318 INDUSTRIAL AND ENGINEERING CHEMISTRY
3
where k , is the rate of cracking into coke xc, and kl is the rate
of cracking into the sum of other products.
The writer does not know of any published experiments
carried out with the express view of studying the formation
of polymerization and allied products, such as coke. Such
experiments, for example, as long cracking runs with a good
charging stock, a t two or more different temperatures, would
be useful.
Literature Cited
(1) Hurd and Spence, J . A m . Chem. SOL.,51, 3561 (1929).
(2) Kiss, IND. END.CHEM.,22, 10 (1930).
(3) Larson and Black, J. Am. Chem. SOL, 47, 1015 (1925).
( 4 ) Leslie and Potthoff, IND.END. CHBM., 18, 776 (1926).
Extraction of Beryllium, Caesium, and Rubidium
from Beryl’
C. James, H. C. Fogg, and E. D. Coughlin
U N I V E R S I T Y OF h-EW H A M P S H I R E ,
A simple and economical method has been developed for
producing pure beryllium oxide from beryl using ordinary
chemicals and apparatus. Any small quantity of caesium
and rubidium which may be present in the mineral can
be obtained a t the same time. The method is applicable
either o n , a commercial or laboratory scale.
The finely ground mineral is fused with approximately
one-half its weight of lime in a furnace of the blast type.
. The resulting slag is ground, decomposed with sulfuric
acid, and heated t o dehydrate the silica. The mass is
stirred with hot water and filtered, and the concentra-
tion of t h e filtrate adjusted so t h a t potassium, caesium,
and rubidium alums will crystallize out when allowed t o
cool.
H E methods of separation and decomposition given in
T this article represent work carried out a t various times
covering a period of a number of years.z The proc-
esses, which are very simple, permit the worker to obtain
several interesting by-products such as caesium, rubidium,
and scandium, if these elements are present even to a very
small extent in the beryl employed.
Decomposition of Mineral
The employment of hydroxides and oxides of the alkali
and alkaline earth elements has been recommended from
time to time for this purpose.
SODIUM HYDRoxrDE-This substance is very commonly
used in the laboratory and for small-scale production. The
usual procedure has been to fuse the mixture of mineral and
hydroxide completely in an iron kettle. Unfortunately the
fused mass attacks the containing vessel to such an extent
that much iron is introduced, making the process of separa-
tion more difficult, and in some cases actually perforating
and destroying the kettle.
It was found that these objections could be overcome and
a decomposition of 96 to 98 per cent still be obtained if a
mixture of beryl, with 11/* times its weight of sodium hy-
droxide, was heated to only a low red heat with thorough
1 Received December 3, 1930.
f Canadian Patent 305,937; United States patent application pend-
ing (Serial 315,595).
1-01. 23. xo.
(5) Lurie and Gillespie, J. A m Chem. SOL.,43, 1146 (1927).
(6) Pease, I b i d . , SO, 1779 (1928).
(7) Poynting, Phil. Mag., [4]12, 32 (1881).
( 8 ) Sackur, “Thermochemistry and Thermodynamics,” p. 219 (1917).
(9) Waterman and Perquin, J. Inst. Petroleum Tech., 11, 36 (1925).
Correction
In my article entitled “Distillate Yields in Cracking,” IND.
ENG.CHEM.,22, 10 (1930), Formula 11 on page 12 should read
as follows:
STEPHEN
A. KISS
DURHAM, h‘. H
Ammonium sulfate is added to the hot mother liquor
which causes ammonium alum to crystallize upon cooling.
After oxidizing the iron in the solution from these crystals
by a suitable oxidizing agent such as potassium bromate
or nitric acid, the major portion of it is removed by the
careful addition of dilute ammonium hydroxide to the
boiling liquid. The last traces of iron are removed by
hydrogen sulfide under slight pressure. The beryllium
is then precipitated from the filtrate by ammonium hy-
droxide or preferably ammonium carbonate.
The principles involved in this method can be adapted to
the extraction of these elements from ,other silicate min-
erals with only a few changes, depending upon the min-
eral used.
stirring. It first softened, then became liquid and boiled
(owing to the presence of water in the commercial hydroxide),
and afterward hardened to a friable bluish solid. The
temperature was then maintained for a short time below that
a t which there is a tendency for fusion with the formation of a
glass.
I n order to obtain the best results it is necessary that the
mineral be ground very fine (200 mesh), thoroughly mixed
with the caustic, and the temperature kept even throughout
the mass.
Although this decomposition n’as efficient, it introduced
large quantities of sodium salts which interfered with the
crystallization of ammonium alum a t a later stage.
CALCIUM OurDE-There are two outstanding advantages
in using this material rather than sodium hydroxide for
opening up beryl: cost, and the removal of calcium sulfate,
which is insoluble, along with silica, aiding in the filtration.
The decomposition of finely divided beryl can be brought
about by intimately mixing it with calcium hydroxide and
heating to a temperature just below fusion. Complete
fusion, however, gives a better decomposition and is more
easily carried out commercially. The furnace for the fusion
was given careful consideration since it was realized that the
temperature of fusion would be necessarily high, the attack
on the container would be rapid, and the cost of fusing large
amounts of material must be kept as low as possible. A small
blast furnace, which could be charged with the fuel (coke,
March, 1931 I.YDCSTRIdL d-1-D EiYGINEERING CHEMISTRY
gas, oil, etc.) and beryl-lime mixture, appeared to be desir-
able This would furnish, a t a low cost, a high temperature
in intimate contact with the substance to be fused. Since
the product of fusion would flow from the base of the fur-
nace, the process would be continuous.
For a rapid decomposition it is essential that 1he beryl be
in a fairly fine state of division and thoroughly mixed with
the lime. It would, of course, be impossible to use a dry and
dusty mixture in a furnace of the above type. Accordingly,
the lime was first slaked, then mixed with the ground mineral
to the consistency of mortar. The ratio of mineral to lime
may be varied within quite wide limits. The best ratios
vary with the type of mineral being decomposed. On the
average, a ratio of 10 parts of mineral to 6 parts of quicklime
gave a very fusable slag which was rapidly attacked by the
sulfuric acid. This was spread out in a layer of about 2
inches in thickness and allowed to dry. The product was
found to be quite firm and easily withstood the burden of the
furnace.
Firebricks were used to construct the furnace of the follow-
ing inside dimensions: 25 cm. diameter a t base, 40 cm. di-
ameter a t middle, 30 cm. diameter a t top, and 110 cm. high.
The main furnace was set upon a base of the same material.
From the center of the base to one side a gradually sloping
channel (5 cm. wide) had been cut to allow the fused product
to flow onto a slightly hollowed steel plate 90 cm. K 60 cm. X
1.5 cm.
The blast was furnished by a centrifugal blower and in-
troduced by three tuyhres placed symmetrically around the
furnace about 10 cm. from the bottom. A brisk fire was
started with wood and coke and when the lower part of the
furnace had reached a white heat, the charge consisting of the
beryl-lime mixture, together with one-half to two-thirds
volume of coke, was added. I n a few minutes the slag began
to flow from the tapping hole in a very fluid state. The
ratio of coke to beryl-lime mixture should be adjusted to
keep the temperature high and yet use the minimum amount
of coke. With a furnace of this size approximatdy a ton of
mineral could be run through in 24 hours.
The short flame that emerged from the tappirig hole was
examined by means of a spectroscope. Lines of the following
elements, arranged in the order of decreasing intemities, were
visible: sodium, potassium, caesium, lithium, rubidium.
Decomposition of Slag
The melt from the furnace was crushed and ground. -The
powder was mixed with an equal volume of water ,md treated
with slightly more than the theoretical amount of sulfuric
acid. The reaction was vigorous, and the whole inass boiled
and gave off fumes of sulfuric acid. The mass was well stirred
to prevent the formation of hard lumps, and then added w t h
thorough stirring to a \-at of water nhich was raised to the
boiling point by mems of steam. After standing for 12
hours it was again heated, stirred. and put through a filter
press. The vaqh water was used for dissolvins the next
hatch.
Since this ground slag was found to be attacked by solu-
tions of aluminum sulfate, etc., any unattacked slag remain-
ing in the center of the small lumps was decomposed by the
solution during the stirring, with the result that a little
silica was dissolved. The solution, consisting mainly of
the sulfates of aluminum and beryllium together with small
amounts of iron and alkalies, was evaporated in lead pans
by means of lead steam coils. The fused sirupy mass was
kept for some time a t 3 25" C., then dissolved in boiling water,
filtered through a large lead Buchner funnel, and diluted
until no simple sulfates separated upon cooling. This con-
centration gave a solution with a specific gravity around 1.36.
It n-as allowed to stand until the crystallization of the alum
319
derived from the potassium, rubidium, and caesium con-
tained in the beryl had ceased. This alum was carefully
removed and saved for caesium and rubidium, the separation
of which is described later.
Separation of Aluminum
The separation of aluminum from beryllium was brought
about by the crystallization of the former as ammonium
aluminum sulfate (ammonia alum) in the following way:
The mother liquor from the potassium, rubidium, and caesium
alums was treated with a 25 per cent excess of ammonium
sulfate. It was found that the best results were obtained
when the ammonium sulfate, in hot concentrated solution,
was added to the boiling mother liquor, after which the whole
was set aside to deposit the alum. When the crystalliza-
tion was completed, the mother liquor was evaporated to
about half the volume of the original s o l u t i o n 4 e., before
the alum crystallized. At this point the solution was al-
lowed t o cool for 24 hours, then placed in a refrigerator a t
approximately 0" C. for 24 to 48 hours. This treatment was
found to remove all but a minute amount of aluminum,
which was removed during the separation of the major portion
of the iron as described below. Unless beryllium exception-
ally free from aluminum is desired, the cooling in the refrig-
erator may be omitted.
In order that no beryllium should be lost, the first alum
crystals were recrystallized from water, wash water from
the beryllium basic carbonate being employed. The crystals
were placed aside and the mother liquors used to recrystallize
the second crop of alum coming from the refrigerator. This
was then treated in a similar manner to the solution and
crystals obtained in the first alum crystallization.
Removal of Iron
The filtrate from the second alum crystallization was heated
in a stoneware vessel, by means of a leaden steam coil, to the
boiling point and thoroughly oxidized by means of a little
potassium bromate or other suitable oxidizing agent. The
liquid was next treated with ammonium hydroxide (about
1 to 1) with very thorough stirring and boiling. The pre-
cipitated hydroxides rapidly dissolved a t first. As the addi-
tion of ammonium hydroxide was continued, the liquid turned
deep red and then finally ferric hydroxide separated as a
very finely divided precipitate. The addition of ammonium
hydroxide was stopped as soon as the supernatant liquid gave
a white precipitate upon dilution with water, indicating
that the liquid was basic. At this point the pH of the cold
solution was 5.5 to 5.7, determined with a quinhydrorie
electrode. The pH of a solution from which the iron was
not quite all removed, as shown by a distinct red color, was 5.2.
With a little practice, it is possible to judge this point quite
accurately, as the precipitate becomes very slightly, though
distinctly, lighter in color.
I t is very important at this stage of the process to add
the ammon'um hydroxide carefully and see that the mass is
thoroughly stirred and well boiled. The precipitated hy-
droxides must be given time to react with the solution so that
the beryllium hydroxide redissolves. The final precipitate
consists mainly of ferric hydroxide together with aluminum
hydroxide, derived from the trace of aluminum which re-
mained from the alum crystallization, and a small amount of
beryllium hydroxide.
The hydroxides were next removed by filtration through
a very large Bt'ichner funnel. Any beryllium accompanying
the ferric hydroxide was largely saved by stirring the residue
with water, boiling, and treating with sulfuric acid until,
after long stirring, the filtered liquid showed the presence
of some iron by its reddish tint. This liquid, after filtration,
 320 INDUSTRIAL AND ENGINEERING CHEMISTRY
was returned to the vessel and carefully treated with am-
monium hydroxide as previously described.
The basic acetate method can be used for separating iron
from beryllium on a large scale. A minute amount of iron
usually remains in solution. The introduction of ammonium
acetate causes a little more bother in subsequent operations,
owing to frothing when the beryllium is thrown out as basic
carbonate. It also remains in the ammonium sulfate mother
liquors which are used over again for the separation of alumi-
num.
The small amount of iron and any members of the copper
group were removed from the basic filtrate from the iron and
aluminum hydroxides by treating with hydrogen sulfide under
slight pressure. After the precipitated sulfides had been
filtered off, the solution was found to give a perfectly clear
solution when treated with solid ammonium carbonate and
warmed. This clear solution did not become discolored on
treatment with hydrogen sulfide, indicating the absence of
iron. Should there be a darkening of the solution owing to
the presence of mere traces of iron, it would be necessary t o
saturate again with hydrogen sulfide. Any difficulty in
removing iron a t this stage indicates that the solution mas
not made sufficiently basic during the removal of the major
part of that element as the hydroxide.
Precipitation of Beryllium Basic Carbonate
The iron- and aluminum-free filtrate was heated to remove
any hydrogen sulfide and the beryllium precipitated by the
addition of ammonium carbonate and ammonium hydroxide
in slight excess. Care must be taken not to add too large an
excess of ammonium carbonate due to the solubility of basic
beryllium carbonate in an excess of that reagent. Sufficient
precipitant has been added when a small portion, after
filtering off the precipitate, shows no turbidity upon the
addition of ammonium carbonate. If the precipitate is not
granular, boiling for a short time will cause it to become so,
thus making it easy to filter and wash. This first wash water
was added to the main filtrate, while that obtained later was
used for recrystallizing the crude alum as described earlier,
By so using all the wash water no ammonium sulfate was
lost,
A%1N.4LYSISOF BERYLLIUM BASICCARBONATE-The product
resulting from the above treatment was pure white as was
also the oxide obtained from the ignition of it. A determina-
tion of the impurities likely to be present gave the following
results :
LOTA LOTB
97 47,
Si02 0 015 0 025
41203 + Fe203 0 05 0 04
Ca (spectroscopically) h-one None
lTolatilematter 60 25 59 80
h-on-volatile matter 39.75 40 20
Approximately 0.25 gram of the original sample contained
in the 2 cc. of solution gave barely a perceptible coloration
with potassium thiocyanate. Accordingly no attempt was
made to determine the iron separately.
The above analyses were made on material from two en-
tirely different runs, showing that a product of consistent
purity may be obtained by this method.
Extraction of Caesium and Rubidium
The material for this work is indicated in Table I a t 7 ,
potassium, rubidium, and caesium alums. These were sub-
mitted to a fractional crystallization, whereupon the caesium
with the minute amount of rubidium rapidly accumulated
in the least soluble portion. These were separated from each
other by further fractionation or/and precipitation of caesium
antimony chloride and crystallization of rubidium acid tar-
trate.
Vol. 23, No. 3
Table I-Schematic O u t l i n e of Separation
(1) Mineral fused with CaO
(2) Slag f Hz0 + HnSO4 extracted with
i. J.
Residue Si02 and Cas04 (4) Solution of sulfates of A1 and Be, etc..
evaporated to render any remaining
SiOr insoluble
4 J.
SiOn (6) Filtrate allowed to crystallize
~
i and R b alums
K , Cs, (8) Mother liquorJ + (NHa)zSO4
3
Alum A
+
(10) Mother liquor concentrated and cooled
V i
Alum B (12) Mother liquor + NH4OH
4
Fe(OH13 and Al(0H)a (14) Filtrate + H2S4
J. +
Filtrate + NH4OH
FeS, CuS, etc. (16) -b (NH4)2C03
+
Be basic carbonate (18) Filtrate evaporated
3
J. c + NazCOa
(KHdZSO4 (20) Mother liquor
+
Lid203
4
Filtrate discarded
Before the fractionation was started, it was decided that
the pink feldspar, which occurred plentifully in the mine from
which the beryl came, should be examined for caesium and
rubidium. d quantity of this was ground and fused as de-
scribed under the decomposition of beryl. I n this case the
slag did not fuse so readily and the furnace had to be run a t
a very high temperature. The ground slag was very easily
decomposed by dilute acid or even by solutions of aluminum
and iron salts. By using insufficient acid, all the alkalies
and only a little of the aluminum were dissolved. The solu-
tion gave a strong test for lithium, showing that the feldspar
contained this metal in addition to potassium and so-
dium. After the solution had cooled, the clear liquid was
siphoned off. and the crystals scraped out and drained. This
alum, which showed only potassium and sodium lines when
examined spectroscopically, was recrystallized many times.
The alums from 9 kg. of mineral v-ere finally reduced to about
200 grams. These crystals gave no distinct test for caesium.
Examination of Ferric Hydroxide for Scandium
Several in\-estigators in the past have pointed out that the
arc spectra of many varieties of beryl show scandium lines.
If the scandium mere present in beryl, it would be expected
to concentrate with the ferric and aluminum hydroxides (see
13 in Table I). This precipitate was dissolved in hydro-
chloric acid, the boiling solution neutralized by means of
ammonium hydroxide, and then thoroughly boiled with an
excess of sodium thiosulfate. The precipitate, after filtering
and washing with a little water, was heated with dilute hy-
drochloric acid and oxidized by addition of nitric acid. The
free sulfur was removed by filtration and the liquid treated
with an excess of sodium hydroxide to remove the aluminum
and beryllium.
The small quantity of insoluble hydroxides, after being
washed, was warmed with water and oxalic acid. This
treatment gave an insoluble oxalate that was colored blue
from the presence of copper derived from the plates of the
filter press. This oxalate was then heated with a slight
excess of sulfuric acid to fumes with careful addition of nitric
acid to oxidize any organic matter. The sulfates were dis-
solved in water, treated with hydrogen sulfide, and filtered.
The clear filtrate was then warmed with oxalic acid, and a
pure white oxalate was precipitated.