CH.
157
W352
!
B 398055
AN INVESTIGATION OF THE BORIDES AND THE
SILICIDES
BY
OLIVER PATTERSON WATTS.
A THESIS SUBMITTED FOR THE DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF WISCONSIN
1905
(REPRINTED FROM THE BULLETIN OF THE UNIVERSITY OF WISCONSIN
ENGINEERING SERIES, VOL. 3, PP. 251-318).
J
MADISON, WISCONSIN
1906
ARTES
LIBRARY
1837
SCIENTIA
VERITAS
OF THE
UNIVERSITY OF MICHIGAN
E PLURIBUS UNUM
TUBEUR
SI-QUAERIS PENINSULAM AMOENAM
CIRCUMSPICE
AN INVESTIGATION OF THE BORIDES AND THE
SILICIDES
BY
OLIVER PATTERSON WATTS
A THESIS SUBMITTED FOR THE DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF WISCONSIN
1905
(REPRINTED FROM THE BULLETIN OF THE UNIVERSITY OF WISCONSIN
ENGINEERING SERIES, VOL. 3, PP. 251-318)
MADISON, WISCONSIN
1906
Chemical Librar
QJ.
157
W 352
CONTENTS.
I. HISTORICAL OUTLINE..
II. REDUCING AGENTS-
CARBON.
ALUMINUM
CALCIUM CARBIDE
•
ALUMINUM CARBIDE
SILICON CARBIDE
CALCIUM....
"MISCHMETALL”
III. SLAGS.....
•
IV. ELECTRICAL EQUIPMENT—
1
PAGE
257
!
}
270
271
275
278
278
278
280
280
SOURCE AND CONTROL OF THE CURRENT
281
ELECTRIC FURNACES
283
V. EXPERIMENTS
GENERAL CONDITIONS
•
PRELIMINARY EXPERIMENTS ON SILICON.
SILICIDES.
•
A NEW SILICIDE OF MOLYBDENUM.
BORIDES
··
BORIDES AND SILICIDES BY ELECTROLYSIS.
SILICO-BORIDES
•
VI. ELECTRICAL AND OTHER DATA CONCERNING THE ARC FUR-
NACE
•
VII. CONCLUSION
VIII. APPENDIX
•
BIBLIOGRAPHY.
• D
161247
287
288
290
295
296
299
300
303
310
312
314
ILLUSTRATIONS.
Figure 1. Electric Furnace
Figure 2. Curves of Resistance..
Figure 3. Curious Electrode Phenomenon....
TABLES.
Table I.
Action of Metals upon Silicon...
Table II.
Table III.
Table IV.
Table V.
Table VI.
Table VII.
Physical Properties of Borides and Silicides
Chemical Properties of Borides and Silicides..
Carrying Capacity of Graphite Electrodes.
Carrying Capacity of Carbon Electrodes
•
Effect of Currents upon the Resistance of the Arc....
Effect of Vapors upon the Resistance of an Arc of
Constant Length…...
Table VIII. General Data Concerning the Arc Furnace
Table IX.
Loss in Weight of Electrodes.....
PAGE
283
306
307
PAGE
259
260
267
286
286
301
307
308
309
PREFACE.
During the last two decades, notable contributions have been
made to the field of chemistry through the use of the electric
furnace. The high temperatures, made available by the con-
version of electrical into thermal energy, have resulted in the
discovery of a host of new compounds, and the duplication by
art of various of nature's processes.
Among the compounds which play a prominent part in this
"new chemistry of high temperatures," are the borides and the
silicides; and it is the purpose of this paper to present the re-
sults of an experimental investigation of some of these com-
pounds, together with a brief, yet comprehensive, account of
these two series of chemical compounds, which have become of
scientific and commercial importance through the advent of
the electric furnace.
The methods most generally used heretofore for the prepara-
tion of the borides and the silicides have been synthesis from
the elements, and reduction of a metallic oxide by excess of
boron or silicon. Either of these methods requires as a pre-
liminary the preparation of the non-metal, a long and difficult
task, particularly in the case of boron, the purification of
which by the method of Moissan¹ took from ten to fifteen days
at a minimum. It therefore seemed to the writer that an ex-
perimental investigation of the possibility of the preparation
of the borides and the silicides from their commercially avail-
able oxygen compounds by a single reaction in the electric fur-
nace would be of value. This is one, of the problems of this
1 C. R., 114: 392–7 (1892). Ann. de Chim., 6: 296-320 (1895).
256
BULLETIN OF THE UNIVERSITY OF WISCONSIN.
investigation. A second problem was the production by means
of the electric furnace of a new series of definite chemical com-
pounds, the silico-borides.
For this investigation all the resources of the electro-
chemical department of the University of Wisconsin were put
at the writer's disposal by Professor C. F. Burgess, who by
this, and by his personal interest and encouragement, has
placed the writer under lasting obligations, which are here
gratefully acknowledged.
Laboratory of Applied Electrochemistry,
University of Wisconsin,
May, 1905.
O. P. W.
AN INVESTIGATION OF THE BORIDES AND THE
SILICIDES.
I.
HISTORICAL OUTLINE.
The history of the borides may be said to begin with the dis-
covery by Descostils in 18082 that platinum could be fused by
heating it with borax and lamp-black. This gave a product
that was hard, brittle, and crystalline, which, dissolved in
acids, left a residue of boric acid. It was a full half century
from this first recorded instance of the combining of platinum
and boron, before a boride of platinum of definite composition
was isolated.
Similarly in regard to the silicides, the first discovery was
the mere fact of the combination of silicon with the metals, fol-
lowed after many years by the isolation of definite chemical
compounds. In 1810, Davy³ discovered that the earths, silica,
alumina, and glucina, when heated with iron and potassium,
yielded a brittle ingot of a crystalline texture, harder and
whiter than iron. In 1812, Berzelius* announced that a mix-
ture of silica and carbon, heated in the presence of iron filings,
gave a white metal which, attacked by acids, yielded silica.
Boussingault found that steel could be made with silicon in
5
2 Ann. de Chim., 67: 68 (1808).
3 Ann, de Chim., I, 75: 152.
4 Ann. de Chim., I, 81. 178.
5 Ann. de Chim., II, 16: 10-16.
258
BULLETIN OF THE UNIVERSITY OF WISCONSIN.
place of carbon; and that platinum, heated with carbon, took
up silicon from the crucible, becoming brittle, harder, and less
dense. The number of such alloys with boron and silicon in-
creased, and finally crystalline compounds of definite com-
position were separated from them.
With the use of the electric furnace for scientific research,
the isolation of new borides and silicides received a tremendous
impetus. It was not the promise of great commercial value
which caused this sudden and remarkable extension of these
series, nor the fact that new borides and silicides were likely
to have strange and marvellous properties; quite the contrary,
—their family resemblance is so strong that nothing short of
a chemical analysis serves to distinguish between certain mem-
bers of both series. It was the fascination of exploring this
new realm of high temperature chemistry that led to experi-
mentation, and since, in addition to carbon, boron and silicon
are the chief elements whose compounds with the metals are
stable at electric furnace temperatures, the inevitable result
was a great increase in the list of borides and silicides. The
phosphides, arsenides, and sulphides are other series of com-
pounds stable at electric furnace temperature, which are as yet
only incompletely investigated.
The bulk of the original literature concerning the borides
and silicides is contained in the German and French chemical
periodicals, and is available in English only through brief ab-
stracts scattered in various chemical journals, and in the two
English translations of Moissan's "Le Four Electrique."
It has been the purpose of the writer to collect from the
original sources and to set forth in the appended tables the
most important physical and chemical properties of the borides
and silicides in such form that the data may be useful to all
experimenters in this field.
WATTS—INVESTIGATION OF THE BORIDES AND SILICIDES. 259
TABLE I.—Action of the metals upon silicon.
Metals which do not form silicides.
Metals in which silicon dissolves
at high temperatures. but sep-
arates again as crystals on
cooling.
Aluminum.
Antimony.
Bismuth.
Cadmium.
Gold.
Lead.
Potassium.
Silver.
Sodium.
Tin.
Zinc.
Aluminum."
Silver.
Antimony.
Bismuth.
Tin.8
Zinc.s
Lead.
Gold.9
6
• Vigouroux, Ann. de Chim., 12: 5 (1897).
7 Announced by Deville and Wöhler, Ann. de Chim., III, 49: 73 (1857).
8 Announced by Deville and Caron, Ann. de Chim., III, 67: 435 (1863).
⁹ Metals in second column, with the exception of gold, are arranged in order
of their solvent power for silicon. Moissan and F. Siemens, C. R., 138: 1299
(1904). Bull. Soc. Chim., 31-32: 1015 (1904).
260
BULLETIN OF THE UNIVERSITY OF WISCONSIN.
TABLE II.-Physical Properties of the Borides and Silicides.
Literature and Discovery.
How Made.
Description.
Hard-
Specific
ness. Gravity.
Al, B4
Wöhler & Deville, Ann. de Chim. 52:63- K₂BF.+KCl+Al
thin coppery, hexagonal
91 (1858).
crystals.
Al,B24
Wohler & Deville,
66
66
K₂BF6+KCl+AI
black, thin leaves.
9+
2.53 at 17°
(Supposed to be Boron.)
Al3B48C2
Wöhler & Deville,
66
66
B₂O+Al
yellow crystals.
9+
2.61
(Supposed to be Boron.)
BaB
Composition and formulae of last two
due to Hampe, Ann., 183:90 (1876).
Moissan & P. Williams, C. R., 125:629-34 BO+BaO+Al+ | small regular cryst.
9+
4.36 at 15°
(1897); Bull. Soc. Chim., 17:1015-20
(1897).
Be 6B6C4
Lebeau, C. R., 126:1347-9 (1898)
BeO+B in carbon brilliant metallic cryst.
crucible.
2.4
CaB
Moissan & Williams, C. R., 125:629-31
(1897); Bull. Soc. Cnim., 17:1015-20
(1897).
B2
3
+CaO+Al+ | transparent cubic
or
9+
2.33 at 15°
C
rectangular cryst.
CB
O. Mulhauser, Z. anorg. Chem., 5:92 B,О3+C
(1891).
bluish black, greasy
feeling, malleable.
CB6
Moissan C. R., 118:556 (1894); Ann. de
B+C
black, brilliant cryst.
9.8
2.51
Chim., 9:280–6 (1896).
Fe+B+C
A. Joly, C. R., 97:456 (1883).
Cu+B+C
CoB
Moissan, C. R., 122:424 (1896); Ann. de
Co+B
Chim., 9:273–85 (1896).
brilliant prisms several
mms. long.
7+
7.25 at 18°
CrB
(1902) p. 14-17.
FeB
Moody & Tucker, J. Chem. Soc. Trans.,
Moissan, C. R., 120:173 (1895); Ann. de Fe+B
Chim., 9:273–85 (1896).
Cr+B
gray, metallic.
8
5
metallic crystals.
WATTS-INVESTIGATION OF THE BORIDES AND SILICIDES.
261
MnB
MnB₂
B. du Jassonneix, C. R., 139:1209 (1904). | MnO4+B
Troost & Hautefeuille, C. R, 81:1263-6 B₂O₂+Mn¸C
(1875); Ann. de Chim., 9:65 (1876).
Mo, B4
H. R. Moody & S. A. Tucker, J. Ch.
Soc. Trans. (1902) 11-17.
Mo+ B
NiB
Pt₂ B
SiB3
Moissan & Stock, C. R., 131: 139–43] Si +B
(1900); Ann. de Chim. 20: 433 (1900).
Moissan, C. R., 122: 424 (1896); Ann. de Ni+3
Chim. 9: 273–85 (1896).
Martius, Annalen 109: 79 (1859); Descos-
tils, Ann. de Chim. 67:88 (1808).
SiB6
SrB。
ThB 4
ThB6
WB.
Zr3 B4
35: 3929-32(1002).
BaSi,
CaSi 2
G de Chalmot, Amer. Chem. J., 18:| CaO+SiO,+C
319 (1896).
Moissan & Stock. C. R 131: 139-43 Si+B
(1900; Ann. de Chim. 20: 433 (1900).
Moissan & Williams, C. R, 125: 629-34
(1897); Bull. Soc Chim., 17: 1015 (1897)
B. du Jassonneix, C. R., 141: 191–3 (1905
B. du Jassonneix, C. R., 141: 191-3 (1905)
Moody & Tucker, (see CrB)
Tho₂+B
ThO₂+B
Moody & Tucker E. Wedekind, Ber., Zr+B
Disc. by C. B. Jacobs, July, 1899.
Chem. News, 82: 149 (1900).
Disc. by Wöhler, Ann. de Chim. 69: Si+CaCl₂+Na
224 (1863); Ann., 127: 255 (1863).
C. B. Jacobs, British Assoc. (1900) p. 699. CaCO3+SiO₂+C
*Moissan & Dilthey, C. R., 134: 503-7 Si+CaO
(1902): Ann de Chim., 26: 289 (1902);
Bull.Soc Chim., 27: 1199 (1902).
Schutzenberger, C. R., 114: 1089-93
W+B
and
treated by chlor-
ine.
brillant metallic powder.
6.2 at 15º
violet gray crystals.
pale brass color, brittle.
9
7.1
brilliant prisms several
mms. long.
7+
7.39 at 18°
brilliant black rhombic
9+
2.52
crystals, yellow if thin
thick, black cryst..
9+
2.47
B₂O3+SrO+Al
black crystalline powder. 9+
3.28 at 15°
+C
yellowish metallic powd'r
7.5 at 15°
reddish violet powder
6.4 at 15°
silvery, metallic.
gray, brilliant cryst.
6.8
2.5
CSi
(1892).
Discov. E. G. Acheson, 1890.
SiO2 +C
Si +C
Moissan, Ann. de Chim.,9: 296-300 (1896). Fe, Si+C
colorless hexagonal crys-
tals.
9+
3.12
Fe+SiO₂+C
262
BULLETIN OF THE UNIVERSITY OF WISCONSIN.
TABLE II-Continued.
Literature and Discovery.
Hard-
How Made.
Description.
ness.
Specific
Gravity.
Ce, Si,
Ullik, Chem. Central., 1865 p. 1045
CeSi2
Cr, Si
3
Cr₂Si
Cr3 Sig
CrSi₂
Cr, AlSi,
61 (1901).
Cr₂ AISI4
61 (1904).
Co. Si
CoSi
CoSi 2
Cu, Si
*J. Sterba, C. R, 135:170 (1902); Ann.
de Chim., 2:229-32 (1904).
Zettel, C. R., 126:833-5 (1898).
Moissan, C. R., 121:621-6 (1895); Arn.
de Chim., 9:289 +1896)
*Lebeau and J. Figueras, C. R, 136:
1329-31 (1903).
G. de Chalmot, Amer. Chem. J., 19:69
(1897).
Cr+Si
Cr₂O+Sio+C
Cu+Cr+Si
W. Manchot & A. Kieser, Ann., 337:353- Cr+K,SiF+Al
W. Manchot & A. Kieser, Ann., 337:353- K₂Cr₂O,+K,SiF6
Vigouroux, C. R., 121:686-8 (1897); Ann. Co+Si
de Chim., 12:175 (1897).
Lebeau, C. R., 132:556 (1901); C. R., 135: Cu,Si+Co
475-7 (1902); Ann. de Chim., 27:271-7|
(1902).
*Lebeau, Ann. de Chim. 27:271-7 (1902) Cu,Si+Co+Si
Vigouroux, Ann. de Chini., 12:181 (1895);| Cu+Si
C. R., 122;318-19 (1896).
G. de Chalmot, Am. Chem. J, 19:118
+AI
steel gray cryst.
Electrolysis of
3KF.2CeF 4
CeO₂+Si
lamellar steel gray cryst.
brittle.
Cu+Al+Cr₂O, in
silicate crucible
gray_arborescent cryst.
6+
6.52 at 18°
powder.
metallic cryst.
9+
brilliant prisms
6+
5.6 at 0°
Cr¸ O¸+Sio+C
2
gray metallic needles.
4.39
5
4.7
5+
4.8
7.1 at 17°
5.67 at 17°
brilliant prisms and oc-
tahedra.
6.3 at 20°
dark bluish cryst.
4.5
5 3 at 0°
steel gray, brittle crys
very
6.9 at 18°
tals: reddens in air.
hard
(1897).
Cu4Si
E. Vigouroux, C. R., 142:87-8 (1906).
Cu+Si
silver white: reddens in hard
7.48 at 0°
air.
Fe₂Si
Disc. by Hahn, Ann, 129:57-76 (1861).
Si+Na+NaCl+
FeCl₂+CaF2
feebly magnetic
6.61 at 23°
WATTS—INVESTIGATION OF THE BORIDES AND SILICIDES. 263
Fe.Si,
FeSi
FeSi
Li, Sig
Mg, Si
and
Mg, Si
Mn,Si
3
Fe+Si
Fe₂O+Si
Fe + Si
Fe+Sio+C
small metallic magnetic
7.0 at 22°
prisms.
brilliant gray octahedra.
6 +
brittle white cryst. fee-
very
6.85 at 20°
6.36
bly magnetic.
hard
SiCl,+Fe
Fe+Cu,Si
metallic tetrahedric cryst
7+
6.17 at 15°
Moissan, C. R, 121:621-6 (1995); Ann.
de Chim., 9:289 (1896).
*Lebeau, Ann. de Chim., 26:5-31 (1902).
G. de Chalmot, J Amer. Chem. Soc.,
17:923 1895).
Encyclopaedie Fremy, Vol. 20: p. 93.
Lebeau, C. R., 128:933 6 (1899); Ann. de
Chim,, 26:5-31 (1932).
Disc by Hahn, Ann., 129:57-76 (1861);
G. de Chalmot, Am. Chem J., 19:122
(1897).
*Lebeau, Ann. de Chim., 26:5-31 (1902).
Moissan, C. R.. 134:1083-7 (1902); Bull.
Soc. Chim., 27:1203-7 (1902).
Wöhler, Ann., 107:119 (1858): Wöhler,
Ann. de Chim., 54:218 25 (1858); Geu-
ther, J. f. pract. Chem., 95:421; Phip
son, Proc. Roy. Soc, 13:217 (1864).
Brunner, Pogg. Ann., 101:261 (1857);
Wöhler, Ann, 106. (1858).
Vigouroux, C.R., 121:771-3 (1895).
4
Fe, Si+HF
Fe₂Si + HF
Fe+Si
Li+Si
brilliant cryst
4+
5.40 at 15°
brilliant indigo blue cr'st
very hygroscopic.
1.12
Mn + Si
Mn30₁+Sio+C
gray metallic crystals,
brittle.
6.6 at 15°
4
MnSi
*Lebeau, C. R 136:89-92; 231-6 (1903)·
Carnot & Goutal, Ann. des. Mines, 18:27
(1900); H. N. Warren, Chem. News,
78:319 (1898);
Mn 3 O + Si + H₂
Mn+Cu+Si
brilliant prisms.
6.2 at 15°
MnSi,
*Lebeau, C. R., 136:231-3 (1903).
Mo, Sig
H. N. Warren, Chem. News.7 8:318(1898);
*Viguoroux, C. R, 129:1238-9 (1899).
Viguoroux, C. R.. 121:686-8 (1895).
*Lebeau, C.R., 136:89-92 (1903).
Mn+ Cu + Si
G. de Chalmot, Am. Chem. J., 18:536 Mn3O4 + SiO2 + Ca
(1895).
-
Ni, Si
Pt,Si
Disc. by Guyard, Bull. Soc. Chim. 25:510
(1876); *Viguoroux, C. R., 123:1 7 (1896):
Ann. de Chim., 12:188 (1897).
O+C
Mn+ Cu +Si
MOO, + Si
Ni+Si
Pt+Si
Pt+Si
steel gray crystals.
brilliant tetrahedric crys-
8-9
5.9 at 1.5°
tals.
small dark gray octa-
hedra.
silver white prisms.
5.24 at 13°
7.2 at 17³
13.8 at 15°
264
BULLETIN OF THE UNIVERSITY OF WISCONSIN.
TABLE II.-Continued.
Hard-
Literature and Discovery.
How Made.
Description.
ness.
Specific
Gravity.
RuSi
Moissan and Manchot, C. R., 137:229-32 Ru+Si
(1903); Ann. de Chim., 2:285 (1904)
metallic white prisms
very
5.40 at 4°
hard.
Bull. Soc. Chim., p. 559 (1901).
SrSi2
See BaSi,
SrCO3+SiO2+C
bluish white
hard.
Ti, Si
2
W₂Si,
L. Levy, C. R., 110:1368 (1890; 121:1148 TiC1,+Si
(1895).
Viguoroux, C. R., 127:393 (1898);
WO,+Si
Moissan C. R., 123:13 (1896); Warren,
steel gray metallic crys-
tals.
10.9
Chem. News, 78:318 (1898).
V₁Si
Moissan & Holt, C. R., 135:493-7 (1902); VC+Si
silver white prisms
5.48 at 17°
Vsi₂
Ann. de Chim., 27:277-88 (1902).
Moissan & Holt, C. R., 135:493-7 (1902);| V
V₂O3+Cu+Si
3+Si
brilliant metallic prisms
4.42
Zr Six
Ann. de Chim., 27:277-88 (1902).
E. Wedekind, Ber., 35:39:29 (1902).
ZrO₂+Si
dark crystals.
The numbers denoting hardness in the above table are according to Mohr's scale; - 6, orthoclase; 7, quartz; 8, topaz; 9,
corundum; 10, diamond.
* In case of several articles upon the same compound, that marked by the asterisk is the more comprehensive.
WATTS—INVESTIGATION OF THE BORIDES AND SILICIDES. 265
Additional references are as follows:
A summary concerning silicon and silicides to the year 1897, Vigouroux,
Ann. de Chim., 12:5-71; 153–197.
The preparation and properties of amorphous silicon, Vigouroux, C. R.,
12v pp. 94, 367, 554, 1161, 1393 (1895).
New properties of amorphous silicon, Moissan, Bull. Soc. Chim., 27:1198
(1902).
Silicon by electrolysis, Gore, 1884, Chem. News, 50: 113-4; Hampe, 1889,
Chem. Zeitung, 12:841; Minet, 1891, C. R., 112: 1215-18.
Controversy concerning the existence of a silicide of silver, H. N. Warren,
Chem. News, 67:303-4 (1893); Moissan, C. R., 121:621–6 (1895);
G. de Chalmot, Am. Chem. J., 18:95 (1896); Moissan, Ann. de Chim.,
9:294 (1896).
The preparation of amorphous boron, Mossian, C. R., 114: 392-7 (1892);
Ann. de Chim., 6: 296–320 (1895).
Mulhauser (Chem. Zeitung, 26:807) claims the discovery of silicon carbide
for E. H. and A. H. Cowles, in 1884.
Unless there is other experimental evidence than that ad-
duced in the original publication, it seems to the writer that
the formulae given by H. R. Moody and S. A. Tucker for the
"borides" of chromium, molybdenum, tungsten and zirconium,
should be accepted by the chemical world with more than the
proverbial "grain of salt." The only experimental evidence
given for these formulae is as follows:
CrB. "10 g. chromium and 2.1 g. boron were heated in the
electric furnace for six minutes by a current of 175
amperes at 60 volts. The product contained 82%
chromium, hence it is a definite chromium boride.
of the formula CrB." It should be noted that the
charge contained originally 82.6% chromium.
Mo₂B4. "6 g. molybdenum and 1 g. boron were heated
twenty minutes by 230 amperes at 70 volts. Analy-
WB2.
sis showed 86% molybdenum, hence the formula.”
This charge contained 85.7% molybdenum.
"4 g. tungsten and 0.2 g. boron, after five minutes
heating by 175 amperes at 65 volts showed 89%
tungsten, therefore the formula is WB." The
charge contained 88.9% tungsten.
ZrзB. “15 g. zirconium and 2.2 g. boron were heated by 200
amperes at 65 volts for five minutes. Analysis
showed 86% zirconium." The charge contained
87.2% zirconium.
266
BULLETIN OF THE UNIVERSITY OF WISCONSIN.
It is probable that these four metals are capable of forming
compounds with boron in definite proportions. It is possible
that the formulae assigned are the correct ones for such com-
pounds, but these experiments prove nothing of the sort.
From the energy and time of heating, the temperatures were
very moderate, even low, for the electric furnace. Under
these conditions all the elements concerned are only slightly,
if at all, vaporized, so that the experiments consisted, in each
case, in heating together two non-volatile elements in pre-de-
termined proportions, and finding them in the same propor-
tions in a homogeneous ingot after cooling. The only deduc-
tion from this is, that the two elements concerned are capable
of alloying in the proportions originally taken. This is an
excellent field for further investigation. E. Wedekind¹º has
obtained compounds of zirconium with boron, and with both
boron and carbon by reduction of zirconium oxide by boron,
but has not as yet been able to purify them or to assign a
formula to either. It is the opinion of the writer that definite
borides and silicides of iridium, osmium, uranium, and many
of the metals of the rare earths, can be isolated, although these
are as yet unknown.
In the following table (Table III), compiled from the
original literature, the vigor of chemical action of reagents
upon the borides and the silicides is indicated as follows:
N =
S
SS
= no action.
slight action.
= very slight action.
Racts readily.
Vacts vigorously.
VV acts violently.
VV V
acts with incandescence.
Numerals indicate the approximate temperature in degrees
centigrade.
10 Ber., 35: 3929-32 (1903).
•
WATTS—INVESTIGATION OF THE BORIDES AND SILICIDES. 267
TABLE III.—Chemical Properties of the Borides and Silicides.
Reagents
H₂SO 4
Al2B4
Al2B24
AlgB 48C2
BaB
SS
SS
V
EONHARD
•
N-dil.
S-conc.
VV
Be, B&C4
R
R
CaB6
N-dil.
VV
S-conc.
CB
CB..
Z Z
N
N
Z Z
N
N
ZZZZZZHCI
HF
HCI+HNO3
KOHsol.
Fused KOH
Na, CO₂+KNO
Fused K₂CO
Fused
Oxygen
Chlorine
HCl gas
Fluorine
H₂O
ર
SS
SS
VVV
VV at 800°
SS
N
V-800°
VVV 20°
•
N
250
Ꭱ
R
•
N-20
SS-800
VVV 450
V-800
Z Z
N
N
N
N
Z Z
N
R
R
•
N
V
V
N-500
V-1000
N
VVV-20
250
N
N
N-dil.
CoB
V SSS
ᏙᏙ
V VV VVV V
VVV 800
VVV 800
S-conc.
❤
CrB
S
N-cold
FeB.
S-hot
Mn B
V
V
MnB₂
R
Mog B 4
S
Es ▷ DROS
V SS
cs 2020 is cs
S
Ꭱ
Z DR
N
V
V
V
ᏙᏙᏙ 800
S
VV
ᏙᏙ
R 700
V 800
V 800
VVV-20
R
R
V
N-dil
NiB.
V SSS
VV
S-hot
VVV VVV | V | VVV 800
VVV 800
VVV 20
a zza is z os
N
N
Pt, B.
SiB
S-hot S
N
N
R
V V
VVV 800
VVV 100?
3
2
268
BULLETIN OF THE UNIVERSITY OF WISCONSIN.
TABLE III.-Continued.
SiB。
N dil.
SrB。
S-hot
H2SO4
HNO3
ZZ HCI
N
Z HF
N
HCI+HNO3
KOH-sol.
S
Fused KOH
<< Fused K,CO;
Fused
NA₂CO₂+KNO,
Oxygen
Chlorine
HCl gas
Fluorine
H₂O
VVV 800
VVV 100
N-20
ᏙᏙ
N
V 800
S conc.
ThB
ThB
N-cold
V
4
N
WB
S
Zr,B+
Basi₂
2
CaSi₂
R dil.
CeSi₂
V-conc
Cr,Si
N
Cr₂Si
N
Cr3 Si 2
N
•
CrSi 2
Cr2 AlSi3
N
Co, Si
SS
CoSi
N
Cosi 2
N
Cu₂ Si.
R
Cu,Si
Fe,Si
N
ZOO PR
PROD LOR Zz z ZZZ ► DZ
ZZ ZZZZ £ UZ
V.hot
N
N
ZZ
N
N
VVV
VVV
S-800
V 500
R 500
VV hot
VVV warm
•
•
S-800
V 500
R 500
VVV warm
V
·
S
Ꭱ
Ꭱ
Ꭱ
R
SS
Ꭱ
VV
R
S
SS 900
•
S 20
N-dil.
VVV 800
N 20
VVV 20
Ꭱ
V
VVV
SS
V 20
ZZ RSS
SSS
SSS
R 100
R-conc.
VVV 800
N
Ꭱ
N
ន
N
S-20
ՇՐ Զ
V
VVV 800
S
R
R 700
•
S-conc.
N
V
V
SS 1100
Ꭱ 400
R
N
100
N-20
N
SSS
RRR
N-20
R
N
N
Ꭱ
N
Ꭱ
V
S 800
R
S
SS
VV R
SS 1200
Ꭱ 800
VVV 800
V 300
SS
V
VV
SSS 1200
V 300
R 800
Ꭱ 700
•
•
VVV 20
VVV-warm
VVV-warm
R
S
N V
R
is p
S
R
Ꭱ
R
SS S
ΤΩ ΤΩ
R
N cold
S 800
•
WATTS—INVESTIGATION OF THE BORIDES AND SILICIDES. 269
FeSi
FeSi
LiSi
ZZ
N
N SS
N
VV
VV
Mg 2
Mg4Si
Mn.Si
MnSi
3
VV
•
V
•
SS
MnSi
ZZ
N
N
Mo₂Si
Ni,Si
SS SS.
Pt,Si
N
RuSi
N
SrSi
Ti,Si
2
W₂Sig
V.Si.
Vsi₂
CSi
ZZ Z
N
N
N
SZZŁ ZZZ Z
SZZZZZ ZZZ Z
SS
SANDR RS
V SSS
V
V
V
R
N
S-100
ᏙᏙ
SS 1200
ᏙᏙᏙ 650 R 700
VVV-warm
VVV 20
Ꭱ
V V
S
UNN
•
VV
VV VV
R 800
VVV-warm V
VVV 500 VVV600 VVV-hot
SS 1000
Ꭱ
N
SSS
Ꭱ
RNRN
R
N
SS
ᏙᏙ
VVV 300
N
R
V
S 800
VVV 800
Ꭱ
R
SS
SS
Ꭱ 20
S 800
Ꭱ
VVV 20
R
S
S
burns
S
R
•
S
R
Ꭱ
R
S
•
N
V 20
V
N
ZZZ Z
SS
VV
VVV 250
N 1000
►
N
N
N
SAA
R
R 800
R 800
N 800
Ꭱ 800
R 800
R-warm
V-hot
N
SS 600
N 1000
R 1200
270
BULLETIN OF THE UNIVERSITY OF WISCONSIN.
In general, the borides and the silicides are brittle, metallic,
harder than the corresponding carbides, and very resistant to
chemical reagents. As seen from Table II, the usual method
of preparing these compounds has been that of direct synthesis
from the elements; occasionally the metallic oxide has been
reduced by boron or silicon, and still more rarely, the oxygen
compounds of both the metal and the non-metal have been
simultaneously reduced by carbon or magnesium.
II.
REDUCING AGENTS.
This investigation requires a survey of the field of possible
reducing agents, their properties, advantages, and limitations.
CARBON.
Carbon constitutes the most important reducing agent in
technical metallurgy, and it has certain advantages over other
substances for reduction in the electric furnace. It is cheap,
it is a powerful reducing agent at high temperatures, so
slightly volatile that it remains in the furnace, while the prod-
uct of its action is a gas, and hence escapes as soon as formed.
This last is an important advantage of carbon over most other
reducing agents, which usually leave in the furnace slags that
are hard to break up and difficult to get rid of by chemical
means.
The great disadvantage of carbon is its tendency to form
carbides with nearly all metals which are produced in the
electric furnace, and also in the presence of silica or borax, to
form the carbides of silicon and boron, extremely resistant to
chemical reagents and difficult to remove from the product de-
sired. In the reduction of the metallic oxides, the amount of
carbon in the product may be made very small, and even neg
ligible in some cases, by using an excess of metallic oxide in
WATTS-INVESTIGATION OF THE BORIDES AND SILICIDES.
271
the charge. In the preparation of borides and silicides from
their oxygen compounds, however, this method of removal of
carbon from the product is not available. If carbon, boric
anhydride, and any metallic oxide are mixed in the propor-
tions necessary for complete reduction, and then an excess of
the metallic oxide be added for the purpose of removing carbon
from the alloy or chemical compound which is desired, after
reaction has occurred the unreduced substance will be, not the
excess of metallic oxide, but its equivalent, as regards oxygen,
of boric anhydride. Unlike the metallic oxide, this will have
little, if any, effect in removing from the product any carbon
which may have been taken up during the reduction.
Hence it will be seen that carbon is not a suitable reducing
agent for use in the preparation of borides and silicides by the
method proposed in this investigation.
SODIUM AND POTASSIUM.
Sodium and potassium, so often employed by chemists for
difficult reductions, cannot be used at the high temperature of
the arc furnace because of their violent action and volatility.
ALUMINUM.
The use of aluminum as a reducing agent began about the
middle of the last century, but has assumed practical import-
ance only within the last six years. The usefulness of carbon
as a reducing agent is limited to the reduction of oxides, except
that it removes the oxygen from ternary compounds such as
the sulphates. Aluminum, however, is an effective reducing
agent for sulphides and chlorides, as well as oxides.
The following outline shows in part the development of its
use as a reducing agent.
1858. Deville and Wöhler¹¹ reduced boron trioxide by
aluminum.
"Ann. de Chim., 52: 63 (1858).
PUL
272
BULLETIN OF THE UNIVERSITY OF WISCONSIN.
•
1858. Wöhler12 formed a chromium-aluminum alloy from
chromic chloride.
1859. Beketoff13 produced a 33% barium-aluminum alloy
from barium chloride and barium oxide.
1860. Michel¹¹ made aluminum alloys with manganese, iron,
nickel, and titanium from their chlorides, also of tungsteǹ and
molybdenum from a mixture of chloride and oxide.
1861. Tissier 15 introduced silver sulphide into fused alumi-
num and reduced it, but failed to reduce the sulphides of zinc,
iron and copper!
1888. Beketoff16 obtained potassium and rubidium from
their hydroxides by aluminum, but found that sodium and
potassium separated aluminum from its chloride:
1888. L. Levy¹7 made an alloy of titanium and aluminum.
1893. Greene and Wahl* produced manganese on a commer-
cial scale by heating a mixture of manganese oxide and gran-
lated aluminum.
1895. Vigouroux18 reduced silica by powered aluminum.
1896. Moissan¹ by introducing a mixture of metallic oxide
and aluminum filings into a bath of fused aluminum, made
alloys of aluminum with molybdenum, nickel, titanium,
tungsten and uranium. He made a chromium-copper alloy by
adding a chromium-aluminum alloy to copper, and then re-
moved the aluminum by copper oxide.
1896. Combes 20 substituted a metallic sulphide or chloride
for the oxide of Moissan's method in order to secure a better
separation of the alloy from the slag.
1898. Goldschmidt 21 heated in a crucible, mixtures of gran-
ulated aluminum with metallic oxides, chlorides and sulphides,
12 Ann., 106 118.
18 Ann., 110: 374.
14 Ann., 113: 248; 115: 102.
15 C. R., 52: 931.
16 Ber., 21: 424.
17 C. R. 106: 66.
* J. Frank. Inst. 135: 218–23.
18 C. R., 120: 1161–4.
19 C. R., 122: 1302-3.
20 C. R. 122: 1482-4.
* Ann., 301:19–28, J. Soc. Ch. Ind., 17: 543-5.
WATTS-INVESTIGATION OF THE BORIDES AND SILICIDES. 273
producing a metal free from carbon and aluminum. In this
way he reduced the sulphides of cobalt, lead, molybdenum,
nickel and zinc, and stated that all sulphides could be reducd
by aluminum except those of barium, calcium, lithium, mag-
nesium, potassium, sodium and strontium (cf. Tissier above).
He next turned his attention to the reduction of the oxides,
from which he separated either in the pure state, or
as alloys of aluminum, the elements boron, calcium, cerium,
chromium, cobalt, copper, iron, lead, manganese, molybdenum,
niobium, potassium, sodium, tantalum, thorium, tin, titanium,
tungsten, vanadium and zirconium, but found that aluminum
would not reduce magnesia. He then developed his now well-
known method of igniting the charge from the top by means
of a cartridge of barium peroxide and aluminum, fired by
magnesium ribbon.
1898. Franck 22 reduced phosphates, with the formation of
aluminum phosphide (Al5P3) and the liberation of phos-
phorus, and reduced the carbonates of barium, calcium, lith-
ium, potassium, sodium, and strontium by heating with pul-
verized aluminum. The metal liberated alloyed with excess
of the reducing agent. He remarked that the oxides of copper,
lead and silver were reduced explosively, and that the presence
of calcium sulphate in calcium phosphate caused explosive re-
duction. Goldschmidt had called attention to explosions
caused by sulphates, in reducing ferric oxide and molybdenum
oxide.
1901. Duboin23 reduced the alkaline earths by ignition with
powdered aluminum or magnesium in hydrogen.
24
1903. Rossi 2* reduced ores of titanium and tungsten in ton
lots, by melting aluminum in the electric furnace, and then
feeding in the ore. The product contained only traces of
aluminum and carbon.
1904. Goldschmidt 25 gave an account of his production of
22 Chem. Zeitung, 22:236-45, J. Soc. Ch. Ind., 17:612.
28 C. R., 132: 826–8.
24 Electrochem. Ind., 1: 523.
25 Electrochem. Ind., 2: 145–7.
274
BULLETIN OF THE UNIVERSITY OF WISCONSIN.
carbon-free metals and alloys. In 1902, Dr. Goldschmidt had
succeeded in obtaining, by his process of firing a charge of the
oxide and pulverized aluminum, only the following metals in
a solid, coherent mass: chromium, cobalt, copper, iron, lead,
manganese, molybdenum, nickel, niobium, tantalum, and tin
Later the niobium was found by Muthmann to contain alumi-
Silica and vanadic acid were not reduced, while bar-
ium, beryllium, boron, cerium, thorium, titanium, tungsten,
and uranium were scattered throughout the slag. Alloys of
these metals have been produced, however, by adding to the
charge some easily reducible oxide like that of iron, to supply
heat enough to secure fusion and a complete reduction.
num.
1905. W. Huppertz 26 in attempts to reduce titanium oxide
by the Goldschmidt process, in order to bring about complete
reaction, combined electric heating with the alumino-thermic
method. With certain precautions, he believes that it will be
possible in this way to produce pure titanium in large masses
from rutile.
As indicated above, aluminum as a reducing agent has re-
ceived much attention from scientific investigators, and has
proved very efficient. Aside from its high price, it has two
disadvantages; one will be mentioned under slags; the other
is its remarkable tendency to form alloys. In respect to the
number of metals with which it will unite, it equals, if it does
not exceed, carbon. When it is used in ingots for the reduc-
tion of metallic oxides, an alloy of aluminum is usually pro-
duced. Goldschmidt has overcome this difficulty by using
aluminum in the form of powder or grains. There are, how-
ever, as mentioned, certain oxides from which only alloys can
be obtained by this method.
Early in the course of this investigation, entirely indepen-
dently of the work of Huppertz, the writer developed a method
of firing in the electric furnace a charge consisting of powdered
aluminum and any other powdered substance or substances that
he desired to reduce, and obtained a product free both from
26 Abstract in Electrochem. Ind., 3: 35, of articles in Nos. 17-22 Metallurgie.
WATTS—INVESTIGATION OF THE BORIDES AND SILICIDES. 275
carbon and aluminum. Besides applying the method to rutile,
as was done by Huppertz, titanic, boric and tungstic "acids"
were similarly reduced, and by the addition of a retarder, to
check the vigor of reaction, ordinary Goldschmidt charges were
fired in the electric furnace without accident. The first such
experiment was made in December, 1904, from which time the
method was regularly used. In all about fifty reductions were
made by powdered aluminum in the electric furnace.
CALCIUM CARBIDE.
Like aluminum, calcium carbide is a convenient medium
for the storage in concentrated form of energy derived from
steam or water power, so that it can be conveniently trans-
ported and applied in a small space.
H. N. Warren27 appears to have first suggested the use
of calcium carbide as a reducing agent, stating that the
oxides of chromium, cobalt, copper, iron, lead, manganese,
molybdenum, nickel, tin, and tungsten yield calcium alloys
when heated with calcium carbide.
Moissan 28 reduced metallic oxides by calcium carbide in
the electric furnace, and obtained the carbides A14C3, Cr₂C::
Mn, C, Mo₂C, SiC, TiC, and WC. With the easily reducible
oxides of copper and lead, he found reaction to occur before
the charge was heated to fusion. The equation of reaction for
lead oxide is
CaC, + 5PbO = 2CO,+ CaO+5Pb.
The reduced metal contained no calcium, contrary to the
results of Warren's experiments. Reduced bismuth and cop-
per were also free from calcium. In the course of Moissan's
article, he said that he had previously 29 indicated that calcium
carbide is a strong reducing agent. This statement, however,
is not borne out by the article referred to, in which no mention
27 Chem. News, 75: 2, Jan. 1st (1897).
28 C. R., 125: 839-44 (1897); Bull. Soc. Chim., III, 19:870 (1897).
20 C. R., 118: 501 (1894).
276
BULLETIN OF THE UNIVERSITY OF WISCONSIN.
is made of calcium carbide as a reducing agent. The only
experiments which might be construed to bear upon its reduc-
ing properties, are heating it with some powerful oxidizing
agents. With fused "chromic acid," calcium carbide became
incandescent. Potassium chlorate and nitrate at their melt-
ing points had no sensible action upon it, but at a red heat,
they produced incandescence. Lead peroxide with calcium
carbide became incandescent at a dull red heat. These experi-
ments were described, but with no deductions as to the reduc-
ing character of calcium carbide.
Tarugi30 confirmed Warren's formation of calcium al-
loys by the reduction of the oxides of copper and lead, and
stated that all salts of antimony, bismuth, cadmium, cobalt,
gold, nickel, platinum, silver, tin, and zinc, when heated with
calcium carbide, yielded alloys (cf. Warren) readily decom-
posable by water. He reduced the oxide, phosphate, chloride,
sulphate, carbonate, and borate of copper.
B. Neumann31 reduced metallic oxides, chlorides, car-
bonates, and mixtures of these, with calcium carbide by heat-
ing the charges in clay crucibles in a furnace. He found that
chlorides are most easily reduced, and gave the following equa-
tions:
M₂ 0+2 MCI+CaC¸=6 M÷Ca¯l,+2 CO.
2
With sodium chloride as flux, copper, lead and nickel are ob-
tained in mass while some metals are scattered in globules, and
others vaporize.
M,SO,+2M_0+CaC,=6M+CaSO4+2 CO.
M¿CO¸+2 M¸O+CaC¸=6 M÷CaCO¿+2 CO.
In a brief article32 Fr. von Kügelgen stated that carbon di-
oxide is formed in these reactions instead of carbon monoxide,
and supported this contention in a long article. He gave as
the equation of reaction:
30 Gazetta chim. ital., 29, i, 509-512; Abstract in Chem. Zeit., 23: 292 (1899);
Abstract in Zeit. f. Electroch., 7: 542 (1899).
31 Chem. Zeitung, 24: 1013-4.
32 Chem. Zeit., 24: 1060 (1900).
33 Zeit fur Electrochem., 7: 441-50; 557: 573 (1901).
WATTS-INVESTIGATION OF THE BORIDES AND SILICIDES. 277
MCI₂+4 MO+CaC₂-5 M+CaCl₂ +2 CO2.
2
2
In addition to a long series of metallic oxides, chlorides, and
mixtures of these (among them sodium chloride and mag-
nesium chloride), he reduced sodium and potassium hydroxides
by calcium carbide. Magnesium oxide is not reduced, but
alumina is partially reduced. Mixtures of carbide with the
following may be kindled by a match: zinc chloride, copper
chloride, copper oxide, silver chloride, tin chloride, bismuth
chloride, chromium chloride.
This difference of opinion in regard to the formation of car-
bon monoxide or carbon dioxide in the reduction resulted in a
lengthy controversy between Neumann³ and Kügelgen.35 The
facts appear to be that with easily reducible compounds, the
product is mainly, and in some cases entirely, carbon dioxide;
at the higher initial temperatures required by compounds
which are reduced with greater difficulty, increasing amounts
of carbon monoxide appear as a product.
Geelmuyden³ used calcium carbide as a reducing agent
in the electric furnace. He reduced boric anhydride with
the production of calcium boride, and also reduced the
sulphides of antimony, iron, lead and magnesium. The metals
vaporized with the exception of iron; this was contaminated by
carbon. Aluminum sulphide was not reduced by calcium car-
bide.
Calcium carbide, then, reduces the same classes of com-
pounds as are reduced by aluminum and is especially advan-
tageous to use for mixtures of an oxide with a choride or a
sulphide. It seems to be settled that calcium alloys are pro-
duced in certain reductions, but the exact conditions which
cause their formation are not yet established. The products
are contaminated by carbon under the same conditions, and
possibly to the same extent, as if free carbon were used for
reduction.
34 Chem. Zeitung, 24 : 1013–14; 26; 176, 1108 (1902); 27: 1026–8 (1903).
34 Chem. Zeitung, 24. 1013-14; 26: 176, 1108 (1902); 27: 1026-8 (1903).
Zeit. f. Electrochem., 8: 772-5; 795; 939 (1902); 9: 699-701 (1903).
35 Chem. Zeitung, 24: 1060 (1900); 27:743 (1903). Zeit. f. Electrochem.,
7:541, 557, 573 (1901); 8: 781 (1902).
3º C. R., 130:1026–9 (1900).
}
278
BULLETIN OF THE UNIVERSITY OF WISCONSIN.
ALUMINUM CARBIDE.
Aluminum carbide would seem especially advantageous for
scientific purposes in the reduction of mixtures of oxide with
chloride or sulphide.
SILICON CARBIDE.
B. Neumann³ states that by using sodium carbonate as a
flux, carborundum reduces metallic oxides. Carborundum has
been tested by the writer in the preparation of silicides, for
which use it has the advantage of serving at once as reducing
agent and source of silicon. Similarly, boron carbide might
be used to advantage for making borides, but so far as the
writer knows, this has not yet been tried.
CALCIUM.38
If we may rely upon the claims of several experimenters, the
problem of the production of calcium has been solved, and we
may soon expect to find it on the market at a price that will
make it available for use on a large scale. Current prices are
as follows:
London, April, 1905:
One pound bars
Five pound tins, per pound
New York, May, 1905:
One hundred grams
One pound
. $3 92
$2 32
. $3 00
.$10 00
As a reducing agent, it may be regarded, like aluminum
and calcium carbide, as merely a convenient medium for the
37 Zeit. f. Electrochem., 8:772 (1902).
38 K. Arndt, Zeit. f. Electrochem., 8: 861 (1902). Borchers and Stockem,
Zeit. f. Electrochem., 8:751-8; 938 (1902). Borchers and Stockem (transla-
tion), Elect. World and Engin., 40: 1002. J. H. Goodwin, J. Amer. Chem.
Soc., 25: 873-6 (1903). J. H. Goodwin, Proc. Amer. Philosoph. Soc., Vol. 43:
No. 178; Abstracted Electrochem. Ind., Feb. 1905: p. 80-81.'
WATTS—INVESTIGATION OF THE BORIDES AND SILICIDES. 279
storage, transportation, and application of the energy of water
power or that derived from coal. Where purity of the prod-
uct of reduction is of importance, calcium would seem as a
general reducing agent superior to any other. In the electric
furnace it should reduce most, if not all, of the oxides, chlor-
ides, and sulphides which we are now able to reduce by other
agents. It contains no carbon to be introduced as an impurity
into the product. In the reduction of oxides alone, the slag
formed, lime, is easily fluxed by certain substances, and even
if it should remain as calcium oxide in the furnace, its density,
tenacity, hardness, and resistance to acids are all less than for
alumina, and render it in these respects a much less objection,
able substance to have in the furnace than the latter.
The metal in granular or powdered form, can undoubtedly
be used as aluminum now is in the Goldschmidt method of
reduction. For this use it will be superior to aluminum for
difficult reductions. For one equivalent of oxygen, its heat of
oxidation is ten per cent. greater than that of aluminum-a
sufficient increase to insure the completion of certain reduc-
tions that by aluminum are only partial. Mixtures of oxides
with calcium will undoubtly have a considerably lower
kindling point than those with aluminum, a quality which will
be a convenience, and will tend to cause more rapid and com-
plete reaction.
A disadvantage of calcium as compared with aluminum is
its oxidation at ordinary temperatures. Another, and possibly
serious objection to calcium as a reducing gent in the electric
furnace, is to be seen in the strong oxidizing action of fused
lime upon silicon, boron, chromium, cobalt, iron, manganese,
nickel, and titanium in the experiments of Moissan.39
In spite of these objections to the use of calcium as a reduc-
ing agent in the electric furnace, the writer regards it as the
most promising of any agent yet suggested for difficult reduc-
tions.
39 C. R., 134: 280 (1902).
280
BULLETIN OF THE UNIVERSITY OF WISCONSIN.
"MISCHMETALL.” 40
The latest reducing agent for oxides is the "mischmetall"
produced from the cerite minerals, and consisting of cerium
forty-five per cent., lanthanum twenty per cent., didymium
fifteen per cent., and the remainder of samarium, - erbiuır,
gadolinium, and yttrium. The advantages claimed for this
over the other metallic reducing agents, are its low kindling
point, 150°, and the ready fusibility of the oxides formed by
its action, as compared with magnesia and alumina. Its low
kindling point causes an exceedingly rapid reaction, and the
fusible slag produced permits the reduced metal to collect to a
regulus. By its use the following metals were obtained from
their oxides in a state of the highest purity: cobalt, chrom-
ium, iron, manganese, molybdenum, niobium, tantalum, and
vanadium. The oxides of lead, tin, titanium, and zirconium
yielded only alloys. With silica the reaction is feeble. Tung-
stic acid is reduced but gives only a dark brown mass. The
authors believe that by working with kilogram lots, a regulus
of tungsten might be obtained.
III
SLAGS.
A matter of nearly as great importance as the reducing
agent, is the nature of the slag formed by its action. Some
of the effects of slags have already been mentioned. The fol-
lowing classification of the action of slags, while based on their
action in an electric furnace of the arc type, will apply more
or less to all high temperature work.
1. A fluid slag conduces to complete reaction; with a solid
slag, reaction will be incomplete. This points to the selection
of different reducing agents according to the maximum tem-
•
40 L. Weiss and O. Aichel, Ann., 337 370-89 (1904). Abstract J. Chem.
Soc., 1905, p. 164.
WATTS-INVESTIGATION OF THE BORIDES AND SILICIDES. 281
perature attained-or at least permits the use at high temper-
atures of reducing agents which could not be used at lower
temperatures. The success of the Goldschmidt method of re-
duction depends upon the attainment of a temperature at which
the slag, alumina, is perfectly fluid. The substitution of a
sulphide for part of the metallic oxide in the charge should
help in the more difficult reductions by the Goldschmidt
method on the core of a more fluid slag, since aluminum sul-
phide is more fusible than aluminum oxide.
2. The slag protects the metal or metallic compound which
is dense enough to sink through it from the direct radiation of
the arc, and yet if fluid, the slag conveys heat very rapidly to
this metallic charge or product.
3. A considerable quantity of slag fixes at its own boiling
point the maximum temperature to which the product is sub-
jected.
4. The slag may exert a purifying or an injurious effect
upon the product, according to its chemical nature.
5. By adding calculated amounts of metallic oxide to the
slag, used as a protection, metals may be melted below the arc
without contamination by the carbon which falls from the
electrodes. This method of fusing carbon-free metals has been
developed in the course of this investigation.
IV
ELECTRICAL EQUIPMENT.
SOURCE AND CONTROL OF THE CURRENT.
The dynamo which supplied current for most of the experi-
ments was a direct current 110 volt machine, rated at 825
amperes. Alternating current with variable voltage was also
available up to twenty kilo-watts, and was occasionally used.
The direct current most frequently used was controlled by
an iron water-pipe rheostat of ten equal sections in series, hav-
282
BULLETIN OF THE UNIVERSITY OF WISCONSIN.
ing a total resistance of 0.5 ohm. With all but one section
of the rheostat short-circuited, 600 amperes was delivered at a
pressure of 80 volts at the furnace terminals. Of the total
energy transformed by the generator, the per cent. delivered
to the furnace was as follows:
Average for an entire experiment....
During maximum load (600 amp. x 80 volts).
Highest per cent for ten consecutive minutes in any experi-
ment...
Highest per cent ever attained, for two minutes, by running
the furnace directly from the dynamo
68 per cent.
73 per cent.
80 per cent.
91 per cent.
The furnace was started with the entire rheostat in the cir-
cuit, which explains the low average. In experimental work,
the convenience of this method of control more than compen-
sates for the waste of energy. This waste of energy might be
considerably diminished by having some sections of the rheo-
stat of one-half and one-third the resistance of the others, with
a generator voltage of eighty or ninety instead of one hundred
and ten.
It is stated by Mrs. Ayrton" "without some external resist-
ance, it is impossible to maintain a silent are between solid
carbons." From the behavior of the arc when the resistance
of the rheostat was reduced to only 0.05 ohm, the writer be-
lieved that even this resistance might be dispensed with, and yet
the arc would run steadily. This was verified by an experi-
ment. The furnace was heated, as usual, with the line voltage
at 110. The voltage regulator was then cut out, the voltage
lowered to 70, and the last section of the rheostat short-cir-
cuited. The only resistance external to the arc was that of
the cables and armature,-between 0.012 and 0.018 ohm.
The arc ran satisfactorily, although not so steadily as with one
section of the rheostat in the circuit. The voltage of the gen-
erator was then raised to 80 with the result that the widest
variations of current for an arc 1% inches in length was from
310 to 450 amperes. It is probable that most of this variation
41 The Electric Arc, p. 250
WATTS—INVESTIGATION OF THE BORIDES AND SILICIDES. 283
was due to the defective form of the anode, developed during
the preliminary heating. The average resistance of the arc
was twelve times that of the remainder of the circuit, and con-
sequently the per cent. of energy delivered to the furnace was
92 +.
In this experiment, it was observed that when the arc be
came silent after it had been "shrieking," the voltage across
the terminals diminished and the current increased. This is
contrary to Mrs. Ayrton's observation upon the open arc."
42
These differences between the action of the arc lamp and the
arc furnace would undoubtedly yield interesting results if
thoroughly investigated.
ELECTRIC FURNACES.
For the production of metallic borides and silicides by the
direct reduction of oxygen compounds, the horizontal arc fur-
nace was decided upon as most easily managed and best suited
to the purpose.
Several different materials of construction were tried and
rejected. Finally magnesite brick was decided upon as being
by far the most resistant material available for the furnace.
Other materials tried were limestone, ordinary fire brick,
chromite, and silica brick. The only limestone obtainable was
very easily broken, and as a furnace material cracked badly,
even when protected from the arc by a graphite lining. It is
quite possible that a harder, stronger limestone, and particu-
larly a dolomite, might prove a satisfactory material for fur-
nace construction. Ordinary fire brick cannot be used for the
inner zone of an arc furnace because of its fusibility. Silica
and chromite brick, while less fusible than fire brick, are still
far from satisfactory. The former fuses and disintegrates by
the heat of the arc, and the latter cracks badly before its fusing
point is reached. Both are distinctly more fusible than mag-
nesite brick, and are more readily attacked by slags.
In point of infusibility, magnesite bricks are very satisfac-
42 The Electric Arc, p. 279.
3
284
BULLETIN OF THE UNIVERSITY OF WISCONSIN.
tory, being melted to a depth of less than 1/4 inch by ten min-
utes exposure to an arc of 600 amperes at 80 volts only an
inch below the brick. They do, however, crack under the in-
fluence of heat, and must be handled very carefully. These
brick were obtained from the Harbieson-Walker Refactories
Co.
In form, the furnaces consisted of a box built of magnesite
brick without cement, surrounded by a second layer of fire
brick. The dimensions of the horizontal arc furnace were:
outside 20 inches square, 26 inches high; and inside-834
inches long, 7 inches wide, 7 inches deep-429 cubic inches.
The outside dimensions included the ventilated base of fire
brick. Although the table on which the furnaces stood had a
thick cement top, it was found necessary to have air circulation
between the body of the furnace and the table top. The fur
nace is shown in Figure I.
The cover consisted of two magnesite bricks, and the elec-
trodes entered through holes drilled in the bricks which formed
the ends. Later, the ends were simply built up around the
electrodes, and the cracks stuffed with asbestos paper. This
type of furnace, with two modifications, was used in about one
hundred and thirty experiments.
Much trouble was experienced from the contamination of the
products by iron. Even after all materials that entered into
the composition of the charges were rendered iron free, the
iron was still found, and its source was finally discovered to be
the magnesite brick. A fused metal resting upon these in the
presence of a strongly reducing slag, will be contaminated by
iron. To remedy this, a bed of powdered magnesia, free from
iron, was spread an inch deep in the bottom of the cavity.
This prevented any contamination from the bottom of the fur-
nace, but there was still occasional contamination from the
side walls. For a part of the work, the entire inside, except
for the bottom and the upper half of each end, was lined with
inch-thick sheet graphite. Although this introduced carbon,
it effectually prevented contamination by iron, and greatly

WATTS INVESTIGATION OF THE BORIDES AND SILICIDES. 285
prolonged the life of the magnesite brick. With the graphite
lining, the inside dimensions were 834 inches by five inches by
six inches-262 cubic inches.
It would have been an advantage if the bottom of the furnace
had been laid in cement, or strapped with iron to prevent the
bricks from separating with the alternate expansion and con-
traction.
FIGURE 1.-The Electric Furnace.
For the short duration of the heating, fifteen to twenty min-
utes, with a powerful current for less than half this time, the
heat insulation was found to be ample. The outside layers of
brick could be removed by the bare hands, if done as soon as
the current was shut off, although an hour later, they might
become red-hot. The nearer air-tight the furnace is, without
actually being so, the better.
The electrodes used were of Acheson graphite. The maxi-
mum current permissible on account of oxidation of the elec-
trodes outside of the furnace has been determined in the course
286
BULLETIN OF THE UNIVERSITY OF WISCONSIN.
of this work to be as given in the following table for twenty
minutes heating:
Table IV.
Diameter.
Amperes.
1 inch.
400+.
11/4 inch.
650+.
11½ inch.
960.
2
inch.
1600.
The capacity of the first two sizes is from direct measure-
ment; that of the last two is calculated from the areas of their
sections, since the maximum current used, 700 amperes, never
heated them to redness. If the heating is prolonged to an
hour or more the current allowable is about fifteen per cent.
less than in Table IV.
Although 600 amperes has been used for a short time with
one inch electrodes, these become red their entire length, and
undergo oxidation. It was found that this oxidation could be
prevented by painting them with a paste of water glass and
carborundum dust.
Moissan ("Le Four Electrique") has given the size of carbon
electrodes suitable for furnaces of different horse-power. As
he ordinarily used fifty or sixty volts, the latter value has been
used in calculating the following table, except in one case where
he specified seventy-five volts. The results as calculated from
Moissan are given in Table V.
Diameter.
m. m Inches.
Table V.
Amperes.
120
16-18
1 1
16
27
1
40
50
112
2
450
1,250 (100 H. P.)
3,700 (300 H. P.)
In this table the current densities per unit area increase
greatly with increase of diameter, although from the fact that
the radiating surface increases less rapidly than the cross sec-
tion, the reverse would be expected. Either the capacity of
1
WATTS—INVESTIGATION OF THE BORIDES AND SILICIDES. 287
the smaller electrodes is greatly underrated, or that of the
larger ones much overrated. The only carbon electrode tried
by the writer was of one inch diameter, and this became red-
hot outside the furnace at 150 amperes, although Moissan
stated that he was able to use 450 amperes. It would be of
interest to the worker with the electric furnace to know
whether the relative conductivities of carbon and of graphite
for heat are in the same ratio as their electrical conductivities,
as is the case for the metals. If this is so, there will be no
advantage of either over the other in regard to the amount of
heat wasted by conduction from the furnace.
On account of their being such good conductors of heat, the
graphite electrodes used should be the smallest that will carry
the desired current. They should be slightly pointed before
use, and set exactly in line in the furnace. The source of any
irregularity in the operation of the arc is usually found in the
anode, which corrodes irregularly, while the cathode keeps in
good order without attention. One advantage of alternating
over direct current for the arc furnace is that both electrodes
tend to keep in good form.
V.
EXPERIMENTS.
After this detailed consideration of the subject of reducing
agents, the general action of slags, and the construction and
operation of the furnaces employed, attention may be directed
to the problems proposed, viz., the production of borides and
silicides from their oxygen compounds by a single reaction in
the electric furnace, and the production, if possible, of certain
silico-borides by the same means.
The experimental work, results of which are here given,
covered a period extending from December, 1903, to May,
1905, and included 160 electric furnace runs, with associated
288
BULLETIN OF THE UNIVERSITY OF WISCONSIN.
analytical work. Each run involved the preparation of the
materials, the weighing and mixing of the charge, the partial
or complete construction of the electric furnace before, and the
dismantling after, a run, and required from one to seven hours,
varying with the time needed to prepare the materials of the
charge, and to construct the furnace.
In addition to this, there was the disintegration of the prod
uct by acids, always requiring a day or more, and in some
cases involving the trial of a number of organic in addition to
the mineral acids. When a crystalline residue remained, this
was tested qualitatively, and if promising, was subjected to a
quantitative analysis.
Several runs of the furnace were for the purpose of prepar-
ing materials needed, such as borax glass, and boric anhydride;
others, although not described in detail, contributed to the re-
sults obtained; while there were many experiments which were
of use only in showing what arrangement of furnace or what
materials were to be avoided.
A most serious limitation upon the work was the difficulty
of carrying on the analytical processes. The resulting prod-
ucts required the working out and application of new methods
of analysis. Boron, especially, presented much difficulty in its
determination, and the available methods for its determination,
especially when combined with various metals, are at best only
partially satisfactory approximations. In several of these
analyses the writer is indebted to Mr. William Hoskins of
Chicago for valuable assistance.
PRELIMINARY EXPERIMENTS.
Beginning in December, 1903, a series of experiments were
undertaken in the preparation of silicon in the electric furnace.
Silica, glass, and calcium and sodium silicates were used as
the source of silicon, and the reducing agents were carbon,
aluminum, and carborundum.
WATTS—INVESTIGATION OF THE BORIDES AND SILICIDES. 289
The following difficulties were noted in the reduction by
carbon:
a. The melting point, 1430° C.,43 and the boiling point of
silicon are evidently near together, and as the reduction
temperature is high, much silicon was often vaporized.
b. The specific gravity of silicon (2.34) is about the same
as that of most of the silicate charges used (S. G.
quartz 2.5), hence the silicon remained diffused in the
slag in small spheres instead of settling to the bottom
in a single mass.
c. Because of its brittleness, the masses of silicon that were
occasionally obtained were broken in removing them
from the slag.
The best results were secured with a charge containing much
sodium silicate, thus lowering the density of the fluid bath and
permitting the silicon to sink in it. By this ineans, a bar of
silicon six inches long, 114 inches wide, and ¾ inch thick
was formed in the furnace, but was broken into three pieces
in removing it from the slag.
Reduction by aluminum was easily accomplished by adding
the metal in lumips, to any melted silicate of the alkali or
alkali-earth metals, containing cryolite or fluor spar as a flux
for the alumina formed. When the action was brief, the re-
sulting lumps of metal on treatment by hydrochloric acid, left
the silicon as a crystalline powder. By prolonged action, the
product appeared to be massive silicon, but still contained a
trace of aluminum.
Nineteen experiments were tried, mainly with some type of
resistance furnace. A method frequently used was to cover
the electrodes with fragments of the silicate, start the arc, and
as soon as a fluid bath was formed to draw the electrodes apart
and allow the melted silicate to serve as the resistor.
By these various methods, a few pounds of silicon were pro-
duced in globules and fragments. This afterward served as a
source of silicon in some of the experiments which follow.
43 F. J. Tone, in Electrochem. Ind., 3:183 (1905).
290
BULLETIN OF THE UNIVERSITY OF WISCONSIN.
SILICIDES.
The first effort in this field was an attempt to prepare a
new silicide, that of phosphorus. Phosphorus unites with
many metals to form compounds akin to the borides, carbides
and silicides. Since compounds of silicon with both boron and
carbon have been obtained, it seemed probable that phosphorus
would also unite with silicon. Further encouragement was
found in the existence of the following compounds: SiS2,
ŞiSe2, BP, and BP3.
The first experiment consisted of the reduction by carbon
of a mixture of phosphoric acid and silica. The charge was
made according to the molecular proportions (2HPO, + SiO2
+70) by dissolving glacial p':osphoric acid in water, and stir-
ring in finely pulverized silica and carbon. The mixture was
dried, ignited, and then heated in an open graphite crucible
in the electric furnace for four minutes at 500 amperes and
50 volts. A greenish-yellow flame was seen at first, but it
quickly died out. The product looked like the original mix-
ture, and gave no indication either of phosphorus or of a phos-
phide when treated with acids.
With the idea that fused copper or copper silicide would
make a good solvent for the interaction of silicon and phos-
phorus, the following charge was prepared:
Glacial phosphoric acid……..
Carborundum.
Silica..
Copper.
Aluminum
•
40 grams,
20 grams.
5 grams.
30 grams.
20 grams.
This mixture was heated in a covered graphite crucible for
13 minutes at 200 amperes and 60 volts. When the metallic
product was treated with hydrochloric acid, a gas was evolved
which burned with a greenish-yellow flame when kindled. The
product appeared to be metallic phosphides only.
Slight variations of this experiment were tried, but with no
better results.
1
WATTS-INVESTIGATION OF THE BORIDES AND SILICIDES. 291
Direct union of the two elements was next tried, with tin
as a solvent. Red phosphorus was dissolved in melted tin, and
a phosphor-tin produced, containing 7 per cent. phosphorus.
This was heated with pulverized silicon by the blast lamp, but
without effect upon the silicon. Then the electric furnace was
used. All silicon had vanished, and the crystalline metal had
a strong odor of phosphorus. Hydrochloric acid acted upon
it readily, and the gas evolved, burned, when kindled, with a
yellow flame and the formation of a white smoke. If any
silicide of phosphorus had formed, it, was destroyed in dis-
solving the tin.
As the results of these experiments were not encouraging,
the attempt to prepare a silicide of phosphorus in the electric
furnace was abandoned.
Production of Silicides from Sulphide Ores.
The possibility of forming silicides by the simultaneous re-
duction of silica and a metallic sulphide was next investigated.
The experiments were confined to such material as was imme
diately available and were directed to the formation of silicon-
metals, not to the isolation of silicides of definite formulae.
Silicon-Copper from Chalcocite (Cu₂S.)
A charge was prepared consisting of 63.2 grams chalcocite,
24 grams ground quartz, and 35 grams calcium carbide, the
proportions having been determined from the equation,
Cu₂S+SiO,+CaC, Cu,Si+CaS+2 CO.
The calcium carbide was estimated as 80 per cent. pure.
The source of heat was a resistor of granular graphite upon
which was placed the crucible containing the charge. This
crucible was cut from a block of Acheson graphite, as it had
been found by previous experiment that the ordinary "graphite
crucibles" used for metallurgical operations, were destroyed in
an electric furnace of this or the arc type. The heat record
was:
292
BULLETIN OF THE UNIVERSITY OF WISCONSIN.
Time.
Amperes.
Volts.
Resistence.
Per cent. of
resistance at
outset.
5:55
160
100
0.62 ohms
100
200
98
.49 ohms
79+
5:59
340
92
.23 ohms
37+
6:00
500
83
.17 ohms
27+
6:01
550
80
.14 ohms
22.5
Although the charge had been intensely heated, no metal was
found. It was reheated in a similar way for 17 minutes with-
out effect. Lime as a flux and fresh materials were added,
giving it the composition,
Sand...
Chalcocite.
Lime....
•
Calcium carbide.
Anthracite coal
•
360 grams.
379 grams.
324 grams.
35 grams.
120 grams.
This was heated half an hour at 300 amperes and 70 volts in
a vertical furnace consisting of a box built of sheet graphite,
5 inches square and 6 inches deep. The charge served as one
electrode. The result was an ingot of brittle white metal
weighing 250 grams. By analysis, the composition was:
81.6 per cent.
Copper
7.7 per cent. Cu,Si....
24.75 per cent.
Copper.
•
9.62 per cent.
.61.35 per cent.
6.4 per cent. or Fe,Si...
Iron..
Silicon
Carbon ...
Undetermined
1.0 per cent.
33 per cent.
100.0 per cent.
As no sulphur was present, it is evident that the ratio of lime
to sulphur used in this case, 4.2:1, was sufficient for its re-
moval. The product contained 16 grams of silicon, or 10 per
cent. of the total silicon in the charge. The yield was 53 per
cent. of the combined weight of metal and silicon in the charge.
WATTS—INVESTIGATION OF THE BORIDES AND SILICIDES. 293
Ferro-silicon from 1errous sulphide.
FeS+2 SiO₂+CaO+4 C=FeSi2+CaS+4 CO.
In the first experiment, 560 grams of ferrous sulphide, 448
grams of lime, and an excess of glass were reduced by anthra-
cite coal in the vertical furnace. The time of heating was 40
minutes by a current of 200 amperes at 50 volts. The prod-
uct, a low grade of ferro-silicon, contained a small amount of
sulphur. The ratio of lime to sulphur in this charge was
2.2:1, yet this is far above the theoretical ratio, 1.75:1. An
explanation offered for the failure of the considerable excess
of lime to remove all sulphur is that the lime entered into
combination with the large excess of glass present, so that the
usual stage of equilibrium between these weights of ferrous
sulphide and lime was not attained.
In a second experiment, ferrous sulphide was reduced in the
horizontal arc furnace at a higher temperature. The charge
was in accord with the equation,
FeS+SIO+CaC₂ = FeSi+CaS+2 CO,
with the addition of sodium chloride and cryolite as a flux.
The heating was for 16 minutes at 450 amperes and 87 volts.
The yield was 80 per cent. of the weight of metal and silicon
only 53.3 per cent. of the combined weight of metal and silicon.
in the charge. Had the charge contained an amount of lime
equivalent to the carbide, the ratio of lime to sulphur would
have been 2.0:1; yet the product was free from sulphur. This
tends to confirm the view that the glass was responsible for the
presence of sulphur in the previous experiment.
Silicon-nickel from Millerite.
The charge of 107 grams NiS, 110 grams SiO2, and 54
grams of powdered aluminum was heated in the horizontal arc
furnace. Only a few globules of brittle, white metal, silicon-
nickel, resulted. A fire brick had melted and flooded the in-
terior of the furnace with slag. The experiment was not re-
peated.
294
BULLETIN OF THE UNIVERSITY OF WISCONSIN.
Silicon-molybdenum from molybdenite.
MOS, +2 SiO2+2 CaC, MoSi, +2 CaS+4 CO.
The charge consisted of 48 grams molybdenite, 86 grams
ground quartz, and 300 grams calcium carbide. The large
excess of carbide was used in the hope that much of it would
remain unchanged and yield a slag that could be decomposed
by water. This charge was heated in the vertical furnace for
25 minutes by a current of 400 amperes and 70 volts. The
result was an ingot of metallic lustre, white, brittle, and harder
than quartz. Its weight was 32 grams. There was in the
charge 28 grams of metal, 40 grams of silicon, and an excess
of carbide over that required theoretically for complete reduc-
tion, so that the yield was only 47 per cent. of the theoretical
amount. Water had but a very slight action on the slag, show-
ing that the carbide was practically all destroyed. As ex-
pected, the slag evolved much hydrogen sulphide when treated
with hydrochloric acid.
Other experiments were tried upon molybdenite using
aluminum powder as a reducing agent, with similar results in
regard to the nature of the product and the per cent. of yield.
Conclusions.
1. Metallic sulphides and silicates can be simultaneously re-
duced by calcium carbide or aluminum, with the pro-
duction of a silicon alloy.
2. The total yield based on both the metal and non-metal in
the charge is little above 50 per cent.
3. The use of carbon or a carbide as reducing agent introduces
carbon into the metallic silicides of those metals capable
of uniting with carbon.
4. The product can be obtained entirely free from sulphur by
the use of lime or aluminum.
•
}
WATTS-INVESTIGATION OF THE BORIDES AND SILICIDES. 295
A NEW SILICIDE OF MOLYBDENUM.
The only silicide of molybdenum heretofore known is that
discovered by Vigouroux in 1899, having the formula Mo₂Si3.
In one of the attempts to prepare a silico-boride of molybde-
num, the writer believes he has demonstrated the existence of
a higher silicide.
The charge consisted of 70 grams molybdic acid, 30 grams
silica, 50 grams boric anhydride, 106 grams copper, and 120
grams aluminum, one-half in the form of powder, with cryolite
as a flux and lime as a retarder to the reaction. This charge
was heated 16 minutes in the horizontal arc furnace by a cur-
rent of 350 amperes at 70 volts. The resulting ingot of metal
was pulverized in an iron mortar, and treated for several days
with hot nitric acid diluted by its own volume of water. After
washing, the residue was treated by dilute hydrofluoric acid,
and washed by water, alcohol, and ether. It was then dried
as much as possible by the filter-pump, and heated for several
hours in a drying oven. The residue consisted of aggrega-
tions of flat, dark crystals with a metallic lustre. Analysis
showed the composition to be:
Molybdenum
Silicon..
Iron
Boron
•
63 4 per cent
31.2 per cent.
1.1' per cent.
2 1 per cent.
99 8 per cent.
Hot hydrofluoric acid dissolved some silicon and all the iron
present. The residue from this treatment was unaffected by
boiling aqua regia, but was completely soluble in a mixture
of hydrofluoric and nitric acids. Table II shows that hydro-
fluoric acid is without action upon the boride of iron, therefore
all the iron must be present as silicide; hence the boron is
present as molybdenum boride. Iron boride was undoubtedly
present in the original ingot, but was destroyed by the treat-
ment with nitric acid. Since crystallization occurred in a
296
BULLETIN OF THE UNIVERSITY OF WISCONSIN.
large amount of copper, there can be no free silicon present.
After deducting the silicon required by the amount of iron
present as FeSi2, and the molybdenum required by the boron,
the residual molybdenum and silicon are far from the propor-
tions required by the formula Mo₂Si,, but correspond closely
with that required by the formula MoSi₂. There is a slight
excess of silicon even over the amount required by this
formula.
The action of several reagents upon this substance is as
follows:
Boiling nitric acid.
Boiling aqua regia.
Boiling hydrofluoric acid.
Fused sodium carbonate.
Fused sodium nitrate.
No action.
No action.
No action.
Action with incandescence.
Slow, but complete decomposition.
Its specific gravity is 6.31 at 20.5° C.
An earlier experiment in which a mixture of molybdenite
and silica was reduced by calcium carbide, gave a similar re-
sult in the ratio of silicon to molybdenum.
While no pure silicide of molybdenum has been isolated in
this experiment, the only way that the writer can explain the
content in silicon is by the presence of a silicide of molybde-
num richer in silicon than Mo,Sis, and probably having the
formula MoSi2.
BORIDES.
The preparation of silicides by the method proposed having
been successfully accomplished, it seemed reasonable to sup-
pose that the borides might be readily prepared by the same
treatment, and a series of experiments was conducted with that
end in view.
Molybdenum Boride.
3 Mo S₂+3B,O,+9 AI-2 Al₂S,+3 Al₂O+3 Mo B₂+Al.
2
2
2
A charge of 94 grams molybdenite, 41 grams boric anhy-
dride, and 47 grams aluminum powder was heated in the arc
"
WATTS—INVESTIGATION OF THE BORIDES AND SILICIDES. 297
furnace for ten minutes with a current of 450 amperes at 60
volts. The result was 39.5 grams of somewhat brittle, white
metal that scratched quartz. It was attacked by hot nitric
acid and was completely dissolved by hot aqua regia. A quali-
tative test indicated that it contained boron in some quantity.
Analysis proved its composition to be:
Molybdenum.
Iron
Silicon...
Boron (by difference)
85.3 per cent.
8.8 per cent.
1.2 per cent.
4.3 per cent.
The product, then, contained only 1.7 grams boron out of a
total of 12.8 grams in the charge, and consisted of molybde
num boride dissolved in a large excess of molybdenum.
Chromium Boride.
A mixture of 300 grams potassium dichromate, 200 grams
of borax, 800 grams of calcium carbide, 100 grams of alumi-
num, and 20 grams of fluor spar was heated 55 minutes with
a current that averaged 360 amperes at 75 volts. The result-
ing slag was not acted on by water. The metallic product,
weighing 66 grams, was of a dull white color, and cut quartz
readily. On standing in air several days, it cracked in many
places. Nitric acid had no visible effect upon it, but in hot
hydrochloric acid there was a slow evolution of a gas of very
disagreeable odor. This gas was probably hydrogen with traces
of acetylene and boron hydride. Chromium and boron were
found in the solution, which, after several days, was complete.
Cold hydrofluoric acid had no visible effect even when nitric
acid was added to it. Analysis indicated 7.0 per cent. iron,
0.9 per cent silicon, 1 per cent. carbon, 4 per cent. boron, 78.9
per cent. chromium, and aluminum which was not determined.
"
As it was thought that the bad odor of the gas evolved was
due to the presence of a trace of an aluminum boride, which
by treatment with acids yielded boron hydride, the experiment
was repeated, with the aluminum omitted and 30 grams of car-
298
BULLETIN OF THE UNIVERSITY OF WISCONSIN.
bon added in its place. The result was 130 grams of a very
brittle, crystalline metal that failed to scratch quartz. Nitric
acid, did not act on it, and hydrochloric acid acted only slowly
when hot. The gas evolved had no noticeable odor. Hydro-
fluoric acid at 60°, either alone or with the addition of nitric
acid, had but a very slight action upon it. Analysis gave iron
9.0 per cent., chromium 78.9 per cent., silicon 0.2 per cent.,
carbon 5.0 per cent., boron 1.8 per cent.
The first of these products was a ferro-chromium-boron alloy,
and the second, a ferro-chromium-carbon alloy. The iron in
each case came from the "carbide," in which 0.6 per cent.
would be sufficient to account for the iron found in the prod-
ucts.
Manganese Boride.
A Goldschmidt charge consisting of manganese dioxide,
boric anhydride, and powdered aluminum in theoretical pro-
portions was placed within a Hessian crucible, and this was
placed inside a strong steel cylinder,** packed with powdered
quartz. The cover was tightly bolted on, and the charge
ignited by the electric current. The result was an ingot of
metal weighing 25 grams, a yield of 72.2 per cent. of the total
manganese and boron in the charge. Analysis proved its com-
position to be:
Manganese
Silicon
Iron...
•
Boron (by difference)
· •
71.7 per cent.
1.0 per cent.
4.0 per cent.
23.3 per cent,
The iron came from the fuse wire by means of which the
charge was fired. The metal was finely crystalline, and strongly
magnetic.
The product was, then, a mixture of the two borides of man-
44This apparatus was devised by Prof. C. F. Burgess for firing Goldschmidt
charges under great pressures, and was placed by him at the disposal of thẹ
writer.
WATTS-INVESTIGATION OF THE BORIDES AND SILICIDES. 299
ganese, MnB and MnB2, with the boride of iron. The boride
of iron is probably magnetic, although no statement in regard
to this is made by Moissan. It is, however, remarkable that
so small an amount (4.5 per cent.) should render the whole
mass magnetic. It should be noted that manganese appears
to be the essential ingredient in the magnetic alloys formed
from non-magnetic metals. 45
Other experiments by this method, yielded similar results.
BORIDES AND SILICIDES BY ELECTROLYSIS.
Several experiments were tried to test the possibilities of
employing an electrolytic method of preparing these compounds.
The first electrolysis attempted was that of borax with the
addition of cryolite, carried out in the vertical furnace. The
container was a section of 4-inch iron steam-pipe, five inches
high, resting upon à magnesite brick in the center of which
was the iron cathode. The charge was melted by striking an
are from a graphite anode, which was changed for a bar of
wrought iron 114 inches in diameter as soon as the charge was
fluid. The electrolysis was conducted with a current of 300
amperes. Metallic globules rose to the surface and burned
brightly. The iron anode was rapidly consumed, and in 25
minutes, about 5 inches of it had melted away.
The resulting metal could be broken by severe pounding with
a hammer. The fracture was quite different from that of
either wrought iron or cast iron. Analysis gave 94.6 per cent.
iron, 1.5 per cent. silicon, 1.5 per cent. carbon, 2.5 per cent.
boron. Moissan has shown that iron containing 10 per cent.
of boron is more fusible than cast iron. In this experiment,
the iron evidently melted as soon as sufficient boron, carbon,
and silicon had united with it to lower its fusing point to the
temperature of the bath.
The experiment was repeated with a weighed iron anode
which in 22 minutes lost 533 grams for 73.7 ampere hours.
45 R. A. Hadfield, Chem. News, 90: 180 (1904). Dr. Hensler, Electrotech.
nische Zeitschr., March 2, 1905.
"
4
300
BULLETIN OF THE UNIVERSITY OF WISCONSIN.
This is seven times the amount of iron dissolved by the same
current by electrolysis in solutions. This current is the
equivalent of 10 grams of boron, so that the product probably
contained 1.8 per cent. boron. Similar results were obtained
by the electrolysis of glass with an iron anode.
Wishing to learn whether the position of the electrodes had
any influence upon the electrolysis or not, an experiment was
conducted with horizontal iron electrodes a foot apart. The
anode lost 121 grams, and the cathode 118.5 grams for 194
ampere hours, a loss per ampere hour of only 8 per cent. of
that when the anode was directly above the cathode.
In these experiments, then, the reduction of the borax or the
silica seems to be due to a secondary reaction of the metal
liberated at the cathode, and not to a direct liberation of the
non-metal at either the anode or the cathode. With an iron
anode, the resulting alloy contains between one and three per
cent. of boron. Metals whose borides and silicides are more
infusible than those of iron ought to yield richer alloys.
SILICOBORIDES.
After three years of experimental research with the electric
furnace, Moissan said:46 "At high temperatures we have a
simple chemistry, and we obtain only a single combination,
always of a simple formula." It is seen from Table II that
this statement is true of the majority of the borides and the
silicides even down to the present.
On the other hand, there is evidence to show that ternary
compounds can exist even at the temperatures of the electric
furnace. In "siloxicon" (Si₂C₂O), discovered by Acheson, and
the double carbide of tungsten and chromium (W₂C.3Cr¸C₂),47
we have what appear to be definite chemical compounds of
three elements. The production of any ternary compound in
the electric furnace renders it at least possible that there shall
ultimately be obtained several series of such substances, just as
there are now known several series of binary compounds.
46 Ann. de Chim., 9: 277 (1896).
47 Moissan, C. R., 137: 292 (1903).
i
WATTS—INVESTIGATION OF THE BORIDES AND SILICIDES. 301
The names silicoboride and borosilicide were applied as early
as 1893 to complex products of the electric furnace contain-
ing among other elements boron and silicon, but no definite
chemical compounds had been isolated. Several facts seemed
to the writer to point to silicon and boron as promising ele
ments with which to form a ternary metallic compound. They
each unite readily with many metals, to produce substances of
remarkable stability, and they also unite with each other.
The attempt was accordingly made to isolate a definite com-
pound containing a metal, boron, and silicon.
Iron, chromium, nickel, and cobalt were selected as the most.
promising and convenient metals with which to experiment.
Ferric oxide was prepared from electrolytic iron. With
100 grams of this was mixed 60 grams of pulverized borax
glass prepared from pure borax, 72 grams silica, 114 grams
powdered aluminum, 60 grams ingot aluminum, 70 grams cop-
per as crystallizing medium, and 100 grams of cryolite. This
mixture was heated 14 minutes in the electric furnace at an
average of 26.7 kilowatts. The result was 135 grams of hard
crystalline metal that could be broken only with much diffi-
culty. As nitric acid acted on it only very slowly, it was
heated with 115 grams more of copper, and 175 grams of
aluminum. The product from this was crushed in an iron
mortar, heated for six days in 50 per cent. nitric acid, repul-
verized, and the treatment with nitric acid continued two days
more. As a qualitative test indicated the presence of copper,
the substance was again ground in the iron mortar, and the
treatment with nitric acid continued. By the microscope it
was found that the resulting powder did not consist of definite
crystals, but of irregular metallic fragments. It was strongly
magnetic and contained boron. Analysis showed its composi-
tion to be:
•
•
·
Iron..
Silicon
Copper.
Chromium
Boron (by diff'ce).
72 3 per cent.
17.2 per cent.
1 0 per cent.
3.6 per cent.
5.9 per cent.
100.0 per cent.
or
Cu,Si....
Cr B...
Fe B...
He₂ Si.
rez
Fe Si...
Si...
1 22 per cent.
4.36 per cent.
35.3 per cent.
26 48 per cent.
31.82 per cent.
99.2
302
BULLETIN OF THE UN.ERSITY OF WISCONSIN.
A
The absence of any free silicon was assured by the large
amount of copper in which it was crystallized. The presence
of chromium was explained by the softening of part of a
chromite brick in the furnace.
A formula for an iron silicoboride can be computed from
this analysis as follows:
Per cent divided by atomic weight.
Iron..
Silicon
Boron...
•
72 3 per ct
17.2 per ct.
5.14 per ct.
76.5 per ct.
18.0 per ct.
5 5 per ct.
1.366
8.1
.643 x6 =
500
3.85 or Fe.Si B¸
(3.00
94.64 per ct. 100.5 per ct.
·
Another portion of the original ingot was treated for a week
with hot aqua regia, being reground during this treatment.
The residue was still magnetic, and contained boron as well
as a trace of copper. The microscope showed that the sub-
stance was homogeneous. It would seem to be either an iron
silico-boride, or a solidified. solution of Fe,Si. 3FeSi.3 FeB.
Its magnetic and chemical behavior support this view. FeSi
is magnetic and is readily attacked by aqua regia. FeB is
probably also magnetic, and is dissolved by either nitric acid
or aqua regia. FeSi is non-magnetic and insoluble in both
acids. From the magnetic properties of the substance, from
its appearance under the microscope, and the treatment by
acids to which it was subjected, it would seem impossible that
these three compounds should exist separately in the residue.
Eight other experiments were tried, mainly with electrolytic
iron as the source of that element. The results varied greatly.
In several cases the product was completely soluble in nitric
acid. In others, the residue obtained oxidized so readily
while drying that it could not be obtained pure, even by wash-
ing with much alcohol and ether. Further experiments, for
which time is at present lacking, are needed to establish be-
yond doubt the existence of a silicoboride of iron.
Nickel oxide, by a similar process, yielded a metallic residue
containing 21.0 per cent. silicon and 11.45 per cent. iron.
The source of the iron was found to be an impure nickel oxide.
WATTS-—INVESTIGATION OF THE BORIDES AND SILICIDES. 303
Five experiments with pure nickel oxide were unsatisfactory,
as the product was completely dissolved by nitric acid.
Tungstic acid, silica and boric anhydride were reduced in
the presence of copper, but the product was entirely dissolved
by nitric acid. This is a surprising result, as the silicide of
tungsten is completely insoluble, and the boride only slightly
attacked by this acid.
Similar experiments were tried with molybdic acid, but with
unsatisfactory results.
More encouraging results were obtained from potassium di-
chromate, but further investigation is needed upon this.
Although these experiments have failed to yield a series of
silicoborides, the writer does not yet regard such a series of
compounds as impossible. The failure may be due to the
method used, which is not suited to the production of com-
pounds of the metals with boron. It is the writer's intention
to take up this investigation again at the earliest opportunity,
using the method of synthesis from the elements.
VI.
ELECTRICAL AND OTHER DATA CONCERNING
THE ARC FURNACE.
In this series of experiments many observations were made
upon the length of are, its variation in resistance, electrode
consumption, etc. Some of these observations are included
here as of interest, and perhaps of value to other experiment-
ers.
The longest arcs obtained were 15.5 and 14.4 cm. (6 1/2 and
5 11/16 inches) at 104 volts, the former at 800 amperes. At
the usual voltage of 80 on the furnace terminals, the length of
are for 600 amperes varied from 6 to 7.5 cm. (23% to 3 inches)
as a rule.
304
BULLETIN OF THE UNIVERSITY OF WISCONSIN.
Resistance of the arc depends upon:
a. Length.
b. Current strength. (See Table VI.)
Increase of current increases the cross section of the
arc.
c. Temperature of the furnace.
1
During the rise in temperature at the beginning of
an experiment the arc must be gradually lengthened
if the current is to be maintained constant. When a
cold charge is fed into a hot furnace the current is
diminished.
TABLE VI.-Effect of varying currents upon the resistance of the arc
No. of Experi-
ment.
Amperes.
Volts.
Resistance.
R₁:R
1
98
600
75
.118
1
360
75
.202
1.71
210
179
.369
3.12
190
82
.425
3.60
150
81
.533
4.51
140
83
.536
4.96
120
79
.651
5.51
120
81
.668
5.66
100
81
.803
6.80
70
80
1.136
9.62
55
85
1.538
13.03
40
92
2.293
19.43
111
580
78
.127
1
400
72
.173
1 36
330
77
.226
1.78
280
77
.266
2.09
200
81
.398
3.14
160
81
.499
3.93
200
76
.373
2.94
180
66
.359
2.83
150
77
568
3.18
120
75
.618
487
130
75
.569
4.48
120
70
.576
4.53
120
70
.576
4.53
110
71
.638
5.02
75
83
1.099
8.66
70
80
1.136
8.95
80
89
1.105
10.97
}
WATTS—INVESTIGATION OF THE BORIDES AND SILICIDES. 305
d. Quantity and nature of vapors within the furnace.
An increase in the amount of vapors of sodium and
of silicon lowers the resistance. Possibly there are
vapors which are capable of increasing the resist
(See Table VII.)
ance.
The data of Table VI were taken at the close of experiments
when the furnace was very hot. Maintaining the length of
arc unchanged, the current was cut down by increasing the
external resistance. The pairs of bracketed readings were made
about ten seconds apart with the same resistance in the circuit.
An entire set of readings occupied about three minutes, so that
the temperature within the furnace was practically constant.
The column R₁: R, shows the increase in resistance of the arc
as the current is diminished and the external resistance is in-
creased. Curves from Table VI are shown in Figure 2.
From the above and other similar data, the writer draws the
conclusion that the arc tends to maintain a constant ratio be-
tween its own resistance and that of the remainder of the cir.
cuit, or an arc of fixed length tends to maintain a constant
voltage.
In experiment 111, the external resistance was varied about
900 per cent., the maximum varation in voltage of the arc was
14.1 per cent., and the average, only 1.8 per cent. The rise
in voltage seen in the last readings, occurred but a few seconds
before the arc died out. One readily pictures the shrinking of
the arc to a mere pencil of carbon vapor as the current is
diminished.
306
BULLETIN OF THE UNIVERSITY OF WISCONSIN.
+
600
60
500
500
400
300
NAMPERES
200
100
20
40 VOLTS 6.0
80
20
400
300
40
60
80
FIGURE 2.—Curves of Resistance, from Table VI.
200
O
100
WATTS-INVESTIGATION OF THE BORIDES AND SILICIDES. 307
TABLE VII.—The effect of vapors upon the resistance of an arc of
constant length.
No. of
experi-
ment.
Am-
peres.
Resist-
Volts.
Change
in resist-
ance.
ance.
Per cent.
20
350
400
8838
75
.214
68
.170
-20
Silicon vapor suddenly
formed.
118
160
300
230
126
250
300
230
132
640
600
500
560
570
LEON 878 7 28
.512
65
.216
-59
75
.239
-53
82
.328
Yellow flame suddenly
appeared.
Little flame.
75
.250
-24
Much flame.
80
.348
+9
No flame.
72
.113
Much flame from furnace
74
.123
+9
Flame continued.
.162
+42
Flame instantly ceased.
77
.138
+22
77
.135
+19
Flame started again.
Flame continued.
The above records were made after the vigor of the chemical
reaction was over, for at the height of the reaction, the fluctu-
ations of the instruments were usually too rapid to be read.
Comment upon these figures is scarcely necessary. It will
be readily seen that the effect of vapors may in some cases de-
mand reckoning upon.
FIGURE 3.-Curious Electrical Phenomenon.
308
BULLETIN OF THE UNIVERSITY OF WISCONSIN.
Exp. No.
Diam. of Electrodes.
Average K. W.
Time in Min.
K. W. x Min.
TABLE VIII.-General Data Concerning the Arc Furnace.
Volts.
Amp.
Inches.
C. M.
Length of
Arc.
Resistance,
129 112 28.0 36
1008
104
800
61%
15.5
.123
.020
.0079
61 114 30.5
12
366
80
600
334
9.5
.126
.034
.0132
121 112 28.6
30
859
81
550
31
Hecko
9.6 .141 .037
.0147
123 112 28.6 61
1521
78
570
3
7.7 .130
.043
0169
88 114 24.2 28
696
78
560
234
7.0 .133
.048
.0190
150 2 21.8 45
980
68
650
2
5.05.098
.049
.0194
86 114 33.3
15
500
75
600
21/4
5.7 118
.055
.0207
•
89 114 18 1
38
687
70
400
28
7.3 .168
.058
.0230
99 114 28.7
14
402 76
600
2
5.05.120
.060
.0238
09 114 26.0 25
670
80
550
218
5.4 .138
.065
0255
04 114 31.1 8
249 79
550
134
4.4
.118 .068
.0268
94 114 24.7
10
247
82
500
225.
3
5.6
.157 .072
.0280
100 114 26.1.
14
414
78
600 111
4.3
123 .073
.0286
16
•
107 114 29.9 23
127* 112 28.1
482 80
300
45 1264 103
150
25
\
278
7.3
.259 .090 .0350
5110 14.4
.679 .119 .0470
* The current was cut down from 400 amperes to 150 by drawing out the
arc instead of increasing the external resistance as usual. This is responsible
for the abnormal resistance found.
The effect of an increase of current in lowering the resist-
ance of the arc has already been pointed out; the other factor
appears to be the temperature, estimating this from both the
average, and the total energy delivered to the furnace.
As has been already stated, the source of trouble when the
arc is unsteady is to be found at the anode, and seems to the
writer to be due to insufficient heat to maintain there a layer
of carbon vapor.
A very interesting phenomenon was twice.
observed with an alternating arc; namely, that the electrodes
interlocked, as shown in full size drawing, Figure 3.
Total.
Per inch.
Per cm.
}
WATTS—INVESTIGATION OF THE BORIDES AND SILICIDES. 309
There is a deposition upon as well as a vaporization of each
electrode. The points of each are coated with smooth, lus-
trous graphite of extremely fine grain. There are two sepa-
rate arcs. Each half of the current wave has its own exclusive
territory on the electrodes. Each arc dies out with reversal
of current; yet sufficient carbon vapor remains in its crater so
that when this electrode is once more anode, the path of least
resistance is from the crater rather than from any other part
of the electrode. This phenomenon occurred only with a low
voltage and large electrodes, and would seem to be possible only
when the crater of the anode during each phase is small in com-
parison with the area of the end of the electrode. į
Observations were made upon the loss in weight of graphite
electrodes during several experiments. The results are given
in Table IX.
Exp. No.
Diameter of electrode
TABLE IX.-Loss in weight of electrodes.
Time.
Average K. W.
K. W. x time.
Anode.
Total.
Ra-
tio.
Cathode.
An.Cath.
LOSS IN GRAMS.
Per
minute.
Per average Per K. W.
K. W.
minute.
Anode.
Cathode.
Anode.
Cathode.
Anode.
Cathode.
Ins.
A. C.....
A. C
150A
1
151B 1
D. C....... 151A 1
139 11/2
136 11/2
138 11/2
137 1½
140
12
38
- 20052
37
6.9
60
229
11.6 695
17
28.8
14.3 1.19 .46 .40
19.6 1.47 .48 .32
9 13.4 120 11.3 2.02 5.13 1.24 .23
58
Average.
23.7 632 16.3 20
25.1 1,457 59.3 32
39 32.2 1,266 54.2 17
55 28.4 1,561 82.2 43
23 7 523 49.0 15
52.3 25
42 53
2.40 2.35
2.48
.074
.062
1.69 .041
.028
.81
.16 .094
.019
.81 .71 .87
1.87 1.02 .55 2.33 1.23
3.19 1.39 .44 1.68 .52
1.93 1.51 .78 2.91 1.51
3.27 2.23 .68 2.07 .63
1.37 .66 1.92 .92
|
.59
72
026
.031
•
.040
.022
.043 .013
.053 .028
.095
.029
.051
.025
152
148
NN
2
229273
57
Average.
37.6 2.72 1.82
20.8 [1,186 103.8
31.3 845 166.2 70.2 2.57 | 6.20
135 53.9
.66 4.98 1.81 .105
2.96 5.31 2.24 .197
4.01 1.81 5.15 2.03 .151
.038
.083
.061
Furnace | 141 12
filled
with
illum-
150
2
155
inating
151
gas.
147
∞ PHON
NNNNNN
38
31.4 1,142 23.5
10.4 2.26 .62 .27 .76 .33
.021
.009
45
21.8
980
42.8
64
26.5 1,894
82
46
14.2 656
56.1
22❘ 19.5 430
40.8
Average.
32
55.4 32.5
15.9 2.69 .95
73 1.12 1.34
9.3 6.03 1.22
1.27 1.85
1.34
.35 2.00
1.14
.73 .043
.016
3.09 2.75 .043
.038
.20 3.95 .65 .085
1.45 2.04 1.60 .095 .074
.014
.78 2.77
1.43 .066 .035
310
BULLETIN OF THE UNIVERSITY OF WISCONSIN.
1
Direct current was used in all except the first two experi-
ments. These are marked A. C. and to them the terms anode
and cathode do not apply. In these the electrode losing the
more in weight extended into the furnace much farther than
the other.
Electrode losses seem to be due to three causes:
1. Volatilization from the crater, confined to the anode.
2. Oxidation.
3. Disintegration-particles fall from the electrodes.
(1) is independent of, (2) and (3) dependent upon, the
length of electrode within the furnace. The total loss is
diminished about 50 per cent. by filling the furnace with il-
luminating gas; this can diminish only (2) and (3). The
losses with 2-inch electrodes were double those with 12 inch
diameter. This points again to the use of the smallest possible
electrodes. It should be stated, however, that only a single
pair of two inch electrodes were tested.
VII.
CONCLUSION.
The results obtained by this series of experiments are the
following:
1. The proposed method of preparing the borides and the
silicides of the metals by the simultaneous reduction of the
oxygen compounds of both elements is not suited to the prepa-
ration of the borides, but can be used successfully for the
silicides.
2. The existence of a new silicide of molybdenum has been
indicated.
3. An investigation of the possibility of a new series of
electric furnace products, the silicoborides, has been made
Although the experiments indicate the existence of such a com-
WATTS-INVESTIGATION OF THE BORIDES AND SILICIDES. 311
pound of iron, they must still be regarded as merely prelim-
inary. The results with chromium and molybdenum were en-
couraging, but by no means conclusive.
4. Data concerning the arc furnace heretofore unpublished,
have been obtained.
In addition to these experimental results, it is believed that
Tables II and III, which have been found most convenient for
reference during these researches, may prove a useful contri-
bution to the literature of the borides and the silicides, and
that the method of firing "Goldschmidt charges" in the electric
furnace, discovered during these experiments, will prove very
valuable in special cases.
312
BULLETIN OF THE UNIVERSITY OF WISCONSIN.
APPENDIX.
49
48
In the interval between the completion of this thesis and its
publication, two new borides, ThB and ThB, and a new
silicide of copper, Cu,Si, have been prepared, and the data
concerning them have been incorporated in Table II. Two
articles have also been published which touch upon the princi-
pal problem of this thesis, namely the preparation of metallic
borides and silicides from their oxygen compounds, thus avoid
ing the difficulties attending the isolation of boron and silicon.
The method used in both cases was that of reduction by pow-
dered aluminum, but without the use of the electric furnace.
In the first case 50 a mixture of the metallic oxide, the non-
metal, and aluminum was ignited by a magnesium cartridge.
By this method two phosphides Fe₂P and Mn.P₂ were pre-
pared. In attempting the more difficult reductions of silica
and boric anhydride, the oxide of copper or of tin with an
equivalent amount of aluminum was added to the charge to
supply the heat required to produce a good fusion. By this
means an ingot of silicon-copper containing 10 per cent. silicon
was obtained, and also the silicide of iron, FeSi. From the
oxides of iron and boron an ingot was obtained which con·
tained boron, concerning which the author remarks, "It is diffì-
cult to isolate a definite boride." All these products contained
traces at least of aluminum. The idea of adding an extra
quantity of aluminum and an equivalent amount of some harm-
less oxidizing material, here published for the first time, so far
as the writer has observed, is a valuable advance in the applica-
48 B. du Jassonneix, C. R., 141: 191-193 (1905).
49 E. Vigouroux, C. R., 142: 87-88 (1906).
50 A. Colani, C. R., 141:33-35 (1905).
WATTS—INVESTIGATION OF THE BORIDES AND SILICIDES. 313
tion of aluminothermy to chemistry. The writer has inde-
pendently applied the same idea to the reduction of tungsten
oxide, using the alkaline-earth peroxides as the source of
oxygen.
C. Matignon and R. Trannoy 51 confirm the experiments of
Colani in regard to the production of binary compounds by
the alumino-thermic method, and, in addition to a number of
phosphides and arsenides, produced the silicides of manganese,
chromium, copper, cobalt, nickel, and iron by inducing re-
action in a strongly heated mixture of metallic oxide, silica,
and aluminum. Borides of manganese and iron were simi-
larly produced. In this case the high temperature necessary
to the separation of the metallic product from the alumina,
was secured by starting the reaction in a mass already heated
to a high temperature.
These two series of experiments confirm the writer's conclu-
sion that silicides may be readily prepared directly from silica
by reduction with aluminum, and they further show that this
can be done without the aid of the electric furnace. Since the
production of silicon on a commerical scale by F. J. Tone of
Niagara Falls this has become of minor importance. The pro-
duction of the borides of manganese and iron by this method
does not confute the writer's conclusion, that, as a general
method, the simultaneous reduction of a metallic oxide
and boron anhydride by aluminum is not suited to the produc-
tion of metallic borides. It can be used successfully in the
case of the easily reducible metallic oxides, but not for those
of titanium, tungsten, and many others. The problem of ob
taining the extra heat needed above that supplied by the re-
action in such difficult reductions is a very interesting one, and
is being approached from different points of view by many
investigators in various parts of the world.
51 C. R., 141: 190.
314
BULLETIN OF THE UNIVERSITY OF WISCONSIN.
BIBLIOGRAPHY.
1808. Sur des expériences de M. Chenevix et M. Descostils.
C. L. Berthollet. Ann. de Chimie, 67:88.
1821. Sur la combinaison du silicium avec le platine, et sur
sa présence dans l'acier. Boussingault. Ann. de
Chim., 16:5–16.
1857.-Ueber Darstellung und Eigenschaften des Mangans.
Brunner. Pogg. Ann., 101:264–271.
1858. Ueber das Silicumwasserstoffgas. Wöhler. Ann. 107:
112–119. Sur l'hydrogène silicie. Wöhler. Ann.
deChim. 54:218-225. Du bore. Wöhler and
Deville. Ann. de Chim. 52:63-91.
1859.-Ueber einige Boroverbindungen. C. A. Martius.
Ann., 109:79-82.
1863. Sur des combinaisons du silicium avec l'oxygène et l'
hydrogène. Wöhler. Ann. de Chim., 69:224-234
Ueber Verbindungen des Siliciums mit Sauerstoff
und Wasserstoff. Wöhler. Ann., 127:255-274.
1865.-Ueber die Darstellung des Siliciums auf elektroly-
tischem Wege und über eine Verbindungen des Cers
mit Siliciums F. Ullik. Chem. Centrallb. p. 104
Ueber das Verhalten des Siliciumcalciums und des
Siliciumsmagnesiums zu Stickstoff und über eine
neue Oxydationsstufe des Siliciums. A. Geuther.
J. für pract. Chem., 95:424-440.
2:1875
875 ----Sur un borure de manganèse crystallisé, et sur le rôle
de manganèse dans la metallurgie du fer. Troost
and Hautefeuille. C. R., 81:1263-1266.
WATTS—INVESTIGATION OF THE BORIDES AND SILICIDES. 315
1876.-Ueber das Bor. Hampe. Ann., 183:75-101. Sur la
silicuration du platine et de quelques autres métaux.
Boussingault. Ann. de Chim., 8:145-175. Note
sur le siliciure de platine. A. Guyard. Bull. Soc.
Chim. 25:510.
Etude calorimetrique sur les carbures, les sil:-
ciures et les borures de fer et de manganèse.
and Hautefeuille. Ann. de Chim., 9:56-70.
1883. Sur le bore. A. Joly. C. R., 97:456–458.
Troost
1890. De l'action du chlorure de titane sur les métaux. L.
Levy. C. R., 110:1368-1370.
1892. Contribution à l'histoire des composés carbosiliciques.
P. Schutzenberger. C. R., 114:1089–1093.
1894. Ueber Borkarbid. O. Mulhauser.
O. Mulhauser. Z. anorg. Chem.,
5:92-93. Préparation et propriétés du borure de
carbone. Moissan. C. R., 118:556-560.
1895.--- Préparation et propriétés du borure de fer. Moissan.
C. R., 120:173–177.
Action du silicium sur le fer, le chrome, et l'
argent. Moissan. C. R., 121:621–626.
Sur les siliciures de nickel et de cobalt. E.
Vigouroux. C. R., 121:686-688.
Sur le siliciure de manganèse. Vigouroux. C.
R., 121:771-773.
Sur le titane cristallisé et les combinaisons de
titane et de silicium. L. Levy. C. R., 121:1148–
1150.
Étude du bore amorphe.
Chim., 6:296–320.
Moissan. Ann. de
1896.-A method for obtaining crystalline silicon. G. de
Chalmot. Amer. Chem. J., 18:536–539.
Étude de quelque borures. Moissan. Ann. de
Chim., 9:273–286.
Étude de quelque composés du silicium. Moissan.
Ann. de Chim., 9:289–300.
5
316
BULLETIN OF THE UNIVERSITY OF WISCONSIN.
Étude des borures de nickel et de cobalt. Moissa.
C. R., 122:424-426.
Recherches sur le tungstène. Moissan. C. R.,
123:13-16.
Sur le siliciure de cuivre. Vigouroux. C. R.,
122:318-319.
Action du silicium sur les mètaux alcalins, le
zinc, l' aluminium, le plomb, l'etain, l'antimoine,
le bismuth, l' or, et le platine. Vigouroux. C. R.,
123:115-118.
Calcium silicide. G. de Chalmot. Amer. Chem.
J., 18:319–321.
1897.-Le silicium et les siliciures métallique. Vigouroux.
Ann. de Chim., 12:5-74 and 153–197.
Sur la préparation et les propriétés des borures de
calcium, de strontium, ei de baryum. Moissan and
P. Williams. C. R., 125:629–634, also Bull. Soc.
Chim., 17:1015–1020.
Silicides of chromium. G. de Chalmot. Amer.
Chem. J., 19:69-70.
Silicides of copper and iron. G. de Chalmot.
Amer. Chem. J., 19:118–123.
1898. On the preparation of the silicides of tungsten, mol-
ybdenum, chromium, and manganese. H. N. War-
ren. Chem. News, 78:319.
Sur un siliciure de tungstène.
R., 127:393–394.
Vigouroux. C.
Sur un nouveau siliciure de chrome. Ch. Zettel.
C. R., 126:833–835.
Sur un borocarbure de glucinium. P. Lebeau.
C. R., 126:1347-1349.
1899. Sur un noveau procédé de préparation du siliciure de
fer Si Fe. Lebeau. C. R., 128:933-936.
Sur le siliciure de molybdène. Vigouroux. C.
R., 129:1238-1239.
WATTS-INVESTIGATION OF THE BORIDES AND STLICIDES. 317
1900.-Préparation et propriétés de deux borures de silicium,
SiB. et SiB. Moissan and Stock.
139-143.
C. R., 131:
Propriétés de deux borures de silicum. Moissan.
and Stock. Ann. de Chim., 20:433–440.
New silicon compounds obtained by the use of the
electric furnace. C. S. Bradley. Chem. News,
82:149-150.
1901. Sur un nouveau silicure de cobalt. Lebeau. C. R.,
132:556-558.
1902. The production of hitherto unknown metallic borides.
J. Chem. Soc. Trans. (1902), pp. 14-17.
H. R. Moody and S. A. Tucker.
1
Recherches sur le siliciure de calcium CaSi₂.
Moissan and W. Dilthey. C. R., 134:503-507;
Ann. de Chim., 26:289-294; Bull. Soc. Chim.,
27:1199-1203.
Étude du siliciure de lithium. Moissan. C. R.,
134:1083–1087; Bull. Soc. Chim., 27:1203–1207.
Préparation et propriétés d'un siliciure de van-
adium. Moissan and Holt. C. R., 135:78–81.
Préparation et propriétés d'un nouveau siliciure
de vanadium. Moissan and Holt. C. R., 135:
493-497.
Étude des siliciures de vanadium. Moissan and
Holt. Ann. de Chim., 27:277-288.
Étude du siliciure de cerium. J. Sterba. C. R.,
135:170-172.
Sur les combinaisons du silicium avec le cobalt
et sur un nouveau siliciure de ce métal. Lebeau.
C. R., 135:475-477.
Sur les siliciures de cobalt. Lebeau. Ann. de
Chim., 27:271–277.
Sur les combinaisons du fer avec le silicium.
Lebeau Ann. de Chim., 26:5-31.
Ueber die Reduction der Zirkonerde. E. Wede-
kind. Ber., 35:3929-3932.
318
BULLETIN OF THE UNIVERSITY OF WISCONSIN.
1903.-Sur deux siliciures de manganése. Lebeau. C. R.,
136:89-92.
Sur les equilibres qui se produisent entre le
cuivre, le silicium et le manganése, et sur le sill-
ciure de manganése Si, Mn. Lebeau C. R., 136:
231-233.
Sur les siliciures de chrome. Lebeau and J.
Figueras. C. R., 136:1329-1331.
Préparation et propriêtés d'un siliciure de
ruthenium. Moissan and Manchot. C. R., 137:229-
232; Ann. de Chim., 2:285-288 (1904); Bull. Soc.
Chim., 31:559–562 (1904).
1904. Sur quelques combinaisons du cérium. J. Sterba
Ann. de Chim., 2:229–232.
Sur la réduction par le bore amorphe des oxydes
du manganèse et la préparation d'un nouveau bor-
ure de manganèse. B. du Jassonneix. C. R., 139:
1209-1210.
1905.-Préparations de composés binaires des métaux par
Aluminothermie. A. Colani. C. R.,
C. R., 141:33-35.
Sur la préparation des composés binaires des
métaux par
l' aluminothermie. C. Matignon and R
Trannoy. C. R., 141:190.
Sur la reduction par le bore amorphe de l' oxyde
de thorium et sur la préparation de deux borures de
thorium. B. du Jassonneix. C. R., 141:191–193.
Sur la réduction des oxydes et sur un nouveau
mode de préparation par l'aluminium du composé
binaire SiMm2. Vigouroux. C. R., 141:722-724.
Ueber die Reduction de Thorerde durch Bor und
durch Silicium. E. Wedekind and K. Fetzer.
Chem. Zeitung, 29:1031-1032.
1906. Sur le siliciure cuivreux. Vigouroux. C. R., 142:
87-88.
1
7
UNIVERSITY OF MICHIGAN
To renew the charge, book must be brought to the desk.
TWO WEEK BOOK
DO NOT RETURN BOOKS ON SUNDAY
DATE DUE
3 9015 07502 9986
Form 7079 3-50 30M S
AUTHOR
BOOK CARD
Watts, O. P.
Chemical Library
QD
157
W352
TITLE Investigation of the
Borides and Silicides.
SIGNATURE
Дно музие
9843
S.VARIANIA!
Drawn
DUE RE
في
Medical
Mudicht Library