INDUSTRIAL AND ENGINEERING CBEMISTRY
January, 1925
15
Alloying Tungsten with Aluminium Containing Ten Per cent Copper' By M. R. Whitmore and F. T. Sisco AIR SERVICE, WARDEPARTMENT, MCCOOK FIBLD,DAYTON,Oxio
HIS paper summarizes the second part of the research ture was 1260' C. (2300' F.1 and the pouring temperature 1040' C. (1900" F.) This was Melt 2. The metal was more with SomeOf the the light Of fluid than Melt 1 and poured better, although again considerable now being conducted by the Metauurglcal undissolved tungsten was found in the bottom of the pot. Branch, Engineering Division, Air service. As stated in a When the ingots of Melt 1 were broken, the appearance of the previous paperta the plan of attack was to make a few ex- fracture indicated that the tungsten was badly segregated. This ploratory alloys to determine (1)if the metal would alloy with melt was then remelted in the Ajax furnace, soaked for 30 minutes a t 1200' C. (2200' F.) and poured. This was designated ahminiurn or aluminium-copper, (2) its solubility, (3) foundry Melt 3. practice for the alloys, (4) prelimi(2) From the appearance of the fracture and from preliminary cheminary physical tests,-&& (5j metalcal analysis it was certain that in all lographic structure. The result three melts made under (1) the A s t u d y has been m a d e of the solubility of this exploration would then detungsten was badly segregated. A of tungsten in a l u m i n i u m a n d a l u m i n i u m termine whether or not the alloys second method of attack was therecontaining 10 per c e n t copper. Difficulties fore attempted. This plan was to merited a more exhaustive investiin alloying the elements a n d casting the make an aluminium-copper-tungsten gation. hardener. An alloy containing 20 alloys a r e described. A method for the In outlining this investigation, per cent copper and 20 per cent tungchemical analysis of the alloy is given and two major difiiculties were anticisten was desired. Accordingly pure the properties and structures of a n y alloy copper was melted in the Ajax furnace pated. The first was in alloying containing 10 per cent copper, 1.25 per cent and preheated to 1540' C. (2800' F.) tungsten with aluminium, due to and the tungsten powder added. (On and balance aluminium are distungsten, the great differencein melting points account of low line voltage it was imcussed. (AI, 657' C.: W, 3267" C.1 and possible to exceed a temperature of 1540' C. with the Ajax induction furspecific gravities (Al, 2.7; W, '18.8). The second was a method for the chemical analysis of the alloy. nace.) The metal was soaked a t this temperature for 30 minutes and the aluminium was then added in the form of bars. After another 30 minutes at 1640' C. (2800' F.) the metal was poured Material
T
T h e chemical analysis of the metals used in this investigation is ag follows: Aluminium Aluminium 99.82 Copper Tungsten powder Tungsten cubes
Iron 0.07
Silicon 0.04
Copper
Tungsten
0.07
99.95
c. P. c . P.
No analysis of the pure tungsten was made. It was purchased as chemically pure and was assumed to contain in excess of 99 per cent tungsten. Metals of very high purity were used in order to prevent the appreciable amounts of silicon and iron occurring in ordinary aluminium from influencing the physical results. With these eliminated, any material change in physical properties of the alloy could be attributed to the added tungsten. Foundry Practice
Five methods of attack were used in attempting to alloy tungsten with aluminium and aluminium-copper
.
(1). In the first experiment it was desired to make a hardener containing 90 per cent aluminium and 10 per ceat tungsten. This alloy could then be used as a source of tungsten for alloys having a tungsten range of 0.50 to 5.00 per cent. It was believed that the best physical properties would be found in the alloys of the lower tungsten contents (0.50 to 2.00 per cent). Aluminium in the form of notched bars was melted in an Ajax electric induction furnace and preheated to 1370" C. (2500" F.). Tungsten powder was then added and stirred well. After the metal (Melt 1) had been soaked at this temperature for 30 minutes, it was stirred again and poured into ingots. The metal was mushy, poured sluggishly, and appeared too cold. Considerable tungsten remained on the bottom of the pot. This experiment was then repeated using tungsten cubes instead of powder and a 5 instead of a 10 per cent tungsten alloy. The same procedure was followed except that the maximum temperaI Received July 18, 1924. Published by permission of Chief Of Air Service. War Department. 2 Sisfo and Whitmore, THIS JOURNAL, 16,838 (1924).
into ingots. Considerable residue, mostly tungstic oxide, was left on the side of the pot. The metal was viscous and sluggish when poured and much slag was present. This alloy was designated as Melt 4. The ingots were hard and brittle and the fracture showed indications of excessive segregation. (3) In the third method it was thought that tungsten might be introduced into molten aluminium and copper by taking advantage of the intense heat generated when these two molten metals were mixed. Aluminium and copper were melted separately in two Monarch oil-fired crucible furnaces according to standard foundry practice for an aluminium-copper hardener. 1 The composition calculated was 40 per cent copper, 20 per cent tungsten, and 40 per cent aluminium. The tungsten was placed in the crucible with the copper in order to preheat it to about 1040' C. (1900' F.) before mixing. When the copper was liquid the crucible was removed from the furnaces and the molten aluminium poured into it. After mixing the tungsten was still undissolved, so the crucible was returned to the furnace and heated to 1370' C. (2500' F.) and soaked a t this temperature for 2 hours. On pouring the ingots the metal appeared mushy and viscous. Most of the tungsten was left in the pot. This was Melt 6. (4) After failing in the foregoing attempts to introduce tungsten successfully into aluminium by using metallic tungsten, a few experiments were made with tungsten compounds on the hypothesis that molten aluminium might reduce these compounds to metallic tungsten and the exothermic reaction be sufficient to form an alloy. The compounds used were C. P. tungstic oxide, sodium tungstate, and ammonium tungstate. For this work a small graphite crucible and an electric crucible furnace were used. Five hundred grams of aluminium were used in each instance and heated to 815' C. (1500' F.) at which temperature the tungsten compounds were added in small quantities (about 5 grams) at short intervals. Only a trace of tungsten was dissolved in any of these trials. In all cases when the metal was poured, tungsten as tungstic oxide was found on top of the metal and metallic tungsten powder on the bottom of the crucible. The results and conditions of this attempt to alloy tungsten with aluminium are tabulated in Table 11. (5) The final foundry work consisted of remelting the various melts previously made in an effort to get rid of segregation. It was believed that aluminium and aluminium-copper would dissolve an appreciable amount of tungsten. In the melts described under (I), (2), and (3) it was certain, from the large amount added and from preliminary chemical results, that most of the
INDUSTRIAL A N D ENGINEERING CHEMISTRY
16
tungsten was held in suspension in a finely divided state, and it was thought that by diluting with aluminium and by additional heating some of this tungsten would go into Accordinalv. Melts 7 and 8 were made. Both of these melts were found to-bi segregated. The final melt, 9,was made from pure aluminium and aluminiumcopper-tungsten ingot, Melt 8. The desired percentages of copper and tungsten were 10 and 1, respectively. The charge was melted in a plumbago crucible in a Monarch oil-fired furnace and was preheated to 1095' C. (ZOOO" F.). The pouring temperature was l0lOQC . (1850"F.) Cryolite was used &s a 5ux. Eight molds of TB-1 test bars were cast in green sand from this melt and B chemical sample was taken from the first, middle, and last molds.
------- -
-
TABLE I-%lMMARK
mired
-As
CoreoSlrroN
Actual
Meit Cu 1 None
W 10
AI 90
C" Not determined
7.80
None
5 10 20
95
Not determined Nclf determined Not soaiyred
2.77
2 3 4
None 20
5
4.54
5.80
6
40
20
7
16
8
16
8
90 Bo
Balance
W*
1.80
9.80
5.51 5.55
40
Not determined
8
Baknce
Not analyzed
8
Balance
Not determined
1.0
Balance
AI
Physical Tests The ultimate strength and yield point in pounds per square inch were determined on an Olsen 20,000-pound standard tensile machine. A@the test bars were cast to size no machining was necessary. The bars were held in wedge grips. Brinell and Rockwell hardness values were obtained under standard conditions, using a 10-mm. ball and 500- and 3ooo-kp. load with the former and a 3.175-mm. ('/&nch) ball with the latter. The physical values at elevated temperature were made on a standard Olsen testing machine of 100,ooO pounds capacity using a 50,OOpoond poise. Pollring
temperature
*F,
C.
2504
1370
19w
1040 12W
22M
28W
Bal.
3.86 0.90 3.80 1.23
No. 1
OP FOUNDRY PRACTICE
7
--
V01:17,
1540
1561
SI5
2500
1370
1470
788
1070
910
RBY*..Sb
Melt poured sluggish. Too cold. Fracture indicated * n e gatioo Poured good. More Buid then Melt 1. Segregated Badly segregated. Remelt of 1 Badly oxidjzed. Heavy slag. Metal sluggish. Badly senegated Remeltofl.2,4. Metaiverycoldwhenpaurrd. Badiyiegrcgated Metal mushy when poured. Tungnten left ill w t . Badly segregeted Remelt ef 5 and part of 0. Metal poured well.
gated Remelt of 7.
Badly -8rc.
Badly zegregated
8.80 1.80 Bal. 1850 1010 Nosegregation. Mdaipouredweiialthou~htooeold 8.72 1.30 9.81 1.33 Where *ne percentage i s given it ir the aveiagt of remits on one sample. Where three percentages are given three S*mPlee vela tilkcn b Metal wured into ingots in ail melts except 9, which w m poured into test bars.
10
The foundry procedure as outlined in Paragraphs 1, 2, 3, and 5 is condensed in Table I. The TB-1 test bar (Fig. 1) is the standard for the U. S. Air Service for sand-cast aluminium alloys. The bars as cast are 0.505 inch in diameter and of a sufficient length between the shoulders to allow for a dctermination of elongation on a 2-inch original gage length. There are three test bars in a mold.
An electric tube furnace 13 inches long with a tube diameter of 2.5 inches was used as the source of heat. This was attached to the fixed head of the testing machine. The test specimens were threaded and screwed into one end of steel pulling bolts (1.5 inches in diameter and 18 inches long). One end of each bolt was attached to the head of the testing machine, the other end entering the tube and holding the specimen in the center of the furnace. A ohramel-alumel thermocouple was placed in the tube with the hot junction in contact with the test specimen. When the assembly was complcte, each end of the furnace was sealed with fire clay. The temperature was raised 28' C . (50" F.) above the desired temperature for testing and the current turnod off. The furnace was then allowed to cool. When tho indicated temperature was 2.8' C. (5" F.) above that required, the specimens were pulled. Cooling the furnace to this temperature required about 20 minutes, which gave suffcient time to obtain temperature equilibrium in the test specimens.
Temp. of Ai at USBDtime of addGrams ing w Rs~crrons RBS LTS 1 Aiuminium 725 845" C. Yellow tungstic Much dag. 20 (1550°F.) acid turned Tungstic oxide No tungorange. Slag sten found ,n.mrrl . .... in mCfd 2 Aluminium 50U 832" C. h ' o ~ l n gformed Sediment in 40 (lM0'F.) oil top but bottom of Sodium tungstate sediment colpot. Tralrcfed at botceroftungtom sten in
--M*ra~rn~
Fro. I-S~~eorao TB-i TESTM o m
FOR
CASTA L U M I N ALLOYS I~
When cold, the bars were sawed from the gate and riser and stamped with the melt and mold number. The test bars from molds numbered 1,4, and 6 of Melt 9 were tested as cast; bars from Molds 2, 3, 5, 7, and 8 were heat-treated as noted below. Heat Treatment The heat treatment operation was conducted in an electric muRle furnace. The temperature was controlled to +9.4" C . (15' F.) bv means of a recordineI .Dotentiometcr and automatic current regulator. The ham were heated slowly to 524" C. (975" F.) *9.4" C. (15'' F.), held a t this temperature for 5 hours, and then quenched in boiling water. They were then aged for 16 hours a t 149" C. (30O0 F.) followed by air cooling. I
"
lixpi.
I_._. ."*,"l
3
Aluminium 480 Ammonium tuiustshte 60
815' C. Coiisiderabie Some metal11500' F.) "lag on te? lietungsten and blackrest- at bottom
duein bottom.
of Pot but
neaftion
trace in metal
parent17 othermi=
sp. ex-
"
Test bars from Molds 2, 5 , and 7 were tested at room temperature. One bar from Molds 3 and 8 was tested at 204" C. (400"F.),onebarat200"C. (WO"F.),andonobar at315"C. (600° F.). The bars from Molds 1, 4, and 6 were tested as cast.
INDUSTRIAL AND ENGINEERING CHEMISTRY
January, 1925
The physical properties of Melt 9 are given in Tables I11 and IV. Chemical Analysis
Two difficulties were anticipated in developing a method for the analysis of aluminium-copper-tungsten alloys; one was the purification of the precipitated tungistic acid; the other, a satisfactory method for the estimation of iron. The customary procedure for the purification of tungstic acid by fusion with sodium carbonate and solution of the fusion in hot water was not applicable to the tungsten-aluminium alloys owing to the formation of soluble sodium aluminate in the carbonate fusion where aluminium would be the principal contaminating element. It was also desired to avoid the volatilization of silicon in the purification of tungstic acid, inasmuch as it was necessary to determine silicon on a separate sample. Silicon can be successfully dehydrated in aluminium alloys only by sulfuric acid and in the presence of this acid tungsten j s not completely precipitated . 3
17
(a) Aluminium sulfate (C. P.)sufficient to give approximately 0.010 gram of aluminium oxide was placed in platinum crucibles and ignited in an electric muffle furnace for 1 hour at 800' C. The crucibles were cooled in a desiccator and weighed. About 10 grams of fused potassium pyrosulfate (C. P.) were added and the contents fused carefully at a low heat over a Bunsen burner. When the fusion had become quiet, the temperature was raised until the melt was clear and fumes of sulfur trioxide were given off. The melt was dissolved in 100 cc. of 10 per cent ammonium carbonate solution at a low heat and not allowed to boil, The precipitate was filtered off and washed with hot 1 per cent ammonium carbonate solution. The paper was returned to the original crucible, ignited, and weighed. The filtrate was made acid with hydrochloric acid, heated to boiling, and ammonia added until the solution was just ammoniacal to methyl red, and boiled for 1 to 2 minutes (Blum).b Aluminium was present in all filtrates. (Table V) ( b ) The procedure in this experiment was the same as in (a) except that the fusion was dissolvedin 100 cc. of ammonium carbonate-ammonium chloride solution (100 grams (NH&COs, 250 grams NH4Cl per liter) and the precipitate washed with hot water containing 50 cc. of the ammonium chloride-carbonate solution per liter. Aluminium was detected in the filtrates but in much smaller amounts than in (a). (c) In (a) and ( b ) care was taken in dissolving the fusion that the solution did not come to a boil. The literature stated that
TABLE 111-PHYSICAL PROPERTIES OF ALUMINIUM-COPPER-TUNGSTEN AND ALUMINIUM-COPPER ALLOYS Gage length, 2 inches Character of fracture, medium granular
r
Mold.. .................................. Type of specimen.. ....................... Diameter, inches.. ....................... Maximum load, pounds.. . . . . . . . . . . . . . . . . . . Ultimate strength, pounds per square inch, Elongation in 2 inches per cent. . . . . . . . . . . . . Brinell, 500 kg.. . . . . . . . . . . . . . .:. . . . . . . . . . . Brinell, 3000 kg.. ......................... Rockwell, 1/8 Ball.. ....................... 0 Each result the average of three tests,
..
-As 1 0,503 3980 19,820 0.67 78 .80
81
cast@4 0.506
4160 20 330 0: 67 78 82 81
CIJ 10, W 1.25, AL BALANCE --Heat-treated"6 2 5 TB-1 Sand-cast round 0.502 0.506 0.506 4130 5630 4670 20 790 27,900 23,260 0: 83 0.67 0.67 77 .. 104 112 126 126 82 97 100 84
Very little work was done on the estimation of iron in aluminium-tungsten alloys, as the preliminary metallurgical experiments produced no alloy of promise. Of the published methods for tungsten, it is believed that only one is applicable to the aluminium alloys. This is the method described in the Tentative A. S. T. M. Methods for alloy steels under tungsten steel. I n this method the purification is based on the solution of the precipitated tungstic acid in ammonia, thereby freeing it from all the iron and the larger part of the silicon. This purification would be more difficult with aluminium on account of the gelatinous nature of aluminiufi hydroxide. It would probably be necessary to reprecipitate the aluminium several times to remove occluded tungsten. This would concentrate the solution with ammonium salts, which tends to prevent the final precipitation of tungsten.4 It is doubtful if a complete separation of tungsten from aluminium could be made by this procedure, owing to the solubility of aluminium hydroxide in ammonia. It is not believed that the concentration of aluminium would be sufficient to introduce an appreciable error in the tungsten results. This was not verified. Therefore, the most promising line of attack for the purification of the ignited tungstic acid seemed to be the fusion with potassium pyrosulfate, which forms soluble potassium tungstate and leaves silicon as the dehydrated oxidee4 This gave a separation of tungsten from silicon. To get rid of the aluminium application was made of the work of Blum and Lundell on the precipitation and separation of aluminium from other elements.6 The following experiments were then made in an attempt to work out conditions for the quantitative separation of occluded aluminium in tungstic oxide: Scott, "Technical Method of Metallurgical Analysis," 1923, p. 605. A, S. T. M. Tentative Standards, 1922, Notes, p. 123. 6 Bfum, Bur. Standards, Sci. Paper 286 (1916); J . Am. Chem. SO&, 88, 1282 (1916); Lundell and Knowles, I b i d . , 45, 676 (1923). 8
7 0.504 5830 29,250 0.50 104 131 96
Cu 10, AL BALANCE Average Average heatas cast treated Sand-cast 0.505 0.505 20,000 1.0 75
30,000 0.50 100
on boiling the tungsten salt would hydrolyze and cause partial precipitation of the tungsten.6 The procedure in this experiment was identical with ( b ) except that 0.5000 gram ignited tungstic oxide was added to the aluminium oxide taken, and after the fusion was dissolved the solution was boiled for 2 minutes and filtered. Aluminium was not detected in the filtrates, but the aluminium recovered was considerably less than the amount taken. It was known that potassium pyrosulfate fusions attack platinum,' and the degree of this attack was then investigated. The contents of the crucibles from Experiment c were carefully brushed out and the crucibles reignited and weighed. The loss in weight in the fusion was almost identical with the difference between the aluminium oxide taken and found. TABLEIV-PHYSICAL PROPERTIES OF HEAT-TREATED CASTALUMINIUMCOPPER-TUNGSTEN AT ELSVATED TEMPERATURES (Cu 10, W 1.25, AI balance) Gage length, 2 inches Character of fracture, coarse-medium granular Type of specimen sand-cast round Hoskins tube furiace: chromel-alumel couDle -Melt 9-Mold 3 -Melt 9-Mold 8Temperature 204' C. 260' C. 315' C . 204O C. 260'C. 315'C. 400' F. 500'F. 600' F. 400' F . 600'F. BOOOF. 0.496 0.499 0.510 0.500 Diameter, inches 0.502 0.505 Maximum load, pounds 5797 5800 4056 4310 5566 4889 Ultimate strength, pounds per square inch 29,280 28,940 20,980 22,020 27,210 24,400 Elongation in 2 inches 0.5 0.5, 1.0 0.5 0.5 0.5 per cent ~
(d) I n this experiment fusions were made in clean, empty, tared platinum crucibles. The melt was dissolved as in (c), the crucibles were thoroughly washed, and then ignited and weighed to determine loss in weight. ( e ) This was a repetition of (c) except that 0.2500 gram of C. P. ignited tungstic oxide was placed in the crucible with the aluminium oxide. The crucibles were weighed after the fusions and the aluminium was calculated from this instead of the original weight of the crucibles. No tungsten was detected with the aluminium, and the aluminium was recovered quantitatively. (f). Owing to the negative results obtained in the foundry experiments on the copper-aluminium-tungsten alloys, little work
4
Bur. Mines, Bull. 212, p. 165. o$. cit., P. 4423.
1 Scott,
INDUSTRIAL A N D ENGINEERING CHEMISTRY
18
was done on a method of analysis for iron in these alloys. On account of traces of tungsten that may remain in solution* after the sulfuric acid dehydration of silicon, it is not believed that iron can be determined by passing the solution from the copper determination through a Tones reductor and titrating the reduced iron with potassium pekanganate. Z t is known that tungsten in solution is reduced by zinc and reqxidized by potassium permanganate. However, very satisfactory results were obtained on a synthetic alloy, as follows (for the synthetic alloys, a h minium ingot (cormer. 0.02: silicon, 0.50; iron 0.55) and C. P. tungsten powder were used) :.
Vol. 17, No. I
acid and place beaker on the hot plate until the copper has dissolved. Add 100 cc. of hot water and digest the solution just below boiling for 2 hours- Add 15 cc. of ci~chonine solution (125 grams cinchonine crvstals in 1000 cc. 1:1 HC1) and filter. washing with hot 10 Der cent hvdrochloric acid containini 25 cc. oyf the solution liter. T ~ fer the Paper and precipitate to a platinum crucible, dry and ignite in a muffle furnace at 800" c. to constant weight. T o purify the ignited tungstic oxide, add about 10 grams of
ier
TABLEV-PRELIMINARY EXPERIMENTS ON TUNGSTIC OXIDE PURIFICATION AlaOa
AhOa taken Gram
WOa taken Gram 0.0097 None 0.0097 None 0,0096 None 0.0092 None 0.0100 None 0.0097 None 0.0095 None
AlzOa found Gram
SOLUTION OR KzSzO? FUSION
0,0085 0.0059 0,0046 0.0060
0,5000 0.5000 0.6000 0.5000
None
None
Dissolved in 10 per cent (NH4)rCOa. Not allowed to boil. Ppt. washed with (NH4)zCOa
0.0078
Dissolved in 100 cc. of solution containing 10 grams (NH4)2 COa and 25 grams NHaC1. S o h . not allowed to boil. Ppt. washed with NH4CI- "(4) ZCOS Dissolved as in b except that s o h . was boiled for 2 minutes just before filtration
Not weighed
Loss in weight of crucible in fusion Gram
foind corrected for crucible loss Gram
0.0048
REMARKS AI detected in all four filtrates. Precipitation according to Blum
0.0069
n nnm
0.0067 0.0048 0.0042 0,0042
Fusion blank on crucibles Fusion dissolved as in (c)
A1 detected in all filtrates but in much smaller amounts than in (a) 0.0014
0.0010 0.0005
0,0007 0,0004 0.0005 0.0004
0,0081 0.0058 0.0047 0,0049
N o AI detected in filtrate, no tungsten in AlzOa. Appreciable loss in weight of crucibles
0.0009 0.0081 0.0074
Same as (c)
0.2500 0,2500 0,0066 0.2500 0.0058 0.2500
0.0083 0.0072
0.0058
Five grams of aluminium and 0.050 gram of tungsten powder were dissolved in acid mixture (300 cc. HCI, 100 cc. "Os, 150 cc. HtSOr, 450 cc. HzO) and evaporated to dense fumes of sulfur trioxide. When cool, the cake was dissolved in 20 cc. of 1:1 sulfuric acid and 200 cc. of hot water. The solution was filtered hot and the precipitate washed with hot 2 per cent sulfuric acid. The paper and residue were ignited in a platinum crucible, fused with potassium pyrosulfate as in (c) and the silicon determined by volatilization with hydrofluoric acid on the ignited residue. After removal of copper by electrolysis from the filtrate of the silicon determination, 10 grams of citric acid were added for each gram of aluminium and the solution made slightly ammoniacal. The iron was then precipitated with hydrogen sulfide. The sulfide precipitate was filtered and washed with hydrogen sulfide water to which a few cubic centimeters of ammonia had been added. The iron sulfide was dissolved off the paper with hot 5 per cent sulfuric acid allowing the solution to run into the beaker in which the precipitation was made. The solution was then boiled for a few minutes and an excess of potassium permanganate added to destroy sulfide sulfur and oxidizable polythionic acids. When cool, the solution was passed through a Jones reductor and titrated with standard potassium permanganate (Table VI). TABLE VI-PRELIMINARYESTIMATION OF SILICONS AND IRONIN ALUMINIUMCOPPER-TUNGSTEN ALLOYS (Experiment fi
Aluminium Tungsten added Gram None None None None
0.050 0.050 0.050
used, 5 grams Silicon Gram 0.50
0.48 0.48 0.49 0.48 0.48
Copper Gram 0.02 Not determined Not determined
0.48
Iron Gram 0.55
0.56
REMARKS Analysis by Standard Method (see A. S.T. M. Tentative Standards on Methods for Analysis of Aluminium Alloys)
(n
0.56 Method as given under 0.65 0.56 Method as given under 0 . Tungsten added was powder (e. P.) 0.56 0.56
Method of Analysis
From data obt,ained in the experiments, the following procedure was used in the analysis of the aluminium-coppertungsten alloys: TUNGSTEN-Dissolve a &gram sample of the alloy (tungsten less than 15 per cent) by adding 100 cc. of 1 : 1 hydrochloric acid, a few cubic centimeters at a time. When the violent action has subsided, add 25 cc. of concentrated nitric 1 Scott,
Crucibles reweighed after fusion
0.0063
09. tit., p. 608.
potassium pyrosulfate to the crucible and fuse carefully at a low heat to avoid spattering. When fusion has become quiet, gradually raise the temperature until the fusion is clear and fumes of sulfur dioxide are given off. When cool, place the crucible in a covered beaker containing 100 cc. of ammo100 nium chloride-carbonate solution (250 grams "&I, grams (NH&COa per liter) and dissolve the fusion slowly with very little heat. When solution is complete, remove the crucible from the beaker and wash well, allowing the washings to run into the beaker. Place the beaker on a hot plate, bring to boiling, and boil for 2 minutes, or until there is just a faint odor of ammonia. Filter a t once, washing the paper, and precipitate well with hot water containing 50 cc. of the ammonium chloride-carbonate solution per liter, Ignite and weigh crucibles after the fusions. Place the paper and residue in the original crucibles, dry, ignite, and weigh. The difference between this and the original weight is tungstic oxide, which contains 79.31 per cent of tungsten. SII,IcoN-Dissolve a 5-gram sample of the alloy in the usual acid mixture (HC1, "08, HzS04) for aluminium alloys and evaporate to copious fumes of sulfur trioxide. Remove beaker from hot plate and when cool dissolve the cake in 20 cc. of 1:1 sulfuric acid and 200 cc. of hot water. Filter the solution while hot and wash paper thoroughly with hot 2 per cent sulfuric acid. Transfer the paper to a platinum crucible, dry, and ignite. The silicon may then be determined, either by direct volatilization with hydrofluoric acid in the presence of tungstic oxide or the residue may be fused with potassium pyrosulfate (as in Experiment e), and the silicon determined on the ignited precipitate from this fusion. I n either instance the residue from the sulfuric acid dehydration should be worked over for traces of iron and copper that may have been carried along and corrections made on the copper and iron determinations given below. COPPER-Using platinum electrodes, determine copper by electrolysis on the filtrate from the silicon determination. IRON-TOthe filtrate from which copper has been electrolyzed add 10 grams of citric acid for each gram of sample taken. Make slightly ammoniacal and precipitate iron with a rapid stream of hydrogen sulfide. Filter and wash with hydrogen sulfide water to which a few cubic centimeters of
~
January, 1925
INDUdTRIAL A N D ENGINEERING CHEMIXTRY
ammonia have been added. Dissolve the iron sulfide with hot, 5 per cent sulfuric acid, catching the washings in the beaker in which the precipitation was made. Boil the solution for a few minutes and add potassium permanganate in excess to destroy sulfide sulfur and oxidizable polythionic acids. Cool and pass the solution through a Jones reductor. Titrate the reduced iron with standard potassium permanganate. Metallography
Following the customary procedure, specimens for metallographic examination were taken from the end of the 0.505 sand-cast test bars. A piece 0.5 inch long was sawed from bar 1-13, which was tested as cast, and from bar 2-B, which was tested as heat-treated. Both specimens were ground flat on a 180 carborundum belt and were rough-polished on French emery paper Nos. 0, 00, and 000. A mixture of paraffin and kerosene was used on the No. 000 paper. The final roughpolishing was accomplished by levigated alumina on a disk covered with broadcloth. The finish-polishing was by magnesia on broadcloth, followed by a final treatment on the broadcloth disk drenched with distilled water. Successful polishing of aluminium alloys is attended by many difficulties in removing scratches and preventing the flow of the relatively soft metal. Details of polishing aluminium and its alloys have been discussed e l s e ~ h e r e . ~ The two specimens were etched 30 seconds in a 20 per cent solution of nitric acid (aqueous) heated to 70' C. followed by quenching in cold water. Resul t s
CHEMICAL ANALYsIs-Tables V, VI, and VI1 summarize the chemical work on the copper-tungsten-aluminium alloys. To obtain a quantitative separation of aluminium from tungsten in the purification of tungstic oxide, it is necessary that the ammonium carbonate solution of the potassium pyrosulfate fusion be made in the presence of a high concentration of ammonium chloride and that the solution be boiled to expel the excess of ammonia. Owing to the loss of platinum in potassium pyrosulfate fusions, the crucibles should be reignited and weighed after the fusions, before igniting the iron, aluminium, and silicon occluded by the tungsten. The one experiment made on the estimation of silicon in the residue from the potassium pyrosulfate fusion and the determination of iron by precipitation with hydrogen sulfide in an ammoniacal citrate solution gave values in close agreement with those obtained by standard methods. The segregation of tungsten, which was early suspected when the ingots and test bars of the alloys were broken, is verified in Table VII. I n all but one instance the tungsten found is much less than that calculated, but with no constant loss. This is probably due to the fact that in some melts the sample was taken from the top of the pot and in others from the center or bottom. The degree of segregation from top to bottom of the pot is shown in Melt 8, where there is a variation of 300 per cent between the top and center and bottom and center, respectively. The same condition is noticed in Melts 1 and 3, where the values may be assumed to be on the same material inasmuch as Melt 3 was a remelt of Melt 1. The one instance (Melt 5) of a high tungsten value is easily expfained by the fact that the tungsten added was calculated from the tungsten values given for Melts 2, 3, and 4, which melts were used as the source of tungsten, and the sample was probably taken from a highly segregated area. But one melt (Melt 9) gave concordant results when samples were taken from more than one portion of the pot, and in this case the value was close to that calculated. I Dix,
Chem.Met. Eng., 27, 1217 (1922).
19
FOUNDRY AND CASTING-From the results of the foundry experiments it is evident that tungsten does not alloy with aluminium or with aluminium containing 10 per cent copper in amounts greater than about 1.5 to 2 per cent. Even a t the comparatively high temperature of 1370' to 1540' C. (2500' to 2800' F.) the solubility of tungsten in molten aluminium and aluminium-copper is limited to 2 per cent or less. When more tungsten (as a powder) than this is introduced into the molten metal, that exceeding the small amount going into solution tends to remain suspended in the melt. Since the specific gravity of tungsten is so much higher than aluminium or its light alloys, it segregates to the bottom of the pot. Even a small amount of tungsten (1 to 2 per cent) has a great influence on the melting point of aluminium or aluminium containing 10 per cent copper. It was necessary to pour Melt 9 at 1O1Oo C. (1850'F.). This is 2GO' C. (500'F.) above the usual pouring temperature for the light alloys of aluminium. The difficulty of getting tungsten into solution, together with its limited solubility in aluminium and very much higher melting point, are the principal obstacles in the way of successfully casting an aluminium-copper-tungsten alloy. On account of its limited solubility, it would be practically impossible to prepare a hardener containing 5 to 25 per cent tungsten; therefore the troubles in founding the alloy would be greatly increased. Negative results were obtained in attempting to use tungstic oxide, sodium tungstate, and ammonium tungstate as sources of tungsten. TABLE VII-CHEMICAL ANALYSIS OF ALLOYS --Tungsten-CopperAdded Found Added Found Melt Per cent Per cent Per cent Per cent RSMARKS 10.00 7.76 None Not de7.80 termined 7.60 5.00 1.75 1.80 1.69 1.67 10.00 2.78 None Not deRemelt of Melt 1 2.77 termined 2.74 20.00 20.00 This melt not analyzed 9.95 4.50 5.80 5.51 Remelt of Melts 1 2 9.80 5.55 and 4. The Gal: 9.86 ues under "Added" were calculated from the analysis on the melts as given above. Segregation readily shown 3.95 40.00 Not deter20.00 mined 3.95 3.94 8.00 N o t analyzed 8.00 0.93 16.00 Not deterRemelt of 7. Sam8.00 mined ples taken when 0.90 (1) metal was poured 3.86 from (1) top of pot 3.90 (2) 1.23 (2) middle of p o t 1.23 (3) (3)bottom of pot. Value shows segregation of tungsten through metal 1.00 1.30 10.00 9.80 The samples taken 1.30 9.72 as in Melt 8 1.33 9.87
STRUCTURE OB THE ALLOY-The structure of Melt 9, containing 1.30 per cent tungsten, 9.80 per cent copper, and balance aluminium, is shown in Figs. 2 to 5. Fig. 2 at 100 diameters and Fig. 3 a t 500 diameters show the structure as cast; and Fig. 4 at 100 and Fig. 5 at 500 diameters show the structure as heat-treated. It should be noted that the structure as cast and as heattreated is practically the same. It is evident that the heat treatment accorded to the specimen had little, if any, effect on the tungsten-bearing constituent.
20
INDUSTRIAL A N D EXGINEERINQ CHEMISTRY
The tungaten has formed a microscopic constitution previously unidentified, It is probably a eutectic of tungsten and copper (copper tungstide?) with aluminium, and assumes characteristic and easily identified forms. At 100 diameters (Figs. 2 and 4) it is grayish white and takes the form of filigree, stars, or flowers. These are shown in detail a t 500 diameters.
FIG. 9-Aveancs S ~ e u d v r aOII Tnmsvaase sKT*ohr--ns CAST. 100 x Etched 30 secoodein2Opveent aqueolls nitric acid at 70'C.. followed by quesching in cold water
Vol. 17, No. 1
Conclusions
Alloys of tungsten with aluminium and With duminium containing 10 per cent copper do not show any indication of being of great commercial importance. This is due principally to the difficulty of dloying tungsten with aluminium and
Fro 8 - A m a n o ~ SInu~l.unzos T B ~ ~ P S Y E R . Pie. B 4 - A v s ~ b c ~Srnunvna 0 s Tnmswnss SearoN--ns CAST. 5ao x Sscrrow-AS HSAT-TBBAIBD.1W X Etched 30 seconds in 20 per cent equeow Etched 80 seconds in 30 per cent aqueous nitric acid at 70"C., followed by quenching in 70-c , foliowed by quenching in nitric acid cold m f e i
(Figs. 3 and 5 ) The tungsten constituent is evidently very hard, as it stands sharply in reliaf after polishing. It is hirly wall distributed throughout the ground mass of aluminium and appa.rently has crystallized with the copper aluminium (Cu AL)-aluminiurn eutectic. ~ I i Y s r C A LPROPERTIES-The physical properties of the tWt bars from Melt 9 as cast and as heat-treated are given in Tahle 111. Each result is the average of three tests. On comparing the properties of the bars from Melt 9 with the average properties of a 10 per cent copper-balance aluminium alloy, it is evident that tungsten bas had but little influence on the ultimate strength and elougation. It hm, however, influenced the Brinell hardness slightly. The heatheated bars are about 10 per cent h a r d e r t h a n t h e corresponding alloy without tungstan. The a d d i t i o n a l hardness is probably due to the harduess of the Fxo. 6 - - A v m % ~ ST~ucruraOF Txmsvsase tungsten-be a r i n g Sacnor*-~s Hsnr-T~anso. 500 x constituent. Etched 30 semods in 20 per =est aqueous As heat treatniuie add at 70' C.. followed by quenching io ment did not affect cold water the solubility or character of the tungsten-bearing constituent, it wm to be expected that the ultimate strength and elongation would not be influenced to any great extent. The physical properties obtained at high temperature are noted in Table IV. Like the results obtained at room temperature, they show nothing exceptiona1. The results are about what would be expected from a 10 per cent copperbalance duminium alloy.
cold water
aluminium containing 10 per cent copper and to the troubles encountered in their founding. The tungsten, which probably forms a eutectic of copper tungstide with aluminium, is unaffeetcd by the customary heat treat.ment, and consequent.lyadds nothing to the value of the alloy, as measured by the ultimate strength or elongation. Tungsten evidently acts as a hardening element. It is entirely possible that, if a suitable heat treatment were discovered that would cause the solution of the tungsten-bearing constituent, the alloys would have properties that would make them of value in aircraft and automotive construction. Acknowledgment
Grateful acknowledgment is made to R. R. Moore, D. M. Warner, Clifford McMahon, and J. H. Heater. who aided in this investigation. References 1-Rnorom, Milo1 Ind., lo, 75 (19%). In e. paper before the American Iristitute of Chemical Engineers at Detroit, Ransom statesthat fherearealloys of aluminiam and tunmien which resist oxidation better than mpperelvmioiumalbyn. These _e used to some Extent in automobile pert8 rtnd for ship propellers. Neither the composition of tba alloy n m reference. were given. zGehirmu4ter, s o f i iU. ~ i ~ r as, n ,650,873,998(1913); J . 1 ~ ~ M 1 . ~ I ~ I ~ . 16, 240 (191B). Tungsten dissolves more readily in aluminium than chromium or mansailere. Tensile strength and hardness of the hot-rolled alIOY inereare s little at first, but become Constant at l per cent, the dvctility falliog dewly. 3-Borehei. "Aluminum." p. 181. Borcher could not Snd the aluminium-tuneten mmwuods (AIrW, AlWs AIW4 described by 0th- invcstigator,. lie states that when ingots of duniniwn-tnngsfen meitn were silowed to eool slowly, the tops of the ingofs were nearly tsngJten free, with the tungsten content rapidly increewiw =B the boitom wpe approached. 4--Guiliet, Cornpi. rend., 18%. 1112 (19011.
Rockefeller Foundation Establishes Institute of Physiology in Copenhagen-The RoCkefeller Foundation has approved the plan outlined by Danish physiologists for a large institute in Copenhagen. The Foundation has given 3 million Danish honer for the building on condition that the Danish Government will maintain and operate the institute. The buiIding is to contain five large laboratories for general physiology, biochemistry, animal physiology, gymnastic theory, and physicaA biology, with the necessary appmtenances of open-air aquarium, stables, bird-cages, ponds, etc.