Ao

MWSa

b

c

d

e

f

g

h

i

a

ARRA

KAACMN

0h

J. of Supercritical Fluids 79 (2013) 314–323

Contents lists available at SciVerse ScienceDirect

The Journal of Supercritical Fluids

journa l homepage: www.e lsev ier .com/ locate /supf lu

pplication of supercritical carbon dioxide–co-solvent mixtures for removal ofrganic material from archeological artifacts for radiocarbon dating

arvin W. Rowea, Jenny Phomakayb, Jackson O. Layc, Oscar Guevarad, Keerthi Srinivasb,. Kirk Hollise, Karen L. Steelmanf, Thomas Guildersong, Thomas W. Stafford Jr. h,

arah L. Chapmani, Jerry W. Kingb,∗

Texas A&M University, Doha, QATAR/Conservation Laboratory - Museum of New Mexico, Santa Fe, NM 87505, USADepartment of Chemical Engineering, University of Arkansas, Fayetteville, AR 72701, USADepartment of Chemistry, University of Arkansas, Fayetteville, AR 72701, USADepartment of Biological and Agricultural Engineering, Fayetteville, AR 72701, USALos Alamos National Laboratory, Los Alamos, NM 87544, USADepartment of Chemistry, University of Central Arkansas, Conway, AR 72035, USACenter for AMS, Lawrence Livermore National Laboratory L-397, 7000 East Avenue, Livermore, CA 94551, USAStafford Research Laboratories, Inc., 200 Acadia Avenue, Lafayette, CO 80026, USADepartment of Anthropology, University of Arkansas, Fayetteville, AR 72701, USA

r t i c l e i n f o

rticle history:eceived 18 July 2012eceived in revised form 2 January 2013ccepted 10 January 2013

eywords:rcheologyrtifactsleaningummiesondestructive radiocarbon dating

a b s t r a c t

Archeological artifacts such as burial and embalmment materials are commonly dated by 14C labelingusing accelerator mass spectrometry (AMS). The presence of contaminant organic matter can interferewith the accurate determination of an object’s age, hence sample preparation is a critical step beforeradiocarbon-based dating. Both harsh acid and base treatments have been applied to remove contami-nation, such as humic acids, resin-based adhesives, and plant oils. Additional removal of carbon-ladenmaterial can also be affected by applying such methods as plasma oxidation. In this study, SC-CO2 withaddition of a cosolvent has been applied to remove the above materials prior to plasma oxidation andsubsequent dating via AMS. Initially, wood/charcoal samples were extracted using a modified-Isco SFX-2-10 extraction unit (Isco Inc., Lincoln, NE). Experiments were conducted using supercritical carbondioxide/10% methanol at a pressure of 20.4 MPa (3000 psig) (pr = 2.80) and 40 ◦C (Tr = 1.36), and car-bon dioxide flow rates of ∼1.4 ± 0.1 ml/min. Comparison of the SC-CO2–methanol cosolvent treatmentwith traditional acid–base–acid sample pretreatment on identical wood-charcoal samples showed com-parable radiocarbon dating results encompassing a period of 10,000 years. In addition, both Russianand Egyptian mummy gauzes and Russian textiles were similarly treated and the extracts analyzed byMALDI-TOF-MS and GC/MS to determine the chemical identity of the extracted material. A polyglycerol-based polymer was positively identified in addition to fatty acid moieties as their fatty acid methyl esterderivatives (FAMES), which potentially formed from the in situ reaction of the triglycerides present inthe embalment materials with the SC-fluid mixture. Model extractions from spiked-linen gauze sampleshave verified removal of such materials as beeswax, coconut oil, frankincense, glycerol, and humic acidsin varying amounts. Application of supercritical fluid extraction (SFE) appears to be a promising method

14

to pretreat small samples for C radiocarbon dating where conservation of the archeological artifact is ofimportance. The SFE pretreatment has the potential to replace harsh acid–base pretreatment methods,and can be coupled with a non-destructive argon or oxygen plasma treatment for microgram carbonremoval prior to accelerator MS isotope ratio age determination of the archeological artifact. This combi-nation of techniques requires as little as 0.05 mg of carbon-equivalent weight for the age determination

mizin

of the artifact while mini

∗ Corresponding author. Tel.: +1 479 575 3835; fax: +1 479 575 7926.E-mail addresses: jwking1@uark.edu, kingjw100@hotmail.com (J.W. King).

896-8446/$ – see front matter © 2013 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.supflu.2013.01.002

g sample degradation.© 2013 Elsevier B.V. All rights reserved.

1. Introduction

The radiocarbon dating of many archeological artifacts, such asEgyptian mummies, can be inaccurate due to contamination fromsoil organic matter. Traditional decontamination methods requirethe use of harsh acid–base pretreatment methods, which can be

critica

ddiditwehtfui

opornPpTtedbrta

b(aSaStfhcawh

tawtlCcvIpiC[ihn[

atom

M.W. Rowe et al. / J. of Super

estructive to these delicate ancient artifacts [1], as well as intro-uce inaccuracies prior to accelerator mass spectrometry (AMS)

sotope ratio determinations [2]. The goal of this research is toevelop a nondestructive radiocarbon dating technique for frag-

le archeological artifacts. Currently, radiocarbon dating utilizeswo destructive steps. Organic-containing artifacts are pre-cleanedith serial strong acid, base, and acid (ABA) treatments, usually at

levated temperatures (∼323 K) to eliminate commonly occurringumic acid, carbonate and oxalate contaminants, followed by highemperature combustion (>1023 K) of the organic carbon collectedrom the samples. Unfortunately samples can be highly degradedsing these chemical washes, and a more benign cleaning method

s desired.A method has already been developed that renders the sec-

nd step virtually nondestructive for many organic materials:lasma-chemical extraction of organic carbon [3–5]. In the coursef conducting that research, the technique developed for datingock paintings [6–9] – plasma-chemical extraction – also elimi-ated the need for the two acid steps required in step (1) [10,11].lasma oxidation may also be preferable for any type of sam-le containing significant amounts of oxalate-containing minerals.wo previous studies found that in unusual circumstances, acidreatment is insufficient to remove offending calcium oxalate min-rals [12,13]. Thus, to be able to obtain nondestructive radiocarbonating requires only a nondestructive substitution for the strongase wash used to extract the contaminating humic acids. Here weeport on the possibility of using a supercritical fluid-based cleaningechnique to extract soil organic matter (SOM) containing humiccids from organic-containing archeological artifacts.

Our approach employs supercritical CO2 (SC-CO2), which haseen found to be benign enough to be used by other conservatorsSousa et al. [14], Kaye and Cole-Hamilton [15,16], Saleh et al. [17])s well in the commercial dry cleaning industry [18]. For example,ousa et al. [14] cleaned silk garments from an 18th century Virginnd Child from Palácio das Necessidades in Lisbon using SC-CO2,C-CO2 + isopropanol, and SC-CO2 + isopropanol + water. The gen-le nature of SC-CO2 was highlighted in this work as the silk wasragile and the fibers could easily suffer disintegration by simpleandling. Sousa et al. [14] also looked at the loss of textile material,olor variation, and dirt removal using gravimetry, spectroscopy,nd optical microscopy. They observed removal of dirt particles,hile the textile structure was not physically damaged even underigh magnification.

Additional studies focusing on pesticide contamination migra-ion on artifacts in museum collections have been reported by Tellond Unger [19], Werner et al. [20], and Zimmt et al. [21] in MCIorkshop proceedings published by the Smithsonian Institute. In

hese studies, removal of dust, grease, DDT, diazinon, mercury,indane, linseed oil, and water from the artifacts using both SC-O2 and liquid CO2 (LCO2) with various cosolvent additives wereited. The cleaning or extraction process is dependent upon the sol-ent characteristics of the supercritical fluid and/or its cosolvent.n addition, the use of SC-CO2 – aside from its adjustable solventroperties – facilitates removal of contaminants from artifacts by

mproving their mass transport out of the sample matrix due to SC-O2’s low viscosity and surface tension. As with critical point drying22], SFE using SC-CO2 plus cosolvent when properly applied, elim-nates any physicochemical perturbation of the sample matrix andence morphological distortion of the artifact – which has beenicely demonstrated on woods and textiles treated with SC-CO219,23].

A key feature of the methodologies described herein is their

pplicability to small archeological samples with minimal destruc-ion of the sample matrix. Plasma oxidation [1] after SFE-treatmentf an artifact supplements the SC-fluid mixture based extractionethod, thereby yielding a non-destructive protocol for prepared

l Fluids 79 (2013) 314–323 315

samples prior to 14C dating. As indicated previously, sample con-tamination with respect to 14C can arise from a number of sources,most notably, soil organic matter. It is essential that a pretreatmentprior to plasma oxidation be employed to yield accurate 14C datingresults. Traditionally, sequential acid–base treatment of the samplehas been applied to remove both inorganic and organic contami-nants [2]. We believe that SFE with SC-CO2–co-solvent mixtures canbe utilized to eliminate naturally occurring matter prior to plasmaoxidation.

2. Materials and methods

The actual archeological samples used in this study came fromdiverse sources. The fiber, charcoal, macro flora, and wood “SR”samples were obtained from the Stafford Research Laboratories inLafayette, Colorado, and were of interest due to their known humicacid content. The Egyptian mummy samples consisted of wrap-pings of a Late Period Egyptian child mummy as well as a bovinemummy from the same time period – both enrobed in a linengauze covered with some type of resin. Their respective ages deter-mined by radiocarbon 14C dating were between AD 137 and 227for the child mummy and between BC 365 and 167 for the bovinemummy. The Russian textile sample was from the 4th InternationalRadiocarbon Interlaboratory comparison standards. The Canopicjar contents were obtained from the collection at the Mabee-GerrerMuseum of Art in Shawnee, Oklahoma.

In our initial studies, wood/charcoal samples were selectedto provide samples that were known to be heavily contami-nated with humic acid. A modified-Isco SFX-2-10 extraction unit(Isco Inc., Lincoln, NE) was used for the reported experiments.We conducted experiments on four samples using supercriticalcarbon dioxide/10% methanol solvent mixture at a pressure of20.4 MPa (pr = 2.80) and 313 K (Tr = 1.36). The gas flow rate used was∼1.4 ± 0.1 mL/min. For the standard reference material, SR 5994,the extraction was run using a total volume of SC-CO2 of 84 mL; forSR 5960, 108 mL; for SR 6097, 117 mL; and for SR 6101, 84 mL. Asaliquots of these samples had already been treated by ABA, com-busted, and radiocarbon dated, they were useful to compare to theSC-CO2 results. After SFE, samples were oxidized using a plasmadischarge at 20 W for 30 min. The collected CO2 from the organicmaterial in the artifact was sent to the Center for AcceleratorMass Spectrometry at the Lawrence Livermore National Laboratory(CAMS-LLNL) for graphitization and radiocarbon measurement.

The solubility parameter for a mixture of 10% methanol in SC-CO2 at 20.4 MPa (3000 psig) and 313 K from theoretical calculations[24] is as follows: dispersion solubility parameter contributionis 10.59 MPa1/2, the polar solubility parameter contribution is5.31 MPa1/2 and the hydrogen-bonding solubility parameter is25.4 MPa1/2, yielding an overall (total) solubility parameter of17.16 MPa1/2 or 8.56 cal1/2/cc3/2. This is similar in solvent strengthto a non-polar solvent.

In a second series of experiments, an Isco SFX-2-10 (Isco, Lin-coln, NE) extraction unit was employed. The extractor consisted oftwo Isco Model 100 DX pumps and a heated module for contain-ing the stainless steel restrictor calibrated to deliver approximately1.5 mL/min CO2 (see Fig. 1). The extraction fluid consisted of99.995% pure CO2 from Air Gas–Specialty Gases (Tulsa, OK) usedin conjunction with HPLC-Grade methanol from EMD Chemicals,Inc., Gibbstown, NJ. The capillary tubing serving as a restrictor wasenclosed in a restrictor heater module set at 60 ◦C (333 K) to pre-vent precipitation of the extracted contaminants in the capillary as

the temperature fell to ambient conditions. The Isco SFX-2-10 unitwas consistently cleaned with the SC-fluid mixture between exper-iments in order to prevent cross contamination between samplesas they were extracted.

316 M.W. Rowe et al. / J. of Supercritical Fluids 79 (2013) 314–323

or sup

ttpaacssmwpsnp

weatavseSU

Fig. 1. Picture and schematic of the apparatus used f

Linen for spiking with model compounds was 18 count, unadul-erated and bore a striking resemblance to linen fibers constitutinghe mummy gauzes when examined by microscopy. Model com-ounds: coconut oil, humic acids, beeswax, frankincense oil, abieticcid, myrrh, glycerol, etc. were obtained from commercial sourcesnd used without further purification. Linen samples and selectedompounds used in embalming and preserving the artifacts werepiked on the linen from n-hexane solutions (1 mg/mL). The linenamples were soaked for 24 h in supersaturated solutions of theixture. After 24 h, the spiked samples were oven dried andeighed. Smaller pieces were then cut to be tested at differentressures of 20.4 MPa and 40.8 MPa at 40 ◦C (313 K). A controlample consisting of linen saturated in hexane with no contami-ant was also extracted using the SC-fluid mixture for comparisonurposes.

In this second set of experiments, the archeological artifactsere weighed and photographed, then cut into smaller samples of

qual dimensions. The weights and photographic evidence beforend after extractions were used to indicate the effectiveness ofhe SFE pre-treatment process. Cleaning experiments were donet 40 ◦C (313 K). Each SFE cleaning used the SC-fluid mixture withariable flow rates and residence times at different pressures. The

amples were dried by allowing 100 mL of CO2 to flow through thextractor after treatment at 40 ◦C (313 K) and 20.4 MPa (3000 psig).FE pre-treated extracts were collected for analysis using a Brukerltraflex II MALDI-TOF mass spectrometer (MS) in the positive ion

ercritical fluid cleaning of the archeological artifacts.

mode. Samples were spotted onto a stainless steel MALDI targetafter mixing 1:1 with a saturated solution of the MALDI matrix.The MALDI matrix was dihydroxybenzoic acid (DHB) obtained fromBruker Daltonics.

GC/MS analysis of the extracted organics was accomplishedusing a Varian Triple Quadrapole GC/MS in the positive ion, electronimpact mode. The GC column was a 30 m capillary column coatedwith an SE-30 stationary phase. A 5 min temperature hold at 50 ◦Cfollowed by a temperature program using a linear heating rampof 10 ◦C/min to 300 ◦C over 25 min was utilized. The column washeld at temperature for 10 min before cooling back to 50 ◦C for thenext analysis. The injector was kept at 250 ◦C and used in both thesplit (50:1) and split-less mode for replicate analysis. The injectionvolume was 1 �L.

FAME analyses were run on an Agilent Model 6890 GC withan Agilent Model 5973 MSD (mass selective detector). The col-umn used was an SGE BPX70 0.23 mm ID, 30 M long. The MSD wasoperated in EI (electron impact) mode; all injection volumes were1 uL. The temperature program was as follows: inlet temperatureof 280 ◦C, initial oven temperature of 60◦ C which was then pro-grammed at 10 ◦C/min to 180 ◦C, held for 1 min, and then furtherprogrammed at 4 ◦C/min to 280 ◦C where it was held for an addi-

tional 30 min. FAMES were identified using mass spectral matchingto the NIST library as well as retention time matching with stan-dards (Supleco BAME-Bacterial Acid Methyl Ester), as well as a 37component fatty acid mixture.

critical Fluids 79 (2013) 314–323 317

sciTtscmp1asS6aum

st(iwlrdtp

mctmahBomr

ftfXnwaAtNcabteswblt

ypwa

Table 1Spiked linen samples – gravimetric removal results after SFE.

%Removed

Matrix 20.4 MPa (3000 psig), 40 ◦C 40.8 MPa (6000 psig), 40 ◦C

Beeswax 79.79 88.89Coconut oil 62.98 76.40Humic acids 36.11 44.94Frankincense oil 41.03 57.14Myrrh 61.81 78.91Abietic acid 57.69 73.68Glycerol 97.31 98.62Myristyl palmitate 89.47 95.16

M.W. Rowe et al. / J. of Super

The original samples of the Egyptian mummy gauzes and Rus-ian textile were weighed and initially photographed and thenut into smaller samples of equal dimensions. These pieces werendividually weighed and photographed (pre- and post-treatment).he contents of an Egyptian Canopic jar were separated accordingo materials as follows: resins, resins with potential tissue, grass,oil, and residual materials. The mummy gauzes and Canopic jarontents were treated using 100 mL of CO2 with 10% (v/v) methanolodifier; the flow rate and residence time varied with the different

ressures tested. SFE-treated samples were then dried by passing00 mL of neat CO2 through the SFE system at 20.4 MPa (3000 psig)nd 40 ◦C, primarily to remove the residual methanol. The textileamples were treated with two consecutive extractions of 100 mLC-fluid mixture at 20.4 MPa (3000 psig), 40.8 MPa (6000 psig), and1.3 MPa (9000 psig) before they were dried with 100 mL of CO2t 20.4 MPa (3000 psig) and 40 ◦C (313 K). Extracts were analyzedsing a Bruker Ultraflex II MALDI-TOF MS using the positive ionode – dihydroxybenzoic acid was used as the matrix.To provide a baseline to compare the efficiency of SFE, small

amples of the linen wrappings taken from the mummy head andhe mummified bovine leg were first treated with dichloromethaneCH2Cl2) and ethanol (C2H5OH) to remove surface organic contam-nants and to reduce the amount of the resin coating applied to the

rappings during the mummification and probably later for stabi-ization. Two other chemical cleaning treatments were employed toemove all resinous layers prior to radiocarbon dating. Radiocarbonating was obtained either by (1) the conventional, totally destruc-ive AMS radiocarbon dating technique, or (2) the nondestructive,lasma-chemical radiocarbon dating technique.

Radiocarbon dating is an extremely sensitive technique that caneasure the approximately one in 1012 of the 14C atoms that exist

ompared to the much more abundant 12C and 13C isotopes; allhese carbon isotopes exist in all living material. In the case of the

ummy linen wrappings here, the linen threads were identifieds being made from linen from plant fibers. When the flax plant isarvested, it is no longer in equilibrium with 14C in the atmosphere.ecause 14C is radioactive, it decays at a known rate after the deathf the flax plant, this allows radiocarbon dating to be used to deter-ine its age. For material that is only a few thousand years old, the

adiocarbon dating method is accurate and reliable.Approximately half of the linen wrappings (∼1 cm × 1 cm) taken

rom each of the two mummies was characterized by the conven-ional AMS radiocarbon dating technique; the other half was usedor nondestructive analysis. Scanning electron energy dispersive-ray analysis showed evidence for substances such as carbo-ates and oxalates. These two aliquots dated using pretreatmentere first washed in strong acid (∼1 M HCl) to remove carbonates

nd oxalates, which are unrelated to the ages of the mummies.fter washing with deionized distilled water, the samples were

hen washed using multiple applications of strong base (∼1 MaOH). The base treatments were originally introduced for radio-arbon dating to remove the humic acid contamination that oftenccompanies archeological artifacts. The treatment is also proba-ly efficacious in removing any resin that may still be attached tohe linen wrappings after the dissolution in dichloromethane andthyl alcohol. Seven strong base treatments were applied to eachample until the base solution appeared clear. The sample was thenashed with de-ionized distilled water. Finally, because the strong

ase washings cause atmospheric carbon dioxide to adsorb onto theinen samples, an additional strong acid wash was used to desorbhat carbon dioxide adsorbed from the atmosphere.

At this point, the samples were sent to the CAMS-LLNL for anal-

sis. Prior to AMS radiocarbon dating, the samples were completelyyrolyzed to produce carbon dioxide from the organic matter of therapping. All steps in the treatment except the dichloromethane

nd ethanol dissolutions of the resins and other hydrocarbons from

Cedarwood oil 79.34 90.69Unspiked linen 0.66 0.76

the surfaces of the wrappings were destructive. The final combus-tion of the sample was totally destructive. After the SC-mixturefluid cleaning was done, the samples were subjected to argon plas-mas (1 torr) to get rid of surface-adsorbed carbon dioxide [7–9].Low-temperature (≤100 ◦C), low-pressure (∼1 torr) oxygen plas-mas oxidize organic components of the sample to carbon dioxide.Decomposition of existing inorganic carbon (limestone rock andcalcite/calcium oxalate accretions) is prevented by running theplasmas at this low-temperature. Carbon dioxide from the sampleis flame-sealed into a glass tube cooled to liquid nitrogen temper-ature (−194 ◦C), after water is frozen out with a dry-ice/ethanolslurry (−58 ◦C). The carbon dioxide was then collected in 6 mmglass tubes and sent for graphitization and radiocarbon analysisat the CAMS-LLNL. AMS 14C measurement is necessary due to thesmall sample size of carbon collected from the samples.

The other half of the samples that were subjected to resin disso-lution with dichloromethane and ethanol were then treated usingthe above described SC-fluid mixture. After SFE, the linen wrap-pings from the two mummies were treated by applying the plasmacleaning procedure prior to sending the samples for AMS assess-ment.

3. Results and discussion

3.1. Spiked linen samples

To better evaluate the effectiveness of SFE as a pretreatmenttechnique applied to archeological samples, a natural linen samplewas spiked with various components known to be used in embalm-ment and burial protocols in ancient Egypt. Their selection waspartially based on the literature [5,14,25,26] and the above acceler-ator mass spectroscopic data. All extractions were performed withthe stated SC-fluid mixture using 103 mL of CO2 with the statedcosolvent level. This was to evaluate the relative effectiveness ofremoving the spiked material from the model linen matrix underthe same extraction conditions – not to conduct exhaustive extrac-tion of any of the spiked materials. These results at two differentextraction pressures are tabulated in Table 1.

The chosen “model” spiked compounds reported to be used andfound in mummification remains are extracted more effectively atthe higher extraction pressure based on weight differences beforeand after SFE. Secondly, overall the extractions are very good (>75%)with the exception of frankincense oil and the humic acids. Modellipophilic components, beeswax and coconut oil, known to be usedby the Egyptians in funeral rites are extracted at ∼80% gravimet-ric level from the linen matrix, and could probably be extractedmore thoroughly by running the SFE longer and at an even higher

pressure. Abietic acid, a common component occurring in resinousmaterial, is removed at about a 75% level at 40.8 MPa (6000 psig).

Using humic acids as a surrogate for “soil” contamination, partialremoval of this contaminant is possible under the stated extraction

318 M.W. Rowe et al. / J. of Supercritica

Fs

cBcniooerwcgtf

3

sootarfl

tfsptspa

Lniam

ig. 2. Radiocarbon dates by pre-treatment with strong acid or base (circles) vs.upercritical pretreatment (boxes).

onditions from these “non-aged samples”. It has been observed byowadt and Hawthorne [27] that SFE results obtained for spikedomponents on model matrices often overestimate the effective-ess of SFE; to quote these investigators, “These examples clearly

llustrate that SFE methods (or any other extraction method) devel-ped on spiked samples will not necessarily yield high recoveriesf native analytes when significant analyte–matrix interactionsxist”. This caveat should be considered in lieu of our reportedesults in Table 1; the results in Table 1 only provide a guideline as tohat certain extraction conditions can yield in terms of removing

ontaminating material from an archeological sample. Note thatlycerol, a component of interest to us based on the mass spec-rophotometric studies presented below, is removed at a ∼98% levelrom the linen fabric.

.2. Wood/charcoal samples

The results for the supercritical fluid treated wood/charcoalamples are summarized in Table 2. There is statistical agreementn three of the four artifacts, suggesting incomplete removal ofrganic carbon in the one case (SR 5994). Radiocarbon dates agreedo within ±2� on two samples, to within ±1� on one sample,nd disagreed at ±2� confidence level for the other sample. Theseesults are very promising even though only a single supercriticaluid–methanol fluid composition was utilized for the cleaning.

Fig. 2 compares the results acquired using supercritical fluidreatment with those results obtained using the ABA method. Theour blue square points are results utilizing SC-CO2 in which theamples were cleaned to remove as much humic acids residue asossible with the SC-fluid mixture. These results correlate well withhe radiocarbon year determinations of Dr Stafford who used thetandard ABA pretreatment to clean the samples (the pink roundoints). Clearly, the SC-fluid mixture treatment provides a moreccurate radiocarbon dating of the wood and charcoal samples.

It should be noted that “removal or cleaning” using SC-CO2,CO2, or dry ice (i.e. snow gun cleaning) can be accomplished based

ot only the solubility of a solute like humic acids in carbon diox-

de, but also on other properties and the method by which CO2 ispplied to a sample matrix. Sorption or adsorption of SC-CO2 byany sample matrices causes them to swell, thereby physically

l Fluids 79 (2013) 314–323

facilitating removal of the targeted contaminant from the matrix[28]. Further competitive adsorption of the dense CO2 on the sur-face of the sample causes displacement of the contaminants [29].Contaminants can also be displaced based on mechanical force athigh flow rates of CO2, which is the basis of some CO2-cleaningmethods [30].

The solubility parameters based on “model” humic or fulvicacid molecular structures are higher than the solubility parame-ters for the SC-CO2–fluid mixtures at the stated treatment pressureand temperature, bringing into question whether “solubility” isthe main or only mechanism responsible for cleaning the artifactsamples. Soil organic matter (SOM) can be dissolved in polar, apro-tic solvents such as dimethylformamide or dimethylsulfoxide, butit should be noted that SOM is in reality a mixture containing anumber of compounds of varying polarity. Estimates of the totalsolubility parameter for fulvic acids based on a polymaleic acidmodel are 27.8 MPa0.5 [31] while Kopinke et al. [32] have provideda detailed assessment of all of the major components which occurin SOM including the humic and fulvic acids fraction. Based on theirpartitioning experiments for solutes of varying hydrophobicity,humic acid was estimated to have a solubility parameter rangingfrom 25.9 to 26.1 MPa0.5 and for fulvic acid, 28.1–29.9 MPa0.5. Whilethese are more “polar” than the SC-fluid mixture, they are of con-siderably lower polarity than very polar liquid solvents.

Multi-sorbate vapor sorption experiments indicate that theSOM’s solubility parameters are highly dependent on the typeof sorbate being adsorbed by the SOM. Values of approximately26 MPa0.5 [33,34] have been reported by several investigators,while Chiou and Kile [35] note that a range of solubility parametervalues between 25.6 and 35.2 MPa0.5 for SOM are due to the inter-action between the sorbate type with either the nonpolar or polarcomponents comprising SOM. Considering the above, the overallpicture that emerges for SC-CO2 treatment of contamination con-taining SOM is that a significant portion of the SOM can be removedby this technique, but it will be dependent on the chemical com-position of a particular SOM on the artifact.

As our results compared well with those of Dr. Stafford (Fig. 2),it is clear that the supercritical fluid cleaning experiments weresomewhat successful. These results suggest that some of theSOM and humic acids were removed, although not likely totallyremoved. Note that from a more homogeneous matrix (linen), ∼45%of a “humic acid” spike was removed using the SC-mixture at 40 ◦Cand 40.8 MPa (6000 psig). Further studies with supercritical flu-ids are almost certain to increase even further the efficacy of theremoval of these type of contaminants, but will require additionalexperiments using varying conditions, such as gas mixtures, tem-peratures, pressures.

3.3. Egyptian mummy gauzes

The gravimetric results from extracting Egyptian mummygauzes showed changes in weight that were not significant (<5%).Because the samples undergo such high pressures and even-tual depressurization, the fiber strands of both the child andbovine mummy samples look “unwoven” and slightly expanded(Figs. 3 and 4). However, no color or significant dimensionalchanges, as well as weight changes before and after, could beascribed to the SFE pretreatment.

MALDI-TOF-MS spectra on the extract from the Egyptianmummy gauze were initially acquired from 500 to 20,000 Da, andthe resulting spectra shown in Fig. 5 represent scans taken from500 to 5000 Da. The MALDI-TOF-MS study revealed the occur-

rence of a polymer in the SC-fluid mixture extract having repeatingoligomeric units differing by 74 Da. This is consistent with thechemical backbone structure of a polyglycerol derivative, possi-bly with alkali salt substitution at multiple hydroxyl positions in

M.W. Rowe et al. / J. of Supercritical Fluids 79 (2013) 314–323 319

Table 2Comparison of SC-CO2 plasma oxidation with acid–base–acid (ABA) and combustion radiocarbon results.

Stafford ID no. Material CAMS ID SC-CO2 plasma 14C date (BP)* CAMS ID Stafford’s ABA combustion 14C date (BP)*

SR 5960 Fiber 130500 8790 ± 100 80532 9070 ± 50SR 6101 Charcoal 130501 3420 ± 35 87581 3550 ± 45SR 5994 Macroflora 130502 860 ± 30 79943 700 ± 30SR 6097 Wood 130503 5915 ± 35 85778 5975 ± 40

* BP, before present.

gauz

tcotaas

orttca

Fig. 3. Sample of the child mummy

he chain structure. This polyglycerol moiety was found in both thehild and bovine mummy gauze wraps. The use and occurrencef glycerol in Egyptian funeral rites is documented in the litera-ure [25], and its appearance can also be due to hydrolysis of oilnd fat triglycerides over an extended period of time. The appear-nce of this polyglycerol derivative in the MALDI-TOF-MS analysisuggests that it has undergone polymerization over time.

Nuclear magnetic resonance (NMR), infrared (IR), and MS datan gauzes from Egyptian mummies that were coated with a “resin”uled out the use of bitumen and other coniferous-based resins on

he gauzes. The accompanying spectroscopic data also indicatedhe absence of aromatic compounds. These spectroscopy studiesonfirmed the possible use of a wax-like substance (ester of a fattycid and a long chain alcohol) and/or animal fat (triglycerides) – the

Fig. 4. Sample of the bovine mummy gauz

e before (left) and after (right) SFE.

wax moiety best fitting the initial data. The NMR data suggesteda fatty acid (or ester), though we suspect that the compound(s)represented by the NMR data are merely one or more extractedcomponents of the resin. The NMR sample appears to contain anoctanoic acid component. IR data clearly indicated the presence ofamine and/or hydroxyl groups that could be consistent with someof the NMR data, but the presence of the amino functionality couldnot be confirmed via the other spectral data. The mass spectral dataalso seem to support the presence of fatty acids with unsaturationor oxidized by-products due to oxygen attack on the double bonds.

This is typically found in aged fatty acids or waxes utilized to coatmummy gauzes.

Based on the above qualitative spectroscopic data, a GC/MSstudy was initiated on the gauze extracts using the conventional

e before (left) and after (right) SFE.

320 M.W. Rowe et al. / J. of Supercritical Fluids 79 (2013) 314–323

Fig. 5. MALDI-TOF spectra of polyglycerol in extract from child mummy gauze sample.

squa

Gqfbti

TDa

T

Fig. 6. Original Russian textile sample on 1 cm

C-FAME (fatty acid methyl ester) approach. Table 3 shows theuality of MS database fit indicative of a probable presence ofatty acid moieties as their methyl esters, primarily in the Egyptian

urial cloths. The FAMES range from C14:0 through C18:3 and arisehrough the use of vegetable oils and wax esters or their derivativesn the preparation of the mummy for burial [36]. The consistent

able 3atabase fit from mass spectrophotometric characterization of FAME content in Egyptiannd 20.4 MPa (3000 psi).

Compound Tr Child mummy

C14:0 FAME 20.3 ndC15:0 FAME 22.2 95C16:0 FAME 24.1 98C16:0 FAME (isomer) 24.7 ndC18:0 FAME 27.5 ndC18:1 FAME 28.2 ndC18:2 FAME 29.3 ndC18:3 FAME 30.6 98

r, retention time in min; nd, not detected.

res background (left); and cut samples (right).

appearance of methyl palmitate in these mummy samples is inagreement with other GC/MS studies [25].

The occurrence of FAME derivatives in the SC-fluid mixture is

indicative of an in situ reaction between the extraction solvent mix-ture and the fatty acids–esters in the burial gauze. The presence ofodd carbon numbered acids is also indicative of ruminant fat. One

mummy gauzes extracted with SC-CO2 modified with 10% (v/v) methanol at 40 ◦C

Child mummy (2) Bovine mummy Russian textile

95 nd nd97 nd nd99 95 94nd 72 nd96 nd nd70 nd nd98 nd nd91 nd nd

critical Fluids 79 (2013) 314–323 321

oeesausnb

3

tEttFptrwtbeaitt(

3

trftwC

Fig. 7. Weight percent of organic matter extracted from the Russian textile sampleextracted at 20.4 MPa (3000 psi), 40.8 MPa (6000 psi) and 61.3 MPa (9000 psi) and

M.W. Rowe et al. / J. of Super

f the authors has in the past reported in situ methylation of veg-table oils during SFE [37,38]; other investigators such as McDanielt al. [39], Wyatt and Haas [40], and others [41] have testified to theynthetic and derivatization utility of gas-expanded alkyl carboniccids. It should be noted that both potash and natron were widelysed by the Egyptians – these could serve as a source of alkali metalalts for possible FAME catalysis. If our rationale is correct, then aumber of materials associated with archeological artifacts coulde analytically identified by their methylated products [25,26].

.4. Russian textile

An ancient Russian textile was treated using the SC-fluid mix-ure under extraction conditions similar to those used on thegyptian mummy specimens. The textile sample was heavily con-aminated with soil; hence the humic acid content is presumedo be high. This precious small sample (note that the grid inigs. 3 and 4 consists of 0.5 cm squares) was cut into four equalieces as shown in Fig. 6 to study the effect of increasing the extrac-ion pressure on sample weight loss, as indicated in Fig. 7. The SFEesults showed a monotonic increase in removal of contaminantsith increasing extraction pressure, as depicted in Fig. 7. Fur-

her evidence of another extracted polymeric moiety was revealedy MALDI-TOF analysis of the Russian textile sample extract; thextracted oligomers differ by 58 Da units, as shown in Fig. 8,lthough the exact structure of this polymeric residue is still pend-ng. It should be noted that the FAME analysis of the Russianextile revealed only the presence of palmitic acid in striking con-rast to the diverse FAME content found in the child mummyTable 3).

.5. Canopic jar contents

Gravimetric weight loss on the partitioned sub samples fromhe Canopic jar is reported in Table 4. It is apparent from theseesults that there is a considerable amount of material extracted

rom the “resinous” samples while very little was removed by SFEreatment from the “soil” sub-sample. A little under 20% by weightas also removed from the “residual material” fraction taken fromanopic jar.

Fig. 8. MALDI-TOF-MS spectra of extracted polymer residue obtained from

40 ◦C using SC-CO2 modified with 10% (v/v) methanol.

The materials separated out from the Canopic jar contentswere treated as described in Section 2 with the SC-mixture, andthe extracts contained in the residual methanol subjected toGC/MS.

A typical profile for the residual material is shown in Fig. 9, whichshows considerable chemical diversity. Chemical compounds asso-ciated with mummification rituals such as camphor verbenone(from pine oil), cuparene (cedarwood oil), and myristic/stearic acids(vegetable or animal fat) have been positively identified based ontheir MS spectra match. The GC/MS profiles of the SC-fluid mix-ture extracts from the resin fraction, resin/tissue, and “dirt” showedsimilar profiles and compounds, although there were subtle differ-ences in chemical composition. The “grass sample” showed only thepresence of an alkane fraction, and palmitic and stearic acids. Thechemical composition of the SC-CO2/methanolic extracts is sim-

ilar to results reported in the literature for embalming materials[42,43].

Russian textile sample extracted at 61.3 MPa (9000 psig) and 40 ◦C.

322 M.W. Rowe et al. / J. of Supercritical Fluids 79 (2013) 314–323

Table 4Gravimetric results before and after SFE on the various Canopic jar fractions.

Sample Initial Final Change % Change

Grass 0.0079 0.0075 0.0004 5.1Resins 0.1310 0.0692 0.0618 47.2Resins with potential tissue 0.0369 0.0213 0.0156 42.3Potential soil 0.0349 0.0342 0.0007 2.0Residual material 0.1935 0.1585 0.0350 18.1

“resid

4

chafsinatogco

rs[cfic

A

–ov

Fig. 9. GC/MS profile of components in

. Concluding remarks

The above reported collective studies suggest that SFE utilizingompressed carbon dioxide with an appropriate cosolvent couldave wide applicability in sample pretreatment and the isolationnd identification of components contained in archeological arti-acts. These preliminary data confirm that in certain cases, SFE mayerve as an alternative, more benign, and non-destructive methodn sample pretreatment prior to 14C dating protocols. More studieseed to be conducted to further evaluate the applicability of thispproach, including additional artifact samples as well as alterna-ive cosolvents or additives with SC-CO2. To date, minimal effectsn the morphology of the artifact samples we have examined sug-est the technique can be applied to small and valuable samples,onsistent also with the repatriation of these materials to theirriginal owners.

An artifact of virtually any size can be cleaned in this way. It onlyequires sample chambers large enough to contain the artifact. Suchcaled up equipment is available and has been used by conservators44]. This study extends the use of liquid CO2 and SC-CO2-basedleaning processes which have been applied with success in diverseelds, including its commercialization in modern day fabric dryleaning [45].

cknowledgments

Support is acknowledged from the National Science FoundationSBE Grant No. 1006001. Sarah L. Chapman of the Department

f Anthropology – The University of Arkansas – Fayetteville pro-ided the Canopic jar specimens. The assistance of Jennifer Glidden

ual material” extract from Canopic jar.

of the State Wide Mass Spectrometry Facility at the Universityof Arkansas and Erik Pollock on the GC/MS studies is gratefullyacknowledged.

References

[1] K.L. Steelman, M.W. Rowe, Potential for virtually nondestructive radiocarbonand stable carbon isotopic analyses on perishable archeological artifacts, in:K.A. Jakes (Ed.), Archaeological Chemistry: Materials, Methods, and Meaning,vol. 831, American Chemical Society, Washington, DC, 2002, pp. 8–21 (Chapter2).

[2] K.L. Steelman, Non-destructive radiocarbon and stable isotopic analyses ofarcheological materials using plasma oxidation, Ph.D. thesis, Department ofChemistry, Texas A&M University, 2004.

[3] K.L. Steelman, R. Rickman, M.W. Rowe, T.W. Boutton, J. Russ, N. Guidon, Accel-erator mass spectrometry radiocarbon ages of an oxalate accretion and rockpaintings at Toca do Serrote da Bastiana Brazil, in: K.A. Jakes (Ed.), Archaeolog-ical Chemistry: Materials, Methods, and Meaning, ACS Symposium Series, vol.831, American Chemical Society, Washington, DC, 2002, pp. 22–35 (Chapter 3).

[4] K.L. Steelman, M.W. Rowe, Non-destructive plasma-chemical extraction ofcarbon from organic artefacts, in: T. Higham, C. Ramsey, C. Owen (Eds.), Radio-carbon and Archaeology, 4th International Symposium, St. Catherine College,Oxbow Books, Oxford, England, 2002, pp. 21–41.

[5] K.L. Steelman, M.W. Rowe, S.A. Turpin, T. Guilderson, L. Nightengale, Non-destructive radiocarbon dating: naturally mummified infant burial bundle fromSW Texas, American Antiquity 69 (2004) 741–750.

[6] J. Russ, M. Hyman, H.J. Shafer, M.W. Rowe, Radiocarbon dating of prehistoricrock-paintings by selective oxidation of organic carbon, Nature 348 (1990)710–711.

[7] M.W. Rowe, Radiocarbon dating of ancient rock paintings, Analytical Chemistry81 (2009) 1728–1735.

[8] M.W. Rowe, Dating studies of prehistoric pictographs in North America, in: L.L.

Loendorf, C. Chippendale, D.S. Whitley (Eds.), Discovering North American RockArt, University of Arizona Press, Tucson, AZ, 2005, pp. 240–263 (Chapter 12).

[9] M.W. Rowe, Reflections on dating of rock art, in: P.C. Reddy (Ed.), Exploringthe Mind of Ancient Man, Research India Press, New Delhi, India, 2007, pp.218–231.

critica

[

[

[

[

[

[

[

[

[[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

M.W. Rowe et al. / J. of Super

10] J. Russ, M. Hyman, M.W. Rowe, Direct radiocarbon dating of rock art, Radiocar-bon 34 (1992) 867–872.

11] S.D. Chaffee, M. Hyman, M.W. Rowe, Radiocarbon dating of rock paintings, in:D.S. Whitley, L.L. Loendorf (Eds.), New Light on Old Art: Recent Advances inHunter-Gatherer Rock Art Research, University of California at Los Angeles, LosAngeles, CA, 1994, pp. 9–12.

12] R.E.M. Hedges, C.B. Ramsey, G.J. Van Klinken, P.B. Pettitt, C. Nielsen-March, A.Etchegoyen, J.O.F. Niello, M.T. Boschin, A.M. Llamazares, Methodological issuesin the 14C dating of rock paintings, Radiocarbon 40 (1998) 35–44.

13] R.A. Armitage, J.E. Brady, A. Cobb, J.R. Southon, M.W. Rowe, Mass spectrometricradiocarbon dates from three rock paintings of known age, American Antiquity66 (2001) 471–480.

14] M. Sousa, M.J. Melo, T. Casimiro, A. Aguiar-Ricardo, The art of CO2 for art conser-vation: a green approach to antique textile cleaning, Green Chemistry 9 (2007)943–947.

15] B. Kaye, D.J. Cole-Hamilton, Supercritical fluids in conservation, in: M. Poliakoff,M.W. George, S.M. Howdle (Eds.), Proceedings of the 6th Meeting on Supercrit-ical Fluids, Institut National Polytechnique de Lorraine, Vandoeuvre, 1999, pp.107–113.

16] B. Kaye, D.J. Cole-Hamilton, K. Morphet, Supercritical drying: a new method forconserving waterlogged archaeological materials, Studies in Conservation 45(2000) 233–252.

17] F.Y. Saleh, P. Shao, Q.H. Jaradat, Step-wise fractionation and recovery of aquaticfulvic acid by modified supercritical fluid CO2–methanol extraction at near crit-ical temperature, International Journal of Environmental Analytical Chemistry73 (1999) 93–108.

18] S.K. Ritter, Green chemistry, Chemical and Engineering News 79 (2001) 27–34.19] H. Tello, A. Unger, Liquid and supercritical carbon dioxide as a cleaning and

decontamination agent for ethnographic materials and objects, in: A.E. Charola,R.J. Koestler (Eds.), Pesticide Mitigation in Museum Collections: Science in Con-servation, Smithsonian Institution Scholarly Press, Washington, DC, 2010, pp.35–50.

20] S. Werner, N. Odegaard, T.K. Moreno, R.A. Turner, M.R. Riley, B. Xie, A.J. Muscat,Pesticide extraction studies using supercritical carbon dioxide, in: A.E. Charola,R.J. Koestler (Eds.), Pesticide Mitigation in Museum Collections: Science in Con-servation, Smithsonian Institution Scholarly Press, Washington, DC, 2010, pp.51–57.

21] W.S. Zimmt, N. Odegaard, D.R. Smith, The potential for adapting some cleaningmethodologies to pesticide removal from museum objects, in: A.E. Charola, R.J.Koestler (Eds.), Pesticide Mitigation in Museum Collections: Science in Con-servation, Smithsonian Institution Scholarly Press, Washington, DC, 2010, pp.59–63.

22] R.J. Willey, A. Radwan, M.E. Vozzella, A. Fatafah, G. Davies, Humic acid gel dryingwith supercritical carbon dioxide, Journal of Non-crystalline Solids 225 (1998)30–35.

23] S.M. Kang, K. Sungmo, A. Unger, J.J. Morrell, The effect of supercritical car-bon dioxide extraction on color retention and pesticide reduction of woodenartifacts, Journal of the American Institute of Conservation 43 (2004) 151–160.

24] K. Srinivas, J.W. King, J. Monrad, L. Howard, C.M. Hansen, Optimization ofsubcritical fluid extraction of bioactive compounds using Hansen solubilityparameters, Journal of Food Science 74 (2009) E342–E354.

25] M. Serpico, R. White, Oil fat, and wax, in: P.T. Nicholson, I. Shaw (Eds.), AncientEgyptian Materials and Technology, Cambridge University Press, Cambridge,

England, 2000 (Chapter 17).

26] M.P. Colombini, F. Modugno, Organic Mass Spectrometry in Art and Archaeol-ogy, John Wiley, New York, 2009.

27] S. Bowadt, S.B. Hawthorne, Supercritical fluid extraction in environmental anal-ysis, Journal of Chromatography A 703 (1995) 549–571.

[

l Fluids 79 (2013) 314–323 323

28] L.N. Nikitin, E.E. Said-Galiyev, R.A. Vinokur, A.R. Khokhlov, M.O. Gallyamov,K. Schaumburg, Poly(methyl methacrylate) and poly(butyl methacrylate)swelling in supercritical carbon dioxide, Macromolecules 35 (2002) 934–940.

29] M.J. Cocero, E. Alonso, S. Lucas, Pilot plant for soil remediation with supercriti-cal CO2 under quasi-isobaric conditions, Industrial and Engineering ChemistryResearch 39 (2000) 4597–4602.

30] J.W. King, L.L. Williams, Utilization of critical fluids in processing semi-conductors and their related materials, Current Opinion in Solid State andMaterials Science 7 (2003) 413–424.

31] Y.-P. Chin, W.J. Weber Jr., Estimating the effects of dispersed organic polymerson the sorption of contaminants by natural solids. 1. A predictive thermo-dynamic humic-substance-organic solute interaction model, EnvironmentalScience and Technology 23 (1989) 978–984.

32] D. Kopinke, J. Porschmann, U. Stottmeister, Sorption of organic pollutants onanthropogenic humic matter, Environmental Science and Technology 29 (1995)941–950.

33] Y.P. Chin, W.J. Weber Jr., C.T. Chiou, A thermodynamic partition model for bind-ing of nonpolar organic compounds by organic colloids and implications fortheir sorption to soils and sediment, in: R.A. Baker (Ed.), Organic Substancesand Sediments in Water: Humics and Soils, Lewis Publishers, Chelsea, MI, 1991,pp. 251–273 (Chapter 14).

34] D.W. Rutherford, C.T. Chiou, Effect of water saturation in soil organic matter onthe partition of organic compounds, Environmental Science and Technology 26(1992) 965–970.

35] C.T. Chiou, D.E. Kile, Effects of nonpolar groups on the solubility of organiccompounds in soil organic matter, Environmental Science and Technology 28(1994) 1139–1144.

36] M.L. Proefke, K.L. Reinhart, M. Raheel, S.H. Ambrose, S.U. Wisseman, Probingthe mysteries of ancient Egypt. Chemical analysis of Roman period Egyptianmummy, Analytical Chemistry 664 (1992) 105A–111A.

37] J.W. King, J.E. France, J.M. Snyder, On-line supercritical fluid extraction-supercritical fluid reaction-capillary gas chromatography analysis of the fattyacid composition of oilseeds, Fresenius Journal of Analytical Chemistry 344(1992) 474–478.

38] M.A. Jackson, J.W. King, Methanolysis of seed oils in flowing supercritical carbondioxide, Journal of the American Oil Chemists Society 73 (1996) 353–356.

39] L.H. McDaniel, M. Ashraf-Khorassani, L.T. Taylor, Supercritical fluid extractionof wood pulp with analysis by capillary gas chromatography-mass spectrom-etry, Journal of Supercritical Fluids 19 (2001) 275–286.

40] V.T. Wyatt, M.J. Haas, Production of fatty acid methyl esters via the in-situ trans-esterification of soybean oil in carbon dioxide-expanded methanol, Journal ofthe American Oil Chemists Society 86 (2009) 1009–1016.

41] K.N. West, C. Wheeler, J.P. McCarney, K.N. Griffith, D. Bush, C.L. Liotta, C.A.Eckert, In-situ formation of alkylcarbonic acids with CO2, Journal of PhysicalChemistry A 105 (2001) 3947–3948.

42] J. Koller, U. Baumer, Y. Kaup, U. Weser, Herodotus’ and Pliny’s embalmingmaterials identified on ancient Egyptian mummies, Archaeometry 47 (2005)609–628.

43] J. Koller, U. Baumer, Y. Kaup, M. Schmid, U. Weser, Ancient materials: analysisof a pharaonic embalming tar, Nature 425 (2003) 784.

44] E.M. Francais, M. Perrut, A.-C. Brandt, Désacidification: renforcement de massedes papiers acides et fragilisés en utilisant un fluide supercritique.La Conser-vation: une science en evolution bilant et perspectives, Actes des Troisièmes

Journées Internationales d’ Études de l’ ARSAG, Paris, 21–25 Avril 1997, ARSAG,Paris, 1997, pp. 131–140.

45] G. Stewart, Dry cleaning with liquid carbon dioxide, in: J.M. Desimone, W.Tumas (Eds.), Green Chemistry Using Liquid and Supercritical Carbon Dioxide,Oxford University Press, NY, 2003, pp. 215–227 (Chapter 13).