Archaeological Chemistry - American Chemical Society

---

Chapter 31

Accelerator Mass Spectrometry Radiocarbon Measurement of Submilligram Samples 1

2

3

1,4

Downloaded by UNIV OF CALIFORNIA SANTA BARBARA on March 2, 2018 | https://pubs.acs.org Publication Date: May 5, 1996 | doi: 10.1021/bk-1996-0625.ch031

D. L. Kirner , J. Southon , P. E. Hare , and R. E. Taylor 1

Radiocarbon Laboratory, Department of Anthropology, University of California, Riverside, CA 92521 Center for Accelerator Mass Spectrometry, Lawrence Livermore National Laboratory, Livermore, CA 94551-9900 Geophysical Laboratory, Carnegie Institution of Washington, 5251 Broad Branch Road, N.W., Washington, DC 20015 Institute of Geophysics and Planetary Physics, University of California, Riverside, CA 92521 2

3

4

14

Previous studies have determined that the total system C background values in catalytically- reduced graphitic carbon samples- are inversely proportional to their weights. We further examine this relationship down to 40 micrograms using both assumed C "dead" background sample and a contemporary standard, Australian National University (ANU) sucrose. Our observations are consistent with those previously reported with respect to the inverse relationship between sample weight and C activity. These observations support the view that a constant addition of modern carbon contamination during the graphitization process explains the observed background C activity in graphitic carbon samples. 14

14

14

Accelerator mass spectrometry (AMS) radiocarbon dating combines the technology of C dating with mass spectrometry and particle acceleration to accomplish high energy mass spectrometry to measure C on an ion-by-ion basis (I). At the outset, researchers anticipated three major advantages to AMS technology for C applications. The first was a reduction in required sample size from gram to milligram amounts of carbon. The second benefit was a significant reduction in counting time from weeks to virtually minutes. The third anticipated advantage was an expected extension in the C timescale from approximately 40/50,000 to 100,000 years (2-4). The anticipation that AMS could extend the temporal range of C dating was based, in part, on the consideration that cosmic radiation, a major source of background in conventional decay counting, is essentially eliminated in AMS technology (5). The previous decade of research has made AMS measurements on one milligram samples of graphitic carbon a routine operation. Subsequent studies that 14

14

14

14

14

0097-6156/96/0625-0434$12.00/0 © 1996 American Chemical Society

Orna; Archaeological Chemistry ACS Symposium Series; American Chemical Society: Washington, DC, 1996.

31.

KIRNER ET AL.

435

Accelerator MS Radiocarbon Measurement

have focused on submilligram samples have reported several problems. One of these issues is a significant increase in blank levels with reductions in sample weights below about 200 to 300 micrograms of graphitic carbon. Vogel et al. (6) suggest that this phenomenon is the result of the introduction of a constant amount of modern carbon during the preparation of the catalytically-condensed graphitic carbon, a step that is required in most of the AMS systems currently in operation. We will focus our discussion on experiments that examine background levels and measured C values in a contemporary standard as a function of decreasing sample size below one milligram.

Downloaded by UNIV OF CALIFORNIA SANTA BARBARA on March 2, 2018 | https://pubs.acs.org Publication Date: May 5, 1996 | doi: 10.1021/bk-1996-0625.ch031

14

Background Levels for Submilligram Samples Various approaches have been used by AMS researchers to reduce background levels (5-10). Currently, in routine operations, standard background values on one milligram samples are reported as rangingfromabout 0.1 to 1.0% modern. In C years, this translated to ages of approximately 35,000 to 55,000 before present (BP). 14

Contamination Sources. Potential sources of contamination resulting in these background levels have been reported by several laboratories. At present, most AMS systems convert samples to CO2 by combustion for non-carbonate organics and acidification for carbonate samples. This is followed by the production of catalytically-condensed graphitic carbon (5-10\ which is introduced into the AMS spectrometer ion source for measurement Five major sources of background contamination can be identified: in situ contamination, the pretreatment chemistry, combustion or acidification, graphitization, and the AMS spectrometer (5, 11). Sample pretreatment requires separation of the datable material from contamination. This separation is accomplished by physical and chemical means. Incomplete separation of in situ contamination of the sample is one source of background contamination. Contamination may also occur during the pretreatment process due to handling, or from C contamination in chemical reagents or adsorbed on glassware. Similarly, C contamination from reagents or reaction vessel walls may be introduced during the production of CO2 by combustion or hydrolysis, or during the subsequent graphitization step (6). Instrument background occurs when C is detected in a sample that should be free of C. This may be the result of a detector anomaly in which a C pulse registers in the detector circuitry when no C is actually present. Additionally, it may result from C contamination of the AMS spectrometer ion source or beam line. This occurs when C that derivesfroma spectrometer component reaches the detector. AMS laboratories reporting experiments designed to quantify their background levels note a hierarchy of values beginning with instrument background. Some facilities define this value exclusively as reflecting activity generated in the beam line with the ion source closed offfromthe remainder of the tandem system. For example, at the Nuclear Physics Laboratory, University of Washington, Seattle, this has been measured at £90,000 BP (0.001% modern), i. e., no counts detected in 30 minutes. An equivalent result was obtained with their ion 14

14

14

14

14

14

14

14

Orna; Archaeological Chemistry ACS Symposium Series; American Chemical Society: Washington, DC, 1996.

Downloaded by UNIV OF CALIFORNIA SANTA BARBARA on March 2, 2018 | https://pubs.acs.org Publication Date: May 5, 1996 | doi: 10.1021/bk-1996-0625.ch031

436

ARCHAEOLOGICAL CHEMISTRY

source open to the beam transport system containing an empty aluminum target holder (8). The University of Toronto group (5) reports an apparent age on an empty aluminum target holder in their system at 85,000 BP (0.0025% modem). In experiments carried on with a FN tandem using a modified GIC Model 846 ion source at the University of California Lawrence Livermore National Laboratory (LLNL) AMS Laboratory (CAMS), we determined that the instrument background with the ion source closed off was >104,000 BP, i. e., no counts detected in 20.5 minutes. With the ion source containing an empty aluminum target holder open to the beam transport system, a C count rate equivalent to 74,000 BP (0.009% modem) was measured (11). More typically, an overall instrument background in a tandem system is measured using some type of geologic graphite. As far as we are aware, the oldest reported C value on geologic graphite is 69,030±1700 BP for a sample powdered and encapsulated under argon. Geologic graphite exposed to ambient air exhibited somewhat higher values, approximately 58,000 to 65,000 BP (8). Typical values reported over the last decade by other AMS laboratories for the C age of geologic graphite rangefrom50,000 to 65,000 BP (5, 7, 9). The average (N=2) C value obtained at LLNL CAMS laboratory on geological graphite used by the University of California, Riverside (UCR) C laboratory is 64,460±3200 BP. The average (N=7) C value on graphite powder used by the LLNL CAMS laboratory is 57,900±1500BP. 14

14

14

14

14

14

Background Mass Dependence. In contrast, graphitic carbon that is produced from combusted CO2 consistently registers higher background values, which increase in relationship to decreasing sample weights in the microgram range During the last decade several laboratories have published studies of their efforts to reduce background levels during their production of graphitic carbon. To date, the most complete published study of the relationship between sample weight and background values was undertaken by Vogel et al. (6). Using anthracite coal, they demonstrated that for samples below 500 micrograms the C background values increased as a function of decreasing sample weight. The best fit of the C activity to the sample weight relationship was interpreted to indicate that a constant amount of contamination, equivalent to 2.2±l.^g of modem carbon, was added to each sample during the combustion or graphitization step. This suggests that as the sample decreased in size, the constant amount of contamination being added resulted in a progressive net decrease in the C age of the sample. The sources of contamination that were evaluated by Vogel et al. (6) for the combustion step were adsorbed CO2 or CO on the walls of the Vycor™ tubing and residual traces of carbon in the CuO used as the oxygen source. In the graphitization step they examined the possibility of memory effects in the vacuum system and traces of carbon in the Fe catalyst. They also considered the adsorption of CO2 by the graphitic carbon and small amounts of contamination that may have been picked up during storage and handling. Their conclusions were that approximately 60%-70% of the contamination occurred as a result of the release of adsorbed C 0 from the Vycor™ combustion tube at the high temperatures used during combustion. 14

14

14

2

Orna; Archaeological Chemistry ACS Symposium Series; American Chemical Society: Washington, DC, 1996.

31.

KIRNER ET AL.

Accelerator MS Radiocarbon Measurement 437

We have examined the relationship between sample size and background activity in submilligram samples prepared from wood recovered from reported Pliocene sediments (11). The pretreatment regimen consisted of an acid wash with 2 M HC1 followed by a wash in 1 M NaOH, distilled H 0 and a final treatment of 2 M HC1. The very low background values achieved on one milligram samples suggested that any measurable C contamination had been removed. Graphitization was carried out at the UCR Radiocarbon Laboratory using methods based on those described by Vogel et al. (6). The mean C age of the 1000 microgram samples is 52,140±439 BP (N=19). The lowest C value obtained on a 1000 microgram sample is 56,150+540 BP (0.09% modern). Figure 1 illustrates the relationship between the graphitized carbon weight and the C activity (% modern) from 10 to 1000 micrograms. The mean for the two smallest samples at 10 micrograms is 20,370±1410 BP (0.9% modern). Accepting the Vogel et al. view (6) that the most probable interpretation of the data is a constant addition of modern carbon contamination, Figure 2 represents the best fit of the data characterizing the constant addition of the equivalent of 1.0± 0.4 micrograms of modern carbon. 2

14

Downloaded by UNIV OF CALIFORNIA SANTA BARBARA on March 2, 2018 | https://pubs.acs.org Publication Date: May 5, 1996 | doi: 10.1021/bk-1996-0625.ch031

14

14

14

Contemporary Standards for Submilligram Samples Most AMS researchers who are pursuing microgram radiocarbon capability concentrate on the preparation and analyses of background material. This allows the researcher to evaluate the sources and amounts of modern carbon contamination in the various preparative steps. Only a small number of studies have examined the relationship using standards such as Oxalic acid or ANU sucrose. Such data is necessary to be able to infer meaningful C age estimates for submilligram samples. A recent study conducted by NIST in cooperation with the National Ocean Science AMS facility utilized materials of known C abundance that spanned the entire range of concentrations from supra modern to effectively zero blanks (12). Sample materials included Oxalic acid I and Π, Carrara marble and graphite. The researchers reported that in modern carbon targets below 500 micrograms there was a significant mass dependence. In addition, they noted a significant processing blank variability that they suggest is the ultimate limiting factor in submilligram sample measurements. Using ANU sucrose, we have examined the relationship between sample weight and C activity for samples from 40 to 1000 micrograms. The accepted values for ANU sucrose is 150.85±0.8% modern (14). Figure 2 is a plot of the relationship between the graphitized carbon weight and the C activity for the ANU sucrose measurements. In each case, the C values have been corrected for background using an algorithm proposed by Donahue et al (15). Above 100 micrograms, the C value is within 1.5% of the expected activity. Below 100 micrograms, there is a reduction in the C values. In order to account for this reduction in C activity, several explanations were considered. The line in Figure 14

14

14

14

14

14

14

14

Orna; Archaeological Chemistry ACS Symposium Series; American Chemical Society: Washington, DC, 1996.

Downloaded by UNIV OF CALIFORNIA SANTA BARBARA on March 2, 2018 | https://pubs.acs.org Publication Date: May 5, 1996 | doi: 10.1021/bk-1996-0625.ch031

438

ARCHAEOLOGICAL CHEMISTRY

Background Activity vs. Sample Size

0

200

400

600

800

1000

Graphitized carbon weight (micrograms) 14

Figure 1. Relationship of sample weight (micrograms) and C activity (% modem) of presumed C-free wood samples: 10 to 1000 micrograms. 14

Orna; Archaeological Chemistry ACS Symposium Series; American Chemical Society: Washington, DC, 1996.

Downloaded by UNIV OF CALIFORNIA SANTA BARBARA on March 2, 2018 | https://pubs.acs.org Publication Date: May 5, 1996 | doi: 10.1021/bk-1996-0625.ch031

31.

KIRNER ET AL.

Accelerator MS Radiocarbon Measurement 439

Background Activity vs. Sample Size

0.0001 10

1

p-J

100

1000

Graphitized carbon weight (micrograms) 14

Figure 2. Relationship of sample weight (micrograms) and C activity (fraction modem) of presumed C-free wood samples: 10 to 1000 micrograms on a common logarithmic scale. The lines represent the fit of the data and the addition of 1.0±0.4 micrograms of modem carbon. 14

Orna; Archaeological Chemistry ACS Symposium Series; American Chemical Society: Washington, DC, 1996.

440

ARCHAEOLOGICAL CHEMISTRY

Downloaded by UNIV OF CALIFORNIA SANTA BARBARA on March 2, 2018 | https://pubs.acs.org Publication Date: May 5, 1996 | doi: 10.1021/bk-1996-0625.ch031

ANU Activity vs. Sample Size 155

η

154 153 -

0

200

400

600

800

1000

Graphitized carbon weight (micrograms) 14

Figure 3. Relationship of sample weight and C activity of ANU sucrose: 40 to 1000 micrograms. The solid line represents a hypothetical relationship between sample weight and C activity of ANU sucrose with the addition of one microgram of "dead" carbon. The dashed line represents ±1% of 150.8% modern.. The C activity was corrected using mass-matched backgrounds produced with a cobalt catalyst. 14

14

Orna; Archaeological Chemistry ACS Symposium Series; American Chemical Society: Washington, DC, 1996.

31.

KIRNER ET AL.

Accelerator MS Radiocarbon Measurement

441

3 represents one scenario - the hypothetical relationship between sample weight and C activity that would resultfromthe constant addition of one microgram of dead ( C-free) carbon contamination. The dead carbon contamination would cause a reduction in the C values for ANU sucrose at the submilligram level while having little effect on the background measurements. Potential sources for this contamination may be dead carbon from pump oil in the backing pump in the graphitization apparatus that has been C dated previously, and appears to be Cfree (Prior, C , University of California, Riverside, personal communication, 1995.) orfrompump oil or vacuum grease in the AMS system. We emphasize that contamination by dead carbon is not the only possible explanation for these results. An alternative explanation is size-dependent isotopic fractionation. Fractionation refers to the selective enrichment or depletion of one isotope at the expense of another. The graphitization reaction may fractionate strongly if it does not go to completion (16) and we suspect that this may be a particular problem for submilligram samples. In addition, changes in the carbon-tocatalyst ratios have been shown to causefractionation(16, 17). Samples with different carbon-to-catalyst ratios may have different thermal properties and thus mayfractionatedifferentially while being sputtered in the ion source. Samples with lower carbon to catalyst ratios generally produce lower beam currents. There may be beam-dependent isotope fractionation in the transmission of particle beams through the spectrometer. Further investigation is needed in this area. 14

14

14

Downloaded by UNIV OF CALIFORNIA SANTA BARBARA on March 2, 2018 | https://pubs.acs.org Publication Date: May 5, 1996 | doi: 10.1021/bk-1996-0625.ch031

14

14

Conclusions This investigation was undertaken to continue the examination of the relationship between C activity and sample weight for catalytically-reduced graphitic carbon in the submilligram range. Our data confirm the previous conclusions of several investigators that in general samples >500 micrograms show little or no measurable mass dependency. However, below 500 micrograms both the background and ANU values reflect the increasing effects of microcontamination orfractionationas a function of sample size. In the case of background contamination, as was previously noted, the interpretation of the data suggests a constant addition of modem carbon on the order of 1.0±0.4 micrograms. The ANU sucrose data can be interpreted in terms of this same background, plus dead carbon contamination or fractionation as previously discussed. Finally, the overall outlook for those pursuing the microgram radiocarbon capability is encouraging. The problem of mass dependence can be compensated for by using mass matching techniques as suggested by Klinedinst et al (12), where submilligram samples are measured against similarly sized standards. However, the key to success is to be able to produce accurate, precise estimates of the size dependent backgrounds for submilligram samples. This is the primary limiting factor for successful submilligram sample measurement. 14

Orna; Archaeological Chemistry ACS Symposium Series; American Chemical Society: Washington, DC, 1996.

ARCHAEOLOGICAL CHEMISTRY

442

Downloaded by UNIV OF CALIFORNIA SANTA BARBARA on March 2, 2018 | https://pubs.acs.org Publication Date: May 5, 1996 | doi: 10.1021/bk-1996-0625.ch031

Acknowledgments The studies reported in this paper were supported, in large part, by a grantfromthe University of California/Lawrence Livermore National Laboratory (UC/LLNL) with additional support by the Gabrielle O. Vierra Memorial Fund, the Dean of the College of Humanities and Social Science and the Intramural Research Fund, University of California, Riverside. Comments of Richard Burky, Lijun Wan, and Christine Prior on an earlier draft of this paper are very much appreciated. This is contribution 95/05 of the Institute of Geophysics and Planetary Physics, University of California, Riverside. Literature Cited 1. Gove, H. E. In Radiocarbon After Four Decades: An Interdisciplinary Perspective; Taylor, R. E., Long, Α., Kra, R. S., Eds.; Springer-Verlag: New York, New York, 1992; pp 214-229. 2. Muller, R. A. Science1977,32, 489-494. 3. Bennett, C. L.; Beukens, R. P.; Clover, M. R.; Gove, H. E.; Liebert, R. B.; Litherland, A. E.; Purser, Κ. K.; Sondheim, W. E. Science 1977, 108, 508509. 4. Nelson, D. E.; Kortelling, R. G.; Scott, W.R. Science 1977, 198, 507-508. 5. Beukens, R. P. In Radiocarbon After Four Decades: An Interdisciplinary Perspective, Taylor, R. E., Long, Α., Kra, R. S., Eds.; Springer-Verlag: New York, New York; pp 230-239. 6. Vogel, J. S.; Nelson, D. F.; Southon, J. R. Radiocarbon 1987, 29, 323-333. 7. Arnold, M.; Bard, E.; Maurice, P.; Duplessy, J. C. Nuclear Instruments and Methods in Physics Research 1987, B29, 120-123. 8. Schmidt, F. H.; Balsey, D. R.; Leach, D. D. Nuclear Instruments and Methods in Physics Research 1984, B29, 97-99. 9. Gillespie, R.; Hedges, R. Ε. M . Nuclear Instruments and Methods in Physics Research 1984, B5, 294-296. 10. Gurfinkel, D. M . Radiocarbon 1987, 29, 335-346. 11. Kirner, D.; Taylor, R. E.; Southon, J. R. Radiocarbon, in press. 12. Klinedinst, D. B.; McNichol, A. P.; Currie, L. Α.; Schneider, R. J.; Klouda, G. Α.; von Reden, K. F.; Verkouteren, R. M.; Jones, G. A. Nuclear Instruments and Methods in Physics Research 1994, B92, 166-171. 13. Vogel, J. S.; Nelson, D. E.; Southon, J. R. Radiocarbon 1989, 31, 145-149. 14. Currie, L. Α.; Polach, H. Radiocarbon 1980, 22, 933-935. 15. Donahue, D. J.; Linick, T. W.; Jull, A. J. T. Radiocarbon 1990, 32, 135-142. 16. Vogel, J. S.; Southon, J. R.; Nelson, D. E. Nuclear Instruments and Methods in Physics Research 1984, B5, 289-293. 17. Arnold, M.; Bard, E.; Maurice, P.; Valladas, H.; Duplessy, J. C. Radiocarbon 1989, 31, 284-291. RECEIVED

October 9, 1995

Orna; Archaeological Chemistry ACS Symposium Series; American Chemical Society: Washington, DC, 1996.