloride [BMIM]Cl, when the dissolved cellulose could be precipitated again by simply adding a water-acetone mixture. This process was recently reported, but still not used in sample preparation procedures for 14C dating.

We have compared accelerator mass spectrometry (AMS) radiocarbon results on wood samples at or near the limit of 14C dating, pretreated with a standard acid-base-acid (ABA) protocol, with those obtained from cellulose prepared from the same samples by several modifications of the Jayme-Wise cellulose extraction method (Green 1963). These tests were carried out to determine the most efficient way to ensure low backgrounds in 14C measurements of well-preserved ancient wood samples.

The reaction conditions for the graphitization of CO2 with hydrogen were optimized for a fast production of high-quality carbon samples for accelerator mass spectrometry (AMS) measurement. The iron catalyst in use is first oxidized by heating with air to remove possible carbon and other impurities and then after evacuation reduced back to iron with hydrogen in several flushing steps to remove any iron oxide. The optimum conditions for a fast graphitization reaction were experimentally determined by changing the reaction temperatures and the H2/CO2 ratio. The resulting graphite samples were measured by AMS to find the smallest isotopic changes (δ13C) at a minimum of molecular fragment formation (13CH current). The improvements are based on thermodynamic data and are explained with Baur-Glaessner diagrams.

The minimum size of radiocarbon samples for which reliable results can be obtained in an accelerator mass spectrometry (AMS) measurement is in many cases limited by carbon contamination introduced during sample preparation (i.e. all physical and chemical steps to which samples were subjected, starting from sampling). Efforts to reduce the sample size limit down to a few μg carbon require comprehensive systematic investigations to assess the amount of contamination and the process yields. We are introducing additional methods to speed up this process and to obtain more reliable results. A residual gas analyzer (RGA) is used to study combustion and graphitization reactions. We could optimize the reaction process at small CO2 pressures and identify detrimental side reactions. Knowing the composition of the residual gas in a graphitization process allows a reliable judgment on the completeness of the reaction. Further, we use isotopically enriched 13C (≥98% 13C) as a test material to determine contamination levels. This offers significant advantages: 1) The measurement of 12C/13C in CO2 is possible on-line with the RGA, which significantly reduces turnaround times compared to AMS measurements; 2) Both the reaction yield and the amount of contamination can be determined from a single test sample.

The first applications of isotopically enriched 13C and the RGA have revealed that our prototype setup has room for improvements via better hardware; however, significant improvements of our sample processing procedures were achieved, eventually arriving at an overall contamination level of 0.12 to 0.15 μg C during sample preparation (i.e. freeze-drying, combustion, and graphitization) of μg-sized samples in aqueous solution, with above 50% yield.

Twenty years after the first International Radiocarbon Database Workshop, and 13 yr after the setup of a preliminary structure for a worldwide database on 14C dates of soils, sound reasons and excuses for not establishing a real and globally accessible database have diminished. Climate change itself is widely accepted as reality now, and the strong demand of the modeling community for reliable data of the carbon pool—especially in soils—has been maintained. With the steadily increasing capacity of 14C dating facilities, these data can be and are produced. Nevertheless, they still cannot be accessed easily and equally enough. Now, decreased costs of hardware and recent developments of the internet enable the IRSDB to be implemented, in a joint effort. As a seed, a test server has been set up, with open-source software, housing the database in alpha-stage, a web interface, and a community portal. Thus, the development of the design as well as the data input of the database is done in close collaboration of the users of the database, the laboratories, soil scientists, archaeologists, modelers, other scientists, and interested laypersons. In order to guarantee the longtime independence of the availability and usability of the database from vendors or changing standards, only widely used open-source software and open standards are used. Therefore, the development of plug-ins for data input from laboratory databases or output to different required formats as well as interfaces to GIS and other software is possible. A version control system takes care of the integrity of the data.

The historical evolution of an ancient forest that developed at Gaoyao, south China, can be divided into 4 stages of radiocarbon intervals (1.1–1.5, 2.0–3.5, 3.6–4.0, and 4.3–4.9 ka) in which the last 3 stages all developed in a wetland and formed humic layers of 2.0, 0.5, and 0.7 m depth, respectively. The humic layers were interrupted by 2 white-gray silty clay layers that most likely formed during climate fluctuations. Four drought events were identified during the evolution of the ancient forest, occurring around 4.3, 3.6, 2.0, and 1.1 ka, respectively, with durations of ∼1000 14C yr. These events are consistent with other records both in low- and high-latitude areas, in particular with the little ice ages occurring since the mid-Holocene. Precipitation likely increased from 5.0 to 3.6 ka in south China, then decreased, which is probably the main cause for the development as well as the demise of the ancient forest.

This paper examines the carbon isotopes (13C, 14C) of soil organic carbon (SOC) and soil CO2 from an evergreen broadleaf forest in southern China during the rainy season. The distribution of SOC δ13C, and SOC content with depth, exhibits a regular decomposition of SOC compartments with different turnover rates. Labile carbon is the main component in the topsoil (0–12 cm) and has a turnover rate between 0.1 and 0.01 yr–1. In the middle section (12–35 cm), SOC was mainly comprised of mediate carbon with turnover rates ranging between 0.01 and 0.025. Below 35 cm depth (underlayer section), the SOC turnover rate is slower than 0.001 yr–1, indicating that passive carbon is the main component of SOC in this section. The total production of humus-derived CO2 is 123.84 g C m–2 yr–1, from which 88% originated in the topsoil. The middle and underlayer sections contribute only 10% and 2% to the total humus-derived CO2 production, respectively. Soil CO2 δ13C varies from –24.7‰ to –24.0‰, showing a slight isotopic depth gradient. Similar to soil CO2 δ13C, Δ14C values, which range from 100.0‰ to 107.2‰, are obviously higher than that of atmospheric CO2 (60–70‰) and SOC in the middle and underlayer section, suggesting that soil CO2 in the profile most likely originates mainly from SOC decomposition in the topsoil. A model of soil CO2 Δ14C indicates that the humus-derived CO2 from the topsoil contributes about 65–78% to soil CO2 in each soil gas sampling layer. In addition, the humus-derived CO2 contributes ∼81% on average to total soil CO2 in the profile, in good agreement with the field observation. The distribution and origin of soil 14CO2 imply that soil CO2 will be an important source of atmospheric 14CO2 well into the future.

A new continuous sediment core (PY608W-PC; 3.8 m length) for reconstruction of climatic and environmental changes in the southeastern Tibetan Plateau was taken from the eastern part of Lake Pumoyum Co in August 2006. Sediment layers of the lower part of PY608W-PC (380–300 cm depth) were composed mainly of relatively large plant residues (up to ∼3 cm in length) with an admixture of fine sand and sandy silt. The large plant residues disappeared at ∼300–290 cm depth in core PY608W-PC and were replaced by silt-silty clay. The large plant residues from the lower part of PY608W-PC could be aquatic, because the plant residues were extremely enriched in 13C (up to –3.0‰, −5.6 ± 2.3‰ on average). On the other hand, the plant residue concentrates (PRC fractions) from the upper part of the core (290–0 cm in depth) could be terrestrial C3 plants (δ13C = –21.8 ± 1.7‰ on average). Radiocarbon dating was performed on the large plant residues and PRC fractions from the PY608W-PC sediment core, which represented the chronology from ∼19,000 cal BP to present.