5150 Vol.

i S

[ C O N l R I U U T I O N FROM C I I E M I S I R Y DIV., S A V A L I<ESEARClI L A B . ]

The Reactions of Molten Sodium Hydroxide with Various Metals1

J. A. GRAND
BY D. D. WILLIAMS, AND R. R. MILLER
RECEIVED
MARCH15. 1956

Under the conditions of the system geometry and a t temperatures up to lOOO”, the primary reaction of sodium hydroxide
with the investigated metals appears to follow a pattern illustrated by the general equation (A) NaOH Metal NazO + +
Metal Oxide (or Sodium Metalite) +
Hz. The various metals differ only in rate of reaction and valence state of the com-
.-f

bined metal, each tending to form its most stable or common oxide. The secondary reactions digress, with only those metals
having easily reduced or unstable oxides reaching a steady-state condition. Hence, chromium, iron, titanium, beryllium,
tantalum, manganese and ferrous alloys exhibit only reaction (A). Nickel, copper, Monel, gold and probably silver and
cobalt exhibit one or more of the following secondary reactions leading to a “steady-state” condition. (B) Metal Oxide +
HZ-c Metal 4-HzO; (C) NazO I/zHz + NaOH + Na; (D) HzO Na -t NaOH + +
‘/zHz. In the presence of nickcl,
copper, Monel and probably cobalt, reactions (B) and (C) are reversible. In the presence of gold and probably silver, reac-
+
tion (C) is not reversible by reason of the formation of a n alloy between the noble metal and the produced sodium metal.
For the same reason and because of the stability of the alloy formed, reaction (D) is prevented. Therefore, only nickel,
copper, and their alloys and probably cobalt will produce definite equilibria, since the summation of equations B, C and D is
the exact reverse of the primary equation A. Only these metals are protected by hydrogen additions prior t o heating.
If reaction A proceeds t o completion, with the hydrogen constantly rcmoved by evacuation, and, if there is a n excess of re-
acting metal present, all of the metals investigated will react further with the sodium oxide fraction of the primary reaction
product (E) NaaO +
Metal +. Metal Oxide +
2Xa. Reaction E proceeds completely to the right only if provision is madc
to remove the metallic sodium from the reaction zone, either by distillation or alloy formation.

The application of liquid materials as high tem- vacuum manifold. Gas pressures were determined to the
perature heat transfer media is a field of growing nearest mm., and were calculated t o volume a t STP. Upon
completion of a run, solid non-volatile products were ob-
importance. Current research has extended the tained by cutting the reaction tube a t the melt line, and
area of interest beyond liquid metals2 to molten volatile products such as metallic sodium were t o be found
salts. Among the obvious requirements for such condensed on the cooler upper portion of the reaction tube.
materials are thermal stability and inertness with Thus, in a given run, the existence and, in some cases the
extent, of a reaction could be shown by pressure measure-
respect to the container material. The work now ment alone. In addition, the establishment of pseudo-
going on is a study of reactions of salts and con- equilibrium or a “steady-state” (since the system was not
tainer materials for possible use in the high tem- isothermal) was made apparent if consecutively reproduc-
perature field. ible constant pressures were obtained. These consecutivc
readings were obtained by successive co!lytion and evacua-
One of the lower melting aiid illore thcrmally tion of the gas. Reactious which gave steady-state” con-
stable materials is sodium hydroxide. While this ditions would be driven to completiou by such incrcinental
chemical has long been handled as a liquid near its gas removal as well as by constant evacuation.
ineltirig point, little is known of its behavior a t Most of the metals studied did not achieve a “steady-
state” condition but continued to react until the reactants
more elevated temperatures. The most significant were consumed or until a protective coating of reaction prod-
data in the literature were published by Le Blanc ucts isolated the reactants. These metals were studied,
and Bergma~iii,~ whose appraisal of their owii work therefore, on a basis of rcaction product valence state by
stipulated the need for further study. ?rSore re- allowing a relatively small known weight of the metal t o react
complctely with an excess of hydroxide. The gaseous yield
cently, an ingenious corrosion study by Lad and (hydrogen) was calculable to only otic valelice state for the
Simon4 has delineated serious phenomena of mass metal in question.
transport in AT systems employing sodium hydrox . Material.-The sodium hydroxide used in the tests was
idc in nickel. Their data show nickel to be resist- ACS reagent grade pellets assaying 97.0% S a O I I , 0.6%
S a z C 0 3 ,plus the usual impurities, the bulk of which was
ant, though not inert, to attack by liquid hydrox- water.
ide. The effect of the carbonate impurity was considered neg-
This paper presents chemical reaction data on ligible inasmuch as no reaction was observed when the metals
metal-sodium hydroxide systems which explain studied were fused in pure sodium carbonate. In addition
to this evidence, the carbonate was always recovered, seem-
corrosion and mass transfer in such systems. ingly unchanged, in the solidified melts. I n the following
data, all weights of original charges and reaction products
Experimental have been corrected for carbonate content.
The water was removed from all charges by evacuation a t
Apparatus .-Wit11 the work of Le Blanc and Bergmatin 400’ prior to higher tcmperature studies. In this conncc-
:is a guide, an apparatus was constructed, the simplicity of tion, it should be noted that water is liclcl tenaciously by
which aided in the interpretation of the data. The appara- molten hydroxide below JOO”, even uiider reasonably high
tus consisted of a vertical closed-end reaction tube made of vacuum.
thc metal to be studied and a glass calibrated volume and a Sodium Hydroxide-Nickel Reactions.-Various grades
mercury manometer. A stopcock was located between thc of commercially available nickel metal ranging in purity
reaction tube and the remainder of the system. A cold trap from “A” nickel t o electrolytic and vacuum cast pure nickel
was used t o check for condensable gases. Temperatures were used in this investigation. Little or no difference was
were recorded and controlled by a Brown Electronik po- found as regards the prime nickel-hydroxide reaction.
tentiometer equipped with a P t , Pt-Rh thermocouple. Some surface leaching of impurities was found, but this had
Loading was accomplished through the glass and metal ball little effect on the final reaction mechanism. For this rea-
joint between the reaction tube and the glass system while son, all nickel r m s are grouped together without regard tcr
purging through B stopcock which connected t o a high minor additives or “inipuritics.”
.~ ~

( 1 ) Sponsored by Materials l.alioratory, Wright Air Ilevelopment Qualitative runs resulted in the estnblisliinelit of the exist-
Center, Wright-Patterson Air Force Base, Ohio. ence of the temperature and pressure sensitive “steady-
(‘2) “Liquid hIctals €Imdbook,” N A V E X O S P-733, U. S. Govt. state” condition shown in Fig. 1. Figure 2 is a plot showing
l’rintiiig Off., J i i n ~ 1U.X
, rate of pressure d~VC~(JplIlC1ltfor three tcinperatures. These
(:i) n1. T X I%l.mc:tnd I.. l { ~ r ~ m i i l i /;cr l, , 4 2 , [rvj ,4725 ( l ! l O ~ , l ) pressures are indepcntlent of the size of the free spacc in thc
( 1 ) I<. A . T.nrl :in11 S. I. h i i i i o i ? , C i , v v system, equal values heiiig o1,taiiicd \villi thc ca1ibr:ttcd
Oct. 20, 1936 REACTION SODIUM
OF MOLTEN HYDROXIDE
WITH METALS 5151

x I ,

I I I
X - COMPOS/T€ DUTA
f f L tH,Q
100 0 Hz O N L Y - COf D TfffP ON SYSTEM -ON€ PUN UNf Y -

so 1
E
2$ 60
&I
L
& 40
1

20 0 20 40 60 80 100 120 140 160

Time (minutes).
0 Fig. 2.-Rate of pressure development over NaOH-Ni
700 800 900 1000 1100 system.
Temp., “ C .
2NaOH + Ni +
Fig. 1.-Steady-state pressure over NaOH-Ni system.
NazO + NiO (or NazNiOz) + H Z (1)
By repeated evacuation a t a given temperature above 725”, The differences between these and previous runs were
the stoichiometric removal of the hydrogen contained in the length of time hydrogen remained in contact with the
the original hydroxide charge was demonstrated. melt, and, in some cases, the extended time of heating after
Since conditions varied in the early runs, particularly
with respect to time, the residual melts did not exhibit con- complete hydrogen removal. I t follows that the appear-
sistent or interpretable analytical results. It was deter- ance of water, nickel crystals and sodium metal must be re-
mined, however, that neither the steady-state pressure nor lated to these differences. T o check these points, the ap-
incremental hydrogen yields were accurate measures of re- paratus was modified by adding an “in-line” drying tube
action extent. (B203)between the reaction tube and the calibrated volume
The most reproducible step through which the reaction and a vacuum line connection above the manometer. The
could be driven was the point of complete hydrogen removal. solid residues from reactions in this apparatus were sub-
This was accomplished by the continuous evacuation of a jected to magnetic analysis on a Gouy magnetic balance.
nickel tube containing hydroxide a t a temperature above This balance was used, not to determine paramagnetism,
725”. A tared drying tube was placed “in-line” to detect but as a qualitative tool for detecting traces of metallic
and measure the water content of the evolved gas. The nickel in the presence of non-ferromagnetic material.
runs were stopped by quenching the system just as the oil The runs were made in the following manner: Sodium
vapor pump in the vacuum line began to reoperate smoothly hydroxide was placed in the reaction tube and dehydrated
after having been flooded by gas evolution. The tube was under vacuum a t 400’. The system was then isolated from
then cut a t the melt line, weighed, leached in water and the the vacuum manifold and the temperature raised to a de-
empty tube dried and reweighed. Thus was obtained a sired level. All evolved gas passed over the tared “in-line”
sample, the dry weight of which was known. The hydro- drying tube. Thus, hydrogen was detected by pressure and
lyzed product was filtered and washed, the filtrate titrated water by the tared desiccant. Solid products were recov-
for alkalinity and the residue analyzed gravimetrically (di- ered, after quenching, by cutting the reaction tube and
methylglyoxime) for nickel. These results calculated as crushing the melt. This crushed sample was transferred
various possible compounds (oxides) of sodium and nickel to a tared glass tube for weighing both in and out of the
which could conceivably have resulted from an NaOH-Ni magnetic field of the Gouy balance. The results of these
reaction were compared to the dry weight of the sample. runs are shown in Table I.
The combination of a sodium compound and a nickel com- TABLE 1
pound that agreed with the original weight was thus shown
to be the probable molecular species of the melt. The M A C N EMEASLKEMENTS
~~C OF Sa0lI-h-i RsacrIoN
tared drying tube was weighed for a check on water content, PRODUCTS
and the upper portion of the reaction tube was checked for -Solid residue--
metallic sodium by hydrolysis in a gasometric system. Wt. of Type
When the reaction is driven rapidly to completion by con-
Time Kesi-
Xi
NiOjn
+ Mag-
netic
of
mag-
stant evacuation, no water, sodium metal or nickel crystals at Hz due rest- deflec- netism
appear in the reaction products. I n addition, the solid T e m p , temp., press., HzO, wt., due, tion, indi-
residues analyze consistently to be an equimolar mixture of OC. min. mm. g. g. g. g. cated
sodium and nickel monoxide. There was no way of deter- 600 5 0 0.0000 0.9928 0 . oouo -
0,0007 Dia
mining from the data how or if these oxides were combined. so0 10 4 .oooo 1,1355 ,0002 -
,0010 Dia
The product may be a sodium nickelite (NarNiOz) or simply 850 10 14 .OOOO 1.1905 .0007 - .0003 Dia
mixed oxides. The NazO fraction is removed readily by 900 80 7 .0007 1.0714 ,0029 f ,0125 Ferro
hydrolysis, and, as will be shown later, the NiO portion is 920 30 10 .0006 1.6313 ,0032 +
,0107 Ferro
quantitatively converted to Ni( 0H)z by the same process.
Samples of the solid residue were subjected to X-ray dif-
950 130 14 0008 1 3629 ,0018 +
,0051 Ferro
fraction, but the lines were too diffuse for identification.
900 240 a .OOOO 1.1261 .0000 -
,0009 Dia
925 20 6 .OOOO 0.5802 ,3174 - ,0003 Dia
It was the opinion of the operator that the material was
amorphous. Irrespective of these doubts as to actual prod- 20 mni. added before heating. * Continuous c”v;icuu-
uct, thc data show the priinc NaOII-Ni rcaction to be tioii.
 Table I1 shows the type of deflection characteristic of Nan0 +~ / z H---+
z NaOH Xa + (3)
dia-, para- and ferromagnetic materials. These data dem- Both reactions tend toward the right in the system used iri
onstrate that the very small nickel-containing fraction of these studies, equation 2 by reason of the distillation of
the melts in Table I had t o contain a dispersed ferromagnetic water and equation 3 by distillation of metallic sodium into
material (nickel) in order t o have produced positive deflec- the cooler parts of the system. Analytical results confirm
tions of the magnitude shown. the stoichiometry of equation 3. Evidently sodium hy-
T A B L E 11
dride did not form under the conditions of this experiment.
In the gold tube runs, the reaction proceeds readily t o corn-
TYPICALMAGNETIC DEFLECTIONS pletion since the sodium produced immediately alloys with
Total the gold tube. From the nickel runs, it follows that in a
deflection, T y p e of steady-state or equilibrated system in which equations 2 and
AI aterinl Weight, g. g. magnetism
3 are progressing, their products would react further
KO2 (SO%) 1.1695 -t0.0088 Para
+
Na --+ NaOH -t ~ / z Hfz
Nio 12.3820 ,1676 + Para
H?O (4)
The sum effect of reactions 2, 3 and 4 is the exact reverse
NaOIH 11.1621 - ,0079 D ia
of reaction 1, thus establishing the mechanism of equilibra-
Si 0.0005 Ferro tion in the non-isothermal, static system used for these ex-
Unweighable-pulled t o pole piece and stuck. pel iments .
Additional proof of equilibrium in the KaOH-Si system
Analyzing Table I, it is shown that no detectable reaction was demonstrated by isothermal capsule runs. The cap-
resulted a t 600”; runs a t 800 and 850” have produced hy- sules were made by inert arc weldirigs 3-in. lengths of “I,”
drogen and some solid residue but no water or metallic nickel tubing after filling half full of predried hydroxide.
nickel in ten minutes. Runs a t 900, 920 and 950” extend The free space in the capsule was filled with argon. These
the time element, and water and nickel appear simultane- capsules were heated in a muffle furnace four times and a t
ously. The long run a t 900” (240 min.) demonstrates that temperatures corresponding to coinpletc attack under
hydrogen added prior t o heating completely suppressed vacuum conditions. The results are shown i n Table 111.
reaction; and the last run shows again that continuous
evacuation has produced only hydrogen and a non-ferro- TABLE111
magnetic residue in accordance with equation 1. Thus, in
a system which allows the evolved hydrogen t o “equilibrate” CAPSULETESTSWITH NaOI1 IN SICKEL
with the melt, the following secondary reaction occurs. Alkalinitv Iieactir,ti
Wt. in melt- product
Xi0 +
Hz --+ Ni HzO+ (2) NaOH,
g.
Temp.,
O C .
Time,
min.
calcd. a s
S a O l i , g.
NiO in
melt, g .
in melt.
wt. %
In these experiments, reaction 2 is driven t o the right by 2.3033 900 120 2,2600 0.0150 0.65
the diffusion of water vapor into the free space of the system.
Earlier work a t this Laboratory5 established the fact that 4.1790 1000 120 4.1310 ,0349 .83
hydrogen and sodium monoxide react a t temperatures up t o 2.7510 800 90 2.7120 ,0027 .10
550’ producing sodium hydroxide and sodium hydride. A
repetition of this experiment a t 700 and 900’ was made in These data (Table 111) show that isothermal conditions
order t o demonstrate a source of metallic sodium in the with a small free space result in limited attack. The rate
NaOH-Ni reaction. The runs were made, in the apparatus is probably dependent upon the rate of diffusion of hydrogen
first described, by placing dry sodium monoxide in the re- through the nickel.
action tube, purging, filling the calibrated volume with The effect of atmosphere control on the KaOH-Ni system
various pressures of hydrogen and heating to temperature mas also investigated. The runs were made in the original
before opening the stopcock between the reaction tube and apparatus, and in crucibles equipped with close fitting lids
the apparatus. The data on the rate of reaction are shown containing inlet and outlet tubes. The data are in Table IV.
in Fig. 3.
TABLEIV
Y1 PDDED 70 475mm.,
- f :L ATMOSPHERE
CONTROL
P,ressure Wt. of
rise as a reaction
Orig. result of residue after
Atmos- press., Temp., Time, reaction, hydrolysis of
400 phere mm. O C . min. mm. melt, g .
Hz 760 950 120 0 0.0000
;i HP 480 950 120 0 . 0000
@- 300 Hz 20 920 120 0 . 0000
?I
Hz 13 910 105 0 . 0000
5 Hz 30 1000 60 7 ,0240
2 ?iA 760 800 13 ,1682
200 N* 23 900 30 1 ,0682
z A 760 900 20 .1351
.\ir 760 800 60 ,4261
100 a Gas flowing over melt in covered crucible.
Table IV demonstrates that initially added hydrogen, in
an amount in excess of the “steady-state” pressure required
L for a given temperature (Fig. 1). will inhibit NaOH-Ni re-
0 60 120 180 240 300 360 420 action. Inert gas has no protective effect.
Time (minutes). Further Reactions of the NaOH-Ni Reaction
Fig. 3.-Rate of reaction between NazO and Hz. Products.-The solid residue produced by equa-
tion l exhibits a t least two further reactions of
The data in Table I and Fig. 3 show that the solid reaction interest. Of primary coiicern to ai1 analyst
product from equation 1 will equilibrate with the hydrogen wrestling with these materials is the fact that the
produced by the same reaction. The NiO portion of the
product reacts readily as in equation 2 with the production NazO-NiO residue is stoichiometrically hydrolyz-
of metallic nickel and water vapor. The NazO portion of able, thus
the residue reacts t o produce NaOH and metallic sodium NarO + NiO -k 21120 ---+
-
( 5 ) Naval Reseairh Laboratory hleuio Keport #33, June 1952 2SaOli + Ni(OH)a (5)
Oct. 20, 1056 SODIUM
OF MOLTEN
REACTIONS HYDROXIDE
WITH METALS 5153

However, if the NazO-NiO residue be heated for Total hydrogen diffusion through the nickel
some time beyond the point of complete hydrogen tubes used, a t an arbitrarily selected average tem-
removal (equation 1) and in the presence of an ex- perature of 900’ and internal pressure of 12 mm.,
cess of nickel metal, a further reaction occurs amounted to less than 0.5 mm. per hour. While
which leaves a non-hydrolyzable residue of crystal- this amount could have a decided long range effect,
line NiO. The reaction involves the NazO por- the times involved in the experiments herein re-
tion of the residue and proceeds as ported make diffusion effects of minor import.
Na20 + Ni +NiO + 2Na (6) No evidence could be found of a reaction be-
tween NaOH and commercial NiO.
The reaction goes Completely to the right only if No evidence could be found of a reaction be-
time and temperature allow for the complete dis- tween N ~ and~ nickel
c at~ temperatures
~ up to
tillation of the metallic sodium produced. loooo.
Table V is a compilation of the results of tests Reaction 6 can be simulated by the use of sepa-
leading to the establishment of equation 6. The rately N ~ ~in Oa ~i tube. Reaction
analyses in Table v are based On the hydrogen times and temperatures are identical to those
yield from the upper portion of the reaction tube shown for the NazO-NiO mixture from equation 1.
when hydrolyzed in a gasometric system* Capsule tests involving NazO in Ni show the exist-
TABLE V ence of an equilibrium tending toward the left
REACTION OF N a 2 0 - x i 0 MIXTUREWITH EXCESSNICKEL of the equation, less than 10% reacting in two
in nacuo hours at 950 to 1000’.
Time a t Na Na NarO The role of certain reactive elements present in
Temp., terpp.8 available, found, reacted, commercial nickel was investigated slightly. Car-
“C. min. g. g. %
830 75 0.575 0.092 16 bon and iron, and probably all other impurities
910 60 .575 ,259 45 except copper and cobalt, are selectively leached
910 105 ,575 ,390 59
from the surface of nickel by sodium hydroxide.
910 180 ,575 .407 70 The effect on the reaction of NaOH with nickel is
910 290 .546 ,424 78 so slight as to be undetectable; however, it is prob-
955 110 .575 .488 85
able that corrosion data would find rate differences,
955 110 ,575 ,501
depending upon impurity concentration and depth
1010 120 ,575 ,560 87
97 of leaching.
1050 120 .590 .585 99
The Reaction of Copper and Monel with NaOH.
-The runs with these metals were performed ex-
Chemical analysis of residues from the last two actly as were the ones with nickel. The copper
runs in Table V confirm the percentage of reaction used was commercial tubing and the &lone1 was
shown. 66Ni-32Cu-2Fe.
Table V I presents data on the hydrolysis of With one exception, these metals exhibit the
residues from equation 1 and equation 6. I n ad- same reactions found for NaOH-Ni. The “steady-
dition, data are given for commercial crystalline state” pressures shown in Fig. 1 apply to both the
NiO. copper and the Monel systems. The one exception
TABLEVI
HYDROLYSIS
OF NaOH-Ni REACTION
RESIDUES
NiO residue wt.
Alkalinity, after treatment
Hydrolyzing Sample calcd. as Ether
Source of samples soh. wt.,g. NasO, 6 . dried llOa 500° Remarks
NaSO-NiO from eq. 1 HzO 0.1905 0.0864 0.1408 0.1293 0.1042 Converted to Ni( 0H)z
NiO from eq. 6 Hz0 .7621 None .7625 .7620 ,7620 No effect
NiO from eq. 6 10% NaOH .5325 .... .5332 .5320 .5321 No effect
NiO( Comm.) 10% NaOH .4765 .. . , ,4775 ,4765 ,4765 No effect
NiO(Comm.) in NaOH (fused) H20 .1417 . ..
, .1420 ,1415 ,1415 Noeffect

The solid hydrolysis product of equation 5 was
pyrolyzed in a semi-micro train equipped with
tared tubes containing P 2 0 6 , LiOH and P z O ~hot
, copper, the oxide formed is CuzO, since this is the
CuO, and hot Cu for determining HzO, Con, HZand
02,respectively. No COZ,HZor 0 2 was detected.
The weight gain of the water absorption tube
was nearly identical to the weight loss of the sample
a t 500’. The chemical analysis of the pyrolysis
residue established it to be NiO. These runs sub-
stantiate the data in Table VI and show the Prod-
uct t o be Ni(0H)z.
Certain tests were Performed to check variables
in the material and equipment used in the preced-
ing experiments* For the sake Of what little
brevity can be they be mentioned
without accompanying data.
is the valence of the copper compound formed by
initial NaOH-Cu reaction. In the case of pure
stable oxide in the presence of excess copper
metal.6 The prime reaction, hence, is
2NaOH + 2cu +
CuzO + NazO (or 2NaCuO) + Hp (7)
I n the case of Monel, the copper compound
formed is bivalent, evidently because the copper
remaining in the alloy is not readily available.
Monel reacts exactly as would be predicted for a
mixture of nickel and copper
(6) J. W. hlellor, “A Comprehensive Treatise on Inorganic and
Theoretical Chemistry,” Vol. X V , Longmans, Green and Co., New
York, N. Y.,1936.
5154 L). J. A. GRANDANL) li. li. A ~ I L I ~ R
U. WILLIAMS, Vul. 78

+
( 2/3Ni.1/~Cu) 2NaOH -+ TABLEVI1
+
( 2 / 3 N i 0 . 1 / ~ C ~ O ) NazO + H2 (8) NdOkl-liU REACTION
I”XJUC.IS
The iron portion of the alloy reacts according to
‘Time
--Reaction pruducts--.
Au in hy-
the equation developed later for pure iron. The at drolysis
total effect is a uniform reaction throughout the Temp., temp., Exptl. residues,a
OC. min. conditions 1120, g. 112 B.
alloy, with little or no sign of pitting. 975 2 Constant 0.0017 Present 0,0005
At the temperatures employed (700-900°), cvacuatioii
both Monel and copper exhibited deep non- 1040 10 Constant 0.0031 Present ,0010
protective external oxidation. The alloy oxidizes cvacuatioii
stoichiometrically and uniformly, with severe 750 10 Static Kor1e Nolle No!lc
spalling occurring upon cooling. vaeuuiu
The Reaction of Gold with Na0H.-In the work ‘310 240 20 ium. HZ Present , ... ,0250
of Le Blanc and B e r g m a n r ~gold
, ~ was presumed to added
be non-reactive with NaOH a t ca. 700’ in spite of 910 240 20 mrn. I& l’rcsciit OlS0
the recovery of macro quantities of gold flakes added
from their melts and certain unexplained gold 1050 2 1x1 air . , . . . . u110
plating effects on the metals tested in gold. The DO0 60 Scaled , . . Konc
present investigation found gold to be resistant but capsule
not inert to NaOH. S00-1U00 120 Static 0 ,U 134 I’rcaciit . 0:j!JO
The runs were made exactly as has been described vacuulll
for the nickel system. Gold was found to es-
a Combined and metallic gold.
tablish a “steady-state’’ pressure of hydrogen and
water vapor somewhat lower than nickel and cop- and the preponderance of gold in the compounds
per. Also, the proportion of water was consider- prevented exact identification. The meager data
ably higher than in the case of nickel. The data indicated, however, that the reaction compound
are shown in Fig. 4. Another important difference was AuzO and that the hydrolysis residue was a
was that the pressure developed is dependent upon hydrated oxide or hydroxide.
the size of the free space in the system, whereas in A study was made of the gold plating on nickel
the case of nickel and copper, no such dependence strips in hydroxide melts reported by Le Blanc
was found. Hence, the pressures shown in Fig. 4 and Bergmann3 I t was found that plating oc-
apply only to a 200-cc. free space. This effect is curred only on strips which remained in the melt
no doubt related to the predominance of water during cooling. The effect, therefore, must be

II0Z,
1
vapor in the gaseous products, and to the solubility
of reaction products in the inelt.
30 r I I I
:5 10
0 An additional reaction affects this system. .lily
700 800 900 1000
Temp., “C.
Fig 4 S t e a d y - s t a t e pressure over NaOH-Au system.
The results of representative NaOH-Au reac-
tions are shown in Table VII.
The data in Table VI1 show that water, hydrogen
and gold compounds occur nearly simultaneously
under all test conditions, except the run a t 750’
which was below the temperature of significant at-
tack (see Fig. 4) and the run in an isothermal
sealed capsule. It should be noted that two runs
produced reaction products even under an excess of
added hydrogen in contrast to the nickel and copper
reactions.
The solid residues resulting from hydrolysis of
the melts were subjected to analysis by pyrolysis
with subsequent chemical analysis of this residue.
In addition t o the hydrolysis residue, a consider-
able fraction of all the total reacted gold was found
as a soluble salt in the alkaline filtrate from the
hydrolyzed melt. This compound separated as
neutrality was approached during the titration of
the filtrate with acid. The small weight involved
related to the solubility and stability of the gold
rcaetion compound in molten NaOH. As the
melt cools, this compound, probably Au20, pre-
I 1 cipitates and thermally decomposes to metallic
gold and oxygen, thus producing the metallic gold
and plating effects reported by reference 3 and sub-
-~
stantiated by this study. The apparent stability
of this gold compound in solution a t these elevated
0
temperatures is somewhat striking.
1100 secondary reaction, such as NaaO with Hz, which re-
leases metallic sodium results in the formation of ai1
Au-Na alloy. The metals combine readily and
the alloy is very stable. The sodium is not com-
pletely removed even with repeated acid washing.
The following reactions can be postulated for
the gold--NaOH system
2NaOH + ~ A --+
u
NaZO + AufO (or 2NailuO) + 1-32 ( 9 )
AunO + H, --+2Au + HzO (or AuzO -+-
2Au + ‘/zOZand HE + +HzO) (10)
1/~02
Pia20 + l/zHz --+ NaOH f Na (11)
Na + Au -+- Na-Au Alloy (12)
The prime reaction, number (9)) is similar to the
nickel and copper reactions. However, reaction
10 is not reversible and the reverse of reaction 11 is
prevented by reaction 12. Thus, the “steady-
state” system is not as predictable as in the case
of nickel and copper, as, indeed, was found to be
the case, particularly with respect to hydrogen
inhibition.
Reaction of NaOH with Fe, Cr, Ti, Ta, Be, Mn,
Ag, Co and Ferrous Alloys.---Except for silver and
Oct. 20, 1956 n I E TITANIUM-HYDROGEN
SYSTEM 5155

cobalt, the above metals were found to react com- None of the runs resulting in the establishment
pletely with molten sodium hydroxide. The rate of the above equations gave metallic sodium or
of reaction was dependent upon the availability of water vapor as reaction products. They were,
unreacted metal. These metals, therefore, were however, short time experiments. Prolonged ex-
studied briefly to establish their valence state in the posure to hydrogen may result in the production
reaction products. Simultaneously, the absence of some water and sodium.
of a "steady-state" and hydrogen inhibition was All of the solid products on the right in the
established. above equations react with water, yielding hydrox-
The procedure used was a modified material ides or hydrated oxides. In the case of tantalum,
balance technique in which a small weighed quan- the hydrolysis product is completely soluble. In the
tity of the metal reacted completely with an ex- case of chromium, when leached in air, part of
cess of sodium hydroxide contained in a nickel tube. the residue forms a soluble chromate.
The hydrogen yielded by the reaction was then Ferrous alloys (304 SS, 310 SS, 36% Ni-6470 Fe
balanced against the original weight of metal alloy, and 64% Ni-17yO Cr-17% Fe alloy) react
(after a chemical check of the residue to ensure com- as might be predicted from all of the foregoing.
plete reaction) and thus interpreted into which The iron and chromium fractions react completely
oxide of the metal could have resulted from the as shown by their individual equations, while the
reaction. nickel portion appears to remain unattacked, prob-
The following equations suminarize the results. ably by reason of the protective effect of the pro-
'The temperatures shown are approximate points duced hydrogen. The over-all effect on the alloy,
of significant reaction however, is one of complete destruction. The
mechanism of attack is one of severe penetration
2Fe+ 6NaOH 600' Fez03 + 3 N a ~ 0f 3Hz (13) in contrast to the more or less uniform removal
650' from the surface of nickel, copper and gold.
2Cr + 6NaOH --+- CrzCh -k 3Na20 f 3Hz (14) Meager data on the reaction of NaOH with silver
325' indicates a similarity to gold. One exception is the
Ti + 4NaOH --+- Ti02 f 2Na20 f 2Hz (15) fact that silver tubes become quite porous to mol-
600' ten NaOH after a few fusions.
2Ta + lONaOH __f T a ~ 0 5f 5Naz0 f 5Hz (16) Cobalt was investigated by a single test which
650"
showed "steady-state" pressures of hydrogen and
Be + 2NaOH --+- Be0 + NazO + Hz (17) water vapor identical to those shown for nickel.
Its further reactions are assumed to be identical.
600"
Mu + 4NaOH MnOZ f %NazOf 2H2 (18) D. C.
WASHINGTON,

[CONTRIBUTION FROM THE ATOMICPOWER LABORATORY,~

KNOLLS GENERAL
ELECTRIC
COMPANY]

The Titanium-Hy drogen System

BY R. M. HAAGAND F. J. SHIPKO
RECEIVED
APRIL 23, 1956

The equilibrium pressure of hydrogen over titanium-hydrogen has been measured a t 300, 400 and 500'. Similar meas-
urements have been made of titanium-deuterium and titanium-tritium. At 500' no isotope effect was observed.

The titanium-hydrogen system is one of several
metal-hydrogen systems which have been studied
both from the practical standpoint because of the
high hydrogen occlusion obtained and from the
academic viewpoint because of the light which may
be shed on the properties of metal systems. The
high hydrogen concentrations (approximately 450
cc. (STP) Hz 18. titanium) obtainable in this system
make it ideal for many storage applications. Among
these are the following: 1. Separation of hydrogen
and its isotopes from other gases; 2. Transfer of
large quantities of gas in a small volume a t low
pressure; 3. Preparation of materials with high
hydrogen density.
In the titaniutn-hydrogen system, as in the simi-
( 1 ) T h e Knolls Atomic Power Laboratory is operated by t h e General
Electric Company for t h e Atomic Energy Commission. T h e work
reported here was carried out under Contract No. W-31-109 Eng. 52.
(2) Presented before t h e Physical and Inorganic Division of t h e
127th National Meeting of the American Chemical Society, Cincinnati.
Ohio, March, 1965.
lar systems zirconium-hydrogen and hafnium-
hydrogen, the occluded gas has apparently less
metallic character than in the possibly unique pal-
ladium-hydrogen system, but more than in the
uranium-hydrogen and thorium-hydrogen systems.
The dissociation pressures of the hydrides are the
most convenient functions and have been exten-
sively studied. Kirschfeld and ,%everts3 and re-
cently McQuillan4 have studied the Ti-H systeni
a t pressures from 1 to 7GO mm., and a t tempera-
tures below 1000". Gibb, et al.,5 have made
measurements above 500' and a t high as well as
low pressures. No results have been reported,
however, for the micron pressure region which in-
cludes low concentrations a t 500' and nearly the
entire range a t 400 and 300'.
(3) I,. Kirschfeld and A. Sieverts, 2. p h y s i k . C h e m . , A146, 227
(1929).
(4) A . D . McQuillan, Proc. R o y . Soc. (London), P04A, 300 (19.50).
( 5 ) T. R. P. Gibb, Jr., J. J. McSharry a n d R. W. Bragdon, THIS
JOURNAL, 73, 1751 (1951).