The preparation of the ultramarine pigment for painting first required isolating of the lazurite present in lapis lazuli by 20–40%. The separation process was long and complex and was probably developed in Venice during the thirteenth century. The process was described in detail by the painter Cennino Cennini (ca. 1360–before 1427), a pupil of Agnolo Gaddi (1350–1396), in ‘The Book of the Art–Treatise on Painting’ [44], written in the first decade of the fifteenth century in vernacular Italian, with some variation of the Venetian dialect (Cennino wrote part of the book during a long stay in Padua). Chapter LXII (‘On the nature of ultramarine blue and how it is prepared’) details the extraction of pigment from lapis lazuli. The incipit of the chapter describes the value of this pigment to Renaissance painters: "Ultramarine blue is a colour noble, beautiful, and perfect beyond all other colours,…And with this colour, together with gold (which adorns all the works of our art) let everything be resplendent, whether on walls or panels." Then, the preparation of the pigment was explained in great detail. To remove mineral impurities and obtain pure lazurite, lapis lazuli was first ground to a powder and then mixed with a paste of resin, beeswax and mastic to form a fist-sized ball (pastello). After a long period (from 2 weeks to 1 month), the ball was put in a warm lye solution, squeezed and kneaded using two wooden sticks. During this operation, lazurite fragments were released, leaving a blue aqueous suspension, while the impurities remained in the pastello. Washing and crushing were repeated until the lye solution became colourless (up to 18 washes). Ultramarine pigment was obtained by filtering the aqueous suspension through a linen cloth and air-drying it. The pigment from the first two extractions, which was more brilliant and luminous, could be sold for 8 ducats per ounce. Since the gold ducat, the currency of the Most Serene Republic of Venice, had the same weight and value as the florin, the cost of the ultramarine blue pigment was ~ €60/g, according to Cennino.

The chapter ends with a questionable statement that today we would call ‘politically incorrect’: "You must also know that it is rather the art of maidens (more explicit in Italian Cennino’s text: ‘belle giovani’, i.e. young beautiful women) than of men to make it, because they remain continually in the house and are more patient and their hands are more delicate'. Now it happens that some young researchers (4 females, 5 males) under the supervision of Katrien Keune, Head of Science at the Rijksmuseum, Amsterdam, recreated the pastello extraction in the laboratory by following Cennino’s recipe as closely as possible [45]. “The men in our team were unable to cleanly isolate the ultramarine, but the female students succeeded in doing just that”, declared Keune at the end of the experiment, adding: “It seems that this requires a certain subtle skill.” This statement clears the Tuscan painter of any accusation of toxic patriarchal statements. Worse and unacceptable was Cennino's next sentence: “Beware of old women.” Cennino died at the age of 70. It is to be hoped that, at least in the last years of his life, the old painter changed his mind.

The students' extraction of ultramarine was part of a complex investigation which showed that heating lapis lazuli before extraction stabilised the pigment and greatly increased the intensity of the blue hue. This technique was unknown to Cennino and his Italian colleagues but had been used by Flemish painters and craftsmen since the fifteenth century, as shown by XANES analysis on samples taken from historical paintings (including the famous 'Girl with the Pearl Earring' by Jan Vermeer) [46]. Sulfur K-edge XANES technique had been previously employed to investigate the mechanism of lazurite extraction from lapis lazuli pigment made according Cennino’s recipe [47].

Ultramarine continued to be used in painting throughout the seventeenth century. In 1706, Dippel and Diesbach in Berlin invented Prussian blue, which was produced industrially within a few years [35], and the more brilliant but infinitely more expensive ultramarine was forgotten by painters. However, it was not forgotten by chemists who, as mentioned above, at the beginning of the nineteenth century turned their attention to the pigments used in ancient paintings. In 1806, Charles-Bernard Désormes (1777–1862) and Nicolas Clément (1779–1841), father-in-law and son-in-law, first separated lazurite from lapis lazuli following the procedure described by Cennino. They then subjected it to qualitative and quantitative analysis [48]. Analytical data are given in Table 2.

The authors observed that calcium carbonate was not essential to the composition of ultramarine. Instead, they unquestionably characterised the other four components. There was astonishment and disbelief among readers that no metals or metal compounds were present in the pigment, thus shattering the so-called metal paradigm of coloration: the colour of any earthy substance must be imparted by metal oxides and salts [49]. However, the authors presented incontrovertible chemical evidence that the colour was due to sulphur or one of its derivatives: when strong acid was added, H2S developed and the pigment lost its colour. Désormes and Clément concluded their memoir with the wish: "May this essay on a substance so little known and so unique be followed by its artificial production."

In 1814, Benjamin Tassaert, director of the soda factory of Saint-Gobain at Chauny (Hauts-de-France), informed his former professor at the École Polytechnique and Academician Louis Nicolas Vauquelin (1763–1829) that blue deposits remained in a stoneware kiln for the production of Na2CO3 (Leblanc process). When the kiln was rebuilt in brick, the blue deposits no longer formed. Tassaert sent a sample for analysis to Vauquelin, who found a composition similar to that of ultramarine and reported it to the Académie. This demonstrated that the ultramarine could be prepared artificially [50].

In 1824, the Société d'encouragement pour l'industrie nationale in Paris announced a prize of 6000 francs (~ 27,000 € today) for the production of an artificial ultramarine pigment that had the same properties as natural ultramarine and whose price did not exceed 300 francs/kg (1350 €/kg today). Note that the natural pigment was 100–1000 times more expensive.

In October 1826, Jean-Baptiste Guimet (1795–1871), an engineer at the Ferme Générale des Poudres et Salpêtres in Toulouse, a state company responsible for producing gunpowder, developed a method for making ultramarine but did not publish it. In July 1827, in contact with the painting world through his wife, the painter Rosalie Bidauld (1798–1876), he gave some of the pigment to Dominique Ingres (1780–1867), who tested it on the oil painting 'The Apotheosis of Homer', which was to decorate the Charles X Museum in the Louvre (Fig. 17, left) and shown to the public at the Salon de l'Académie royale de peinture et de sculpture (4 November 1827).

Left: Jean-Auguste-Dominique Ingres, 'The Apotheosis of Homer' (1827), oil on canvas, 386 × 512 cm, Louvre, Paris. The blue cloak of Apelles, on the left of the painting, was painted with artificial ultramarine. Right: Raphael, 'The Parnassus' (1509–1511), fresco, 670 cm wide, Apostolic Palace, Vatican Museums, Vatican City. Ingres’ 'Apotheosis' was probably inspired by Raphael's fresco; in this painting, Homer's cloak, on the left, was painted with ultramarine (but natural)

When the great Neoclassical painter saw the result, he declared he preferred it to any other commercial blue. At the end of 1827, Guimet sent a sample of ultramarine to Joseph Louis Gay-Lussac (1778–1850), Président de l'Académie des Sciences, without revealing the synthesis but stating that he had followed the indications implicit in Désormes and Clément's analysis. Gay-Lussac announced at the 4 February session of the Académie royale des Sciences that "Mr. Guimet, Deputy Commissioner of Poudres et Salpêtres, had succeeded in producing ultramarine by himself by combining the principles discovered by chemical analysis. "The new product is richer in colour and brighter than natural lapis lazuli" [51].

Guimet never published the method of preparation, nor did he want to patent it so as not to give any useful information to possible industrial competitors. However, we can deduce the synthesis based on the following scientific information available to him: (1) the soda furnace at Saint-Gobain factory was made of gres, composed of clay, feldspar, kaolin (= Al2Si2O5(OH)4 = (Al2O3·2SiO2·2H2O)) and sand, thus containing two of the components of ultramarine found by Désormes and Clément: Al2O3 and SiO2; (2) the Leblanc process for synthesising soda, carried out at the Saint-Gobain factory in a kiln at 900 °C, proceeds trough two steps:

$$}_} }_} + } \to }_} } + }_}$$

(3)

$$}_} } + }_} \to }_} }_} + }$$

(4)

Guimet probably hypothesised that the blue deposits found in the kiln were the result of reactions involving components of gres as well as some reactants and products of the Leblanc process. In 1878, Émile Guimet, Jean-Baptiste's son and successor, who had consulted his father's laboratory notebooks, reported a poorly detailed description of two different synthetic procedures [52], which are summarised below and illustrated with hypothesized corresponding chemical equations.

(1) A crucible containing kaolin, sodium carbonate, sulphur and sodium sulphate is placed in a furnace and heated to red heat; Al2O3, SiO2, Na2CO3 and Na2SO4 probably react to form Na8[SiAlO4]6SO4, the sodalite containing the sulphate ion (corresponding to the mineral noselite, Na8[SiAlO4]6SO4·H2O, which has the same structure as sodalite, with sulphate ions occupying the truncated octahedron cavities, one doing so and one not) [53]. The four sodium ions interact with the four oxygens of each SO42–, along the axis of the tetrahedron, whose Na+–Na+ edge is 5.20 Å:

$$}_} }_} + }_} + }_} }_} + }_} }_} \to }_} \left[ }_} } \right]_} }_} + }_}$$

(5)

(2) Simultaneously, sulphur dissolved in the molten soda disproportionates to form sulphide and thiosulphate [54]:

$$} + }_} }_} \to }_} } + }_} }_} }_} + }_}$$

(6)

(3) As heating proceeds in the presence of air, oxygen oxidises the sulphide to the anion radicals S2·– (yellow) and S3·– (blue), so that the melt appears green; when all S2·– has been oxidised to S3·–, the melt turns blue:

$$}_} } + }_} \to }_} } + }_}$$

(7)

$$}_} + }_} \to }_} + }_} }$$

(8)

Sulphur, if present in excess, may react with NaS2 to give NaS3 [55]:

On further heating, the melt becomes white. Presumably, S3·– and any remaining sulphur derivative are finally oxidised to sulphate (white).

(4) This white powder is mixed with charcoal and the mixture is heated to red heat: the melt turns blue again: charcoal reduces the sulphate to sulphide (Eq. 3, as in the Leblanc process) and again the oxygen oxidises the sulphide to S3·–, (7) and (8). Heating is stopped and the cooled blue mass is washed with water (and Émile reports finding sulphate and thiosulphate in the washing solution) to give ultramarine.

The process described above may seem unnecessarily elaborate and 'redundant'. In the same note, Émile succinctly describes a simpler alternative process in which ultramarine is prepared in a single step by heating a mixture of kaolin, sodium sulphate and charcoal to red heat in the presence of air. It is likely that this is the original process devised by Jean-Baptiste. This is suggested by the existence of a rough copy of a letter from J.-B. Guimet to Barthélémy Bérard, Marseilles, dated 28 October 1826, asking him to send 500 kg of sodium sulphate and 100 kg of sodium carbonate [56]. This also proves that J.-B. Guimet had begun to produce ultramarine on an industrial scale by the end of 1826, using the single step process. É. Guimet adds in the note that charcoal could be replaced by hydrogen or ammonium chloride, reducing agents suitable for laboratory synthesis but too expensive to be used in industrial production [57]. Guimet’s artificial ultramarine showed a more intense and brilliant colour than that obtained from lapis lazuli, using Cennino’s procedure, which was contaminated with pyrite and calcite impurities that were not completely removed during purification.

Gay-Lussac's four-line communication to the Académie was taken up and disseminated by French and German newspapers (Schwäbischer Merkur in Stuttgart on 28 February). One particularly interested reader was Christian Gottlob Gmelin (1792–1860), professor of chemistry at Tübingen. Gmelin had been studying ultramarine since 1826, analysing commercial and natural pigments, thus developing a laboratory method of synthesis. When he read about the production of artificial ultramarine, he rushed to publish his findings. First, he reported a detailed description of the synthesis in the German journal Hesperus − Encyclopädische Zeitschrift für gebildete Leser (Encyclopaedic journal for educated readers), which was translated into French by Justus von Liebig (1803–1873). He had been a student of Gay-Lussac in Paris and sent this as a note to the Académie [62]. The synthesis proceeded according to the following steps: (1) hydrated silica and alumina were dissolved in an aqueous solution of caustic soda. The solution was dried to give a white powdery mass. (2) Then, a molten mass of sulphur and soda was slowly added to the white mass in a crucible at red heat. (3) Heating in air was maintained for 1 h. After cooling and repeating the washing with water, ultramarine was obtained in crystalline form.

Notably, Gmelin used reactants derived from Désormes and Clément's analysis (SiO2, Al2O3, S, Na2CO3) but not Na2SO4. He also did not use the charcoal introduced by Guimet as a reducing agent. Thus, Gmelin’s synthesis is similar to the first part of a ‘redundant’ Guimet’s process, the S3·– radical anion being produced by stepwise air oxidation of the sulphide formed through the disproportionation of sulphur in molten soda (Eq. 6).

The note began with a polemical premise: Gmelin reported that during a stay in Paris in the spring of 1827, he had informed a number of chemists, including Gay-Lussac, that he was working on the synthesis of artificial ultramarine, and he complained that Gay-Lussac had not informed him at the time that others in France were working on the same subject. Gay-Lussac replied, in a following note in the same issue of the Annales [63], that he could not know what was going on 200 km away (i.e., in Guimet's laboratory in Toulouse) and that, in any case, everyone was aware of the possibility of synthesising ultramarine after Vauquelin-Tassaert's communication, particularly because 4 years earlier the Société d'encouragement pour l'industrie nationale had announced a prize for those who developed a convenient industrial preparation method for artificial ultramarine. Interestingly, the minutes of the 23 June meeting of the Académie mentioned that Gay-Lussac had presented a sample of the ultramarine made by Guimet [64].

The matter concluded on 3 November 1828 when the Société d'encouragement pour l'industrie nationale awarded the prize to Jean-Baptiste Guimet. On this occasion, Léonor Mérimée (1757–1836), a painter, chemist and distinguished member of the Chemical Arts Committee of the Société, presented the report giving the reasons for the award [65]: (1) the ultramarine produced by the applicant had the same chemical properties as natural ultramarine and had already been used to the satisfaction of several painters, including Ingres; (2) the retail price (600 fr/kg) was twice the price limit set by the Société, but, as the applicant pointed out, the intensity of the colour was at least twice that of commercially available ultramarine of natural origin; (3) at the request of the Committee, the applicant, who was entitled to keep the synthesis process secret, had disclosed the details of the preparation to an authoritative expert of his choice (Vauquelin), who expressed his firm belief that the pigment had been entirely artificially prepared from scratch.

Guimet soon realised that ultramarine, an unsurpassed blue pigment for painting, could have less artistic but more profitable uses, such as in blueing paper, textiles and linen, replacing smalt (cheaper but much less intense). In 1830, he built a factory in Fleurieu-sur-Saône, Lyon, for large-scale production. In the same year Friedrich August Köttig (1794–1864), laboratory director at the Royal Porcelain Manufactory in Meissen (Dresden), devised a way to manufacture artificial ultramarine blue (Lasursteinblau), inspired by Gmelin’s method [49].

The synthesis of the ultramarine blue pigment has not changed much in 200 years. See, for example, the recipe used by undergraduate students in the inorganic chemistry laboratory exercises [66]: a porcelain crucible containing a mixture of kaolin (100 parts by weight), anhydrous sodium carbonate (100 parts by weight), charcoal (12 parts by weight) and sulphur (60 parts by weight), finely divided and lightly pressed, is heated for 3 h on a Bunsen burner, first with the lid on and then in the open air. After cooling, the coloured part is separated from the uncoloured part, washed thoroughly with water, dried and ground. The yield is quite low: from 15 to 18 g of reactants, 1–3 g of product is obtained. However, the yield can be increased if the heating time is increased (beyond the time normally allocated to laboratory exercises).

Ultramarine, both natural and artificial, is a permanent pigment, but unlike Egyptian blue, it is not acid-resistant, a feature discovered by Désormes and Clément in 1806 [48]. In acidic media, S3·− is unstable and disproportionates according to Eq. (10), a process causing the disappearance of the blue colour:

$$}_}^ + }^ \to }_} } + }$$

(10)

Thus, paintings containing ultramarine blue can be damaged by acid rain when outdoors. The destruction of ultramarine by acids has been monitored through colorimetric techniques and solid-state 29Si and 27Al NMR studies, which have demonstrated that colour fading (i.e., the reaction of S3·− with H+) is preceded by the acid-induced collapse of the crystalline framework [67]. Discolouration of the pigment in mural paintings can be induced by biocontamination by microorganisms (fungi, yeast and bacteria) capable of releasing acidic toxins [68].

In the nineteenth century, artificial ultramarine (called Bleu Guimet in France, French blue in other countries) enriched the painters' palette, joining other industrially produced blues, notably cobalt blue and Prussian blue. Technological advances had made life easier for painters, who no longer have to prepare their colours from pigments and binders but can buy them directly from merchants in convenient tin tubes and carry them around with canvas and easel to capture the ‘impression’ of landscapes. Figure 18, left, shows a painting by Claude Monet (1840–1926), a modernist and avid supporter of artificial colours. Hence, he enthusiastically used ultramarine. However, most painters used different types of blue on the same canvas. Among them was Vincent Van Gogh (1853–1890), who used both ultramarine and cobalt blue in his famous painting 'Starry Night' (Fig. 18, right).

Left: Claude Monet, 'The Gare Saint-Lazare: Arrival of a Train' (1877), oil on canvas, 83 × 101 cm, Harvard Art Museum; Cambridge, Massachusetts; all the blue is ultramarine. Right: Vincent Van Gogh, 'The Starry Night' (1889), oil on canvas, 74 × 92 cm, Museum of Modern Art, New York; the deep blue surrounding the stars is predominantly ultramarine; brighter blue of the swirling sky and the area surrounding the moon is predominantly cobalt blue

Ultimately, the ease and versatility of ultramarine synthesis depends on (1) the tendency of sulphur to concatenate and to the easy formation of polysulphides by reduction of sulphur or its derivatives; (2) the reducing tendency of polysulphides whose oxidation in air produces unstable radical intermediates such as the blue S3·– chromophore. Fortunately for painting and humans, sodalite is there, ready to welcome S3·– and offer it a safe haven, where it can shine for centuries and millennia to come.