304 DUL Y-CORROSION OF STEEL J.S.C.I.

, 6g, October, 1950

Acknowledgments After filtration through a Berkfeld candle, sea water was

Acknowledgments are due to Prof. Sir Ian Heilbron; with evaporated until crystals separated and the evaporation was con-
whose cooperation the work was carried out a t The Imperial tinued on a water bath until a fairly dry mass of crystals was
College of Science and Technology; to Mr. J. E. Youell, obtained, the temperature at no time exceeding 80"c. T h e
who assembled the results and prepared the paper for publica- crystals were then dried for 36 hr. in a vacuum desiccator over
tion; and to the Directors of The Distillers Company Ltd. concentrated sulphuric acid at room temperature, then powdered
for permission to publish the results. in a mortar and dried in a vacuum for a further 6 hr. 20 g.
of the sample was weighed out iuto a glass tube and sealed
T h e Distillers Company Ltd. at once.
Research and Development Department The same technique as above33 .I was used to determine the
Great Burgh, Epsom percentage of moisture absorbed by the dried sea salt after it
Surrey had reached equilibrium with air of known relative humidity.
Ikccired Xorcmher 7, 19.19 Samples of the salt (about 2 g.) were weighed out on crucible
lids suitably wired up to be suspended from the beam of a
References Beeker Chainomatic balance. T h e samples were then suspended
1 Reboul, Aiai. Chim 1878,(s), 14, 462 over sulphuric acid of appropriate densities to maintain atmo-
2 Heincmnnn, U.S.P. 1,180,947 ; C h i . Abs. 1916, 10, 1693
Du Pont d e Ncmours, U.S.P.1,477,047
spheres of 10, 20, 30, 35,40, 50,60,70 and 80 per cent relative
I Klebanski and Wolkenstcin, Chew. Zeiitr. 1935, 11. 3298
humidity in Kilner jars, the jars being kept in a room maintained
Carbide and Carbon Chem. Corpn., U.S.P. 1,75s,o4g constantly at 24O c. (75' F.). At intervals the samples were
(I Shell Development Co., U.S.P. z,207,1g3; Clieiw. Abs. 1940,34. reweighed in the constant temperature room without removing
19:4
Mugdan, M.and Barton, D. H. R., B.P. 573,532
them from the jars. The experimental results are shown in
h Goodcrham, W. J., J. SOC.c H ~ h i IND.
. I935,54,297T Table I and graphically in Pigs. I and 2. T h e sample exposed
to air at 307" R.H. lost weight in reaching equilibrium, whereas
that at to:{, R.H.lost slightly more. These losscs probably
correspond to the amount of moisture taken up by the sample
during its initial transfer from the sealed tube at the time of
THE CORROSION OF STEEL BY SEA the first weighing, but samples kept a t 357" R.H.and above
SALT OF GIVEN MOIS'PURE CONTENT increased in weight rapidly during the first 24 hr.
When the samples of sea salt prepared in this way are brought
Dg S . J. DULY into contact with bright steel plate and kept a t the relative
T h e moisture content of the residue obtained by evaporating sea humidity appropriate to the condition of the salt, the tempera-
water is shown herc to be determined by the rclativc humidity of ture being kept constant, the steel is corroded away fast by the
the air In which it is subscquenrl kept, provided that the storage damp salt but remains almost (and perhaps entirely) unaffccted
temperature does not vary : sampLs of such sea salt corrode mild
steel in proportion to their moisture contcnts. It follows that the by the salt that is quite dry.
w e n t ofcorrosion of mild steel by sea salt is governed by the relative
humidity of the ambient air and that keeping the moisture content
of the salt-complcx 81 S",, or less subsrantially prevents corrosion
from this cause.
s
One of the urgent problems confronting those responsible $45- -
for tanker design and operation is the prevention of corrosion d
of the interiors of the tanks.' The corrosion is due in part
to the action of sea water used as ballast or for the purpose of
cleaning out the tanks, and in part to the nature of the cargo.
The present paper is concerned with that part of thc corrosion
due to the sea water. Air dried artificially by large silica-gel
driers has already been used to dry out empty tanks after they
have been cleaned with sea water, or emptied of sea-water
ballast.' The effectiveness of this procedure depends, among
other things, on the amount of moisture remaining in the dried-
out sea saIt when conditions of equilibrium are reached.
!
X 2+/
a
W
Whereas the moisture content of many natural products
kept at constant temperature has been determined in its relation
to the humidity of the air to which the product is exposed, sea
salt has not hitherto been the subject of experiment. T h e tech-
nique for this purpose is, however, well-established and has
been described by L. H. N. Cooper and H. Daynes' and used
by various investigators? This determination was thc first
step in the present investigation.
Sea salt for thc purpose was prepared by the Plymouth Lab-
oratory of the Marine Biological Association, using sea water
from the English Channel. Bearing in mind the complex nature
of the residue when sea water is evaporated,5 the following
procedure nas adopted to secure a product likely to react to
moisture in the same way as the residue which was. left when
sea water dried out on the surface. Thc method adopted by
the Association for the preparation of sea salt was as FIG. I. Absorptioii of iiioisriire by oacino,r-dried sea salt at a coiIsIaiiI
follows : rettiperarcire of 75" P.
J.S.C.I., @, October, 1950 DUL Y-CORROSION OF STEEL 305
the corresponding constant relative humidities in a constant
, temperature room. At the end of 13 days, the salt was trans-
*' ferred from the driest samples into standard 50 ml.-flasks
without washing. A glass rod was used to assist in transferring
the damper samples, and the cups washed out with distilled
water intoa IOO m1.-flask.
Immediately after transferring the contents of the cups to the
flasks, a solution of 15% by volume sulphuric acid was run in to
dissolve the corrosion products. The acid contained 0.05% di-
o-tolylthiourea to inhibit any attack on the steel itself. The
samples were then placed in an oven at 40" c. After 30 min. the
acid was run off into the appropriate flask, and the cup washed
with distilled water. The solutions thus obtained were analysed
for iron colorimetrically, with the results given in Table 11. The
results are also expressed graphically in Fig. 3. A comparison

Table I1
Wtiglir of irou corrodedJrum bright steel plate iii 13 days by sea salr con-
taitiiiig varyiiig perceiifager of vioistiire, wheir kept iii air at the relative
hiiniidities required to itiaiiiraiii eqiiilibriiiiii at 75" F.
Temperature 75" F.
R.H. of Moisture Iron removed Iron Iron
air, content of sea after 13 days removed, removed,
perccntagc salt (basis, dry contact with sea g./m' oz./ft.'
wt.), pcrccntage salt, mg./S*lr cm.'
80 231' 0 7'00 12'05 0'0394
70 38.4 1.55 4'39 0.0143
60 18.5 1'15 1'99 0.0063
50 8.1 0.92 1.58 0 . 0 0 5 2
45 5'9 0.56 0.96 0.0032
35 3 '45 0.48 0.82 0.0027
I 30 0'2 0'35 0.60 0'0020

MOISTURE
' d
ON DRY
0 2 3 h 5 ' 49 e'l
WEIGHT OF SEA
le'5 3d4
SALT, oer cent
2ilO 20 0 '0 0'17 0.47 0.0016

PIG.2. l'drceitragc iiicrease in rueiglir uf vaciiioli-dried sea salt exposed to air

of aarioiis relative Iiiotiidiiies wider static coiidirioiis at a coiistaiit rerupera-
trire of 75" P.

Table I
Change iit weigh of vacriiiiii-dried sra-salt wlteir exposed to air at variorrs
relative Aiieiidirier at cuiistuuf rctnpernture ticider static coitditioiir
Tempcraturc 24' C. (75" P.)
K.H. of Percentage increase or Appearance of salt
illr, dccrcnsc after after 27 days
Pcrccntogr I day 8 days 27 days
80 +40.70 +155'0 j 1 3 2 . 0 Clear liquid; one particle
of undissolved salt rc-
mained
70 i-19.81 + 37.4 -t 39'2 Mainly damp solid, but
with free liquid
+
60
50 +i 1 37. .7 58 ~ -i--I- 19'3
8.9 +
J9.3 Slushy
8 . 9 Damp and cakcd
1; $ ;:;g $ 4:: ;;:; ;;;fK-::y:,$ilere [ O h
glass rod
35 + 2.99 -1- 4.25 + 4'25 Dryglassbutrodwill adhere to a
30 - 0.08 - 0.1 - 0.1 dry andfree-running
20 - 0.50 - 0.69 - 0.8 Dry and frcc-running

The steel samples for the above purpose measured approsi-

matcly 2.8 cm. square by 0.3 cm. thick. They were polished
bright on all sides with fine emery, cleaned and lacquercd.
Thc lacquer was then reniovcd from one side with emery and
a glass ring I cm. deep and 2 . 72 cm. internal diameter cemented
to the steel with a plastic cement. The bright steel outside the
ring was also cemented, so that only that within the ring was
free to rust. The samples of salt from the previous experiment
were transferred to the cups thus formed so that they were FIG,3. IVekIir of iroii reinoved froui brkh steel plate bi 13 days at 75" F.
in contact with the steel. They were then suspended in air at rvkeri kept iii coiitact with sea salt coiiruiriitig krroruir anioiim of ittoistiire
306 THOMAS A N D IVILKINSON-CARBON DIOXIDE IN C O A L
with the curve in Fig. 2 shows that the amount of corrosion
depends upon the moisture content of the salt which in turn
depends on the relative humidity of the air under the conditions
of the experiment. The amount of corrosion taking place at
60O,:,R.H. is only one sixth that at 80% R.H.
The samples of steel were examined under a binocular
magnifier when the experiment was finished. T h e sample
which had been kept at 20'1: R.H. showed no sign of corrosion.
T h e small quantity of iron found in the acid solution was
probably dissolved as iron in spite of the inhibitor. T h e weight
of iron shown as removed at zo'z, R.H. in Fig. 3 and Table I1
would then be nil and the values of other percentages relative
humidity would require to be adjusted accordingly. T h e sample
exposed to air at 30"/~ R.H. showed no sign of corrosion either.
But that exposed to air at 3 5 7 , R.H. showed slight etching in
specks, visible also to the naked eye. That at 45y;, R.H.
showed similar etching but more general. The sample at 50°<,
R.H. showed deeper sporadic etching. That at 60%~ R:H.
had lost iron from the whole surface; so had the remaining
two samples at 70",;, and So":, R.H.
The curve (Fig. 3) corresponds to the case examined by
W. H. J. Vernon'; in which rapid corrosion of iron was shown
to set in at a temperature above the dew-point through thc
presence of liquid water owing to the hygroscopic nature of
the initiating solid and the resulting corrosion product.
The conclusion that may be drawn from the foregoing investi-
gations is that corrosion of the inner surface of a tank by sea
salts would be materially slowed down by maintaining an artifi-
cially dry atmosphere within it. Although the curve in Fig. 3
is smooth, so that strictly there is no point which could be
described as corresponding to a ' critical ' R.H. value, the rate of
corrosion under the conditions of the experiment increases
rapidly as the moisture content of the salt rises above St%,.
T h e practical aim would, therefore, appear to be to dry out the
surface until these conditions are rcachcd.
Cargocaire Limited
I 55 Fenchurch Street
London, E.C.3
Hrrci%cd.\kip 1 5 , 1950
References
' Lamb, J.,J. l i t s f . l'crrol. 1949,35, 339
Colvin. 0. D., SOC.,taw. Archirccrs i w r . Ewrs.,
- . Philadelphia Chaptcr
OCt. 1949
3 Dames. H.. 7. Riibb. Res. 1936,5, 131
Gnhc, R., J:"Soc. CHEM. IND. 194r, 5 & ~ 4
j Phillios. H. C.,C/teiit. Soc. Qriarr. Rev. 1947, 1, 9
Vern&; W. H.. J., 1'ram. FaradaJI. Sot. 1927,23, 162 ; 1931,27,2Ss all tests, or to avoid in-leakages of air if all the acid is added.
THE DETERMINATION OF CARRON
DIOXIDE IN COAL
Ug W. C. THOMAS and )I. C. \I'ILYINSON
In the present Br;tish Standard method for the determination of
carbon dioxide in coals, thc carbon dioxide evolved by rcaction of
the coal with dilute hydrochloric acid or phosphoric acid is absorbed
in barium hydroxide, and the excess barium hydroxide is titrated
with hydrochloric acid. In Fuel Research Survey Paper No. 44,
a method is described in which the carbon dioxide liberated IS
absorbed in soddime and determined gravimetically. In this
paper a third method based on the prcssomctric principle is des-
cribed. T h e amount of a r b o n dioxide is determined by measuring
the pressure developed in an apparatus of constant volume, the
pressure and volume relationship having previously been determined
using a known amount of sodium carbonate.
The present B.S. method for the determination of carbon space V , the volume of air present at the initial pressure of PI
dioxide in coals uses Sinnatt and Harrison's modification of will decrease to 2
the Schrotter apparatus; in this method the carbon dioxide
J.S.C.I., 6g, October, 1950
obtained by reaction of the coal with dilute hydrochloric acid
or phosphoric acid is drawn into a 24-1. evacuated flask where
barium hydroxide is added, and the excess barium hydroxide
is titrated with hydrochloric acid. This method specifies
heating at 50" C. for 30 rnin., and a t 100' C. for 10 min. ; the
reaction with barium hydroxide takes 30 min.
In Fuel Research Survey Paper No. 44 a gravimctric mechod
is described in which the carbon dioxide liberated is absorbed
in indicator-type soda-lime and determined gravimetrically.
Heating is for 15,min. to the boiling point, with gentle boiling
for 3-35 min.
Hughes' has described a method for determining carbon
dioxide in baking powder, in which the pressure of the carbon
dioxide, enclosed in an apparatus of constant volume, was
measured and the percentage of carbon dioxide read from a
calibration curve after 3 min. With the materials that were
investigated this method gave rapid estimation in the cold.
The solubility of carbon dioxide in the acid is minimized by
using low pressures in the apparatus. This method was applied
to dusts from mine roads by the Monmouthshire and South
\Vales Coal Owners' Association? and was found to take about
54 min. per sample.
Bee? applied Hughes' method to fluorspar, which was used
as a flux in open-hearth furnaces, as it was required to find out
the quantity of calcite present. The method is extensively
used in the Sheffield steel works for this purpose. Beet also
applied it to coals and found good agreement with values
obtained by the B.S. method for three coals. He reduced the
volume of the flask containing the coal by using a 50-ml.
suction flask (nominal volume to the side arm) ; he used glass
beads to standardize the volume of a series of flasks to use with
the same U-gauge ; he put indicating-type silica gel as a desic-
cant in a 4-mm. bore tube instead of the wider tube used by
other workers, and added a silica gel guard-tube before the
two-way tap connected to the water pump and the atmosphere.
He also increased the bore of the upper part of thc capillary
gauge from I to 4 mm., thus nearly doubling the reading.
He also advised heating of dolomite materials over a smalL
bunsen flame for 2-3 rnin., and cooling subsequently in a
beaker of water for 7 rnin., the whole determination taking
15-20 min.
The effect of a number of different factors has been studied
using the apparatus illustrated in Fig. I. This is the same in
principle as that used by Hughes and the other workers men-
tioned, but is a true constant-volume apparatus in that it
omits the funnel with tap for adding the acid. From a funnel
of this type it is difficult to obtain the same volume of acid in
I n the apparatus described here the acid is spilt from an internal
container as in the well-known method for determining CO,
in carbonates. Since it has been found that heat has to be
applied to break up the small amount of carbonates present in
coal, it is easier to manipulate the small reaction flask without
a tap funnel. A convenient reaction flask is a 50-1111. CO,
flask fitted with a rubber stopper which is always inserted to
the same level, marked by a scratch. Some of the tests were
carried out using a Ioo-ml. specific-gravity bottle with a wide
neck and a ground joint ending in a capillary ;with this appara-
tus the joint had to be lubricated freshly for each test. I n all
tests a !-in. diameter flat-ended sample tube reduced to about
I in. in length was used to hold the acid.
Tire effect of the voltirtre o/ the apparatus.-It was pointed out
by the research workers of the Monmouthshire and South
Wales Coal Owners' Association that, in an apparatus of gas
VP
when the pressure has increased to P..
P,