Background

The aim of the Stuttgart RaCoPhInO programme (‘Rare Corrosion Phenomena of Inorganic Objects’) is to identify hitherto unknown phenomena and compounds occurring during the deterioration of heritage objects. Even after nearly 240 years of scientific studies on heritage objects since the identification of cuprite on a Roman bronze leg in 1779 [1] it is still possible to detect new copper corrosion products. This has been demonstrated recently in our GIMME research (Glass Induced Metal-corrosion on Museum Exhibits) for dicopper formate trihydroxide, Cu2HCOO(OH)3 [2, 3], and sodium copper formate hydroxide oxide hydrate, Cu4Na4O(HCOO)8(OH)2∙4H2O [4].

But in this case, it is the other way round: we present experimental and theoretical evidence that a rare copper compound identified on metal heritage in the literature is non-existing: basic sodium copper carbonate, Na3[Cu2(CO3)3(OH)]∙4H2O.

The evidence for basic sodium copper carbonate

Reports

Nandi and Sengupta [5, 6] reported the synthesis and characterization of the basic sodium copper carbonate Na3[Cu2(CO3)3(OH)]∙4H2O (compound ①, sodium tricarbonatohydroxodicuprate(II) tetrahydrate) in 1972/4. The only other occurrence of ① in the literature identifiable with a Chemical Abstract search (CAS no. 55521-72-7) is our first review of the GIMME phenomenon in 2010 [7]. In the discussion of chalconatronite (Na2Cu(CO3)2∙3H2O, a joint corrosion product of soda glass in contact with copper alloys in clean air free from formaldehyde) it was noted that ① has been mentioned by Barger and White [8] to occur on a cover glass of a daguerréotype (copper plate with photo-sensitized silver layer). Identified with X-ray Powder Diffraction (XRPD), it was not included in their earlier original research paper on corrosion products [9]. Bellendorf [10] also identified ① by XRPD together with chalconatronite in five corrosion samples from fourteen to eighteenth cent. burial plates in Franconian chapels in Bamberg and Rothenburg ob der Tauber cast from quaternary copper alloys (Cu/Zn/Sn/Pb). Here the sodium most likely originates from migrating salts in the wall plaster, not from soda glass.

Therefore, we tried to synthesize ① as reference for our GIMME research (e.g., for Raman spectra).

Synthesis

① was prepared by Sengupta and Nandi ‘by adding 50 ml of copper(II) acetate (3.2 g) solution to 250 ml NaHCO3 (10 %). The light blue needle-shaped crystals were filtered, washed and dried’ [6]. Our analyses by XRPD of products obtained that way surprisingly found chalconatronite as only crystalline phase. Rietveld refinement of the pattern proved the presence of amorphous material as well (estimated to 10 to 14 %, depending on washing).

As ① is a basic salt (i.e., it contains a hydroxide anion) in opposite to the ‘neutral’ chalconatronite, one might assume that raising the pH might help to precipitate the basic compound. But by using a sodium hydrogencarbonate/carbonate buffer chalconatronite is definitely synthesized as Sengupta and Nandi report themselves [6] (see also [11: 353] for a similar synthesis). At even higher pH, by using sodium carbonate (soda) alone, we yielded a totally amorphous product. Gettens and Frondel [12], the discoverer of the mineral chalconatronite, obtained ‘a pale blue-green precipitate closely similar to chalconatronite in chemical composition… and other properties…simply by grinding in a mortar copper acetate in a saturated solution of sodium carbonate, allowing to stand overnight, filtering, washing and drying.’ Erdös [13] grew chalconatronite on copper wetted with 10 % soda solution.

By no way, ① could be synthesized. Therefore, the physico-chemical data reported for ① were compared to those for chalconatronite.

X-ray powder diffraction

Sengupta and Nandi used Cu-Kα radiation, no further instrumental details given [6]. As they started their measurement from 2Θ = 20° (common experimental value: 5°) on, reflections for lower angles (i.e., d > 4.5 Å) are simply missed. Their data have been entered as ICDD 28-1048 in the reference database with the comment ‘o = doubtful’. If compared with our own measurement of a synthetic chalconatronite sample or with diffraction values derived from the crystal structure ([14], ICDD 01-71-1490, Table 1) it can be seen that the positions of all measured reflections for ① are in excellent, and the intensities in reasonable good agreement with chalconatronite, see also the experimental chalconatronite entry ICDD 22-1458 (Guinier camera, Table 1, right column). The agreement can be visualized by simulating a diffractogram from the structure [11] with a peak half width of 0.3°/2Θ (typical for measurements in the 1970’s) and superimposing ICDD 28-1048 (grey bars, Fig. 1). There can be no doubt that Sengupta and Nandi also synthesized chalconatronite and not the basic sodium copper carbonate ①. The identity of the diffractograms has not been noticed as the important characteristic peaks at higher d-values were missing (because not measured) and the chalconatronite card 22-1458 of 1969 (with d > 1.8 Å only) is of low quality (in modern standards) with intensities only estimated visually from a film.

Table 1 Comparison of XRPD data of ① (ICDD 28-1048, middle) with chalconatronite: data calculated from the crystal structure ([14], left) and ICDD 22-1458 (right). Grey shaded fields: no measurement in that range
Full size table
Fig. 1
figure 1

Diffractogram simulated from structure [11] in comparison to ICDD 28-1084 (grey bars)

Full size image

It is no wonder then that in mixtures were chalconatronite is present also the basic sodium copper carbonate, ICDD 28-1048, can be proposed by peak matching software (as happened for the burial plate samples), and in that way enters analytical reports.

IR-spectroscopy

The authors measured IR-spectra for ① [5, 6] and chalconatronite [5], but only published peak lists, which look very similar for both compounds (Note that lists for ① vary slightly in the earlier conference contribution [5] and the journal article [6], Table 2). For many peaks, the chalconatronite spectrum [5] is even more similar to ① than to a contemporary spectrum for chalconatronite [15] from another group (Table 2, right column). Therefore, there is no IR-spectroscopic proof that anything else than chalconatronite was measured and that ① exists at all.

Table 2 Peaks in IR-spectra of ① (left [5, 6]) compared to chalconatronite [5] and [15]
Full size table

Other physicochemical data reported by Sengupta and Nandi [6]

  • The authors found the same effective magnetic moments for ① and for chalconatronite.

  • The thermogravimetric curves for ① and for chalconatronite differed somewhat, but one cannot assess from the publication if this difference is really beyond normal experimental error or could be caused by the presence of some amorphous material.

  • Quantitative analytical data for Na+, Cu2+, and CO2 for their precipitate ① led to the supposed formula, they do not fit the stoichiometry of chalconatronite. But we were unable to get a phase pure crystalline compound by their synthesis for ①, 10–14 % amorphous material was present as well. From such a mixture, no meaningful stoichiometric formula can be derived by quantitative analysis of the mixture.

  • The same holds true for the isothermal weight loss measured for ① and chalconatronite (at slightly different temperatures (150 °C/160 °C).

Conclusion

Our synthetic replication experiments and a closer look at the diffraction data and the IR spectra show that Sengupta and Nandi [5, 6] have synthesized chalconatronite and not the basic sodium copper carbonate ① they postulated. There is no experimental proof of any kind that the latter compound exists at all, neither as lab chemical, nor as corrosion product on heritage objects. Consequently, ICDD 28-1048 should be deleted.

The chalconatronite reference set ICDD 22-1458 of 1969 is still often used in metal heritage corrosion studies (e.g., [16–18]) for identification. To analyse complex mixtures, it should better be replaced by data derived from the crystal structure (e.g., ICDD 01-71-1490 or [19]) with quantitative values for intensities.