My path to the invention of accelerator mass spectrometry—now just called AMS—was quirky and extraordinary, and it is a saga worth telling, particularly for young people who may have an oversimplified image of how progress in science is actually made. It was an adventurous journey, and like many adventures, it was often uncomfortable, haphazard, and frequently characterized by a feeling of being lost. In retrospect, the only reason I set out on this journey was my belief that I didn't know much about finding my path in physics, and that the best I could do was to follow the lead of the great physicist Luis Alvarez. He had an incredible ability to create new and ingenious projects that led in new directions, and I wanted to understand how he did that. So I had decided that I would work with him on any new idea he came up with, even if it inconvenienced the rest of my life. And true to my expectations, while working with Luie (that's what he wanted everyone to call him), I often felt like Odysseus, tossed between distant shores by capricious gods.

This is the written version of a retrospective talk given at the 20th Radiocarbon Conference in Hawaii, 2009. As requested, this is not a scholarly work, but simply a personal, subjective account of how a good experiment in the field of radiocarbon dating came about. It differs from the talk itself only in that I have eliminated some sideline commentary so that I can add a few thoughts originally intended for the end, when I had talked myself out of time.

A new and unique radiocarbon accelerator mass spectrometry (AMS) facility has been constructed at the Woods Hole Oceanographic Institution. The defining characteristic of the new system is its large-gap optical elements that provide a larger-than-standard beam acceptance. Such a system is ideally suited for high-throughput, high-precision measurements of 14C. Details and performance of the new system are presented.

The suppression of interferences from atomic and molecular isobars is a key requirement for the extension of accelerator mass spectrometry (AMS) to the analysis of new cosmogenic isotopes and for increasing the range of applications for small AMS systems. In earlier work, it was shown that unwanted isobars can be eliminated by anion-gas reactions (Litherland et al. 2007). Recently, a prototype system in which such reactions could be applied to ions from an AMS ion source, the Isobar Separator for Anions (ISA), was described (Eliades et al. 2009). This system decelerates the beam of rare anions from keV to eV energies, guides them through a single radiofrequency quadrupole (RFQ) gas cell, and re-accelerates them for further analysis in a 2.5MV AMS system. Tests of this system with Cl and S anions and NO2 gas showed a suppression of S with respect to Cl of over 6 orders of magnitude, with a transmission of ∼30% for the Cl beam. In this work, results of the analysis of a range of standard reference materials are reported; these show the linearity of the system for measuring the 36Cl/35Cl ratio over a span of 2 orders of magnitude. Further tests, using the AMS system as a diagnostic tool, have provided clues about the loss of Cl at higher cell pressure and the nature of the residual low level of S transmission. These lead to the assessment of various gases for cooling the Cl– beam. Suppression measurements for 41K in the analysis of 41Ca, using NO2 as a reaction gas, are also discussed. These preliminary measurements have provided data for the development of a more advanced system with separate cooling and reaction cells.

A 1MV AMS was installed in KIGAM (Korea Institute of Geoscience and Mineral Resources). After 4 months of installation, the AMS started normal operation from January 2008. This multi-element AMS was developed by HVEE to measure 14C, 10Be, and 26Al. The results of an acceptance test demonstrate that this machine is capable of routine 14C age dating and of measurements of other radioisotopes in terms of accuracy and precision as well as the background level. After installation, an investigation aimed at determining the stable operating conditions was conducted, and background levels were determined to be as low as 10–15 for 14C and 10–14 for 10Be and 26Al.

The prototype mini carbon dating system (MICADAS) at ETH Zurich has been in routine operation for almost 2 yr. Because of its simple and compact layout, setting up a radiocarbon measurement is fast and the system runs very reliably over days or even weeks without retuning. The stability of the instrument is responsible for the good performance in highest-precision measurements where results of single samples can be reproduced within less than 2‰. The measurements are described and the performance of MICADAS is demonstrated on measured data.

In 2003, a National Electrostatics Corporation (NEC) 5MV tandem accelerator mass spectrometer was installed at SUERC, providing the radiocarbon laboratory with 14C measurements to 4–5‰ repeatability. In 2007, a 250kV single-stage accelerator mass spectrometer (SSAMS) was added to provide additional 14C capability and is now the preferred system for 14C analysis. Changes to the technology and to our operations are evident in our copious quality assurance data: typically, we now use the 134-position MC-SNICS source, which is filled to capacity. Measurement of standards shows that spectrometer running without the complication of on-line δ13C evaluation is a good operational compromise. Currently, 3‰ 14C/13C measurements are routinely achieved for samples up to nearly 3 half-lives old by consistent sample preparation and an automated data acquisition algorithm with sample random access for measurement repeats. Background and known-age standard data are presented for the period 2003–2008 for the 5MV system and 2007–2008 for the SSAMS, to demonstrate the improvements in data quality.

In accelerator systems, beam lines are generally equipped with diagnostic elements, such as Faraday cups and beam profile monitors (BPM), to optimize beam transport. These diagnostic elements, or at least commercial ones, are designed to only work with continuous beams, and their typical maximum sensitivity is about few tens of pA. Thus, in the case of diagnosis of rare isotope beams in accelerator mass spectrometry (AMS), Faraday cups and BPMs are not suitable on the high-energy side of the tandem accelerator, after energy-mass-charge analysis. For example, in 14C AMS, even for a modern sample, the expected counting rate is a few tens of Hz; in these conditions, a commercial BPM cannot be used. On the other hand, checking the shape and the position of the rare isotope beam hitting the detector can be important in order to better identify signals in the detector itself, thus also helping in reducing the measurement background.

This paper presents a prototype BPM especially designed for low-intensity beams. The BPM is based on a multiwire proportional chamber characterized by 2 grids of anode wires, oriented perpendicular to each other in order to measure both the x and the y coordinates of the particle impact point. Details about the design and the electronics of the device are given, and the first test measurements are discussed.

Submicron-sized airborne particulate matter (PM) is not collected well on regular quartz or glass fiber filter papers. We used a micro-orifice uniform deposit impactor (MOUDI) to fractionate PM into 6 size fractions and deposit it on specially designed high-purity thin aluminum disks. The MOUDI separated PM into fractions 56–100, 100–180, 180–320, 320–560, 560–1000, and 1000–1800 nm. Since the MOUDI has a low flow rate (30 L/min), it takes several days to collect sufficient carbon on 47-mm foil disks. The small carbon mass (20–200 μg C) and large aluminum substrate (∼25 mg Al) present several challenges to production of graphite targets for accelerator mass spectrometry (AMS) analysis. The Al foil consumes large amounts of oxygen as it is heated and tends to melt into quartz combustion tubes, causing gas leaks. We describe sample processing techniques to reliably produce graphitic targets for 14C AMS analysis of PM deposited on Al impact foils.

Actinide isotopes are present in environmental samples at ultra-trace levels (236U concentration is quoted to be on the order of pg/kg or fg/kg). Their detection requires the resolution of mass spectrometry (MS) techniques, but only accelerator mass spectrometry (AMS) has the sensitivity required. In order to perform the isotopic ratio measurements of actinides, such as 236U/238U, an upgrade of the Center for Isotopic Research on Cultural and Environmental Heritage accelerator (CIRCE) in Caserta, Italy, has been performed. The system was originally equipped for radiocarbon AMS measurements. The main difficulty of AMS measurement of 236U is the intense neighboring beam of 238U. Although most of the 238U ions are suppressed by means of magnetic and electrostatic elements, a small fraction of this intense beam can interfere with the rare isotopes. This paper reports the preliminary results of the 236U/238U isotopic ratio measurement limit (<5.6 x 10–11), aimed also to better understand the origin of background ions. For this purpose, a large 16-strip silicon detector providing spatial resolution has been used. In addition, calculations to assess the performance of the system obtained by adding a high-resolution time of flight-energy (TOF-E) detector are discussed.

Gas-accepting ion sources for radiocarbon accelerator mass spectrometry (AMS) have permitted the direct analysis of CO2 gas, eliminating the need to graphitize samples. As a result, a variety of analytical instruments can be interfaced to an AMS system, processing time is decreased, and smaller samples can be analyzed (albeit with lower precision). We have coupled a gas chromatograph to a compact 14C AMS system fitted with a microwave ion source for real-time compound-specific 14C analysis. As an initial test of the system, we have analyzed a sample of fatty acid methyl esters and biodiesel. Peak shape and memory was better then existing systems fitted with a hybrid ion source while precision was comparable. 14C/12C ratios of individual components at natural abundance levels were consistent with those determined by conventional methods. Continuing refinements to the ion source are expected to improve the performance and scope of the instrument.

We present a status report of the accelerator mass spectrometry (AMS) facility at the University of California, Irvine, USA. Recent spectrometer upgrades and repairs are discussed. Modifications to preparation laboratory procedures designed to improve sample throughput efficiency while maintaining precision of 2–3‰ for 1-mg samples (Santos et al. 2007c) are presented.

We have developed accelerator mass spectrometry (AMS) measurement techniques for ultra small-size samples ranging from 0.01 to 0.10 mg C with a new type of MC-SNICS ion source system. We can generate 4 times higher ion beam current intensity for ultra-small samples by optimization of graphite position in the target holder with the new ionizer geometry. CO2 gas graphitized in the newly developed vacuum line is pressed to a depth of 1.5 mm from the front of the target holder. This is much deeper than the previous position at 0.35 mm depth. We measured 12C4+ beam currents generated by small standards and ion beam currents (15–30 μA) from the targets in optimized position, lasting 20 min for 0.01 mg C and 65 min for 0.10 mg C. We observed that the measured 14C/12C ratios are unaffected by the difference of ion beam currents ranging from 5 to 30 μA, enabling measurement of ultra-small samples with high precision. Examination of the background samples revealed 1.1 μg of modern and 1 μg of dead carbon contaminations during target graphite preparation. We make corrections for the contamination from both the modern and background components. Reduction of the contamination is necessary for conducting more accurate measurement.

In the last decade, small and compact accelerator mass spectrometry (AMS) systems became available operating at terminal voltages of 1 MV and below. This new category of instruments has become competitive for radiocarbon detection to larger tandem accelerators and many of these instruments are successfully used for 14C dating or biomedical applications. The AMS group at ETH Zurich has demonstrated that small instruments can be built, which allow measurements also of other radionuclides such as 10Be, 26Al, 129I, and the actinides. 41Ca measurements can be performed with sufficient sensitivity for biomedical applications. A summary of recent developments made at the 500kV Pelletron in Zurich is given and its performance is compared with that of a commercial compact instrument from the company High Voltage Engineering Europe (HVEE) in Amersfoort, the Netherlands, operating at 1MV at CNA in Seville, Spain, as well as with that of larger AMS facilities. It turns out that the ion optics, stripper design, and the detection system are critical for the performance.

This study follows on from previous research at Perth, Scotland, in which we dated carbonized food residues removed from the external surface of rim sherds of cooking pots of London Sandy Shellyware pottery (Museum of London Pottery Fabric Code SSW). The 15 residues that were dated produced 14C ages between 910 ± 35 and 1085 ± 40 BP. We have now carried out radiocarbon measurements on similar residues from the same fabric obtained from the Billingsgate excavations in London and the Bryggen excavations in Bergen, Norway. The London and Bergen measurements gave age ranges of 905 ± 35 to 1115 ± 35 BP and 920 ± 35 to 1055 ± 35 BP, respectively, both very similar to the Perth age range. The measurements at each site are in agreement with our Bayesian model assumption that they belong to a single phase of activity. The model estimates the introduction of London Sandy Shellyware in London to cal AD 820–1020, in Perth to cal AD 930–1020, and in Bergen to cal AD 980–1030 (95% probability). Further modeling predicts that it fell out of use in the reverse order.

Recent efforts to precisely date the florescence of the Clovis culture in North America have been hampered by both practical and theoretical problems: 1) The era of Clovis expansion (about 11,200–10,700 BP or 13,200–12,700 cal BP) coincides with the gap between the anchored central European tree-ring sequence (back to 12,400 cal BP) and the floating Bølling-Allerød sequence; 2) Clovis seems to immediately precede the onset of the Younger Dryas (YD) stadial. The “black mats” of the US Southwest appear to mark the regional occurrence of this climatic downturn. However, the timing and means of long-distance propagation of this climatic event are not yet well understood. Greenland ice cores (GISP2, GRIP, and NGRIP) remain poorly synchronized, with a discrepancy of 100 to 250 yr for the date of onset (as late as 12,700 cal BP, or as early as 12,950 cal BP); 3) The YD onset was accompanied by a rapid drop of radiocarbon ages from 11,000 to 10,600 BP in less than a century. The mechanism causing this was probably a change in overturning circulation in the North Atlantic. Do variable Clovis ages, often from what appear to be single-occupation contexts, reflect this “cliff” effect, slightly earlier minor reversals during the late Allerød, or simply the practical limitations of precision of the 14C method? 4) Dates for Fishtail or Fell I sites (with fluted, stemmed points) in southern South America are statistically indistinguishable from Clovis dates in North America. Does this imply very rapid population expansion, diffusion of tool-making techniques through long-established local populations (as argued by Waters and Stafford 2007), or abnormally large interhemispheric 14C offsets? 5) Are recent ostensibly high-precision collagen-derived dates for Paleoindian-associated fauna (e.g. horse and mammoth) reliable? Are interlaboratory blind tests of the new filtration processes necessary?

Many of the Loch Tay crannogs were built in the Early Iron Age and so calibration of the radiocarbon ages produces very broad calendar age ranges due to the well-documented Hallstatt plateau in the calibration curve. However, the large oak timbers that were used in the construction of some of the crannogs potentially provide a means of improving the precision of the dating through subdividing them into decadal or subdecadal increments, dating them to high precision and wiggle-matching the resulting data to the master 14C calibration curve. We obtained a sample from 1 oak timber from Oakbank Crannog comprising 70 rings (Sample OB06 WMS 1, T103) including sapwood that was complete to the bark edge. The timber is situated on the northeast edge of the main living area of the crannog and as a large and strong oak pile would have been a useful support in more than 1 phase of occupation and may be related to the earliest construction phase of the site. This was sectioned into 5-yr increments and dated to a precision of approximately ±8–16 14C yr (1 σ). The wiggle-match predicts that the last ring dated was formed around 500 BC (maximum range of 520–465 BC) and should be taken as indicative of the likely time of construction of Oakbank Crannog. This is a considerable improvement on the estimates based on single 14C ages made on oak samples, which typically encompassed the period from around 800–400 BC.