Because of the extremely low natural levels of radiocarbon in the Earth's atmosphere (about 1 x 10-10 %), accurate measurement of 14C is not an easy task. These difficulties are further compounded, by the influence of cosmic and environmental background radiation, other radioisotopes being present, electronic noise and instability, and other factors. These background factors limit the accuracy, precision, and range of the radiocarbon dating method because finite ages can only be calculated where sample activity is at least 3 standard deviations above background activity (Gupta and Polach, 1985). Consequently, each of the components of the LS counting system needs to be totally optimised for low-level counting, so as to maintain high counting efficiency whilst significantly reducing the background. The 'total optimisation concept' (Polach, 1987; p.8) has been applied in the design of the Perkin Elmer 1220 Quantulus. The counter components requiring optimisation are '... shielding and electronics, data processing, data evaluation, and validation' (Polach, 1987; p.2).
The Quantulus achieves ultra-low background levels by both passive and active forms of shielding. The passive shielding consists of a 650 Kg asymmetric lead block that encloses the sample PMTs. It is constructed from low residual activity lead (from Boliden Mine, Sweden) and the internal cavity surrounding the PMTs is lined with high purity copper (Fig1). A layer of cadmium, lined with copper, surrounds the sample PMTs ('S' in Fig 1) to shield against neutrons (Kojola et al., 1984). The passive shield is thickest at the top, where the cosmic ray flux is most intense. Active shielding in the Quantulus is in the form of a liquid scintillation anticoincidence guard, which completely surrounds the sample phototubes, within the lead shield. This active guard has an additional pair of 'guard' PMTs ('G' in Fig 1), which normally operate in anticoincidence with the sample PMTs, thus rejecting cosmic induced pulses.
The counting vial is placed into one of three trays located beneath the shield, and counting data (tray position number,1-60; sample identification number; counting requirements etc) entered into a laboratory computer. The LS vial is housed in a stepped stainless steel sleeve, which forms a light seal when the sample is loaded through the bottom of the shield assembly, by a low activity copper piston. The sample changer and shield areas are refrigerated to 12°C to reduce benzene evaporative loss, and also ventilated, to prevent condensation on vials and to eliminate possible radon build-up.
The Quantulus has been designed with state-of-the-art electronics that includes short coincidence resolution time, two-position coincidence bias, and pulse comparison analysis.
(i) Coincidence resolution time. All LS counters with horizontally-opposed PMTs record pulses initiated within each PMT only if they fall within a set coincidence resolution time interval. This reduces the possibility of pulses generated within either tube being recorded as genuine decay events, and results in a lowering of background levels. The Quantulus has a very short coincidence time (20 nsec) compared with the average of 25-40 nsec in conventional counters (Polach and Singh, 1985).
(ii) Selectable coincidence bias. In LS counters, pulses generated within each PMT are summed, so long as they occur within the specified coincidence resolution time, as noted above. However, some events not originating in the counting sample, can generate coincident pulses and contribute to an increase in background levels. For example, an event originating in one PMT may result in a very large pulse in one tube, and a coincident, small pulse in the other. Events of this nature can be recognised and rejected, by accepting for coincidence summing only those pulses which originate in either PMT, but have an amplitude larger than a certain value, defined as the 'coincidence bias'. The result is that the smaller of the two time coincident pulses, if also smaller than the set coincidence bias, will not be accepted for pulse summation. The Quantulus, unlike most conventional counters, incorporates a two-position, coincidence bias threshold, allowing counting optimisation for either 3H or 14C.
(iii) Pulse amplitude comparison (PAC). The PAC circuitry compares the amplitude of the pulses from each PMT and, if the pulse heights differ by more than a selected amount, the pulses are rejected as background events. PAC levels are software selectable, and determine the amount of pulse amplitude variation that may be tolerated. The optimum PAC setting is experimentally determined, and results in lower background levels without significant reduction in counting efficiency.
Other electronic controls include : high voltage spectrum stabiliser, negative ion generator to remove static charges on counting vials, pulse shape analysis (PSA) capabilities to allow differentiation between alpha and ß emitters, and RF pickup.
Twin multichannel analysers (MCAs) record both coincident and anti-coincident events in both sets of tubes, with the data not only used for age calculation , but also for quality control purposes.
* Benzene weights used, for 0.3 ml = 0.2637 g; 3 ml = 2.637 g; 10 ml = 10.0 g
B = background
No = derived net cpm for 14C reference standard, 0.95 oxalic acid
fM = factor of merit (No/sqrtB)
FM = figure of merit (E2/B)
tmax = maximum determinable age (using 3000-min count time, and 2-sigma criterion)
tmin = minimum determinable age (using 3000-min count time, and 1-sigma criterion)
The data above is derived from a wide window (generally 100 - 550) from typical background and modern count rates in a number of Quantulii. It does not indicate the maximum performance data. The scintillator used is butyl-PBD, at a concentration of 15g/l. Hi coincidence bias is selected for all measurements.
The counting performances of the Perkin Elmer Quantulus and Perkin Elmer Packard Tri-Carb 2050 instruments were compared by Hogg and Noakes (1992) for numerous LS vials. Counting performance data is shown in Fig 2.