Proceedings of the Tomsk Technological Institute of Emperor Nicholas II.
Vol. XXIV, 1911, No. 4.

Rational Analysis of Clays.

A. E. Sabek.

Part II.

A Critical Study.

“The procedure of theory and practice is at first quite the same; the point is to get to know an object or process so precisely that one can call it forth or direct it at will.”

Prof. W. Ostwald.
Theory and Practice, lecture.

In the first days of May 1900, in the laboratory of Professor Witt at the Berlin Polytechnic School, I was asked to undertake the technical evaluation of two clays of Russian origin. Being sufficiently familiar with the significance of rational analysis in the technical evaluation of clays, I began my tests precisely with the application of this method and soon became convinced that both one and the other of the indicated clays present certain exceptions to the general rule, or else the method is described in insufficient detail, for the application of the general method to both varieties yields only negative results.

I. Here is a brief description of the clays mentioned above: one is a kaolin of white color, with a noticeable amount of sandiness, not greasy to the touch; on ignition it loses about 1% in weight; its color does not change from ignition, and it is distinguished by neither any particular odor nor taste.

II. The clay is dark gray, approaching black in color; after being wetted with water it darkens and has a black color; to the touch it is greasier than the preceding one; it contains considerably fewer sandy particles than the preceding; its odor is somewhat reminiscent of chernozem; on ignition it loses up to 15%, which consists of the loss of water and partly of the combustion of carbonaceous particles.

Both clays were taken near the village of Malevka Muravnya, Ranenburg district, Ryazan province, on the estate of the Ilyshevs. The white clay

lies close to the surface, whereas the black one is at a greater depth, being situated between a layer of coal and the Devonian formation.

Proceeding according to Seger’s instructions in “Böckmann’s chemisch-technische Methoden,” both clays were ground in a porcelain mortar, but not so much that the sandy admixtures were completely pulverized. Drying of both clays was carried out at a temperature of \(120^\circ\) C in an air bath, to constant weight; moreover, a certain loss in weight was still noticed after 3–4 hours of drying and ceased only after five hours of drying.

A repeated analysis of the same clays with an increased content of sulfuric acid, calculated as 15 cubic cm for each 1 gram of clay. After neutralization, treat with 15 cubic cm of NaHO, and then 10 cubic cm of HCl. Dissolution proceeded quickly and easily.

Comparison of the Results Obtained.

For one and the same clay, the following residue undecomposed by sulfuric acid was obtained:

Residue of feldspars from 2 experiments for each clay:

From these determinations it is easy to see how unsatisfactory the results are. The difference in the two extreme determinations reaches 3.3%, while the extreme limits diverge by as much as 1.5%. Such inadequate results were obtained despite the most careful conduct of all operations. Supposing at first that I was overlooking some instructions, I was compelled to consult a whole series of manuals, such as, for example: Seger’s Gesammelte Schriften. Kerl, Handbuch der Thonwaaren Industrie. Chemisch-Technische Untersuchungsmethoden von Bökmann, editions of 1891 and 1896, and also Chemisch-Technische Untersuchungsmethoden von G. Lunge. Further, Hermann Zwick, Die Natur der Ziegelthone; Post, Chemisch-technische Analyse; Hecht in Dammer’s Handbuch der chemischen Technologie.

As a result I found that the articles are for the most part borrowed from one another and the methods are described so briefly that, with the aid of these directions alone, there is no possibility of obtaining better results. Having then reviewed the history of the origin and development of the question of rational analysis, and having convinced myself of the importance of the analysis itself for technology, I set myself the task of making a fundamental verification of the method, since, in consequence of the small number of critical experiments, the thought arose as to whether the inaccuracies indicated above might not depend on the inaccuracy of the method itself.

The first matter was to dwell on clarifying the question of the origin of clays, so that it would be possible to indicate with certainty which elements precisely could be expected in the residue from clays after their treatment with sulfuric acid.

In this respect almost all the literature is exhausted by the following authors: Forchhammer, Senft, Neumeier, Seger, and, among the newest, Fibelkorn.

Forchhammer pointed to the origin of white clays and kaolins from feldspars and feldspathic rocks, such as ortho-

...clay, albite, oligoclase, granite. Senft is interested in the origin of all the remaining clays, such as brick clays, pottery clays, etc., and asserts that their origin is explained by contamination from all possible sources with iron, lime, magnesia, alkalis, organic substances, and humus and, depending on such admixtures, classifies them as brown clays, marls, magnesites, alkaline clays, chernozem, and so on. Heimayer relies on an entirely opposite assumption concerning the origin of clays: he regards as primary clays those which are contaminated with all possible impurities, from which, with the aid of water and CO₂ from the air, leaching takes place until, as a result, there remains pure white clay, no longer leachable, which is kaolin. Fiebelkorn objects to both of these assumptions, finding that, under such an assumption as Senft’s or Heimayer’s, there ought to be observed a gradual transition from white clays to brown ones in deposits, or conversely from brick clays to white ones. However, in known deposits such transitions have nowhere been observed. The boundaries of the transitions are usually quite sharply outlined, and it is possible to observe transitions of plastic clays into yellow and red ochres, but not into brick clay. In order to connect the actually observed relations of clays to the primary rocks, one is left to make the latter assumption: that most clays may derive their origin from various rocks—white clays from feldspathic rocks, granites, etc., which is also confirmed by Prof. Zemyatchensky from the study of kaolinitic formations in southern Russia (dissertation). The other clays, however, originate from such rocks as already formerly contained Fe, Ca, Mg, alkalis, and from simple rocks—mica and other fragments of the primary rocks from which these clays arose.

A rather significant question presents some difficulty: whether mica can be found in white clays. In this respect there exists a whole series of contradictory indications; thus, for example, Prof. Heim admits the possibility of finding mica in kaolins and white plastic clays. Chermak and Kossmann cite several such cases, whereas Zeger, Bischof, and others deny this. Prof. Zemyatchensky points out, on p. 11, that the kaolins of the villages of Blagodatka and Vladimirovka, on the bank of the river Kashlagach in Yekaterinoslav Governorate, do not contain mica; mica is encountered very rarely (p. 138); feldspar is contained in white clays (pp. 17, 21, 22). Mica is encountered near the town of Aleksandrovsk in variegated clays (p. 24); near the settlement of Kamyshevakha, Izium district (p. 134).

Having reviewed the principal material available concerning the origin, predominantly, of the newest pure clays, we...

we are inclined to think that in kaolins and white plastic clays, apart from kaolinite, quartz, and feldspathic residues, mica is almost never encountered¹; in other clays, however, the presence of mica is rather frequent, which should be taken into account when applying the method of rational analysis. As for brick clays, which are often situated in Tertiary deposits, the latter are very often contaminated by the presence of mica (muscovite). The application of rational analysis to such clays, as Kaul does in his dissertation entitled “Beitrag zur Kenntniss der Elsässer Thone 1900,” seems, owing to the absence of sufficient observations on the action of sulfuric acid on muscovite, to be a priori; as Focht’s preliminary observations already indicate, this procedure seems insufficiently substantiated in its application to brick clays.

It is especially important here to point out also that Seger, in testing the accuracy of the method itself, permits the use of a brick clay from Schwargotte, treating it with sulfuric acid.

A comparison of Focht’s observations, Seger’s application of rational analysis to brick clays, and other considerations made it necessary for me to review once again the basis of rational analysis of clays and the limits of its validity. This seemed to me all the more necessary because, thanks to the constant use of rational analysis for about 20 years by the school of Aron and Seger, it has gained acceptance both in fundamental and in periodical literature, and has especially acquired significance in the control of ceramic production. On the basis of rational analysis, which shows what quantity of quartz and feldspathic residues is contained in a given clay, one can decide what quantities of quartz and feldspar must be added in order to obtain a mixture with definite physical properties, as well as pyrometric ones, i.e., in order to observe the following conditions:

1. To have in the body a sufficient quantity of tempering substances, which is extremely important for preserving a certain degree of plasticity of the body, and is also important for the regularity of drying of the given article.

2. To have a sufficient quantity of fluxes, in the form of feldspar, so that on firing the proper relation of the body to a high temperature of the flame may be obtained.

Having become acquainted with the theoretical side of the question that interests us and with the practical significance of rational analysis, let us now turn to a brief enumeration of the methods of the latter that are found in the literature (see Appendix No. 1).

In 1879 there is a brief description of rational analysis in Bruno Kerl, Handbuch der Thonwaaren Industrie. The description, however, is made so briefly, without indicating details, that, without personal experience, it is hardly possible to carry out the analysis itself on the basis of the description.

Jul. Post in 1891 gives the following description of rational analysis: 1–2 g of clay are carefully boiled for 10–12 hours with concentrated sulfuric acid until it evaporates. After cooling, the residue is washed, transferred into a dish, and treated with a boiling solution of soda. The treatment with sulfuric acid and alkali is repeated until $\mathrm{SiO_2}$ no longer passes into solution. Filtration is carried out with the aid of a hot filtering funnel. The residue is then washed with water and hydrochloric acid and weighed on a paper filter. In the filtrate the content of silica is determined.

Seger in 1893, in Böckmann’s chemisch-technische Untersuchungsmethoden, describes the same method as follows: 5 g of clay are mixed with 100–150 cc of water, 2 cubic cm of caustic soda are added, and the mixture is boiled until uniformly turbid. After cooling, 25 cubic cm of sulfuric acid are added, and boiling is continued until fumes of sulfuric acid appear. A gelatinous mass is obtained; after dilution with water it is separated from the greater part of the sulfuric acid and soluble aluminum sulfate by decantation, and is then leached twice, alternately, by boiling with caustic soda and hydrochloric acid. The residue is collected on a filter and washed with hydrochloric acid, ignited, and weighed. The quartz and feldspar residues are treated with hydrofluoric and sulfuric acids, the alumina is precipitated, and from it the content of feldspars is calculated.

Textor proposes taking 5–6 g of coarsely powdered clay, dried at $120^\circ$ C, moistening it with water in a capacious porcelain dish, adding 300 cubic cm of water and, for each 1 g of clay taken, 10 cubic cm of sulfuric acid, and boiling until fumes of sulfuric acid appear. After cooling, about 200 cubic cm of hot water and about 5 cubic cm of HCl are added, and the mixture is boiled once more for about 10 minutes. After settling, the clear liquid is decanted; an excess of caustic soda is added, the mixture is diluted with water and boiled. After settling, the transparent liquid is decanted. The residue is treated with an excess of hydrochloric acid, diluted with water, and boiled again. After settling, it is decanted. The treatment with caustic soda and HCl is repeated once more. The residue consists of quartz and feldspar, which are ignited and weighed. Further treatment is identical with that described above.

Bischof proposes carrying out the analysis as follows: 5 g of unignited clay are stirred in a dish with 100–150 g of water, and after addition to the wa—

RATIONAL ANALYSIS OF CLAYS.

where 2 g of caustic soda are added, and the mixture is boiled until the clay particles are uniformly distributed. After cooling, 25 cubic cm of sulfuric acid are added, the vessel is covered with a watch glass, and it is boiled until vapors of sulfuric acid appear. In treatment with caustic soda, quartz partly dissolves and therefore is not determined completely. The resulting mass is diluted with water and decanted, whereby the greater part of the sulfuric acid is separated from the alumina sulfate; the residue is then subjected, not without error, to a double treatment successively with caustic soda and HCl. The precipitate is collected on a filter, ignited, and weighed. The subsequent treatment is carried out in the same way as Seger proposes. Bischof points out, however, that both treatment with sulfuric acid leads to the dissolution of feldspars, and caustic soda—to the dissolution of silica, and in general what is obtained is not an exact but an artificial picture of the composition of the clay. Moreover, the lack of control in this method, in order to know whether the decomposition of the clay has been completed, makes it imperfect.

Applying rational analysis to two clays of Russian origin (Ryazan, Ranenburg district, village of Torki), according to Seger’s directions, it was established that, with ordinary decantation, fine particles of undecomposed rock are easily carried away, or, if one wishes to retain the latter by filtration, the filter is easily clogged by silica hydrates and filtration is extremely slowed. Washing the filter free of silica hydrates is also very tedious.

At the same time it was found that, even with vigorous boiling with sulfuric acid, only in some exceptional cases is it possible to digest all the clayey substance completely. Despite the difficulties accompanying filtration, I used the method of filtration (and not decantation) for separating the solid undissolved particles of the rocks from the wash liquors. The results obtained by me from 4 parallel determinations were:

From these data it is easy to see that the determinations give numbers that are far from coincident and cannot be recognized as exact. In fact, a difference in two determinations reaching 3.3%, which decreases when compared with the average result to 1.5%, cannot be regarded as satisfying even the demands of practice.

Having reviewed all the references in the literature, I was unable to find any indication of what the inaccuracy of the results depends on and how it might be eliminated. It is appropriate to mention here that in many scientific works, such as, for example, the dissertations of Kaul, 1900, and A. Müller, 1892, dealing with German clays, no duplicate determinations of rational analysis are given anywhere, and no features are indicated of the method used by the authors; therefore, no idea can be formed of the degree of accuracy of the results.

On making inquiries in special laboratories that often employ rational analysis as to the accuracy of the determinations they carry out, I was able to establish that their requirements in this respect are very modest, and that they are satisfied with errors of 1%—4%¹.

The failure of the first experiments, the magnitude of the errors that might be expected as a function of the action of concentrated reagents, \( \mathrm{H_2SO_4} \), \( \mathrm{NaHO} \), etc.; Bischof’s statement, in one of his polemical articles with Seger, that rational analysis is in essence a very crude method, associated with a whole series of large errors; the absence in the specialized literature of indications of what should be attended to in order to reduce these errors to the possible minimum; and, finally, the magnitude of the error inherent in this method—all this occupied me, and to all these questions I found no proper answers in the literature.

I had to solve a difficult problem: first, it was necessary to compare with one another the various proposals that already existed and to settle on the one that gives the most satisfactory results; then to make corrections in this latter one in order to obtain the least error; and, having finally settled on such a method, to apply it to the individual elements that may enter into the general composition of clay, in order to determine the degree of error that could accompany the complete process of treating these elements with acids and alkalis.

I. Grinding of the Clay.

Experiments on white clay, the analyses of which are given above, containing 21—24% coarse sand, already at first glance forced one to attribute large errors to the fact that, during sampling, a grain of greater or lesser size would probably fall into one of the two samples, and thereby conditions are immediately created that disturb the possib—

…of obtaining identical results. Therefore, all clays subjected to testing were subsequently ground in an agate mortar to such a degree that grains of quartz or rock could no longer be detected either by touch or by eye, while with a glass rod the scratching of quartz could still be noticed. Later, our supposition was confirmed that this principal condition for maintaining the homogeneity of the entire mass by grinding is absolutely necessary and, moreover, does not at all lead to any appreciable increase in the magnitude of the errors, as Seger believed in the description placed in the first edition of Böckmann’s chemisch-technische Methoden; other authors say nothing about this at all. Here is an example: two determinations without grinding the grains gave the following results: insoluble particles—24.08% and 19.21%—on average—21.65%. After grinding the entire mass to complete homogeneity, two determinations gave: insoluble particles 21.936 and 21.418—on average—21.627%; as is evident, the results in the latter case differ little from one another, and the average results almost coincide in both cases.

II. Choice of the concentration of sulfuric acid and features of boiling.

It was next necessary to give due attention both to the choice of the most suitable concentration of sulfuric acid and to how the process of boiling with the acid itself should be conducted. Already after the first observations it became clear that vigorous boiling of clay with sulfuric acid cannot lead to accurate results, since sulfuric acid during boiling gives such considerable jolts that some portion of the contents of the dish is thrown both onto the glass covering the dish and onto the walls of the dish. Subsequent treatment is thereby carried out under conditions that exclude the uniform action of the acid on the clay substance. Even if one assumes that some portion of this clay that has settled on the walls falls back into the dish with acid, nevertheless this portion, having been exposed to the action of the acid for a shorter time, will not be decomposed to the same extent as the portion remaining in the dish. For complete uniformity of decomposition under such conditions, boiling would have to be continued for several days, not hours, as is done. Proof of this may be the very inconsistent amount of Al₂O₃ found in the preceding four determinations.

To avoid inaccuracies arising from jolts during the boiling of sulfuric acid, and also from the lowering of the level of the boiling mass, whereby some portion of the clay remains on the walls of the dish, it was necessary to carry out a comparative experiment on the various proposals made in this direction.

For every 5 g of clay the following quantities of concentrated sulfuric acid were proposed:

First of all, it was necessary to settle upon a definite amount of sulfuric acid. In principle, the smaller quantity of sulfuric acid should be preferred, i.e., if possible, not 50 cu. cm, but 25 cu. cm. In practice, however, the smaller quantity—25 cu. cm—proves inconvenient, since with this amount the water evaporates very quickly and the sulfuric acid begins to fume; meanwhile, the interval of time is so short that the clay does not yet have time to decompose completely. The digestion proceeds correctly, however, only on the condition that it is completed before the sulfuric acid begins to evaporate.

It will be shown below that, if prolonged fuming is allowed, the field spars are attacked more strongly than in the case when the acid boils without fuming.

Moreover, by the time the sulfuric acid begins to fume, its quantity must be such that the supply of liquid covers the level of the clay in the dish; this is not the case with the smaller quantity, i.e., 25 cubic centimeters.

Thus we arrive at the conclusion that, for 5 g of clay, the smaller of the proposed quantities should not be 25, but 50 cu. cm. Next, it remains to choose the proper quantity of water from the proposed 300, 150, and 100 cu. cm per 50 cu. cm of sulfuric acid. In making the latter choice it is necessary to pay attention to the following two circumstances:

1. The dilution of the sulfuric acid must be chosen so that boiling proceeds calmly, without any jolts. From comparative experiments it appears that the best ratio of water to acid is 300 : 50.

2. It is necessary to avoid too great a lowering of the level of the liquid in the dish during boiling, since the particles that remain behind from the liquid on the walls of the dish are subjected to briefer treatment by the acid than the rest of the clay, and under such circumstances one may already expect less successful results in advance. From this point of view, the most unsuccessful proposal is the ratio 300 : 50. If, however, the ratio 150 : 50 is taken, then at this dilution of the acid the boiling is accompanied by constant strong jolts.

...with water. From the point of view of the two circumstances indicated above, the dilution 100 : 50 proves to be the most successful.

It is necessary, at this concentration, to pay attention to the fact that the flame of the heating burner should not be very large and should not touch the cup; the heating must be conducted so that the combustion products—the gases—only wash over the cup, otherwise, during interruptions, the possibility of jolting during boiling is not excluded.

In parallel, experiments were carried out on what concentration is most convenient for the digestion of clay in the sense of saving time. For this purpose, 5 g of clay and 50 cc of concentrated sulfuric acid were placed in each of 5 porcelain cups and diluted respectively with 0, 50, 100, 200, and 300 cc of water. After 2 hours of heating, all 5 cups were removed from the fire, cooled, diluted to one liter, mixed into a homogeneous liquid, and 40 cc of liquid were taken for determination in it of the amount of alumina, after separation of SiO₂. From the table given below it is seen that more concentrated solutions decompose clay more quickly than diluted ones.

Since in kaolins the alumina is rarely more than 37%, it is clear that in 2 hours, in the first 3 cases, almost all the kaolin was dissolved, whereas with dilution by 200 cc it dissolved only up to 60%, and with dilution by 300 cc—up to 50% in the same time. It is clear that, for greater saving of time, preference should be given to one of the first three dilutions, and since, on the basis of the preceding considerations, we settled on the dilution 100 : 50, and such a dilution does not contradict the new experiment, then in all further experiments the dilution of acid 100 : 50 was preferably adopted.

At such a dilution sulfuric acid begins to fume after 4–4½ hours. In some manuals it is indicated that it is useful to maintain the fuming of sulfuric acid longer, but, relying on the experiments cited by us, it is necessary to recognize further fuming as harmful for obtaining accurate results in determinations of the quantity of feldspar residues. Feldspars suffer greatly from fuming sulfuric acid, and such a method would lead to significant errors. Seger also draws attention to the fact that one should not continue the evaporation of sulfuric acid for a prolonged time; however, he does so on the grounds that aluminum sulfate is insoluble in fuming sulfuric acid, i.e., in other words, the dissolution of the clayey substance stops altogether. To verify...

…of this indication, the following experiment was to be undertaken. It was necessary to determine the actual solubility of \( \mathrm{Al_2(SO_4)_3} \) in fuming sulfuric acid. For this purpose, an excess of \( \mathrm{Al_2(SO_4)_3} \), in the form of a chemically pure powder, was introduced into two portions, each of 100 cc, of fuming sulfuric acid. After an hour the dishes were removed and cooled. Two parallel experiments gave, in each 5 cc of sulfuric acid, the following content:

which corresponds to a solubility of aluminum sulfate of 0.19%; i.e., the solubility of \( \mathrm{Al_2(SO_4)_3} \) in fuming sulfuric acid is in fact very slight. Nevertheless, his conclusion that under these conditions the clay is only slightly decomposed is refuted by the experiment given on the preceding page.

III. Double Decantation.

In individual descriptions of the method of rational analysis, it is not mentioned how the dissolved part of the clay is separated from the undissolved part. I have already pointed out that filtration is almost inapplicable in this case, since it proceeds too slowly and does not give accurate results because of the dissolved salts of \( \mathrm{SiO_2} \). It remained to use decantation, adapting it, however, so that the errors would be minimal; simple decantation is also of little use, since, when the clear portion is poured off, part of the turbidity always goes with it, and with it the losses increase. Double decantation gives considerably better results.

Into a thick-walled glass beaker with a spout, of about 2 liters capacity, the first portion of the liquid, diluted with water, is poured without taking any special precautions, notwithstanding the fact that flakes of undissolved hydrosilicates are floating in it. Part of the solid residue may be carried along with the clear liquid, but this presents no danger in the subsequent work and, meanwhile, considerably accelerates the operation of pouring off the liquid. All the remaining liquids from repeated treatment of the solid residue—namely, with alkalis: 10 cc of 33% \( \mathrm{NaHO} \), and then 5 cc of concentrated hydrochloric acid—are poured, while still hot, into one and the same beaker, without paying special attention to the fact that some part of the solid residue passes into the beaker at the same time. If the pouring in of the hot alkaline liquid is carried out carefully, then in the upper layers of the liquid a precipitate of alumina hydrate is formed, which in the diluted solution subsequently dissolves only in part; and on standing in warmth (about \(60^\circ\) C), this precipitate, settling…

to the bottom, carries with it all the finest particles suspended in the liquid. After 1–3 hours have elapsed, the liquid becomes perfectly transparent, and a thin layer of precipitate, formed by alumina hydrate together with the entrained particles of the solid residue, settles on the bottom. The liquid prepared in this way is very easily decanted. In the dish there remains the greater part of the solid residue treated with alkalis and acids; this is washed once more with hydrochloric acid on the filter, to prevent loss of the finest particles of the precipitate.

The liquid that has stood above the settled layer in the beaker can easily be poured off without the slightest loss of any turbidity. Usually 50–100 cubic cm of turbid liquid remains; this is transferred to a porcelain dish and subjected to the same treatment with alkali and hydrochloric acid as the first portion of the precipitate. The acid and alkaline wash liquors are now collected separately, so that it may be checked whether solid particles have been carried away with the washings. Hot alkaline solutions readily precipitate the solid particles suspended in them, and in part also hydrate of alumina; subsequent treatment with hydrochloric acid frees the precipitate from flakes of alumina hydrate. Repetition of this operation with NaHO and HCl is done only to guarantee the unconditional correctness of the treatment. For ordinary technical purposes the repeated treatment with caustic soda and hydrochloric acid may be omitted. The residue washed with hydrochloric acid is placed on the same filter that already contains the first—and principal—part of the precipitate of solid particles insoluble in sulfuric acid. The residue is then ignited in a platinum crucible and weighed. The examination of the solid residue for its content of quartz and feldspar, according to Seger’s directions, is carried out with hydrofluoric acid.

To calculate, from the found content of alumina, the quantity of feldspar, Seger proposes the coefficient 5.41, which corresponds to the theoretical formula of feldspar \(K_2OAl_2O_3 \cdot 6SiO_2\). In fact, however, feldspars of such composition are very rare.

Below I give a comparative table of analyses of feldspars made by various authors, borrowed from Traité des industries céramiques by E. Bourry, 1897, pp. 80–81.

Potash feldspars—per 100 parts contain \(Al_2O_3 + Fe_2O_3\):

Potassium-sodium feldspars:

Sodium feldspars:

If one takes the average of the entire series of data given, it turns out that it will be more correct to take as the multiplier not 5.41, but from 4.78 to 5.10, i.e. closest of all, 5.00.

IV. Oxidation of organic substances included in clay by nitric acid.

Clay often contains, in greater or lesser quantity, organic impurities, namely: particles of coal, plant roots, etc., which, sometimes reaching several percent in content, considerably hinder all operations carried out by the method of rational analysis; thus, for example, they do not decompose completely with boiling sulfuric acid, float to the surface, and impede both treatment with alkalis and with hydrochloric acid, and especially become a nuisance during decantation.

1. Oxidation with potassium chlorate.

The first attempts at oxidizing organic substances were carried out with the aid of \(KClO_3\). Berthollet’s salt was finely pulverized and added in small portions to boiling sulfuric acid. One might have expected explosions from the rapid decomposition of \(KClO_3\) with the instantaneous evolution of oxygen; however, no phenomena resembling an explosion were observed—the oxygen was evolved rapidly in the form of bubbles, without the slightest explosion.

The oxidation of the organic substances of a clay containing up to 15% carbon in the form of coal dust was completed within an hour. A drawback of this oxidation may be considered the neutralization of the sulfuric acid by potassium, as a result of which a mass of crystalline potassium sulfate precipitate is obtained, hindering further work.

2. Oxidation with nitric acid.

The oxidation of organic substances with the aid of nitric acid is carried out very easily; for example, with a carbon content of up to 15%, \(15\) cc of concentrated nitric acid is sufficient. The oxidation is performed in a dish covered with a watch glass.

The disadvantages of oxidation with Berthollet’s salt are absent here. Nitric acid is easily distilled off from boiling sulfuric acid, but is immediately absorbed by drops of distilled water collecting on the watch glass. When a drop reaches a certain size, it falls back into the sulfuric acid, carrying nitric acid with it. The oxidation, despite the excess of carbon reaching up to 15%, is completed in \(1\)—\(1\frac{1}{2}\) hours. The liquid, containing dissolved organic constituents, acquires a brownish tint, but remains perfectly transparent; particles floating on the surface are no longer noticeable. Below it will be shown that treatment with nitric acid introduces scarcely perceptible changes into the amount of the determinable mineral constituents of the clay, i.e., the use of \(HNO_3\) has no appreciably harmful influence on the course or degree of accuracy of determinations in the method of rational analysis.

Summarizing all that has been said above, the process of rational analysis may be set out in the following manner: a 5 g sample of the given clay is placed in a porcelain dish; 100 g of distilled water and 5 g of caustic soda are added to it; the liquid is boiled until the mixture is brought to complete homogeneity; after cooling, 50 cc of concentrated sulfuric acid are carefully added, and the mixture is cautiously heated over a gas flame through a wire gauze for 2–3 hours, until sulfuric-acid vapors appear; the liquid is allowed to cool,

about 100 cc of water is again added from the wash bottle, washing down with this water all the clay particles adhering to the walls of the dish, and the dish is again placed for heating over a gas flame for a period of from 2 to 3 hours.

Then, after the liquid has cooled, it is diluted with 200–300 cc of water and the solution is poured from the dish into a two-liter beaker, without paying special attention to the fact that part of the turbidity is carried along by the hydrosilicates; only the larger particles of the precipitate remain in the dish. Next, 10 cc of caustic soda and 150 cc of water are added to the dish, boiled for no more than 10–15 minutes, allowed to settle, and the hot alkali is poured into the beaker in which the first acid decantate is placed; then 5 cc of strong hydrochloric acid is added to the dish, diluted with water to 150 cc, boiled for 10 minutes, and again poured into the glass beaker. The residue is collected on a filter, washed with hydrochloric acid, ignited in a porcelain crucible, and weighed.

All the mixed decantates poured into the glass two-liter beaker are placed for 1–2 hours on a warm plate (drying oven) at a temperature of 60–70° C. Then the precipitated hydrates of alumina and hydrosilicates settle to the bottom in the form of a dense crust, carrying with them all solid particles from the liquid. The supernatant liquid in the beaker is completely transparent and can be poured off without the slightest losses.

The subsequent treatment with alkalis and acids is similar to the preceding one, with the difference that decantation is carried out carefully, and the acid and alkaline decantates are collected separately; from them it is easy to judge how accurately the decantation operation has been performed. In the event of turbidity slipping through, it is possible to make a correction by repeated decantation, but usually this is entirely unnecessary.

The further treatment of the solid residue is carried out with hydrofluoric acid, with precipitation of the alumina and traces of iron by ammonia, as proposed by Seger and Lindhorst.

In calculating the amount of feldspars from the alumina, one should take the multiplier 5.00, and not 5.41, as proposed by Seger. If it is possible to extract grains of feldspar from the clay, it is more correct to establish this ratio beforehand for the given feldspar by analysis.

Limits of Error in the Method of Rational Analysis

Now, when the first difficulties and inaccuracies, which depended on the method itself, had been examined and overcome, it became possible to establish the limits of error as a function of the size of the individual grains of the constituent parts. For this purpose, an artificial mixture was first taken from the constituent parts of clay, in which it was proposed to determine the constituent parts again by analysis. The mixture consisted—

of 3 g of Zettlitz kaolin, 1 g of finely ground sand, sifted through the finest sieve, from “Hohenbocka,” which had previously been treated repeatedly with hydrochloric acid and sodium hydroxide, washed, and dried at \(120^\circ\) C; to the mixture was added 1 g of feldspar, ground and treated in the same manner.

A preliminary determination of the solid residue insoluble in sulfuric acid in the washed Zettlitz kaolin gave

\[ \begin{array}{cc} \mathrm{I}. & \mathrm{II}.\\ 0.47\% \ (0.0140\ \mathrm{g}) & 0.49\% \ (0.0149\ \mathrm{g}) \end{array} \]

\[ \text{on average } 0.48\%. \]

The insoluble residue in the synthetic mixture contained feldspar + quartz:

\[ \begin{array}{cc} \mathrm{I}. & \mathrm{II}.\\ 2.0014 & 2.0033 \end{array} \]

\[ \text{on average } 2.00235 \]

Subtracting that part of the solid residue which belongs to the Zettlitz kaolin, we find

\[ \begin{array}{r} 2.00235\\ -\,0.01445\\ \hline 1.98790 \end{array} \]

To determine the magnitude of the error, it is necessary to recall that feldspar, on ignition, gives an error of \(0.48\%\), and consequently in the residue from 2 g, after ignition, one may expect not 2 g, but \(2.000 - 0.0048 = 1.9952\) g. The mean error from the two determinations will, on this basis, be

\[ 1.9952 - 1.9879 = 0.0073, \]

which, relative to the 5 g quantity of clay taken, will amount to only \(0.15\%\).

Although the above already shows that by such a method the limits of error of the method of rational analysis could be determined, nevertheless this method proved to be very laborious, taking up too much time, and, most importantly, the sources of error were not separated and could be attributed to different causes. It therefore seemed more advisable to determine the limits of error for the individual constituent parts, and therefore the problem posed above breaks down into the following two, which must be studied:

I. The influence of the totality of the operations used in rational analysis on silicic acid, depending on the size of the particles introduced.

II. The influence of the totality of the operations employed in rational analysis upon the residues of rocks, namely: upon potassium and sodium feldspar, and also upon mica, depending on the size of the particles introduced.

As is known, acids, with the exception of hydrofluoric acid, do not act upon crystalline silica. As regards the action of alkalis upon crystalline silica, in the detailed article by Lunge and Millberg entitled “On the Behavior of Various Kinds of Silicic Acids toward Caustic and Carbonate Alkalis,” a sufficient number of examples are given, from which only those most suitable for us remain to be extracted.

Lunge calls coarse grain those particles which have passed through a sieve with 2000 openings per sq. cm.; dust-like grain—particles settling out of water only after two days’ standing; and finest powder—those particles which, from a turbid state in water, settle only after 14 days.

By two hours’ boiling with caustic alkalis the following losses are obtained:

Taking into account that, in the proposed analysis, treatment with alkali lasts not 2 hours, but is repeated only twice for 5 minutes, and assuming that solution proceeds proportionally to time, we arrive at the conclusion that the losses must be reduced 12-fold; therefore, for 5% NaOH the table will take the form:

Taking further into account the circumstance that the quantity of silica only in rare cases reaches 20%, the magnitudes of the errors with respect to the whole mass must therefore be further reduced 5-fold and will amount, in percent, for:

The latter number refers to the finest powder in so finely divided a state that it can be expected in clay only in the most insignificant fractions of a percent. In fact, in clay, the precipitate settles after agitation comparatively easily, and after 14 days even liquid clays give scarcely perceptible turbidity in water. From this it is easy to conclude that the error resulting from the dissolution of quartz in alkali is generally of barely perceptible magnitude. Moreover, it must be noted that 5–10 cc of NaHO is diluted with water to 150–200 cc, i.e., in practice we are dealing not with 5% alkali, but with 2%–2.5% NaHO.

In fact, a whole series of experiments confirms the preceding considerations, namely that the action of alkali on the finest particles plays only a very limited role in the entire process; for, despite the grinding of quartz to the finest state, in all parallel tests the results obtained were almost in agreement with one another.

In view of the complete impossibility, in practice, of carrying out parallel experiments under identical conditions, it would not have been possible to obtain coincident results if the solubility of silica in alkalis played a noticeable role.

From the work of Lunge and Millberg it follows, however, that if one wishes to obtain still more exact analytical results, sodium carbonate should be preferred to caustic soda.

Turning to the second part of the problem stated above, concerning losses from the action of the sum total of all analytical operations on the mineral constituents, it should be mentioned that no observations in this direction have been published, with the exception of fragmentary indications by Seger and Focht, which, however, are distinguished neither by sufficient detail nor by accuracy. It therefore seemed entirely necessary to undertake a series of experiments to establish the possible losses from the totality of the operations on the ground rocks orthoclase, albite, and mica. The rocks were ground in a steel mortar, sifted through a sieve, treated with weak solutions of caustic soda and HCl, and, after washing, dried to constant weight. The grinding of the feldspar powders was as follows. Very coarse grains, coarse grains, and fine grains were distinguished. By very coarse grains were meant those grains which did not pass through a sieve with 4000 openings per 1 sq. cm.; coarse grains were those which passed through a sieve with 4000 openings per 1 sq. cm. and did not pass through a sieve with 5000 openings per 1 sq. cm. The finest grains were all that passed through a sieve with 5000 openings per 1 sq. cm.

All operations on each type of grain were carried out in the strict sequence described above, and in this way the following figures were found:

I. Losses for orthoclase:

II. Losses with albite:

III. Losses with mica (muscovite).

Considering the three tables given, one first of all encounters the circumstance that the losses with 25 cc of sulfuric acid are greater than in the case when 50 cc of it are taken, while the action in both cases lasts the same amount of time. This is explained solely by the fact that with 25 cc of sulfuric acid the evaporation of the sulfuric acid continued for about 4 hours, whereas with 50 cc of sulfuric acid the evaporation continued only about 2 hours, the total duration of boiling being 5 hours. Since the temperature of evaporation of sulfuric acid is higher than during its boiling, this also accounts for its increased action on the mineral constituents of the fraction during the evaporation of the sulfuric acid. Clearly, in order to avoid great—

to avoid losses one should prevent evaporation of sulfuric acid, i.e., boiling should be brought only to the moment when vapors of sulfuric acid begin to be evolved. Seger explains why boiling should be carried only to the evaporation of sulfuric acid by the circumstance that fuming sulfuric acid acts very weakly on clay; we have already earlier convinced ourselves of the incorrectness of this assertion.

CONCLUSION.

The conclusion that can be drawn from the experiments presented above amounts to the following: feldspars, when treated with sulfuric acid, hydrochloric acid, and caustic soda, are attacked and the losses fluctuate around 2% even with the finest comminution. If, however, one takes into account that feldspars are rarely present in the composition of clay in amounts up to 10%, then the error with respect to the entire mass of the clay is reduced 10-fold and does not exceed 0.2% in the most unfavorable case. Thus, notwithstanding Bischof’s assertion, rational analysis is in no way inferior in accuracy to many methods of technical analysis. From the same observations it also follows that, using 50 cubic cm of sulfuric acid per 100 cubic cm of water, with 5 g of clay, for the same boiling time with acid, the possible error becomes smaller than in the case when 25 cubic cm of sulfuric acid per 100 cubic cm of water are taken with 5 g of clay.

As regards the stability of mica, the analyses cited show that it withstands the action of the indicated reagents very weakly. It was stated earlier that mica occurs very rarely in kaolins and white plastic clays; in clays, however, that fire to red colors, inclusions of mica are more common. Therefore the method of rational analysis in the form presented here may be applied without reservation only to white clays serving as material for the finer grades of ceramic products. In the analysis of brick clays, however, the method of rational analysis is of little or no applicability.

This conclusion is identical with that of Foch and deserves attention, because up to now the method has been applied to the investigation of brick clays not only by Seger and many others, but a number of technical studies have also appeared, such as, for example, the dissertation of G. Kaul, 1899, in which the author gives a comparative assessment of the merits of an entire series of brick clays, knowingly accompanied by particles of mica. The author not only does not mention the unsuitability of the method, but reduces all calculations to the determination of the amount of mica (muscovite), and not of feldspars. Obviously, the circumstance that mica suffers greatly from the action of sulfuric acid has been completely overlooked. Also deserving of surprise is the circumstance that all

Kayul’s determinations are not accompanied by control analyses, which would undoubtedly have drawn his attention to the impossibility of obtaining identical results.

For white plastic clays or for kaolins, rational analysis leads to very satisfactory results, provided only that the precautions indicated above are observed. Here are examples:

I. Russian white clay from the village of Topki, Ranenburg district, Ryazan province, containing a significant quantity of coarse quartz grains:

II. Russian black clay, taken at the boundary with coal-bearing strata from the village of Topki, Ranenburg district, Ryazan province, with an inclusion of very fine sand:

III. Halle clay (Halle’sche Erde) with an inclusion of fine sand:

IV. Washed clay (China Clay):

V. Zettlitz kaolin of the Geibach brothers, from the pit near Wallendorf:

These analyses are sufficient to convince one that, by means of rational analysis, one can obtain exactly the same results as in quantitative analysis. Therefore all of Bischof’s objections to the roughness and inaccuracy of the method must be rejected; for, if quartz were in fact so strongly soluble in alkalis, and feldspars strongly soluble in sulfuric acid, then, given the impossibility of carrying out an entire series of operations in two parallel cases under completely identical conditions, it would have been unthinkable to obtain coincident results. If, however, coincident results are obtained, then this once again indicates the absence of sources of large errors, or else the low solubility both of \(SiO_2\) and of feldspars.

In the analysis of brick clays containing mica, rational analysis cannot be regarded as an exact method; for kaolins and white plastic clays this method gives sufficiently accurate results.

For control of the work it should be pointed out that, with proper and sufficiently prolonged boiling of the clay with sulfuric acid, a transparent liquid is obtained, in which flakes of hydrosilicates float, but clay particles in the form of turbidity are absent. The operation succeeds only if all the precautions indicated above are observed.

By an entire series of theoretical considerations, subsequently confirmed by practical data, the following may be considered proven:

1. That the totality of the actions adopted in the method of rational analysis does not exert any appreciable influence on the quartz entering into the composition of the clay.

2. That feldspathic residues, by the totality of the same actions, although affected to a more noticeable degree, nevertheless even in the most unfavorable case the limit of errors does not exceed \(0.2\%\).

3. In clays containing mica, the latter, under the influence of the totality of the actions of rational analysis, suffers severely and may even be decomposed completely.

4. The oxidation of organic substances in clays is best accomplished with the aid of nitric acid; quartz and feldspars are almost not affected by this additional treatment.

5. For the proper course of the operation of rational analysis, the following conditions must be observed:

6. Bringing the average sample to homogeneity by grinding in an agate mortar.

7. Observance of the proper concentration of sulfuric acid \((100 : 50)\) and the corresponding boiling.

8. Oxidation of organic substances with nitric acid.

9. Use of double decantation.

In this form, in its main features, this article appeared in print in the Zeitschrift der chemischen Industrie, edited by Otto Witt. The

...of its appearance, professors of the Berlin Polytechnic School—Karl Schoch, who lectured on the technology of building materials, and Hermann Hecht, who lectured on ceramics, glass manufacture, etc.—expressed great interest in my work and asked that separate offprints of the article be sent to them; in Tomsk I received a written request from Professor Gustav Schultz of the Munich Polytechnic to send him a separate offprint of the article. In addition, I received a similar request from K. Torn, head of the private ceramic laboratory in Voronezh. The latter laboratory, as it later turned out, was under the supervision of Professor Lisenko of the Mining Institute.

Quite by chance, an offprint of my article that had found its way into Russian literature elicited a critical review from Prof. K. Lisenko (Mining Journal, 1903, book 1, pp. 387–406), in which he, while not objecting to the substance of the article and reproducing it almost in translation, confines himself to comparing at the end of the article some data from the work of Prof. Zemyatchensky with my conclusions. In general he considers the question still far from completed. A whole series of questions is opened up by the outlined issues; I have set their resolution as the task of future laboratory work, both my own and that of my students. The problem I posed proved, on the one hand, very interesting, but on the other hand so extensive that exhausting it in the present work does not seem possible, for it may take a whole series of years. Nevertheless, wishing to hear competent opinions on what has been done, I intend to extract from critical remarks much that might by chance have escaped my attention. The very appearance of the second part of the work is already due to several remarks made in due time by Prof. Lisenko concerning the first part of this work. It is quite possible that several competent remarks with respect to the second part of the work will give me greater confidence in the correctness of what has been begun and, in the future, may warn me against one-sided enthusiasms.

In the foreign press my article “Beiträge zur Kenntniss der rationellen Analyse der Thone” is mentioned by B. Zschokke in the article “Ueber technische Analyse der Tohne,” Baumaterialienkunde, Nos. 10, 11, 1902; in Handbuch der gesammten Thonwaarenindustrie von Bruno Kerl, III Auflage, 1907, p. 1368; in Jahres Berichte der chemischen Technologie 1902–1903; and in the new (third) edition of Treadwell (quantitative analysis).

Since the article by the honored professor of the Mining Institute K. Iv. Lisenko served as the occasion for further development of the question of the rational analysis of clays that I had initiated, I place here, as a preface to the third part of my work, without any alterations, the concluding words of Prof. Lisenko’s critical note, published in the Mining Journal for 1903, vol. I, pp. 387–406.

1. Oral communication of Professor Hecht. ↩↩