INDUSTRIAL A N D ENGINEERING CHEMISTRY
344
Literature Cited (1) (2) (3) (4) (5) (6) (7) (8) (9)
Alyea and BackstrBm, J . A m . Chem. SOL.,61, 90 (1929). Anderson and Moore, Ibid., 4S, 1944 (1923). Bond, Brit. M e d . J . , 1927, No. 3483, 637. Eibner and Pallauf, Chem. Umschau, 32, 81, 97 (1925); through C. A , , 19, 2420 (1925). Engler and Weissberg, Ber., SI, 3046 (18981, and later papers. Greenbank and Holm, IND. EKG. CHBM.,17, 625 (1925). Greenbank and Holm, I b i d . . Anal. Ed., 2, 9 (1930). Holm, Greenbank, and Deysher, IND.END. CHEM.,19, 156 (1927). Kerr and Sorber, I b i d . , 16, 383 (1923).
Vol. 22, No 4
(10) Kerr and Sorber, J . Assocn. Oficial Agr. Chem., 8, 90 (1924). (11) Lewkowitsch, “Chemical Technology and Analysis of Oils, Fats, and Waxes,“ Vol. I, p. 437. (12) Mattill, J . A m . M e d . Assocn., 89, 1505 (1927). 31 ‘13 (1926). (13) and Dufraisse8 (14) Powick, J . Agr. Research, 26, 323 (1923). (15) Triebold, Thesis in Agricultural Biochemistry, University of Minnesota, 1926 (16) Tschirch, Ckem. Umschau, 32, 29 (1925); through C. A , , 19, 1557 (1925). (17) Woodman, “Food Analysis,” p. 167 (1924).
Volatilization of Phosphorus from Phosphate Rock Robt. D. Pike 4069 HOLLIS ST., EYRRYVILLR, CALIF.
11-Experiments in Volatilization of Phosphorus and Potash in a Blast Furnace1 A detailed description is given of two continuous runs of an experimental blast furnace supplied with an oxygen-enriched blast. In one run the furnace charge consisted of phosphate rock, a siliceous flux, and coke; and in the second run the siliceous flux was replaced by a potash-rich flux. The average extraction of PzOswas about 70 per cent, and of K,O, 47 per cent. Further experimental work with the slags on a small scale led to the conclusion that, with a proper furnace design and correctly proportioned slags, extractions of 97 and 92 per cent of P106and KzO, respectively, can reasonably be expected. A good slag to use closely resembles Equipment
CROSS SECTION of the experimental furnace is shown in Figure 1. The general arrangement of apparatus is shown in Figure 2. The breast of the furnace with a 200-pound slag buggy in place is shown in Figure 3. The phosphate rock, coke, and flux were charged by hand through a door in the top of the furnace and the shaft was kept approximately full of charge. This left a large empty chamber a t the top into which air for combustion of phosphorus and carbon monoxide was blown in two semi-tangential streams. Sufficient air was introduced to burn all the carbon monoxide to carbon dioxide and all the phosphorus to phorphorus pentoxide, That portion of the top gases which went to the precipitator was humidified by the introduction of a little steam. The major part of the gas passed up through a stack issuing from the top as a dense cloud of phosphoric acid. (Figure 4) The flow of oxygen and air to the blast was accurately measured by orifice meters and controlled from a central control board. It will be noted that, with the exception of the enlarged top and the central tuy&re, the interior lines of the furnace are of conventional type. The tuysre is a long, water-cooled copper tube extending downwardly along the central axis of the furnace with a four-holed nozzle a t the bottom. The blast of oxygenated air was introduced a t high velocity through these nozzles and a t an angle of about 30 degrees below the horizontal. This type of tuykre was originally designed for the use of highly oxygenated blasts, the idea being to keep the intense heat of combustion away from the walls of the furnace. It was successfully employed with a blast of pure oxygen, but when a cold blast containing 40 to 45 per cent of oxygen by weight was adopted, it was found that
A
1 Received
December 16, 1929.
that employed in an iron blast furnace, and free-running properties should be the chief consideration in determining its composition. The slag should be tapped at from 1450 O to 1500 O C. Under strongly reducing conditions phosphorus will be almost completely evolved from slags of a basic nature similar to those employed in iron blast furnaces, and relatively basic slags are required for the high extraction by volatilization of potash. Blast furnaces operated along the lines indicated by the experiment with hot oxygen-enriched blast for the joint volatilization of P201 and KzO may be expected to operate smoothly and continuously.
this tuy6re was faulty and caused the only persistent operating difficulty that was encountered. This difficulty arose from the fact that as the cold blast entered the heart of the furnace it chilled the dag immediately below it, forming a frozen crust of slag. It was therefore almost invariably necessary to tap the furnace with an oxygen lance, and the top of the crust gradually arose until, a t the end of about 5 days’ continuous operation, the hearth was completely filled with solid slag and a tap could only be effected by burning a hole with the oxygen lance through into the bottom of the bosh. Obviously this operating difficulty pertains to the type of tuy&rc.employed and to the cold blast and not to the process itself. By employing the usual form of side tuv&reswith hot blast containing 30 per cent of oxygen by weight, no trouble of this kind need be expected, nor should there be any more than ordinary difficulties of operation which pertain to the iron blast furnace. I n considering the extractions actually obtained in this experiment, one should give due consideration to the influence of the difficulties of operation which resulted from the type of tuyPre employed. Obviously these were of such a nature as to impair the normal working of the furnace and to reduce the extraction of phosphorus and potash which might otherwise have been reasonably expected. Materials Used
For the purpose of this paper the description will be confined to two continuous runs of 31.5 and 21 hours, respectively. In the former, which will be called run 1, the fluxing material was low in potash and was made up of a mixture of Green River sandstone and firebrick grog; and in the latter, or run 2, Wyomingite was employed as flux. I n other respects the two runs were substantially the same. The analyses of the materials used are given in Table I.
INDUSTRIAL A N D ENGINEERING CHEMISTRY
April, 1930 MATERIAL
PzO
Phosphaterock Green River sandstone Firebrick grog Coke Wyomingite
30 2
%
T a b l e I-Materials Used K I O NazO Si02 A1203 FeO
7;
%
0 33 0 . 6 5
R
R
4.2
1.1
..
15.77 28.92 3.17 13.8
4.8 2.0 1.55 5.3
. . . 2.00 6.17 58.5 0.20 . . 67.7 0:60 0 . 3 0 0 . 1 1 4 . 3 3 2 . 2 11.57 0 . 7 5 5 1 . 4
ASH
CaO MgO
%
%
%
%
45.2 0 . 9 8 92.18 3.1 0.55 1.16 5.6
3.2 0.43 0.15 7.6
93.4 100.0 11.39 98.40
The materials were carefully sized. The largest particle of phosphate rock would pass a 2-inch ring. Its screen analysis follows: Pcr cent 70
Mesh +2 +3 +4
12 2
Mesh + 6 +10 10
to Blast F u r n a c e R U N1 Pounds
Phosphate rock Sandstone Firebrick grog Wyomingite
100 50 20
Total, carbon-free Coke
170 97.1
P10s in charge P10s in carbon-free charge
T a b l e IV-Temperatures DATE
2 2 12
6/27/21
in T o p of F u r n a c e Six I n c h e s below Top of B u r d e n TIME
c.
850 772
5:30 P. M . 8:30 P. M. 12315 A. M. 12:40 A. M . 12:45 A. M. 9:oo A . IS. 1o:oo P. M . 2:20 A M.
6/23/27 a
TEMPERATURE
6250
755 813b 765 722 723
Just after charging.
b Just before charging.
R C N2 Pozcnds 100
80 180 107
---
Total charge
IV) .
6/26/27
The coke was sized between a 21/4-inch ring and I-inch mesh. The sandstone and Wyomingite, both relatively hard materials, were sized the same as the coke. The firebrick grog was sized between 11/2 inches and l/2 inch. The coke was charged separately in lots of 1 5 to 20 pounds followed immediately by a corresponding quantity of the other ingredients mixed together. The charges are given in Table 11. T a b l e 11--Charges MATERIAL
slag temperatures could have been maintained betweeii 1450" and 1500" C., as would be the case in a large properly designed blast furnace, volatilization of both phosphorus and potash would have been substantially complete. Temperatures were measured in the furnace top by inserting a thermocouple 6 inches under the top of the burden (Table
P e r ceul
-
345
~
267.1 Per cent 11.5 1 s . s5
287 Per cent 11.4 1S.iO
Operating Procedure and Data
A rQuni6 of the principal features of the two runs is given in Table 111. T a b l e 111-Operating D a t a R C N1 31 5 4580
Length of run, hours Total charge, pounds Coke charged, pounds 2618 Slag out, pounds 3785 Number of taps of slag 22 .4verage weight per tap 172 .inalysis of composite:" P?O6 6 . 55 (fusion) Si0 37.90 hl2OJ 1 4 . S5 CaO 36.20 FeO 2.35 MgO 1.10 Kz 0 0.62 Nag0 1.66 Oxygen in blast, per cent by weight 41.0 Slag, pounds per pound PzO6 4.44 Coke, pounds per pound PzO:. charged 3.16 Oxygen from cylinders, pounds per pound PnOs charged 2.20 SlagZemperature read direct by optical pyrometer, C.: 1385 Average 1452 Maximum 1327 Minimum Cooling water to tuyi.re, p o u ~ l d zper 111 minute 14,4 Temperature in, Temperature out: ' k . 30 a Sample of slag made up from samples from each tap.
Ruv 2 21 2760 1641 2345 15 156 4 7 35 (fusionj 35 so 10 40 35 15 2.80 4.25 3.57 0.29 45.0 4.51 3.30 2 76 1360 1388 1308 111 14.4 30
This table brings out very clearly the general trend to more unfavorable conditions toward the latter part of the operation, brought about by the filling up of the hearth with frozen dag, as explained above. In run 2 the average slag tapped is less, average slag temperature is lower, and average volatilization of phosphorus is lower, as evidenced by a higher P205 content in the slag; a t the same time the percentage of oxygen in the blast in run 2 is greater than in run 1. The experiments described below on volatilization of phosphorus and potash from the slag leave little doubt that, if
F i g u r e 1-Cross
'
f/ow Section of E x p e r i m e n t a l Blast F u r n a c e
Samples of the top gas were taken through a silica tube and analyzed in an Orsat apparatus (Table V). The silica tube was inserted an inch or two under the top of the charge and the phosphorus combustion air was shut off while the sample was being taken. T a b l e V-Analyses of S a m p l e s of F u r n a c e Top G a s (Per cent by volume) DATE 6/27/27 6/28/27
TIME
co2
0 2
co
Nz
Per cent Per cent Per cent Per cent 1:15A.M. 10.6 0.7 39.2 49.5 9:OOA.M. 6.8 0.1 53.1 40.0 12:OO Mid. 5.1 0.1 52.8 42.0 4:05~.M. 9.6 0.5 46.7 43.2
OXYGEN
BLAST Per cent 40 45 45 4.5
When the gas sample was first taken from the top of the shaft, the phosphorus combustion air was not shut off. This
INDUSTRIAL AND ENGINEERING CHEMISTRY
346
sample showed a very high carbon dioxide content, which disappeared when the air was shut off, proving that the combustion air, which was introduced into the furnace top under a considerable pressure, penetrated downwardly into the charge. I n an earlier run of the furnace, not recorded here, an attempt was made to blow the phosphorus combustion air into the charge below its surface, in order to render available in the shaft of the furnace some of the latent heat of the top gas, but this had caused an accretion to form in the charge where the air entered, which soon held up the burden. After run 2 when the shaft was emptied, the depth of burden above the tuyt3-e ports was 63 inches and for 26 inches down from the top of the shaft there were solid accretions on the wall extending 6 inches out in some cases. It seems evident that these accretions were caused by oxidizing conditions due to the penetration of the phosphorus combustion air into the charge, and that it is necessary to maintain strongly reducing conditions in the shaft in order to keep the walls free from accretions of phosphates. An analysis of this accretion follows: Per cent 25.5 28.2 5.85 3.15
PZOS(fusion)
SiOz
A1203
FeO
CaO
Kz0
NazO
Vol. 22, No. 4
Likewise, m [l - (m‘ m[l
+ n’ + + b ‘ )1 - m’A 0’
- (m’ + n’ + 0’ + p ’ ) ]
- (m’ + e’ + 0’
X 100 =
% total extraction of FeO
X 100 =
% total extraction of KzO
0[1
+ b’)]- o’A o[l - (m’ + n’ + 0’ + p ’ ) ]
p[1
- (m’ + n’ + 0’ + +’)I
P [ 1 - (m’
- b’A
+ n’ + + P’)I 0‘
X 100 =
% total extraction of NazO
The PZOSwhich goes to ferrophosphorus is the equivalent attached to the iron equivalent of the extracted FeO. Therefore, the equivalent percectage of P?Oawhich went t o ferrophosphorus containing 15 per cent phosphorus and 85 per cent iron is m[l
- (w’ + n‘ + 0’ + P’)lnt’A n(m’
+ n’ + + p’) 0’
31,5
The above formulw do not take into account the addition of any material from the walls of the furnace. This is not a factor in the operation of a large furnace, but is likely to be of some importance in a small furnace.
Per cent 29.40 1 84 1.68
Calculation f o r Run 1
A typical analysis of the acid collected during run 2 follows:
The percentage of materials in the carbon-free charge (Table 11) is as follows: Per cent 31.37 10.90 31.20 1.73 2.72 18.85 0.89 2.34
SiOz AI203
CaO
MgO
FeO PSOK KzO
Extractions
METHODOF CALCULATION-The extractions were determined by comparing the analysis of the charge with that of the slag. Extractions can be roughly estimated almost a t a glance from the data given in the tables, but it is desirable to employ a precise method for these calculations. Let A
= m’ = n’ = 0‘ = p’ =
A
(m‘
+ n’ + + p‘) 0’
The weight of P105 in the slag is n’A
1
-
(m’
+ n’ + + p ’ ) 0‘
+ + + +
The sum of A vi n o p = 1. Therefore, the weight of PzO5 in the original charges is n and total extraction of PzOsas per cent of total is n -
n‘A
1 - (m’
+ n’ + o‘ + p‘) n
or
-
+
n[l (m’ n’ S o ’ f p ’ ) ] - n‘A n [ l - (w‘ n‘ 0’ pol x 100 = % total extraction of PzOa
+ + +
n
o
p
p‘ = 0.0166 0.1118
Total extraction PZOS:
-
0.1885 (1 - 0.1118) 0.0655 X 0.7520 0.1885 (1 - 0.1118)
=
~ o , ~ %
Total extraction K20: 0.0089 (1 - 0,1118) - 0.0062 X 0.7520 X 100 = 41.0% 0.008Y (1 - O.ili8) Total extraction Na20: 0.0234 (1 - 0.1118) - 0.0166 X 0.7520 0.0234 (1 - 0.1118)
-
m
m‘ = 0.0235
If all iron is extracted as ferrophosphorus and all of PzO~, KzO, and NazO are liberated by volatilization, the weight of slag resulting from 100 pounds of carbon-free charge is A .
1
0.0272 = 0.1885 = 0.0089 = 0.0234 =
E‘‘ == 0,0655 0.0062
Altoa, CaO, MgO, in the carbon-free charge per cent FeO in carbon-free charge/100 per cent P z O in ~ carbon-free charge/100 per cent K20in carbon-free charge/100 per cent N a 2 0 in carbon-free charge/100 per cent FeO in slag/100 per cent P206in slag/100 per cent KtO in slag/100 per cent NaZO in slag/100
But the actual weight of the slag is
0.7520 = A
From Table 111, the figures for the slag are
= sum of percentages/100 of “permanent” oxides Si02,
m = n = o =
p
NazO
Per cent/100
= 39.90Jo
PzObequivalent which goes to ferrophosphorus: 0.0272 (1 - 0.1118) - 0.0235 X 0.7520 31,5 = 1,22%, 0.1885 (1 - 0.1118) Slag P206 charged: _ per _ pound 0.7520 = 4.55 pounds 0.1885 (1 - 0.1118)
Calculation f o r Run 2 Total extraction P 2 0 5 : 0.1870 (1 - 0.1401) - 0.0735 X 0.7242 = 66.8970 0.1870 (1 - 0.1401) Total extraction KzO: 0.0566 (1 - 0.1401) - 0.0357 X 0.7242 loo = 46,9QJo 0.0566 (1 - 0.1401) Total extraction NazO: 0.0078 (1 - 0.1401) - 0.0029 X 0.7242 = 68.6% 0.0078 (1 - 0.1401) Per cent of P z O equivalent ~ which goes to ferrophosphorus:
INDUSTRIAL AND ENGINEERING CHEMISTRY
April, 1930 0.0244 (1 - 0.1401)
- 0.028 X
0.7242
0.1870 (1 -- 0.1401)
was 1530' C., but as calibration of the disappearing-filament optical pyrometer against the melting point of electrolytic iron showed only a very small temperature correction, the temperature was reduced to 1470' C., since there was a greater attack on the crucibles a t the higher temperatures. In all the experiments the slag was completely melted and some coke remained on its surface at the end.
31,5 = Nil
0.7242 = 4.51 pounds 0.1870 (1 - 0.1401)
Slag per pound Pa06 charged:
Effect of Time, Temperature, and Composition of Slag on Volatilization of P20b,KzO, and NazO
As has been pointed out, the normal, orderly accumulation of slag in the crucible of the furnace did not occur in either run, because of the chilling effect of the cold central blast. Furthermore, the slag was flushed every hour and a half and the temperature was low compared with standard blastfurnace practice. Normal slag temperatures in blast furnaces run around 1525' C. ( d ) , which is higher than the temperatures shown in Table 111. I n a commercial furnace the time between fluches can be extended to from 3 to 6 hours as compared with 11/2 hours in the experimental furnace. It is therefore apparent that in experimental practice important factors tended to reduce volatilization, for cwtainly it T o d d seem that both P r o 5 and KrO would continue to volatilize to some extent from a pool of moltell slag a t a temperature around 1500° C. and the Pz05 would be assisted in such volatilization by maintaining contact with solid carbon a t the surface of the slag pool. Regarding the volatilization of K20 in blast-furnace practice, Wysor (5) brings out the essential facts that of 22.4 pounds of KzO charged per ton of pig, 17.9 pounds are volatilized and the slag contains 0.45 per cent KzO. On the other hand, the work of Ross and Merz (3) shows 13.1 pounds of KzO charged and only 4.9 pounds volatilized as an average for the United States. If slag is placed in a graphite crucible together withsome coke, and the crucible is covered, and placed in the DeVille furnace and heated to temperatures corresponding to blastfurnace temperatures for various lengths of time, the slag will be subjected to conditions similar to those that surround it in the crucible of the standard blast furnace which is being operated under strongly reducing conditions. Accordingly a sample of slag was selected which had been produced during run 2, and which analyzed as follows: Per
lent
FeO
38 i 10 35 31.60 4 50
SI02
AlzOa CaO MgO
PzOs (fusion) Kz0
Nan0
Time, min. Av. temp.,
a
C.
I 1
Kz0
Figure 2-Set-Up for Blast Furnace, Run 2
The experiments are divided into four main groups: (1) original slag; (2) original slag plus 8 per cent CaO; (3) original slag plus 8 per cent synthetic calcined dolomite; (4)original slag plus 16 per cent CaO. The results are given in Table VI. The volatilization of Pz05 is in all cases between 80 and 90 per cent of the P r o r in the slag used in the experiment, regardless of time, indicating that it is dependent on factors other than chemical composition of the slag or time of heating in excess of 11/2 hours. It is noteworthy that this applies to the strongly basic slag of group (4). Evidently the solid carbon has been instrumental in the volatilization of PzO6 by first reducing it to phosphorus. Fluidity of slag, regardless of composition, is no doubt of importance. The volatilization of KzO increases with the basicity of the slag and time of heating, but in group (4),where the basicity is highest, extraction is almost independent of time after l l / z hours. From the practical standpoint two facts stand out. The composition of the slag of group (4), which shows the greatest extraction of KZO, is practically the same as the blast-furnace slag having the lowest free-running temperature according to Johnson (1) and is therefore an ideal slag for blast-furnace practice. The composition of this slag is close enough to that of run 2, Table 111, to justify calling them the same for
%:ACTION5
%
1800 1538
%
(')
90 1473
180 1471
270 1472
%
%
%
2.85
4.15
SLAG
I
-k '%
120 1520
90 1472
180 1471
%
%
70
270 1472
%
1.61 1.50 1.43 1.07 0 . 5 8 None None 0.40 0.80 LO5 0.80 0.70 39.8 40.5 40.2
...
3.4
3.4
3.3
...
48.3 73.6 87.3
55.3 28.7 87.3
63.9 43.0 90.5
67.7 41.3 87.2
2.85
2.55
2.45
i I
+
(4) SLAG
120 1532
90 1472
%
I
NazO 1
43.4 24.3 80.0
70.5 82.4 82.6
a Bottom of crucible recovered and part of slag button.
-
69.9 71.3 1 0 0 . 0 100.0 82.5 87.2
78.5 59.6 88.7
+ 16%
(3) SLAG 8% CALCINED DOLOMITE 90 1471
180 1472
%
%
1 . 6 0 1 . 6 0 1:40 0.27 0.40 0 . 4 5 0 85 1.1 0.85 39.80 39.7 37.5 2.75
2.45
1.9
FROM SLAG:
K2O
PZOK
of Time, Temperature, and Composition of Slag on Volatilization of L O , NazO, and PzOj
3.03 1 . 5 9 2 . 7 8 2 . 4 0 1.194 0 . 8 1 0 . 1 9 0 . 2 8 0 . 7 6 0.60 1.35 1.15 0.85 0.85 0.65 41.6 42.8 41.9 42.15 41.6
NazO PnOs (fusion) Si02 AlzOa CaO FeO
Furnoce
fiL