Full Text
On the Influence of the Chemical Nature of a Substance on the Magnetic Properties of Bodies
D. Reichinstein.
1. FERROMAGNETIC ELEMENTS.
Until recently, iron, nickel, and cobalt were considered the only strongly magnetic substances. These metals, owing to their magnetic properties, occupy a distinctive position among the other elements. Iron, which has found wide application in electrical engineering, gave its name to this whole group—the ferromagnetic metals. Between ferromagnetic and paramagnetic bodies there exists not only a quantitative but also a qualitative difference. A characteristic feature of the magnetic properties of the former is residual magnetism—a phenomenon wholly unobserved in paramagnetic bodies. In addition, in ferromagnetic bodies the magnitude of magnetization at first increases rapidly with the field strength and then reaches a limiting value, whereas in paramagnetic bodies (as also in diamagnetic ones) the magnetization is proportional to the field strength, i.e., their susceptibility does not depend on the field strength. Nevertheless, it is impossible to draw a sharp boundary between ferromagnetic and paramagnetic bodies, since gradual transitions from one to the other exist.
The magnetic properties of iron vary within wide limits depending on the preliminary treatment of the metal, on the degree of heating, and on the presence in it of foreign impurities. The softer the grades of iron, the more strongly they are magnetized, but the less constant are their magnetic properties; and, conversely, hard grades of iron, which are magnetized comparatively weakly, possess strong residual magnetism and great coercive force. The following table ^1) gives an idea of the relative strength of the temporary magnetic properties of various grades of iron:
| Grade of iron | Relative strength |
|---|---|
| Wrought iron | 100 |
| Cast iron | 88 |
^1) E. Wedekind. Magnetochemie, Berlin p. 20, 1911.
^2) Wedekind loc. cit.
Mild steel . . . . . . . . . . . . . . . . . 66
Hard steel . . . . . . . . . . . . . . . . . . 33
Mild cast steel . . . . . . . . . . . . . . . 74
Hard cast steel . . . . . . . . . . . . . . . 49
Of the influence of impurities, the best studied is the influence exerted by the presence of carbon. With an increase in the amount of carbon, the magnetizability of iron decreases. Small quantities of carbon increase the permanent magnetic properties of iron. The maximum of the coercive force corresponds to a carbon content of 1.2%, and that of residual magnetism to 0.5%. The coercive forces for certain kinds of iron and steel have the following values ²):
Swedish wrought iron . . . . . . . . . . . . 0.8
Cast iron (annealed) . . . . . . . . . . . . 4.9
Tungsten magnetic steel (unhardened) . . . . 27.5
Magnetic steel, hardened . . . . . . . . . . 52.6
In their magnetic properties, elements related to iron, nickel and cobalt, are close to iron. In a strong magnetic field the degree of magnetizability of cobalt reaches the value observed for cast iron. Nickel is magnetized somewhat more weakly than cobalt.
The magnetic properties of iron are influenced, just as by carbon, by paramagnetic elements: chromium, molybdenum, and tungsten, i.e., their presence increases the coercive force. Aluminum, on the contrary, lowers the magnetic properties of iron; manganese, added in an amount of 12%, makes iron entirely nonmagnetic; the action of chromium on nickel is analogous. It is remarkable that an alloy of two ferromagnetic elements—iron and nickel (25%)—is entirely nonmagnetic; when cooled below 0°, it becomes magnetic and retains its magnetic properties upon subsequent heating up to 580°. Weiss and Foex ¹) showed that alloys of iron and nickel form two continuous series of solid solutions; according to their observations, the molecular magnetic moments are additive for the solutions; upon the transition of a solid solution into a chemical compound, the additive character of the magnetic properties is destroyed.
Tammann’s investigations ²) on the structure of metallic alloys clarified the question of the dependence of the magnetic properties on the composition of alloys of ferromagnetic metals. It turned out that all binary compounds of ferromagnetic metals with other metals are nonmagnetic. Mixed crystals in which the ferromagnetic metal plays the role of solvent possess magnetic properties, whereas mixed crystals in which the solvent is
¹) Weiss et Foex. Arch. sc. phys. et nat. Genève, [4] 31, 5 et 89 (1911).
²) Tammann. Ztschr. f. phis. Chem. 65, 73, (1909).
nonmagnetic metal are nonmagnetic. If an alloy forms a double compound, then the magnetism of such an alloy varies from that of the pure ferromagnetic metal to that of its compound with the other. If the alloy does not form chemical compounds, then the magnetic properties of the ferromagnetic metal gradually change as the content in it of the nonmagnetic metal increases up to that metal, or up to the formation of saturated mixed crystals. The exception is the case when the nonmagnetic metal exerts an influence on the temperature at which the ferromagnetic metal loses its magnetic properties, as will be discussed below.
Fig. 1.
Honda1 carried out quantitative investigations of the alloys: $Ni—Cr$, $Co—Cr$, and $Ni—Al$; the first (Fig. 1) is nonmagnetic already at 10% chromium; the second has a maximum at 15% chromium, and at 25% loses its magnetic properties. The third alloy (Fig. 2), with a minimum—the curve has three points of inflection corresponding to the compounds: $NiAl$, $NiAl_2$, and $NiAl_3$. The alloy $Co—Cr$ is remarkable in that the presence of a small amount of a nonmagnetic metal causes an increase in the magnetic properties. From all that has been said above it follows that the magnetic properties of alloys are a linear function of their composition only if this linear dependence is not disturbed by the interaction of the constituent parts of the alloy.
Fig. 2.
If iron is heated to $760^\circ$2, it loses its magnetic properties (on further heating to $910^\circ$ it becomes paramagnetic). Simultaneously with the change in magnetic properties, iron undergoes structural changes. At the magnetic transformation point (the Curie point) a sudden change is observed in the thermoelectric properties and in the temperature coefficient of electrical con-
resistance. At \(910^\circ\) an abrupt change in the properties of iron is again observed. Thus, iron exists in three allotropic forms—\(\alpha\), \(\beta\), and \(\gamma\). Upon reverse cooling the magnetic properties return; at this moment the phenomenon of “recalescence” is observed—heat is evolved, cooling ceases, and the red heat turns into white heat, while the volume increases. The temperatures at which the magnetic properties disappear upon heating and reappear upon cooling do not always coincide; in ordinary iron the second point lies \(10\)—\(12^\circ\) below the first. The presence of carbon lowers the transition point. Similar transformation points also exist in nickel and cobalt.
Honda investigated the influence of temperature on the magnetic properties of nickel; he found that the transition point lies at about \(350^\circ\). The curves for rising and falling temperature coincide (Fig. 3). The presence of \(2\%\) chromium already considerably lowers the transformation temperature, and the temperature of return of the magnetic properties lies \(40^\circ\) below that at which they disappear. With a greater content of chromium the transformation temperature is lowered still more and, at a sufficiently high concentration, falls below room temperature. The presence of copper has the same influence on the transformation point of nickel. The influence of temperature on the magnetizability of cobalt, and the influence of the presence of chromium, are shown in Fig. 4. An alloy of cobalt with copper is magnetic only at temperatures below room temperature.
Fig. 3.
Tammann concludes1, from the influence that foreign metals exert on the magnetic transformation point, and from the dependence of this influence on the structure of the alloy, that the loss of magnetic properties is associated with the transformation of the magnetizable form of the crystal into other, non-magnetizable ones.
Fig. 4.
He sees the cause of the displacement of the transformation point in the fact that in the nonmagnetic \(\beta\)- and \(\gamma\)-crystals, stable at high temperatures, foreign metals dissolve and, owing to this, lower their transformation point. Hence one may infer that, if foreign metals do not dissolve in the modifications of ferromagnetic metals that are stable in higher temperature regions, then their presence should not exert an influence on the magnetic transformation point. The latter was in fact observed for alloys of iron and cobalt with silver, thallium, and lead. The dependence of the magnetic properties on temperature and on the concentration of the alloys becomes intelligible if the capacity for magnetization is ascribed exclusively to the \(\alpha\)-modification of ferromagnetic metals.
As regards compounds of ferromagnetic metals with metalloids, first of all the mineral magnetite \((Fe_3O_4)\) should be mentioned. The ability of this mineral to attract iron was known already to the ancient Greeks and Egyptians long before the birth of Christ. Its magnetizability is equal to half the magnetizability of iron (Becquerel), but its residual magnetism may be almost three times greater than that of steel. It is interesting that the magnetic properties of magnetite are not the same in the direction of different axes, despite the fact that it crystallizes in the regular system. Among other natural compounds of iron, iron pyrites \((FeS_2)\) and pyrrhotite2 possess magnetic properties; their magnetizability is of the same order as that of magnetite. Iron pyrites, which also belongs to the regular system, is magnetized only in one direction.
Iron oxide by itself is not magnetic; its compounds with bases in some cases possess strong magnetic
properties—these are the so-called ferrites, of composition $MeO \cdot Fe_2O_3$1. Alkaline and alkaline-earth ferrites: $K_2O \cdot Fe_2O_3$, $Na_2O \cdot Fe_2O_3$, $CaO \cdot Fe_2O_3$, $BaO \cdot Fe_2O_3$, $MgO \cdot Fe_2O_3$, as well as $ZnO \cdot Fe_2O_3$ and $PbO \cdot Fe_2O_3$, are nonmagnetic; on ignition they become only weakly magnetic. Strongly magnetic are $CuO \cdot Fe_2O_3$2, $CoO \cdot Fe_2O_3$, and $FeO \cdot Fe_2O_3$; their magnetic properties as a function of temperature are analogous to those of ferromagnetic metals, i.e., they too have a critical temperature at which they lose the ability to be magnetized. The magnetic properties of these compounds are determined by the acidic character of ferric oxide3. For the compound of ferric oxide with ferrous oxide, Hilpert showed by a special investigation that the ability to be magnetized depends on ferric oxide. A magnetic form of ferric oxide is known, obtained upon oxidation of $FeO \cdot Fe_2O_3$, but it is unstable and on heating passes into a nonmagnetic oxide. Hilpert gives the following scheme for this oxidation reaction:
$$ (2eO)F^{+}(2Fe_2O_3)^{-} + O = (Fe_2O_3)^{+}(2Fe_2O_3)^{-} $$
i.e., in the magnetic form, ferric oxide possesses both basic and acidic character. On the other hand, with increasing $FeO$ content the magnetizability is at first almost constant up to $66\%$ $FeO$, and then falls, and at $82\%$ the magnetic properties disappear completely—which indicates the formation of a saturated solid solution. Ferrous oxide (like the ordinary oxide) is nonmagnetic, and, consequently, the magnetic properties of $FeO \cdot Fe_2O_3$ are caused by the combination of both oxides.
Also of interest with respect to magnetic properties are the amalgams of ferromagnetic metals4. Amalgams of iron and cobalt are strongly magnetic, and despite the fact that their residual magnetism is insignificant, the coercive force reaches an exceptional value. For an amalgam with $2.3\%$ iron it is equal to 370, whereas for tungsten steel the maximum value is 80. Nickel amalgam is weakly magnetic, whence it may be concluded that, in contrast to the first two, it represents a chemical compound.
The carbonyls of iron $[Fe(CO)_5]$ and of nickel $[Ni(CO)_4]$ are diamagnetic5.
2. MAGNETIC COMPOUNDS OF NONMAGNETIC ELEMENTS.
As we have seen, compounds of ferromagnetic metals with other metals are nonmagnetic, and even the compound of two ferromagnetic metals (iron—nickel) is nonmagnetic. The question is whether
a magnetic compound of nonmagnetic elements may be formed. Such a compound was first found by Wöhler upon heating vapors of chromyl chloride \((CrO_2Cl_2)\); the oxide \(Cr_5O_9\) thereby obtained possesses strong magnetic properties. Wöhler expressed the supposition that, like magnetic iron ore, the composition of this oxide corresponds to the compound \(2Cr_2O_3 \cdot CrO_3\). Recently Zhukov1 obtained magnetic chromium oxide by heating chromic acid to \(510^\circ\); the maximum of magnetization was observed by him at a loss of oxygen of \(13.3\text{–}14.1\%\), which corresponds approximately to the formula \(2CrO_3 \cdot Cr_2O_3\) \((Cr_4O_9)\). The magnetic properties of the oxide obtained by Wöhler are three times weaker. The remaining compounds of chromium are more or less strongly paramagnetic. Metallic chromium is weakly paramagnetic; its specific susceptibility2 is \(\chi \times 10^6 = +3.7\).
Still more interesting with respect to magnetic properties are the alloys and compounds of manganese. The magnetic properties of the former were discovered by Heusler3 in 1900; he found that an alloy of manganese with tin possesses strong magnetic properties, even in the case where it is dissolved in an equal quantity of copper. When tin is replaced by aluminum, an alloy with still stronger magnetic properties is obtained, which at first glance seems especially surprising if this fact is compared with the data concerning the influence of manganese and aluminum on the magnetism of iron. Magnetic alloys are also obtained if aluminum is replaced by one of the trivalent metals: arsenic, antimony, bismuth. The alloys \(Mn + Al + Cu\) were studied in detail by Richard and his pupils. The influence of temperature on the magnetization of these alloys proved to be very complex; the maximum of magnetization was obtained upon preliminary heating for several hours in boiling toluene \((110^\circ)\). All these alloys have a definite critical temperature \(\theta\), above which they are nonmagnetic. The critical temperature varies, depending on the composition, from \(60^\circ\) to \(350^\circ\); the presence of lead lowers it. Richard and Geisler ascribe the magnetic properties to definite chemical compounds. The maximum of magnetism is observed for a composition corresponding to the content of one atom of manganese per one atom of tin in the alloys—\(Mn + Sn + Cu\)—and of one atom of aluminum per one atom of manganese in the alloys—\(Mn + Al + Cu\). In reality the compound \(MnAl\) is unknown, and Geisler assumes in these alloys the existence of the compound \(Al_xMn_yCu_{3x-y}\) \((Al_xMn_{3x})\).
Among the chemical compounds of manganese, the following are magnetic: $MnB$, $MnP$, $MnSb$, $Mn_2Sb$ (Wedekind), $MnAs$, and $MnBi$, i.e., manganese gives magnetic compounds with the most diamagnetic elements—antimony and bismuth. It is sufficient to add manganese to bismuth in an amount of $1/4\%$ in order to impart clearly expressed magnetic properties to the latter. The permeability of manganese compounds with phosphorus, antimony, and boron was investigated by Wedekind1; in Fig. 5 the values of the magnetization as a function of the field strength are given in comparison with soft iron and cobalt. Cast iron is magnetized 29 times more strongly than $MnP$, and $10^{1/2}$ times more strongly than $MnSb$. Hilpert2 investigated the transformation temperatures for a whole series of manganese compounds; it turned out that, with an increase in the atomic weight of the element entering into the compound with manganese, the transformation temperature rises, as is seen from the following data:
Fig. 5.
$MnP$ . . . . . . . . . . . . . . . . . . . . . . . . $18—26^\circ$
$MnAs$ . . . . . . . . . . . . . . . . . . . . . . . $40—45^\circ$
$MnSb$ . . . . . . . . . . . . . . . . . . . . . . . $320—330^\circ$
$MnBi$ . . . . . . . . . . . . . . . . . . . . . . . $360—380^\circ$
The coercive force and residual magnetism in some of the compounds and alloys of manganese reach large values, e.g., the coercive force of $MnB$ is 33.4 and that of $Mn_2Sb$ is 30.93 (Wedekind), i.e., of the same order as in cast iron and tungsten magnetic steel. Honda4 investigated the dependence of residual magnetism on concentration in alloys of manganese with tin (Fig. 6); mixed crystals with a manganese content above 92% are nonmagnetic,
maximum corresponds to the compound \(Mn_4Sn\); with a decrease in the concentration the magnetization of manganese rapidly falls to \(Mn_2Sn\) (\(48\%\,Mn\)).
Manganese peroxide is weakly magnetic; the nitrides \(Mn_3N_2\), \(Mn_5N_2\), and \(Mn_7N_2\) are clearly magnetic—their magnetic properties increase with increasing manganese content.
[Figure: graph of residual magnetization, \(J=11^\circ\), for the system \((Sn)\)–\((Mn)\). Labels on the diagram include: \(Sn + MnSn\); \(MnSn + Mn_2Sn\); \(Mn_2Sn + Mn_4Sn\); \(Mn_4Sn + \mathrm{con.}\ \mathrm{Hp}\); \(\mathrm{Con.}\ K_2\); vertical labels \(Mn_2Sn\) and \(Mn_4Sn\).]
Fig. 6.
Compounds of manganese with divalent elements—sulfur, selenium, and tellurium—are very weakly magnetic (E. Wedekind and Feith1).
It remains also to point out here the place occupied by the ferromagnetic metals and by the metals that give magnetic compounds in the periodic system of the elements: all of them have a closely similar atomic weight, from 52.1 to 59, and stand in the 4th horizontal row. Vanadium, which stands to the left of them, is paramagnetic, while its oxides and sulfur compounds are weakly magnetic.
A separate group of elements possessing magnetic properties is represented by the rare-earth metals. Unfortunately, the difficulty of obtaining them in pure form has until now prevented their more thorough investigation. Their oxides and salts have been studied somewhat more. The rare earths include in their group both diamagnetic (lanthanum) and strongly paramagnetic members: dysprosium, neoytterbium, erbium, samarium. Urbain and Jantsch2 found for certain oxides the following values of the susceptibilities—\(k \cdot 10^6\):
| Atomic weight | \(k\cdot 10^6\) | |
|---|---|---|
| Neodymium | 144.3 | 33.5 |
| Samarium | 150.4 | 6.5 |
| Europium | 152 | 33.5 |
| Gadolinium | 157.3 | 161 |
| Terbium | 159.2 | 237 |
| Dysprosium | 162.5 | 290 |
Urbain¹) developed a method for determining the composition of mixtures of rare-earth oxides from the magnetic properties of these mixtures.
3. WEAKLY PARAMAGNETIC AND DIAMAGNETIC ELEMENTS.
The study of the magnetic properties of the remaining elements presents great difficulties in view of the small magnitude of their magnetic constants. Even minute quantities of impurities of ferromagnetic metals—chiefly iron—have a large influence on the magnitude of the susceptibility. Many elements formerly considered paramagnetic, according to more recent investigations, have proved to be diamagnetic. The metallographic state of iron present as an impurity in a given element also plays a large role: if it is present in the form of a compound or of diluted mixed crystals, then it exerts no influence.
Wedekind²), on the basis of Honda’s data, arranged the elements in a series according to decreasing susceptibility, beginning with strongly paramagnetic \((+)\) and ending with strongly diamagnetic \((-)\): \(+\) (ferromagnetic metals) \(Mn, Pd, Cr, Ce?, La?, Ti, V, Nb?, Rh, Pt, Ta, U, Al, Ru, Mg, Na, K, W, Th, Zr, Mo, Os, Sn\) (metallic) \(+;\ — Cu, Cd, Pb, Si, Au, Zn, Hg, Ag, Tl, Sn\) (gray), \(As, Se, Te, I, Br, C\) (diamond), \(Sr, S, B, Sb, Bi, C\) (arc-lamp carbon) \(-\).
Elements occurring in two allotropic modifications have two different susceptibilities and even change sign (tin). For gases, data are almost absent; oxygen is rather strongly paramagnetic \((k\cdot 10^6 = 0.117—0.157)\).
Honda³) represents the susceptibilities of the elements obtained by him as a function of atomic weights; a curve is obtained with a clearly expressed periodic course (Fig. 7⁴). The curve is divided
¹) Urbain. Compt. rend. 150, 913 (1910); 149, 37 (1909).
²) Wedekind, Magnetochemie, 71, Berlin (1911).
³) Honda, Ann. d. Phys. [4] 32, 1054 (1910).
⁴) Atomic weights are plotted on the axis of abscissas, susceptibilities on the axis of ordinates; on the lower horizontal line the series of the periodic system of elements are separated from one another by dashes.
into three parts by two large maxima.^1) In the first of these are placed the ferromagnetic metals and chromium, manganese, vanadium, and titanium (hidden magnetic metals, giving magnetic alloys and compounds)—\((A_1)\); in the second—the rare-earth metals \((A_2)\). In addition, there are three sharp maxima \(B_1, B_2, B_3\), to which correspond the minima \(C_1, C_2, C_3\), and three secondary maxima \(a_1, a_2, a_3\), with the corresponding minima \(b_1, b_2, b_3\). In the three most sharply expressed minima \(D_1, D_2, D_3\), there are three analogous elements: phosphorus, antimony, and bismuth. With increasing atomic weight the distance between the secondary maxima and minima decreases. In some cases analogous elements lie on the corresponding parts of the curve. An irregularity is represented by the position of nickel, which stands before cobalt.
Fig. 7.
Pascal^2) introduces the concept of atomic susceptibility; by atomic susceptibility he means the product of the specific susceptibility and the atomic weight—\(\chi.a.\). The atomic susceptibilities calculated by him for related groups of metalloids increase with increasing atomic weight:
^1) The absolute magnitude of the maxima on the curve is not indicated.
^2) Pascal, Compt. rend. 147, 1290 (1908), Ann. d. Chim et phys. [8] 19, 5 (1910).
| Element | Element | Element | |||
|---|---|---|---|---|---|
| \(Cl\) | 209.5 | \(S\) | 156 | \(P\) | 274 |
| \(Br\) | 319.2 | \(Se\) | 240 | \(Sb\) | 775 |
| \(I\) | 465 | \(Te\) | 389 | \(Bi\) | 1896 |
Honda, in addition to the above-cited investigations at room temperature, investigated a whole series of elements at high temperatures. It turned out that the susceptibility of only a few elements, chiefly weakly paramagnetic or weakly diamagnetic, does not change with increasing temperature; in the others it either decreases or increases1. Thus, Honda’s investigations do not confirm Curie’s law, according to which the susceptibility of diamagnetic bodies should not depend on temperature, while the susceptibility of paramagnetic bodies is inversely proportional to the absolute temperature.
Fig. 8
In the following diagram are shown the changes of susceptibilities with change of temperature for certain elements (Fig. 8). In paramagnetic elements the rapid fall of susceptibility at the beginning of the curve can be explained by the presence of iron; for example, in \(Mg\), which above the transformation point ceases to exert an influence. In bismuth the susceptibility decreases linearly with temperature up to the melting point (\(268^\circ\)), at which a sudden decrease is observed; above the melting point the susceptibility of bismuth does not depend on temperature. Tin presents an interesting case: at ordinary temperature it is weakly paramagnetic, susceptib—
its susceptibility does not change with a change in temperature up to the melting point (233°); at the melting point the susceptibility decreases, tin becomes diamagnetic, and with a further rise in temperature it does not change. Honda1, from a comparison of curves 7 and 8, derives the following rule: an increase in temperature causes a change in the susceptibility of an element in the sense corresponding to a small increase in atomic weight. From this rule it follows that the susceptibility of elements situated at maxima and minima must numerically decrease with increasing temperature. The susceptibility of elements situated on the ascending branches must increase, and on the descending branches decrease. In fact, the susceptibilities of aluminum, iron, cobalt, nickel, palladium, platinum, antimony, and bismuth decrease with increasing temperature; the susceptibilities of sodium, titanium, vanadium, chromium, manganese, ruthenium, radium, and iridium increase with increasing temperature. The susceptibilities of weakly paramagnetic or weakly diamagnetic elements situated at secondary maxima and minima do not change with increasing temperature.2
As for the magnetic properties of substances at low temperatures,3 the magnetization of iron, nickel, and cobalt, as the temperature is lowered to that of liquid hydrogen (20° abs.), increases by approximately 1%. The susceptibilities of chromium, manganese, and vanadium change only insignificantly, whereas according to Curie’s law they should have increased, on cooling to the temperature of solid hydrogen (14° abs.), by approximately 20 times. The susceptibility of ferric oxide up to 64° absolute temperature is inversely proportional to the absolute temperature; with a further lowering of the temperature this dependence is violated. Liquid and solid oxygen do not follow Curie’s law.
In order to conclude the survey of the magnetic properties of the elements, it remains to say a few words about the magnetic properties of compounds of weakly paramagnetic and weakly diamagnetic elements. The magnetic properties of compounds do not depend on the properties of the elements entering into their composition. As we saw above, compounds of ferromagnetic metals are non-magnetic, and, conversely, compounds of metals close to them—chromium, manganese, vanadium—are more or less strongly magnetic. Exactly the same phenomena are encountered in compounds with weakly expressed magnetic properties; a compound of two para-
magnetic elements may be diamagnetic, e.g., \(MgO\), \(Al_2O_3\) ¹); a compound of two diamagnetic elements may be paramagnetic, e.g., \(CuBr_2\) and \(CuCl_2\). Copper occupies an exceptional position among the other elements: it is itself diamagnetic, while its oxidized compounds are strongly paramagnetic, e.g., \(CuO\) \((k \cdot 10^6 = +2.9)\), \(CuBr_2\) \((k \cdot 10^6 = +6.1)\), \(CuSO_4\) \((k \cdot 10^6 = +6.72)\).
S. Meyer ²) establishes a connection between susceptibility and atomic volumes: maxima of atomic volumes correspond to diamagnetic bodies, minima to strongly magnetic ones. If, upon the combination of two elements, a decrease in volume occurs, their paramagnetic character is strengthened; if the combination is associated with an increase in volume, the diamagnetic properties are strengthened ³).
Most liquid compounds are diamagnetic: mineral acids, carbon disulfide, etc.; water is also diamagnetic; the mean value of the susceptibility, from determinations by various investigators, is \(k \cdot 10^6 = -0.75\). Of the gaseous compounds, nitrous oxide is paramagnetic, while most are diamagnetic (carbon monoxide, carbon dioxide, hydrogen sulfide, cyanogen, etc.) ⁴).
4. MAGNETISM OF SALT SOLUTIONS.
The magnetic properties of salt solutions have been studied by a whole series of investigators ⁵). Most of the investigations were carried out by means of Quincke’s “magnetic manometer” ⁶), consisting of a U-shaped tube, one limb of which (the narrow one) was placed in a transverse magnetic field in such a way that the meniscus of the liquid under investigation was in a uniform magnetic field of intensity \(H\). The level of a paramagnetic liquid then rose, and that of a diamagnetic one fell. The susceptibility of the solution was calculated by the formula:
\[ \frac{1}{2}(k-k')H^2 = hgs, \]
where \(k\) is the susceptibility of the liquid, \(k'\) that of the gas above the liquid, \(h\) the change in the level of the liquid, \(g = 981\), and \(s\) the density of the liquid. As a result of these investigations it turned out that the susceptibility of solutions is a linear function of the concentration and depends only on
¹) St. Meyer, Wied. Ann. 69, 247 (1899)
²) St. Meyer, Wied. Ann. 69, 261 (1899); see also Königsberger, Wied. Ann. 66, 731
³) Jäger and St. Meyer, Wied. Ann. 63, 83 (1897); 69, 236 (1899).
⁴) Khvolson, Course of Physics, vol. IV, part 1, p. 803.
⁵) Quincke, Wied. Ann. 24, 347 (1885); Du Bois und Liebknecht, Ann. d. Phys. 1, 189 (1900); St. Meyer, Ann. d. Phys. 1, 668 (1900); Königsberger, Wied. Ann. 66, 731 (1898).
⁶) Quincke, Wied. Ann. 24, 369 (1885).
cation1. According to later studies, the anion also exerts an influence on the susceptibility of the solution, but its influence, in comparison with that of the cation, is small2. Koenigsberger performed the calculations of his observations according to the formula:
\[ 10^{6}k' = k'_1 \frac{p}{100} - 0.80\left(1-\frac{p}{100}\right), \]
where \(k'\) is the susceptibility of the solution, \(k'_1\) is the susceptibility of the dissolved solid substance, \(p\) is the number of grams dissolved in 100 g of solution, and \(-0.80\) is the susceptibility \(k \cdot 10^{6}\) for pure water.
Some investigators calculate the magnitude of atomic and molecular magnetism3. According to Wiedemann’s formula:
\[ K=\frac{p}{m}K_m+\left(\frac{s-p}{s}\right)K_w, \]
where \(k\) is the susceptibility of the solution, \(K_m\) is the molecular magnetism of the dissolved salt, \(K_w\) is the molecular magnetism of the solvent, \(p\) is the number of grams of dissolved salt in 1 cubic cm of solution, and \(S\) is the specific gravity of the solution4.
Centnerszwer5 introduces the concept of the molecular work performed by a gram-molecule of the dissolved substance in a magnetic field; the work performed by the solution is composed of the work of the dissolved substance and of the solvent and is calculated by the formula:
\[ \frac{h_0 S}{H^2} = \frac{pS}{100M_1}A_1 + \frac{(100-p)S}{100M_2}A_2 \]
whence
\[ A_1= \frac{M_1}{M_2} \frac{100M_2h_0g-(100-p)A_2}{p}, \]
where \(A_1\) is the molecular work of the dissolved substance, \(A_2\) that of the solvent, \(M_1\) and \(M_2\) are the molecular weights of the dissolved substance and the solvent, \(h\) is the change in the height of the liquid level, \(H\) is the field strength
\[ \left(h_0=\frac{h}{H^2}\right), \]
\(p\) is the number of grams of salt in 100 g of solution, \(S\) is the specific gravity of the solution, and \(g=981\).
Jaeger and Meyer found for the atomic susceptibilities, assuming that 1 gram-atom of metal is dissolved in a liter of solution, the series1:
\[ \begin{aligned} Ni\ \chi.a.10^6 &= 2\text{--}2.5\ \text{c.g.s.},\\ Co\quad "\ &= 4.\quad "\quad " \\ Fe\quad "\ &= 5.\quad "\quad " \\ Mn\quad "\ &= 6.\quad "\quad " \end{aligned} \]
Du Bois, Liebknecht, and Wills obtained the following series in decreasing susceptibilities: iron (in ferric salts), manganese, iron (in ferrous salts), cobalt, chromium, nickel, copper2. Here one should note the influence of the transition into the ionic state on the magnetic properties of metals. In the series cited above, paramagnetic manganese occupies a place corresponding to a higher value of the atomic susceptibility than that for iron (the data of Jaeger and Meyer) and close to the ferric compounds according to the data of Du Bois, Liebknecht, and Wills. Chromium, which gives only weakly magnetic compounds, falls between cobalt and nickel; diamagnetic copper also gives strongly paramagnetic solutions of salts.
Paramagnetic solutions are given by the salts of the rare-earth metals \(V, Pr, Eu, Nd, Ib, Sa, Gd, Er, Ho\). The relative magnitudes of the molecular susceptibilities of the salts of these and of the ferromagnetic metals are represented by the numbers3:
| \(V\) | \(Pr\) | \(Eu\) | \(Ni\) | \(Md\) | \(Ib\) | \(Cr\) | \(Co\) | \(Sa\) | \(Fe\) | \(Mn\) | \(Gd\) | \(Er\) | \(Ho\) |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1.3 | 3.3 | 4.9 | 5 | 5.2 | 6 | 6.3 | 10 | 11.2 | 12.5 | 15 | 27.3 | 38.2 | 50 |
The magnetic properties of salts vary depending on the valence of the metal; thus, for example, ferric salts are more magnetic than ferrous salts; chromic salts are less magnetic than chromous salts4. The same is observed for manganese salts; copper in this respect is analogous to iron—cuprous salts are either weakly paramagnetic or, like copper itself, diamagnetic. The influence of the transition of metallic ions into complex ions was studied by Pascal on ferricyanide, ferripyrophosphate, and ferrimetaphosphate salts. It turned out that the transition of a metallic ion into a complex ion is always associated with a decrease in the paramagnetic properties5. Potassium ferricyanide is diamagnetic (Oxley)6; also diamagnetic are com-
platinum and cobalt compounds \([Co(NH_3)_6]Cl_3\). Complex compounds of chromium and nickel are weakly paramagnetic.
When a metallic cation passes into an oxygen-containing anion, e.g., \(Mn^{\cdot\cdot} \to MnO_4'\), \(Cr^{\cdot\cdot\cdot} \to CrO_4''\), there is likewise a considerable weakening of the magnetic properties; the same phenomenon is observed upon passage into the colloidal state ²).
5. MAGNETIC PROPERTIES OF ORGANIC COMPOUNDS.
The magnetic properties of organic compounds were first investigated by Henrichsen ³). All organic compounds are diamagnetic. Henrichsen found that in homologous series each \(CH\) group increases the molecular magnetism by 163 (taking the susceptibility of water as \(=10\)) and that the molecular magnetism of primary and secondary derivatives is the same for alcohols, aldehydes, acids, and ethers.
In recent years Pascal ⁴) has undertaken a series of works on the question of the influence of structure on the magnetic properties of organic compounds (analogous to Brühl’s work on molecular refraction). In view of the brevity of the present survey, here it is possible only to indicate in general terms the results of Pascal’s investigations. Pascal calculates the molecular susceptibility of compounds containing no nitrogen or oxygen by the formula: \(χM = \Sigma p \cdot χ \cdot A + \lambda\), where \(p\) is the number of atoms of one kind in the molecule, \(χ \cdot A\) is the atomic susceptibility, and \(\lambda\) is a quantity depending on the structural features of the given compound (double bond, benzene nucleus, etc.); for normal saturated compounds \(\lambda = 0\). The presence of a benzene nucleus causes an increase in molecular susceptibility by \(-15 \cdot 10^{-7}\); for an ethylenic bond \(\lambda = +57 \cdot 10^{-7}\), and for a compound with two ethylenic bonds \(\lambda = 110 \cdot 10^{-7}\).
In the following tables are given the values of the atomic susceptibilities of the elements most frequently entering into the composition of organic compounds, and the values of the molecular susceptibilities calculated from the atomic ones and found directly:
¹) Pascal. Compt. rend. 147, 241 (1908).
²) Pascal. Compt. rend. 147, 742 (1908).
³) Henrichsen. Wied. Ann. 34, 207 (1888).
⁴) Pascal. Compt. rend. 149, 342, 508 (1909); 150, 1054, 1167 (1910); 152, 862, 1010, 1852 (1911); 156, 323 (1913); Pascal: Recherches magnétochimiques: Ann. d. chim. et phys. [8] 16, 531 (1909); 19, 5 (1910), 25, 289 (1912); 28, 218 (1913).
Atomic susceptibilities \(\chi \cdot A \cdot 10^7\).
| \(C\) | \(-62.5\) |
| \(H\) | \(-30.5\) |
| \(N\) | \(-53\) |
| \(Cl\) | \(-209.5\) |
| \(Br\) | \(-319.2\) |
| \(J\) | \(-465.0\) |
| \(S\) | \(-156.0\) |
Molecular susceptibilities \(\chi \cdot M \cdot 10^7\).
| Hydrocarbons | Calculated | Found |
|---|---|---|
| Hexane | 802 | 796 |
| Decane | 1296 | 1297 |
| Di-allyl | 570 | 574 |
| Benzene | 573 | 574 |
| Toluene | 696.5 | 699 |
| Styrene | 702 | 700 |
| Halogen derivatives | Calculated | Found |
|---|---|---|
| Chlorobenzene | 752 | 749 |
| Bromobenzene | 861.5 | 856 |
| Iodobenzene | 1007.5 | 1000 |
| Trichlorobenzene | 1109 | 1109 |
| Benzyl chloride | 875.5 | 877 |
The atomic susceptibility of oxygen entering into the composition of organic compounds varies depending on the nature of its bond with other elements. If an oxygen atom is bound to two different atoms, as, for example, in alcohols and simple ethers, then its atomic susceptibility is \(\chi \cdot A = -48 \cdot 10^{-7}\). If oxygen is bound to carbon by a double bond, as in aldehydes and ketones, then \(\chi \cdot A = +18 \cdot 10^{-7}\). If two oxygen atoms are bound to one carbon atom (acids and compound ethers), then \(\chi \cdot A = -35 \cdot 10^{-7}\). In addition, the magnitude of the susceptibility of oxygen is influenced by the structure of the side chain, as, for example, by the presence of a benzene nucleus, multiple bonds, and tertiary and quaternary carbon atoms.
The susceptibility of nitrogen also has different values in different compounds; in compounds of the fatty series the atomic susceptibility of nitrogen is \(\chi A = -58 \cdot 10^{-7}\), and in compounds of the aromatic series \(\chi A = -48 \cdot 10^{-7}\). Here, too, the influence of multiple bonds between nitrogen and carbon atoms is manifested.
The observations of various investigators set forth above, namely: that ferromagnetic metals in their compounds are non-magnetic, that paramagnetic metals can give magnetic alloys
and compounds, that different magnitudes of susceptibility correspond to allotropic modifications of one and the same element and, finally, the connection that exists between susceptibility and the molecular volume of compounds—all this makes it possible to say that the capacity for magnetization is inherent not in the atoms of ferromagnetic metals as such, but rather is conditioned by the distribution of matter in the molecule. Weiss’s theory of magnetons1 is an attempt to explain magnetic phenomena by means of elementary molecular magnets, analogous to the elementary quantities of electricity—electrons.
Cecilia Reichenshtein.