The State of Ammonia Synthesis at the Turn of the Twentieth Century: The Arena for Discovery

Abstract

When Fritz Haber took up the problem of ammonia synthesis in 1903 there were several methods of obtaining fixed nitrogen, both naturally and through synthesis in chemical reactions. The main natural sources were Chile saltpeter (sodium nitrate from South America) and ammonium sulfate from gas works and coking (and still some guano).

When Fritz Haber took up the problem of ammonia synthesis in 1903 there were several methods of obtaining fixed nitrogen, both naturally and through synthesis in chemical reactions. The main natural sources were Chile saltpeter (sodium nitrate from South America) and ammonium sulfate from gas works and coking (and still some guano). Although production was rising, both were limited stock supplies, not considered viable for increased future needs of the fertilizer, dye stuff or, explosives industries. Furthermore, the Chilean exports were, from the perspective of a country like Germany, precarious in that they could be cut off by a sea blockade (Schwarte 1920, pp. 537–543), (Haber 1971, pp. 84–85), (Tamaru 1991), (Stoltzenberg 1994, pp. 137–138), (Szöllösi-Janze 1998b, pp. 155–158 and citations therein), (Smil 2001, pp. 39–51), (Cushman 2013, Chapter 2).

There were two synthetic options for producing fixed nitrogen in 1903, which still contributed only small quantities to the nitrate market. They were briefly examined in Part I but are presented here in more detail to supply context for the comprehensive discussion of ammonia synthesis that follows.

The first was the electric arc in which oxygen and nitrogen were directly combined to NO or higher oxides using high temperatures from an electric discharge. That is: N₂ + O₂ ↔ 2 NO at temperatures between 2000 and 3000 ^∘C. It was an imitation of the natural fixation process by lightning in the atmosphere (Lunge 1916, pp. 231–237, 257–269), (Mittasch 1951), (Haber 1971, pp. 85–88), (Grossmann 1974, pp. 371–372), (Holdermann 1960, pp. 48–50), (von Nagel 1991, pp. 9–12), (Stoltzenberg 1994, pp. 135–137), (Szöllösi-Janze 1998b, p. 158), (Smil 2001, pp. 53–55). The process itself had been studied since the end of the 1700s but did not advance to viable industrial application for another century. Early efficiencies were low because the arc was unstable and localized in space so only a small portion of the surrounding air was oxidized. Furthermore, the reaction is reversible and the equilibrium at high temperatures needed for appreciable oxidation rates of N to produce NO becomes unfavorable for the oxide upon cooling (as decomposition rates increase) until temperatures fall below 1000 ^∘C. A technique was needed by which a fast reduction in temperature could preserve the oxidized nitrogen. Between 1902 and 1903, the Norwegians Kristian Birkeland, a physics professor, and Samuel Eyde, an engineer, developed the first industrial process by using a magnetic field to disperse the electric arc into the form of a disc. Air could be blown perpendicularly through it so that not only was the NO swept away quickly from the point of high temperature, but a larger volume of nitrogen could be oxidized. Implementation of the Birkeland-Eyde process began at Notodden in 1904 and by 1906, three ovens were producing fixed nitrogen in the form of calcium nitrate (also known as Norwegian saltpeter). Before Birkeland and Eyde’s collaboration, the Badische Anilin- und Soda-Fabrik (BASF) had begun work on the electric arc in 1897 led by Otto Schönherr, a chemist, and Johannes Hessberger, an electrical engineer. While they made slower progress than the Norwegians, a process was developed between 1905 and 1907 in which a meters-long arc was surrounded by tangentially spiraling air. It was also able to overcome the difficulties of rapid cooling and oxidation of large volumes of gas which had beset earlier processes. A trial plant was built in Ludwigshafen, but due to limited success, BASF replaced it in 1907 by a larger plant in Christiansand, Norway, this time in cooperation with the owners of the Birkeland-Eyde process under the name Norsk Hydro. Norway had a decisive advantage for the electricity-intensive electric arc process: hydropower was plentiful and cheap. Only under such conditions could the process be made economical. The NO created in the arc oxidized further upon cooling to form NO₂ and treatment with water resulted in the formation of nitric acid (HNO₃). This was combined with limestone to produce calcium nitrate, Ca(NO₃)₂, which, when concentrated to ∼75%, solidified and could be brought to market. It was as effective as Chile saltpeter.

In 1911, BASF pulled out of Norsk Hydro–for several years their focus had been shifting toward Haber’s direct synthesis of ammonia. While the decision caused the Norwegian company considerable difficulty, it was able to prosper during World War I and into the late 1920s, at which point it converted its facilities to the Haber-Bosch process. Fritz Haber and Walter Nernst were also active in determining thermodynamic equilibria under the physical conditions required for the electric arc. It was Haber’s work in this field with Adolf König that first attracted the attention of BASF in 1907 and he recounted his activities in a lecture to the German Chemical Society in 1913 (Haber 1913a), (Stoltzenberg 1994, pp. 144–151), (Szöllösi-Janze 1998b, pp. 171–175).

The second synthetic option was a multi-step approach called the (calcium) cyanamide (Kalkstickstoff) process which found its natural analogue in the bacterial fixation of nitrogen. Metals such as calcium, barium, aluminum, or manganese were reduced with nitrogen at temperatures between 700 ^∘–1500 ^∘C to produce fertilizer or ammonia in a process that was first investigated in the mid-1800s. Like the electric arc, an economically viable process did not emerge until the turn of the century (Frank 1903), (Schmidt 1934, pp. 177, 246–251, 337–339), (Mittasch 1951, pp. 35–38, 57–61, 87–90), (Haber 1971, pp. 88–90), (Grossmann 1974, p. 371), (Holdermann 1960, pp. 53–56, 62–65), (von Nagel 1991, pp. 13–15), (Stoltzenberg 1994, pp. 138–139), (Szöllösi-Janze 1998b, p. 159), (Smil 2001, pp. 51–52). In 1895, Adolph Frank and Nikodem Caro began experiments on calcium carbide in order to produce calcium cyanamide, a fertilizer. The availability of carbides had recently increased through production in high-heat electric arc ovens and could possibly provide an alternative production method for cyanide compounds, which were in high demand but were also patented. They were joined by Fritz Rothe in 1897 and, now on the basis of barium,¹ found that barium carbide (instead of barium carbonate) could be used to produce barium cyanamide via hydrogenation. Using this concept, Rothe developed a process in 1898 on the basis of the cheaper calcium carbide to produce calcium cyanamide via the reaction

$$\displaystyle \begin{aligned} \mathrm{CaC}_2 + \mathrm{N}_2\rightarrow \mathrm{CaCN}_2 +\mathrm{C} \end{aligned}$$

In 1899, Frank, Caro, and Rothe founded the Cyanidgesellschaft with other industrial partners, and in 1905 they began producing calcium cyanamide, a fertilizer, with the (Rothe)-Frank-Caro process at temperatures around 1100^∘C. Also in 1905, just before Frank and Caro, production began at the Gesellschaft für Stickstoffdünger Westeregeln at reduced temperatures using the catalytic ability of calcium chloride, discovered by Ferdinand Eduard Polzeniusz in 1901. It was, however, the Frank-Caro process that caught on. From the calcium cyanide, ammonia could be produced by steam treatment:

$$\displaystyle \begin{aligned} \mathrm{CaCN}_2 + 3 \mathrm{H}_2\mathrm{O}\rightarrow \mathrm{CaCO}_3 + \mathrm{NH}_3 \end{aligned}$$

Another form of the multi-step process was developed by Carl Bosch at BASF. He knew of the earlier experiments on the reduction of barium carbonate, and in 1903, he began producing barium cyanide from barium sulfate. In 1904, Alwin Mittasch came to BASF and worked closely with Bosch on the production of ammonia by treating the barium cyanide with steam. Because barium is toxic, Bosch and Mittasch investigated other nitrides and hydrides including titanium nitride, silicon nitride, titanium hydride, tantalum hydride, and uranium hydride. Experiments with aluminum nitride were also undertaken, similar to those of Ottokar Serpek, who began pilot production based on aluminum nitride with the Société Générale des Nitrures in 1909. At BASF, the use of nitrides and cyanides represented the main attempts to synthetically produce ammonia until 1907, when their interest moved to direct synthesis from the elements over the following year (Mittasch 1951, p. 90).

Like the electric arc, the cyanamide process required high temperatures so that a proximity to hydropower was advantageous (for this reason hydropower was closely linked to fertilizer manufacturing before the availability of direct ammonia synthesis). Another difficulty was the recovery of the raw materials after ammonia production in order to close the material cycle. Thanks to its relative simplicity, however, the cyanamide processes spread across North America and Europe as well as to Japan and became the main source of synthetic nitrogen on the eve of the First World War (Haber 1971, p. 90). But it, too, was eventually replaced by direct synthesis from the elements.

The three processes were not completely independent. Over the course of the cyanamide process studies, Bosch and Mittasch gained valuable experience with catalysts, mixed materials, and high-pressure synthesis, which later aided them in upscaling Haber’s laboratory results. It is also evident from the longer history of both the electric arc and multi-step processes that they, like direct synthesis from the elements, could only be successfully understood and implemented on an industrial scale in the context of physical chemistry (Timm 1984). Emphasizing this fact while looking back on the prior century of empirical research, Mittasch described “the thicket in which pure empiricism finally lost itself.”² (Mittasch 1951, p. 57). It was knowledge from physical chemistry that revealed the way out.

When Haber took up ammonia synthesis in 1903, there had already been attempts at direct synthesis, including some based on physicochemical principles (Mittasch 1951, pp. 44–45, 48–53), (Sheppard 2020, pp. 87–91).³ In 1876, Marcelin Berthelot reported finding the same amount of ammonia in both generation and dissociation experiments using an electric spark. It was the observation of what we today call equilibrium between ammonia, hydrogen, and nitrogen. However, the discussion of whether an equilibrium actually existed would last another thirty years. A more advanced step in this direction came in 1884, when William Ramsay and Sydney Young observed the reversibility of the reaction: they showed that ammonia never completely disappeared in dissociation experiments at high temperatures. They investigated the effect of gas flow-rates, different catalytic materials and surface areas, and also attempted experiments under pressure (Ramsay and Young 1884). In 1901, Henry Le Chatelier, an early advocate of physical chemistry, investigated the synthesis of ammonia under pressures up to 100 bar at 600^∘C using an iron catalyst (Chatelier 1888), (Chatelier 1936, pp. 73–76). His detonation experiments with a mixture of hydrogen and nitrogen (he ended up destroying the experimental apparatus in the process) were used to obtain a patent in 1903, but his positive results were erroneous. Nevertheless, his work is notable for its reliance on physicochemical principles [compare: Silverman (1838), Tamaru (1991), Mittasch (1951, pp. 52, 54)]. Later, Le Chatelier would, like Mittasch, distinguish strictly between the new chemistry and the approach that had preceded it.

Wilhelm Ostwald and his assistants also took up a serious investigation of ammonia synthesis from the elements in 1900 at his institute in Leipzig (Figs. 9.1 and 9.2). They used a 1-to-3 nitrogen-to-hydrogen ratio, under pressure, with temperatures between 250 and 300 ^∘C, and mainly iron and copper catalysts with large surface areas (Ostwald 1927, pp. 279–287), (Farbwerke Hoechst 1964), (Holdermann 1960, pp. 40–42). After they believed to have synthesized ammonia—according to Ostwald he achieved the impressive yield of 8% although lesser amounts had also been observed (Ostwald 1927, p. 286)—he filed for a patent in March and entered into negotiations with the Badische Anilin & Soda Fabrik (BASF), the Farbwerke Hoechst in Frankfurt, and the Elberfelder Farbwerke to industrially upscale the process. Ostwald received several lucrative offers and test production began. In the ensuing investigations, Carl Bosch, who had come to BASF in 1899, found the iron in Ostwald’s process was not acting as a catalyst, but instead contained atomic, reactive nitrogen left over from production (Farbwerke Hoechst 1964, pp. 22–25). As a test, Bosch placed two different iron wires in glass pipes and passed pure hydrogen over one, and a 1:3 N₂:H₂ mixture over the other between 350 ^∘C and 450 ^∘C. The vessel containing pure hydrogen yielded 6.2 mg ammonia while the vessel containing the N₂/H₂ mixture yielded less, only 3.6 mg. In other words, the configuration that ideally should have produced no ammonia yielded more than the configuration of the process Ostwald purported to have developed. The ammonia had been generated when hydrogen passed over the heated iron already containing nitrogen. According to literature, Bosch reported, red-hot iron absorbed between 0.2% and 11.5% nitrogen when ammonia was dissociated over it or when nitrogen was passed over it. When cooled in the presence of hydrogen, the absorbed nitrogen reacted slowly to ammonia. Bosch requested Ostwald’s original experimental apparatus and tested it directly. He heated it to 400^∘C, passed pure hydrogen over the iron catalyst for 25 hours and achieved “relatively high amounts of ammonia.”⁴ This result sealed the deal for Bosch and similar conclusions were soon reported by Hoechst. It was another in a long line of mistaken detections of ammonia synthesized from the elements (Mittasch 1951).

Fig. 9.1

Wilhelm Ostwald in 1902. Source: Archiv der Berlin-Brandenburgischen Akademie der Wissenschaften, NL W. Ostwald, Nr. 5292

Fig. 9.2

Portrait of Ostwald in 1898 at the time Haber was inquiring about a post in the senior scientist’s institute. By David Vandermeulen from Fritz Haber: L’Esprit du Temps (Vandermeulen 2005); Ⓒ David Vandermeulen/Guy Delcourt Productions

While Bosch sought clarification from Ostwald to help with the continuation of the experiments, Ostwald decided not to continue in the field of ammonia synthesis. In light of Bosch’s work, he withdrew his patent application and backed out of the agreements with industry. He turned to the synthesis of nitric acid from ammonia, hastened by a personal belief that if cut off from Chile saltpeter, Germany could find itself in a precarious situation on the world stage (Ostwald 1903), (Ostwald 1927, pp. 287–288).

Despite the negative result of his experiments, we consider here several passages from his patent application. “I have found,” Ostwald wrote (Ostwald 1927, pp. 284–285),

that the binding of free nitrogen with hydrogen through a suitable contact substance or catalyst at low temperatures of 250 ^∘ to 300 ^∘ can be realized at measurable rates. The rate increases quickly with increasing temperature. Metals for example, mainly iron and copper, having large surface areas, can be used as catalysts. The binding is never complete, but rather leads to a chemical equilibrium and the amount of ammonia formed is, therefore, dependent on the ratios of the precursors. In order to achieve complete conversion, one must remove the ammonia from the reaction mixture, which can be realized through dissolving the ammonia in water or acid. For this, the mixture can be circulated, if necessary while cooling it to re-use the heat contained therein.⁵

Later he continued: “Because the relative amount of ammonia in the gas mixture increases with increasing pressure, it is advisable to conduct the synthesis under high pressure.”⁶

With these statements Ostwald succinctly wrote the “recipe” for ammonia synthesis from the elements and as a result, identified himself as the “intellectual father of [the nitrogen] industry (Ostwald 1927, p. 285).”⁷ While the recipe contained the same concepts used later by Haber and Nernst in their successful experiments, a look at the breadth of attempts to synthesize ammonia during the nineteenth century shows that all of these factors, if only separately, had already been discussed at that time in the literature in terms of physicochemical principles. One possible exception was a system of heat exchange (Mittasch 1951). This illustrates again that a holistic (or global) approach was still missing in attempts to synthesize ammonia before 1900. Although Ostwald brought these elements together, he lacked both a quantitative theoretical underpinning of chemical equilibria, later provided by Nernst, and a sensitivity to the reproducibility of small quantities of ammonia in equilibrium with hydrogen and nitrogen at normal (or even higher) pressures, an achievement of Haber’s. Several pages from Ostwald’s laboratory book show values from a review of ammonia experiments obtained in pressure oven experiments from March and April of 1900 while he was in negotiations with BASF and Hoechst (Figs. 9.3 and 9.4) (Ostwald 1900, pp. 23–25).⁸ The equilibrium values for ammonia fluctuate widely compared, for example, to Ramsay and Young (although they, too, complained of their own “want of uniformity”). The latter study, on the other hand, reported values for the percentage of decomposed ammonia at raised temperatures which are too high—at 830 ^∘C/1100K and 1 atm well over 99% should be decomposed (see Ramsay and Young (1884) and Fig. 9.5). As is evident from his publications, it was Haber who was first able obtain the accuracy and reproducibility needed for reliable measurements.

Fig. 9.3

Wilhelm Ostwald’s 1900 laboratory book for nitrogen experiments in a pressure oven, pages 23 and 24. Source: Archiv der Berlin-Brandenburgischen Akademie der Wissenschaften, NL W. Ostwald, Nr. 4397–3

Fig. 9.4

Wilhelm Ostwald’s 1900 laboratory book for nitrogen experiments in a pressure oven, page 25. In the experiments with a platinum catalyst marked “Pt a.” and “Pt b.” the conditions are nearly the same: 500 ^∘C at 49$\frac {1}{2}$ Atm and 48 Atm, respectively. In the former, no ammonia was dissociated; in the latter, 3% was generated from a starting mixture of N₂ and H₂. Theoretically, the values should have been nearly the same. In the iron experiments, “Fe a.” and “Fe b.”, at 450 ^∘C, 50% of the ammonia was dissociated at 70 Atm while at 53 Atm no ammonia was formed from the N₂+H₂ mixture. The latter result illustrates the difficulty of combining nitrogen and hydrogen. Source: Archiv der Berlin-Brandenburgischen Akademie der Wissenschaften, NL W. Ostwald, Nr. 4397–3

Fig. 9.5

Two pages from Ramsay and Young’s 1884 publication “The Decomposition of Ammonia by Heat” showing their results using iron filings (Ramsay and Young 1884). The consistency of the measurements is evident. Reproduced by permission of The Royal Society of Chemistry

There was one other modern effort to synthesize ammonia from the elements at the turn of the twentieth century: that of Edgar Perman. His work will be discussed with Haber’s first publications in the context of their exchange in literature.

We know today that none of these attempts to synthesize ammonia from the elements could come to fruition and no other sources of fixed nitrogen, neither natural nor synthetic, would prove as viable as the Haber-Bosch process at that time.⁹ Some (estimated) numbers by L.F. Haber put the world’s sources of fixed nitrogen into perspective in both 1900 and in 1913 after synthetic production emerged on the market (Haber 1971, pp. 101–104).¹⁰ In 1900, fixed nitrogen production from coking (ammonium sulfate) was 85,000 metric tons while 230,000 tons came from nitrate deposits. In 1913, out of a total consumption of 750,000 metric tons of fixed nitrogen, 280,000 tons (37%) were from coking, 430,000 (57%) tons from nitrate deposits, 24,000 tons (3%) from the cyanamide process, and 15,000 tons (2%) from the electric arc. The first Haber-Bosch plant at Oppau began production in late 1913 and by 1914 was producing ammonium sulfate at the rate of 36,000 tons per year, which amounted to about 8,000 tons of fixed nitrogen. By 1917/1918, 105,000 tons of nitrogen were being produced annually in Germany using the Haber-Bosch process, representing about 40% of total production; this number rose to three quarters over the next decade. In 1930, the German chemical industry was producing 1.2 millions tons (Mittasch 1951, pp. 136–137), (Szöllösi-Janze 2000, pp. 119–121). In comparison, at the end of World War I, saltpeter exports were also in the millions of tons (Eucken 1921, p. 170). In 2005, total global man-made nitrogen production was 140 million tons, of which 85 million (∼60%) was manufactured with the Haber-Bosch process (Erisman et al. 2005).