the exception of hydrofluoric acid gas, direct destruction of firebrick by gas does not occur t o any extent. Where gases are primarily responsible for trouble they usually act as conveyors of the destructive influences which are themselves either liquid or solid. A familiar example is t h a t of salt-glazing. The articles t o bc glazed are cxposed to the vapors of sodium chloride. I n contact with the hot fireclay sodium silicate is formed and the chlorine is libcrated. If the operation is unduly prolonged the clay is caten away by the accumulating glaze. A similar action is somctimes mFt with i n furnaces where alkaline chlorides are found in the furnace ~ a s c ~ s . A very curious casc of fircbrick destruction by gases has recently bccn dcscribed by Prof. Osann ("Reduktion und Kohlnng in Hochofen, in Zusammenhange mit Hochofenstorungrn," S f a h l zrlzd Eisen, S o . I 2 . ,912). The trouble is usually met with in the top lining of blast furnaces. The bricks weaken and crumble away, more or less rapidly, without exhibiting any sign of fusion. I n fact the low temperature a t the top makes fusion impossible. I n a case which came under my observation three years ago, the facts were madc more striking because, owing to certain furnace troublcs, the hcat in the top had been excessive and the inner ends of the bricks wcrc burnt hard. The outer ends ncxt t h e shell appeared to he sheared. They scaled off i n laminae, parallel to the shell, and the faces of the laminae vere covered with patches of carbon. From the centers of the patches a minute grain could be withdrawn by a magnet. We therefore had the curious phenomcnon of the hot ends of the bricks being in good shape and the cold ends destroyed. Prof. Osann in the paper referred t o above explains very clearly what is taking place. It is well known that carbon dioxide in contact with red hot carbon bccomcs more or less rompletely reduced t o CO, the equation bcing co, C = ZCO The complcteness of the reduction to CO is a iunction of the temperature. At 1000' C. the reduction is practically complete. A t s o o n C., a temperature frequently met with in blast furnace tops, i t is r r r y far from complete. NOWt h e foregoing reaction is reversible, and in t h r presence of a catalyst, the reaction proceeds rapidly. The composition of the gas coming from the lower part of the furnace corresponds to a temperature a t which the reducing reaction is complete. When the gas arrivcs a t the top, and comes in contact with particles of iron oxide contained in the brick, the reaction reverses and carbon deposits. This carbon accumulates in the substancc of the brick until the latter is completely disintegrated. It is noteworthy that this carbon deposition is particularly noticeable in the bond clay which has been used for laying up the bricks. When an old lining is being taken out it is not, uncommon to see masses of brick adhering together and showing, where broken across, the bond clay like a geometrical pattern drawn in charcoal.
+
A solncwhat parallel case of destruction by gas is afforded in rhe case of blast furnaces running on ores containing zinc. Vnder these conditions the vapors of metallic zinc are liable tci enter the brick and, passing t o the cooler parts of thc brick, oxidize a t thc expense of the C O ? and solidify. The veins of crystallized zinc oxide which result from this action bavr a powerful effect in rupturing t,he brick and causing crumbling. Fig. 7 is a photograph of a piccc of the uppcr inwall lining in which a vein oi mctallic zinc has condensed.
F ~ 7G ~
This is very unusual. A s 8 gcncrai rule the vcin would he zinc oxide as described above. I t is noticeable in these cases oi gas destruction that the trouble is set up in the interior of the brick. and in the clay bond in xhich the bricks \sere laid. I t is by no means confined t o the siirfare; morcover. if the bricks mcre entirely impenetrable the destructive cffccts could not occur. The effect is one of penetration by the destructive influences much as in the case ol destruction by slags. Other things bcing equal I a m convinccd that the mor? compact and close-textured a brick is, the better it mill stand up against ccrrrosivc slags and gases. And thc truer i t is to shape, thc closer i t can he laid up in the ~ ~ r penetration k ; into the bond between th? bricks i; facilitated by wide joints and is highly destructive t o t h r Ivork. .4s regards the risk of spalling v i t h changrs of temperature in the case of closetextured brick, experience shoxs that brick of this type can be made to stand up perfectly. If they spall, the triinhle is due t o unsuitable clay, or poor workmanship, or both. KESSICI,
IAUOXATORi
,L-rseu n n c CO. I'*I\IER?OV. P*.
XEW
THE INFLUENCE OF CINDERS ON THE CORROSION OF IRON IMBEDDED IN CLAY' UT \>-ALTER u. ScxuLrE Investigations of electrolytic corrosion of underground structures occasionally reveal examples of marked corrosion under certain conditions which preclude a consideration of stray currents from electric railways as the cause. Filled ground secms t o be especially harmful t o iron pipes, espccially where the filling is composed largely of cinders, coal, and furnace products. Oi the various explanations vhich have been offered to account for rhis corrosion the more commonare: 1 pager presented at the Eighth Inremntionnl Conpress of I m l i e d Chemistry, N e w l'ork, Se~iembtr,1912.
July, 1913
T H E J O L - R S 2 4 L O F I - Y D L - S T R I A L A N D EIZ-GIXEERILVG C H E J f I S T R Y
t h a t the sulfur, which is a constituent of most coals, probably forms sulfuric acid, with the resultant chemical corrosion; also t h a t cinders and particles of conductive carbon in contact with iron produce a local couple, and the pits which are formed are attributed t o this effect. The objection t o the former theory is that after a severe heat treatment the sulfur is usually expelled from the cinder material, and t h a t the latter explanation is not completely adequate is indicated by the fact t h a t the corrosion occurs even where there may be a n intermediate layer of clay between the cinder bed and the pipe. Certain cases of deep pitting of iron pipe have been called to the attention of the writer where a n overlying bed of cinder filling out of direct contact with the pipe seemed to be in a measure responsible for the damage noted. I t is well known t h a t if a mass of carbon and a mass of iron are imbedded in a n electrolytic conductor, such as in moist earth, a difference of potential is established between these two bodies, the iron being electrochemically positive t o the carbon. This potential does not establish a flow of current unless metallic contact is made between these two bodies. While a layer of carbon material may be separated from a n iron pipe in one locality b y a bed of sand or clay, corrosion may be produced a t t h a t locality by reason of contact between the iron and the carbon bed a t some other position, such for example as occurs where a service pipe or other portion of the underground system may pass through the carbon bed. I t is evident, therefore, t h a t contact between the cinders and the iron is not necessary a t the exact location where the corrosion is observed in order t o account for such corrosion. An active couple produced by carbon and iron results in the iron being the anode and the carbon being the cathode. Polarization tends t o reduce the flow of current but this polarization is minimized if air has ready access t o the carbon bed ; therefore the nearness of the carbon bed t o the surface of the earth is a factor which influences the corrosion caused by such material. The object of the investigation here described was t o reproduce, in the laboratory, conditions which were found in the field and t o make laboratory measurements of potential and current flow, and determination of the amount of corrosion. The experiments were carried out in the Chemical Engineering Laboratories of the University of Wisconsin. APPARATUS
The apparatus, as set up in the laboratory, included a sheet of iron buried in a bed of clay on the top of which rested a layer of granular carbon which represented the cinder bed. Electrical contact between the plate and cinders was provided for by a wire connected between them. The arrangement of the apparatus is shown in Fig. I . I t is of course evident t h a t the conductive layer of carbon here used has a higher electrical conductivity t h a n that produced by cinder filling, as found in practice. and the reason for using this was to accelerate
555
the action so as t o produce measurable results within a short time. The iron plate was of No. 24 sheet iron which was pickled, dried and weighed before being placed in the clay which was taken from a fresh excavation and was moist when placed in the box. The layers below and above the plate were well tamped in place. The carbon, which had been used as a resistor in electric furnaces, was granular, the particles ranging from I/~" to in diameter. A graphite block, with a stem, was placed in the carbon t o make electrical contact between the carbon and external connections, and insulated wire connected t o the plate was brought out through the clay. The carbon lead was connected t o a switch, also connected to the iron plate terminal I
W
S W I T t PI
I
FIG.1
so that b y closing it the cell was short-circuited. The switch was provided with short leads so that a n ammeter or voltmeter could be connected t o them and measurements be made with the switch open. The carbon was covered by a I I// layer of sand a t the beginning of the test but this was removed on the fourth day. To measure the single potentials of the iron and carbon against the clay a calomel electrode was employed with a long nose reaching down through a glass tube into the clay half way between the carbon and iron. A few drops of normal KCI were poured into the bottom of the glass tube t o moisten the clay and assure good contact between the clay and electrode. M E A S U R E 31E N T S
Before closing the switch, the potentials of the iron and carbon against the clay were determined by the potentiometer method, taking -0.56 volt a s the potential of the calomel electrode. The results are given below: Total E. M. F., 0.842 volt, carbon cathode Clay to carbon, -0.638 volt Iron to clay, 0.204 volt
+
Total, 0 . 8 4 2 volt A voltmeter indicated 0.83 volt.
After making the single potential measurements, the circuit was inadvertently closed for a minute or two. after which the voltage, a s indicated b y a voltmeter, had fallen t o 0 . 7 2 volt. Further short-circuiting caused the voltage t o drop from 0.68 to be-
THE J O VR.YL4L OF I-1-D C S T R I A L A.\-D
5 56
tween 0.45 and 0.50 after a few minutes. One and onehalf minutes after opening the switch the voltage increased t o 0.68 volt. The cell was short-circuited for twenty minutes and the switch then opened, after which single potential measurements were made a t various intervals. After being on short-circuit for seventeen hours, similar single potential measurements were made, during which time the average current, as measured by a milliammeter, was o . o j x j ampere. The data of these measurements, as tabulated in Table I and plotted in Fig. 2 , show that the rate of depolarization is decreased after the cell has been short-circuited for the longer time. The cell was then allowed to stand short-circuited and readings were made of the voltage and current a t various times during the run, the results of which are tabulated in Table 11. The instruments were of standard make and in good calibration and of the following scales and resistances:
EA-GISEERI-VG CHEMISTRY
Vol. 5 , KO.7
from the iron to the carbon, the depolarization, due presumably to the action of the air on the carbon. taking place at a constant rate. The loss of iron by one ampere-year is about 2 0 pounds so t h a t 0.03 ampere, the assumed average current with the clay moist, would be about 20 x 0.03 = 0.6 pound. The area of the plate was 1 . 4 5 square feet or 2.9 feet total surface, so that the theoretical corrosion per square foot would be 0.23 pound per year or about
53.5 ohm resistance 0.09 ohm resistance
Voltmeter 0-15 volt: Ammeter 0-500 mi!-ampere:
During the fourth day the layer of sand above the carbon was removed as noted in the table. The TABLEI-DEPOLARIZATION Single potentials of iron
Single potentials of carbon 2
Short-circuited 20 minutes
-
r
Short-circuited li hours
__A__ _---A__
Time Min. E. M. F.
.. 1,'s
1 2
3 4 5 7
.. -0,432 -0.478 -0.485 -0.508
Time >fin. 0
1 1'/2 2
..
...
3 4 5 8 10 16 23 43
..
...
60
-0.510
-0.516 -0,524
10
-0,530
24 42
-0 -0
540 564
Short-circuited 20 minutes
Short-circuited 17 hours
7 - 7 - F
E. M. F.
Time Min.
...
..
-0 394 -0.424 -0.434 -0 438 -0.442 -0.452 -0.452 -0.458 -0.468 -0.470 -0.474 -0.484 -0.499
0 3/4
2 3 4 6 24
.. .. .. ..
E. M. F.
Time Min.
FIG.2
E.
M. F.
,..
..
...
+O
184 +0.184
0 1
+0.186
2 3 4 5 8
+0.162 '0 160 +0.160 +0.160
+0.180 i-0.182 +O.l80 t0.178
+0.160
+O.l60 i-0.158
+0.180
10
+0.158
... ... ...
16 23 43 60
+ O 158 i-0.156 t0.156
...
+0.158
.. ... .. . ,. ... clay began to dry out, so that beginning with the tenth day about half a liter of water was sprinkled over the carbon and allowed to soak in before a set of readings were made. No readings were taken between the 33rd and 58th days, nor was the cell moistened, with the result that the current on the 58th day was almost nil, and the potential 0.22 volt. Three minutes after the cell had been moistened the observed values increased t o o . 0 1 7 j ampere and 0.32 volt. Rough integration of the curve plotted between current and days, with the current as zero on the j 8 t h day, gives. 35 ampere hours. If the cell had been moistened regularly each day throughout the test i t is fair to assume that the average current would have been 0.03 ampere: 0.03 ampere for 5 8 days is equivalent to about 42 ampere hours. The average voltage over this period can be taken as 0.35 volt. The interesting point t o be noted is that as long as the clay m-as moist a current flowed continuously
I/4 pound per year. The loss of I , ' ~pound per foot from each surface of a plate would destroy the equivalent of a No. 30 sheet in one year o r half of a KO. 24 sheet. If this corrosion were concentrated in six square inches area on every square foot, the metal would be pitted t o the depth of 0.1j inch-which would pierce a three-inch cast iron main in a little over two and one-half years.
DAP 1
2 3 4 5 7 8 9 10 11 12 14 15
17 24 25 29 30 31 32 33 58
TABLE11-CURRENT VOLT
AMPERE 0.052
0.0518 0.050 0.051 0.051 0,0495 0 035 0.030
0.040 0.035 0.0475 0.0475 0.040 0.041 0.030 0.0375 0.023 0.020 0,035 0 028 0 028 0.015
0.45 0.50 0.45 0.45 0.44 0.38 0.40 0.20
VOLTAGEREADINGS REMARKS
AKD
Sand removed from top
....
Began wetting
0.40 0.42 0.42 0.42 0.40 0.35
Dried out since last reading
0.36
0.30 0.20 0.40 0.35
0.35 0.32
CORROSIOK O F PLATE
After the j8th day the cell was dismantled. Examination of the clay near the plate showed t h a t the rust had penetrated into i t as much as I/4 inch in some places. The plate was spotted with oxide
and greenish iron conipounds. Corrosion x a s noticed both on the upper and lower sides of the plate and the corner diagonally opposite the wire connection mas eaten through. The plate was washed, the greater part of the rust scraped off and the remainder then dissolved off by hot ammonium citrate, after which i t was weighed. The data follow: Grams U‘eight of plate before t e s t . . . . . . , . . . . . . . . . . 401 6 Weight of plate after test. . . . , . . . . . . . . , . . , . 346.5
__
1.0s~..
..... . , . , . . .. . . .. . .
55.1
Thirty-five ampere hours theoretically would corrode 36.4 grams of iron so that the efficiency of corrosion in this case would be about I j o per cent.; that is. i t may be said that two-thirds of the loss by corrosion was due to the current produced by the difference of potential between the carbon and the iron. These experiments were intended as preliminary work for further investigation, which, however, was not undertaken, but a description of this first, rather incomplete work, it was thought, would prove of interest and draw attention t o the influence of carbon and cinders on the corrosion of iron and perhaps lead to further experimentation on the subject. Further work al?ng this line should be undertaken with the conditions more closely approaching those actually found; t h a t is, using cinders instead of carbon and putting them, with the iron plate, in the earth. Similar plates not connected with the carbon should also be buried so that the amount of natural corrosion can be determined. Under these conditions, with the higher resistant cinders. i t is to be expected that less current would be produced. An interesting case of corrosion of a gas pipe which may be explained by action similar to that discussed above has been observed recently by the writer. Along. a certain street the gas mains were paralleled by a water main buried a short distance beneath. The soil is mainly clay or silt b u t at the lowest portion of the street the water main dipped into a peaty layer. The gas main lay in the clay above the peat at this point and was corroded through at several places. The main apparently was in a locality not subjected to stray currents from the street railway or grounded power circuits so t h a t the currents causing the corrosion must have come from other sources. A length of gas main about 400 feet long was disconnected from the adjoining portion. Potential measurements between this disconnected length and a n iron rod driven in the black layer showed that the gas main was 0.3 volt positive, and readings between gas and water connections showed that current was flowing from the gas to the water main through the earth. The corrosion can be explained by the fact that peat sets up a different potential toward an iron surface than does clay, and con,nections between the gas and water services on the consumer’s premises makes the couple electrolytically active, thereby producing electrolytic corrosion. MADISOX. \VIS.
RECENT ANALYSES OF THE SARATOGA MINERAL WATERS. I11 B y LESLIE RUSSELLMILFORD Received April 5 . 1913 A C Q U I R E M E N T O F T H E G E N E R A L C A R E O S I C COA.IP.LXT
Following the passage of the “Anti Pumping Act ” in 1908, by the legislature of the State of New Tork, the gas companies situated a t Saratoga Springs, which Tvere pumping the mineral water and extracting the gas from i t for commercial purposes, were the defendants in a great amount of litigation. One of the largest gas-producing companies was known as the General Carbonic and was situated in the Geysers district. This company derived about one-half of its supply from one spring with natural flow (the Adams) and the remainder from several wells b y pumping. This plant continued its operations until the Reservation Commission, in July, 191I , acquired the property which contained about one hundred
ISLAND SPRING
acres of land and 2 j existing bores, of which 19 varied in depth from 2 7 0 to 533 feet to the bottom of the well. This acquisition was necessary for the control of the large area of the mineral water basin, adjacent to the fault from which the wells and springs have their origin. I n order to obtain all of the information possible relating to the condition of the wells upon this property and the effect of pumping on this land and the lands of other proprietors arrangements were made for taking scientific observations while the previous operations of this company were continued without disturbing existing conditions. The General Company continued pumping from July 2 , 1911, t o July 2 0 , 1911. During this period six pumps worked continually, pumping from wells ranging from 316 t o 397 feet in depth, an average of 120.000 gallons of water in twenty-four hours. Observation of water levels, flow of water, gas pressure, and other physical conditions were noted. Com-
