INTRODUCTION
Since Arnold and Libby’s first 14C measurements in the early 1950s (Arnold and Libby Reference Arnold and Libby1949), the major concern of “radiocarbonists” has been to reduce the background level of metering facilities and/or to obtain the best stability possible of the counting background as this is more important than its magnitude. The background level results of several contributions: the electronic noise that depends on the counting system itself and a part due to external variables that will both increase the electronics noise and produce some variability to the background level. The background depends firstly on the specific radioactivity of the environment including radon in air that diffuses into cavities in the shield and also that of the components of the metering installation, i.e., the counter, the anti-coincidence system and the shielding; and secondly on disturbances induced by cosmic radiation i.e., secondary radiation formed by muons and protons in the shield, neutrons, and muon leakage (muons that escape detection in the guard counter).
To mitigate the effects of cosmic radiation, the first strategy of the laboratories was to reduce the size of the installations in order to reduce the flux of cosmic radiation passing through them and, at the same time, to increase the amount of carbon in the counters. Thus, liquid scintillation counting that allowed to concentrate more carbon in a smaller volume, seemed to bring some progress from the sixties. Another approach was tested with underground facilities, namely a mushroom house, 60 m deep (Delibrias and Rapaire Reference Delibrias and Rapaire1967; Rapaire Reference Rapaire1968). Its main aim, however, was to study background noise components rather than to develop an underground laboratory. The use of massive physical shielding to reduce the cosmic-ray component of background in underground counting labs has been reported by many authors (Vogel and Marais Reference Vogel and Marais1971; Grootes and Stuiver Reference Grootes and Stuiver1979; Stuiver et al. Reference Stuiver, Robinson, Yang, Berger and Suess1979; Schotterer and Oeschger Reference Schotterer and Oeschger1980; Loosli et al. Reference Loosli, Heimann and Oeschger1980; Calf and Airey Reference Calf, Airey, Ambrose and Duerden1982; Bowman Reference Bowman1989). Nevertheless, these laboratories operated a few meters to a few tens of meters below the surface of the ground and only offered an intermediate solution since the influence of cosmic radiation was merely attenuated, which we knew how to do in the laboratory thanks to increasingly sophisticated counting systems. Measurements in natural cavities were all the more difficult the deeper they were, the more delicate were their accessibility. In 1990, the opportunity to use the Laboratoire Souterrain de Modane facility afforded us the opportunity of developing new metering installations with total protection from cosmic radiation and at a minimal cost because it did not involve the use of an anti-coincidence system, thereby limiting, in fact, the quantities of lead required for shielding. We present here the main results.
Modane Underground laboratory
The Laboratoire Souterrain de Modane (LSM) is a joint French laboratory of the National Center for Scientific Research (CNRS) and the Atomic Energy Commission (CEA). It is located below Pointe du Fréjus (2932 m; 45°07'38"N 6°40'58"E). At an altitude of 1263 m, its entrance is located on the French–Italian border, in the middle of the 13-km-long Fréjus tunnel. The rock cover, mainly composed of lustrous shale, is about 1700 m thick, equivalent to 4400 m of water. The flux of charged particles from cosmic radiation is reduced by a factor of 3 × 10–6 compared to that at the ground level. The muon flux is of the order of 4 per m2 and per day, while at sea level it is of the order of 1.4 × 106 muon/m2/day. The fast neutron flux is (4.0 ± 1.0) × 10–6 neutron/s/cm2 and the thermal neutron flux has been measured to be (1.6 ± 0.1) × 10–6 neutron/s/cm2 (Chazal et al. Reference Chazal, Brissot, Cavaignac, Chambon, de Jesus, Drain, Giraud-Heraud, Pastor, Stutz and Vagneron1998).
Ventilation is provided by an air intake at an altitude of about 2100 m below plant B (French side of the ventilation plant) and transported by a 2250-m-long duct (630 mm) to the LSM. Radon levels commonly vary between 20 and 40 Bq/m3; improvements in the ventilation system in 1999 reduced it to below 20 Bq/m3. Hygrometry is low and the temperature is stable around 26ºC.
Proportional Counters at LSM
In the fall of 1990, we obtained a few square meters in the LSM to test, without any anti-coincidence system, one of our 14C-CO2 proportional counters running routinely at Gif (Delibrias Reference Délibrias, Roth, Potty and Ménager1985). A lead castle with 20-cm-thick walls was quickly built with bricks; the interior was lined with 5 cm of copper ingots, which constitutes the last shield against the residual radioactivity of the lead of the castle, that of the metal part of the valves and the glass of the vacuum line, and of the pre-amplifiers and others electronic components inside the lead castle. Immediately, the results of the first count showed a reduction of more than 80% in the background level. We decided to set up four chromed copper proportional counters: two (C24 and C31) of a volume of 1.2 L and two (CD1 and CD2) of 2.4 L filled to a pressure of 148 cm of Hg. The shielding consisted of 15 cm of lead bricks and 10 cm of copper ingots. Several batches of copper ingots were tested at LSM for their radioactivity using an HP Germanium detector from the Gamma spectrometry facilities (Reyss et al. Reference Reyss, Schmidt, Legeleux and Bonté1994). We placed the counters in an airtight plexiglass box tested for radioactivity in which the atmosphere was nitrogen from a compressed gas cylinder, dried by passing over silica gel to remove any moisture. This controlled atmosphere also freed us from the radon problem. The four quartz tubes bringing the gases to the meters pass straight through the cover. The seal between the cover and the cables on the one hand and the cover and the tubes on the other hand is best achieved with a silicone gasket; however, a slight leak must be left for the nitrogen to escape. For each counter, the pulses from the preamplifier enter a discriminator before controlling a time constant set to 10 ms to suppress the secondary impulses. The output feeds a CAMAC standard storage scale which is read periodically by a computer. The high voltage on the proportional meters is 0–6 kV and 0–10 kV for small and large respectively, typically running at 4900 V and 5600 V, respectively.
RESULTS
Immediately after its arrival at LSM, a proportional counter (C24) filled with dead CO2 at the Gif-sur-Yvette laboratory was placed in a lead and copper castle prepared about two weeks in advance. This first test showed a rapid decrease in the count rate, reducing the background level of this counter from 1.2 cpm at Gif-sur-Yvette to about 0.2 cpm three days later.
Such a rapid decrease in the counting rate corresponds to the deactivation of 64Cu (half-life 12.7 hr), which is produced through the reactions 63Cu (n, γ) 64Cu due to the formation of neutrons by the interaction of muons and protons within the shield (see de Vries Reference de Vries1956, Reference de Vries1957).
Based on these encouraging results, the experiment was continued, and four counters were implanted in a new geometry of the lead castle. The two counters (CD1-CD2) with a volume of 2.4 L had not been in operation for a long time at the Gif sur Yvette laboratory, their performances having been judged too poor with background levels greater than 2.5 and 2.7 ± 0.2 cpm. The last counter (C31) presented performances lower than those of the C24 counter (C24) with a background of 1.67 cpm. Background level and NIST Oxalic Acid I activity (Karlen et al. Reference Karlen, Olsson, Kallberg and Kilicci1964) at LSM are reported in Table 1. The comparison of the background levels shows a reduction of 65 to 85% of this parameter at LSM compared to Gif-sur-Yvette. The background levels of these proportional counters were the lowest obtained, at the time, as evidenced by the compilation of background in gas proportional counters by Theodorsson (Reference Theodorsson1991a, Reference Theodorsson1992), Loosli et al. (Reference Loosli, Heimann and Oeschger1980) or by liquid scintillation by Kalin and Long (Reference Kalin and Long1989). From another point of view, the main interest of this installation was the stabilization of the variability of this background noise. The standard deviation was divided by a factor of 23 to 36 for small counters and over 80 for 2.4-L counters. With a limit value at twice the standard deviation of the background, the dating limits of these counters were between 67,500 and 69,300 years. Obviously, the background was more stable because barometric corrections were no longer significant.
Technical performances of the four beta counters that were implemented in the Underground Laboratory of Modane (LSM in French). The first columns are for the background comparison between the Gif-sur-Yvette and LSM laboratories. The subsequent columns are for Oxalic Acid I activity measured at LSM. The factor of merit and dating limit are given in the last two columns.
The activity values of NIST Oxalic Acid I are reported in Table 1. The calculated factors of merit (FOM= S/√B with S, the signal, B the background; Theodorsson Reference Theodorsson1991b) may seem relatively modest (between 19 and 38) but this is without taking into account the low filling pressure of these counters (1 or 2 atmospheres). We did not attempt to increase the amount of carbon in the counters because a few grams of carbon were sufficient with these counters to obtain good precision and the counting time (a few days) remained reasonable even for very old samples. The stability of the counting parameters of these counters made it possible to obtain a routine precision of a few per mil.
Routine Use and Scientific Interest
As the main 14C laboratory is at Gif-sur-Yvette and the LSM is more than 600 km away, refills daily or even every other day were not possible. The CO2 remained in the counters for 2 to 3 weeks, thus enabling very long counts and very good measurement statistics, pushing back the dating limits. However, to obtain a stable measurement over such long periods of time, the CO2 had to be perfectly pure, free of all electronegative impurities that could bias the measurements. Two purification methods were used: “purification on calcite” and “yoyo on Cu” (Olsen Reference Olsson1958; Delibrias Reference Délibrias, Roth, Potty and Ménager1985). The “purification on calcite” consisted of adsorption of the sample CO2 on CaO, resulting from the high temperature decomposition of a carbonate from a stalagmite of Eemian age. By lowering the temperature, a calcium carbonate was formed from the sample CO2, leaving behind non-adsorbable contaminants that were then pumped out. The clean sample CO2 was then released by raising the temperature again. The method was later adapted to small samples for AMS measurement (Hatté et al. Reference Hatté, Hodgins, Jull, Bishop and Tesson2008). The “yoyo on Cu” consisted of 3–4 slow CO2 transfers on Cu and silver wool maintained at 500ºC, under vacuum, from a –80ºC point (dry-ice with ethanol) to a –173ºC point (liquid nitrogen). This resulted in the binding of halogens on silver, the adsorption of oxygen and sulfur compounds on the Cu in the form of copper oxide and sulfide, and the reduction of nitrogen oxides to dinitrogen, thus decontaminating the gas from these compounds. Both these purification methods are highly time-consuming, from 2 to 5 days depending on the volume of CO2, but ensured an effective measurement for 2 weeks.
Despite their performance, the LSM installations did not arouse the enthusiasm of prehistorians studying the Middle Paleolithic. Indeed, the large quantity of carbon necessary for the analysis cannot often be provided, and archaeologists have therefore preferred AMS technology. Nevertheless, a few important Middle Palaeolithic sites such as Zaffaraya, Baume Néron, Oued El Akarit, Barbas, Kozarnika and South American Pleistocene settlements (Pedra Furada, Toca do Meio, Santa Elina in Brazil) could be dated to the LSM, and the comparison with AMS dates is rarely disadvantageous (see respectively Hublin et al. Reference Hublin, Barroso Ruiz, Medina Lara, Fontugne and Reyss1995; Defleur et al. Reference Defleur, Bez, Cregut-Bonnoure, Fontugne, Jeanet, Magnin, Talon, Thinon and Combier1994; Fontugne and Hatté Reference Fontugne, Hatté, Roset and Harbi-Riahi2007; Boëda et al. Reference Boëda, Fontugne, Valladas, Ortega, Carbonell and Vaquero1996, Guadelli et al. Reference Guadelli, Sirakov, Ivanova, Sirakova, Anastassova, Courtaud, Dimitrova, Djabarska, Fernandez, Ferrier, Fontugne, Gambier, Guadelli, Iordanova, Iordanova, Kovatcheva, Krumov, Leblanc, Mallye, Marinska, Miteva, Popov, Spassov, Taneva, Tisnérat-Laborde, Tsanova, Molines, Moncel and Monnier2005; Boëda et al. Reference Boëda, Fontugne, Hatté, Clemente-Conte, Santos, Rocca, Lourdeau, Villagran, Lucas, Gluchy, Ramos Marcos, Lahaye, Felice, Griggo, Pino, Viana, Da Costa, Guidon, Pessis, Borges and Bruno2016; Vialou et al. Reference Vialou, Benabdelhadi, Feathers, Fontugne and Vilhena Vialou2017). This unusual and exceptional installation also allowed us to provide key information to constrain climatic dynamics based on the investigation of European loess sequences (Hatté et al. Reference Hatté, Antoine, Fontugne, Rousseau, Tisnérat-Laborde and Zöller1999) and sedimentary Pleistocene deposits in central and eastern Turkey (Kuzucuoglu et al. Reference Kuzucuoglu, Pastre, Black, Ercan, Fontugne, Guillou, Hatté, Karabiyikoglu, Orth and Turkecan1998; Fontugne et al. Reference Fontugne, Kuzucuoglu, Karabiyikoglu, Hatté and Pastre1999) and to point out the impact of 14C-free volcanic CO2 on geochronology (Pasquier-Cardin et al. Reference Pasquier-Cardin, Allard, Ferreira, Hatté, Coutinho, Fontugne and Jaudon1999).
CONCLUSION
The transfer of old proportional CO2 gas counters from the Gif-sur-Yvette laboratory to the LSM under Pointe du Fréjus made it possible to measure 14C activity in the absence of cosmic radiation. The first results showed that a large part of the background level of these counters came from the activation of the copper. The reduction in the background noise by between 65–85% depending on the counters, and its stability made it possible to reach a theoretical dating limit close to 70,000 years. The precision obtained, in routine operation, for samples less than 20,000 years old is a few per mil. However, the ability to measure very old ages is certainly more limited by our ability to decontaminate samples and prepare high purity CO2 gas than by the ability to use high performance counters. LSM 14C dating facilities ceased in 2000.
