INTRODUCTION
Usually, paper is a cheap mass product. However, as a medium for historic or important documents and works of art, paper can be of high value. Hence, paper objects are more and more subject to art forgeries and fraud (Bredekamp et al. Reference Bredekamp, Brückle and Needham2014). In most cases, the detection of forgeries cannot be done solely on the basis of expert textual, calligraphic, or art-historical analyses. Therefore, scientific analytical methods also have to be employed to determine the real age and even the provenance of a paper (Pigorsch et al. Reference Pigorsch, Finger, Thiele and Brunner2015).
Paper as we know it today is a composite, consisting of several organic and inorganic components. For example, until the end of the 19th century the main components of paper were rag fibers (linen or wool fibers), which were than replaced by wood-based cellulose fibers (e.g. Beneke Reference Beneke1999). In addition to fibrous material, paper also contains fillers, pigments, and sizing agents (organic and inorganic), whose introduction and use can be assigned to certain date ranges (e.g. Clapp Reference Clapp1972). However, the detection of chemical components with a known date of invention will only give a terminus post quem date for the production of the paper.
A significant change in paper and cardboard production occurred in the mid-1960s by adjusting the pH from an acid to a neutral pH (~6.8–7.5) environment. This resulted in numerous changes also in material composition of paper and cardboard material, such as the use of synthetic sizing agents in exchange of rosin-sizing, the use of calcite as a filler, and further changes of used additives. An important production change with respect to radiocarbon (14C) was the increased application of starch to the paper pulp and surface coating of paper (Gullichson and Paulapro Reference Gullichsen and Paulapro1998; Götsching et al. Reference Göttsching and Katz1999). The starch sources are annual grown plants like potatoes and cereals, with an assumed short time lag between harvesting and final use as binder.
14C measurements on short-lived plant tissues formed after 1955, i.e. during the so-called bomb-pulse, will allow a comparatively precise age estimate within a few years due to the large changes in atmospheric 14C concentrations (e.g. Harkness and Walton Reference Harkness and Walton1969,Reference Harkness and Walton1972; Stenhouse et al. Reference Stenhouse and Baxter1977; Geyh Reference Geyh2001; Canosa et al Reference Canosa, Hodgins and Weaver2013; Al-Bashaireh et al Reference Al-Bashaireh, ElSerogy, Hussein and Shakhatreh2015). As the source of the cellulose fibers in modern paper are trees with a complex chronological signature, thus limiting the usefulness of radiocarbon measurements of paper cellulose, our proposition is to use starch for 14C measurements.
In this case study, we performed 14C measurements on post-bomb known-age paper and paper-like samples on bulk-paper, i.e. cellulose fibers, and starch extracts to assess the possibilities and limitations for estimating paper production dates.
MATERIALS AND METHODS
Materials
For this study, 20 paper and cardboard samples from a collection of the Papiertechnische Stiftung (PTS), Germany, have been selected (Table 1). The majority of the samples originate from the so-called Weissenborner Musterzimmer, archived in the German National Library in Leipzig, Germany, which contained the whole collection of the Weissenborn paper factory. From around 1930 until the mid 1990s, this paper factory (now belonging to Schoeller Technocell™) collected annual archetypes of current paper productions. Additional paper samples, partially printed, came from private collections and the Zentrum für Bucherhaltung in Leibzig, Germany (ZfB; a center for book preservation). For the latter samples, the production year was approximated from the year of publication (see Table 1).
Samples for 14C analysis and results. Numbers given in parenthesis give measurement amounts of specific fractions.
* Average NH1 14C concentration (Hua et al. Reference Hua, Barbetti and Rakowski2013) at the time of production;
† extrapolated average NH1 14C concentration;
** source of paper samples (A: “Weissenborner Musterzimmer,” archive of German National Library; B: private collection of art historian Georg Dietz (www.papierstruktur.de); C: ZfB; D: paper prototype collection PTS).
All paper samples are in principle so-called technical paper, i.e. writing and printing paper (with and without printing), cardboard for drawings, packaging-paper, and packaging-cardboard. All were produced using cut wood-fibers over Fourdrinier wire and cylinder mold paper machines, respectively. Overall, the selected samples are assumed to mirror a representative spectra of manufactured paper over the last 65 years within Central Europe.
Spectroscopic analysis (FTIR, Raman) of the paper samples were performed at PTS and used to identify paper samples containing starch (Table 1; not quantitative). Aside from cellulose, other ingredients such as kaolinite, gypsum, barite, styrene-binder, and resin-glue were detected (for a detailed overview see Pensold and Pigorsch Reference Pensold and Pigorsch2015).
Sampling for subsequent sample preparation occurred outside printed areas along the paper edges. Eight paper samples were analyzed on bulk-paper material and extracted starch, nine samples on bulk-paper material alone, due to limited starch content, and two samples only on their starch extracts in a preliminary phase of this study.
Sample Preparation and AMS-Measurements
Bulk Paper Samples
To remove hydrophobic components such as resin or styrene binder as indicated by material analysis, bulk-paper samples (~20–60 mg original sample material, cut in mm-sized stripes) were subjected to a soxhlet-type serial extraction. In sequence, samples were extracted several times each with boiling tetrahydrofurane (THF), chloroform, petroleum-ether, acetone and methanol and then rinsed with deionized water (Bruhn et al. Reference Bruhn, Duhr, Grootes, Mintrop and Nadeau2001). Subsequently, samples were extracted with 1% HCl at 85°C, washed in deionized water and finally dried at 60°C.
Starch Samples
Three grams of the starch-containing paper samples were crushed in a grinding mill and left to soak 12 hr in 50 mL of distilled water. The resulting pulp was beaten in a mixer and left 30 min in an autoclave at 137°C. The final suspension was centrifuged to separate the fibers completely from the aqueous phase. Calcium carbonate was precipitated by adjusting the pH value to about 8 and by adding 1M disodium carbonate solution. The solution was again centrifuged and the calcium carbonate discarded. Finally, starch was precipitated by adding drops of ethanol to the aqueous solution. After decanting the liquid, the precipitate was dried, leaving between 2 and 30 mg of starch for 14C measurements. Due to the high vapor pressure, ethanol is assumed to be removed quantitatively. The final extraction product was qualitatively inspected by Raman spectroscopy to verify that its main component was starch.
AMS Measurements
For AMS-measurement, an aliquot of prepared sample material (cellulose, starch extract) estimated to contain 2–3 mg carbon was converted to CO2 by sealed quartz-tube combustion with preconditioned copper(II) oxide and silver wool at 900°C. Sample CO2, equivalent to 1 mg carbon, was graphitized by the Bosch reaction with an iron catalyst (Vogel et al. Reference Vogel, Southon, Nelson and Brown1984; Nadeau et al. Reference Nadeau, Grootes, Schleicher, Hasselberg, Rieck and Bitterling1998). The resulting mixture of graphite and iron powder was pressed into aluminium target holders for AMS 14C measurements with a 3 MV HVEE Tandetron AMS system at the Leibniz Laboratory in Kiel, Germany. Sample 14C measurements are normalized to modern oxalic acid II standard (NBS SRM 4990C) and corrected for isotopic fractionation and background effects (Nadeau et al. Reference Nadeau, Grootes, Schleicher, Hasselberg, Rieck and Bitterling1998, Reference Nadeau and Grootes2013). Final 14C concentrations, expressed in percent modern carbon (pMC; Stuiver and Pollach Reference Stuiver and Pollach1977), are shown in Table 1.
With a few exceptions (see Table 1), multiple 14C measurements of selected paper fractions were performed and a weighted mean calculated. For quality assurance, measurements were controlled by parallel measurements of IAEA C3, -C5, and IAEA C6 standards.
For comparison to measured paper cellulose and starch 14C concentrations, we used the NH_zone1 (Hua et al. Reference Hua, Barbetti and Rakowski2013) and Levin datasets (Levin et al. Reference Levin, Kromer and Hammer2013).
RESULTS AND DISCUSSION
Table 1 and Figure 1 give the results for the 14C measurements on paper cellulose and starch samples. Measured 14C concentrations are plotted versus paper production year.
14C concentrations measured in starch (rectangles) and solvent-cleaned paper cellulose (circles) with respect to paper production year and compared to Northern Hemisphere NH1 dataset (Hua et al. Reference Hua, Barbetti and Rakowski2013) and western European atmospheric 14C concentration (Levin et al. Reference Levin, Kromer and Hammer2013) concentration.
With the exception of sample 2010_FK (year 2010; multilayer, leveled folding boxboard), all paper starch extracts gave 14C concentrations close to or slightly above contemporaneous atmospheric 14C concentrations. Visually fitting apparent measured starch-14C concentrations to equivalent atmospheric 14C values indicates a lag time up to 5 yr between harvesting of the source material (mostly potatoes) and its use as a paper additive.
The starch 14C concentration in sample 2010_FK is significantly lower than atmospheric 14C concentrations in 2010, by about 4 pMC, and may indicate contamination during the starch extraction process. Further studies are needed to evaluate the applied starch extraction method with respect to efficiency, purity of the starch extract, and possible contamination problems. For example, the basic pH 8 step during the starch extraction process may pose a risk of atmospheric carbon contamination, emphasizing the need for improving the extraction protocol. Nevertheless, the close agreement of the other measured starch extracts with the expected atmospheric radiocarbon concentration at their production time shows that other, unidentified organic substances still abundant in starch extracts (not quantifiable during Raman inspection), or atmospheric carbon (which may have been introduced during the basic phase of the starch extraction process), seem not to have had a significant influence on measured 14C concentration.
In contrast to the short-lived starch extracts, paper cellulose 14C concentration give a poorer fit with the atmospheric 14C data, obviously caused by the complex multi-seasonal cellulose mixture.
In similar studies with radiocarbon measurements on paper cellulose of known-age paper material, Zavattaro et al. (Reference Zavattaro, Quarta, D’Elia and Calcagnile2007) and Fedi et al. (Reference Fedi, Caforio, Mandò, Petrucci and Taccetti2013) estimated apparent tree-ring cellulose admixtures of a few decades.
Adopting the approach of Zavattaro et al. (Reference Zavattaro, Quarta, D’Elia and Calcagnile2007), we modeled hypothetical paper cellulose 14C concentrations with tree-ring admixtures of 10, 25, and 50 years using
$${}_{{}}^{{14}} C_{{paper{\_}year\,}} }} \,{\equals}\,{1 \over {N^{2} }}\mathop{\sum}\limits_N^{n{\equals}1} {\left( {\left( {2n{\minus}1} \right){\asterisk}{}_{{}}^{{14}} C_{{cellulose\,growth\,year\,n}} } \right)} $$
with 14Cpaper_year being the apparent 14C concentration of bulk cellulose at the time of paper production, N = maximum number of tree-rings (e.g. 10, 25, and 50) and 14Ccellulose growth year n being the atmospheric 14C concentration of the tree-ring growth at year n, approximated by calculating mean annual 14C concentration using the NH1 data set (Hua et al. Reference Hua, Barbetti and Rakowski2013). The term n – 1 refers to the assumption that there is a minimum time-gap of 1 yr between felling of trees and utilization of tree-ring fibers for the paper.
Figure 2 shows the measured bulk paper 14C concentrations in comparison to atmospheric and modeled 14C concentrations. With respect to the samples discussed here, paper produced before 1982 seem to contain fibers with tree-ring mixtures spanning >50yr. For later produced paper samples, however, no clear distinction can be made between short or long-lived tree-ring mixtures.
Measured paper cellulose 14C concentration in comparison to atmospheric 14C data and modeled 14C for paper-pulp containing tree-ring mixtures of 10–50 yr.
Fedi et al. (Reference Fedi, Caforio, Mandò, Petrucci and Taccetti2013) inferred tree-ring cellulose mixtures in paper fibers spanning 10–25 yr for newspaper material before 1970. Later produced newspaper material seemed to indicate longer-lived tree-ring admixtures, comparably to our results. An obvious explanation could be changing production procedures and source materials of paper manufacturers. The use of recycled paper material, either in part or completely, would resemble the utilization of longer-lived tree-ring cellulose mixtures.
Clearly, 14C dating of multi-yearly paper cellulose will allow an only vague estimate of paper production year in contrast to a more precise estimate by radiocarbon measurement of short-lived organic substances such as starch. However, even then one has to accept uncertainties in production-year estimates, which results from the form of the bomb-spike, that is one solution lying on the increasing and one on the later decreasing part.
Using two or more sample fractions with different residence times may allow further discrimination. Provided that paper fibers consist of multi-yearly tree-ring cellulose mixtures, a positive 14C difference may be expected between short-lived organics and paper fibers during the early and increasing phase of the bomb-pulse. Depending on the date range of mixed tree rings, negative 14C differences could be expected on the decreasing side of the bomb-pulse.
The timing of this cross-over point, i.e. the change from positive to negative 14C differences between contemporaneous starch (assuming a minimum 1-yr time lag between harvesting and utilization during paper manufacturing) and a certain cellulose mixture, is illustrated in Figure 3. A short, 10-yr paper cellulose mixture will give negative 14C differences for paper manufactured after 1969, for a larger cellulose mixture, e.g. 50 yr, this transition occurs after 1979.
Modeled 14C difference between short and long-lived organic paper components (starch–paper cellulose).
The combination of radiocarbon measurements of short-lived and long-lived paper fractions therefore could increase the precision for dating paper materials from the post-bomb period, particularly in the near future when atmospheric 14C concentration will fall below the reference value in 1950 of 100 pMC (Graven et al. Reference Graven2015).
CONCLUSIONS
By 14C analysis of 20 paper samples from the post bomb-period on paper cellulose material, containing long-lived fibers, and short-lived starch extracts, the possibilities and limitations of absolute dating of paper has been investigated.
Measured 14C concentrations of paper cellulose show an only poor fit with atmospheric 14C concentrations of the past 65 yr. In comparison with modeled 14C concentrations of hypothetical 10–50-yr tree-ring cellulose admixtures, measured 14C concentrations seem to indicate mean apparent mixing periods of >50 yr. Cellulose in recycled paper material will therefore also represent a large temporal cellulose mixture. In a forensic investigation, 14C measurements on bulk paper cellulose would inevitably allow an only imprecise age estimate before or after 1955.
In contrast, 14C-AMS measurements on short-lived starch extracts generally are in reasonably good agreement with atmospheric 14C concentrations and indicate an only small time-lag up to 5 yr between growth and harvesting of source material and subsequent utilization during paper manufacture.
With respect to 14C dating of post-bomb paper material, combining 14C measurements from single- and multi-seasonal grown material could potentially increase the precision of production-year estimates by helping to differentiate between periods in the early or later phase of the bomb-spike.
Acknowledgment
This work has been financially supported by the German Federal Ministry of Economics and Technology BMWi to SP and EP.
