Professional Documents
Culture Documents
Diffusion of Caron in Alpha Iron
Diffusion of Caron in Alpha Iron
C. G. HOMAN?
The diffusion of carbon in alpha iron using a tracer technique was performed in the temperature range
of 616°C to 844°C. 9 plot of the In D versus l/T in this temperature range deviates from the usual
straight line behaviour. An empirical model which includes the possibility of a carbon-vacancy interac-
tion is proposed to explain this deviation.
1. The diffusivity of carbon in alpha iron is assumes that the carbon atoms are the more mobile
independent of carbon concentration. defect, then equation (4) ma,y be written:
2. Equations derived from the model should be
dC E
consistent with changes of vacancy concentration -2 = Gv,C,C, exp - *g - C?,v,
at different temperatures. clt ( !
3. A good empirical fit be obtained for measured
diffusi~ities over the entire ~m~)erature range. x exp - (Em;, “) (5)
where C,, C, and C, are the pair, total carbon and B - E”,, - E,,
ex!? 112)
total vacancy concentrations, respectively. If one ! RT
HOMAN: DIFFUSION OF C IN ALPHA Fe 1073
A plot of In (11, - D,) vs. l/T together with the evacuated to about 2 ,u, radiomethane was ad-
literature value of Bc8) and Efvcg) should yield the mitted to a pressure of 0.5-4.0 cm of Hg and the
value of [SCv,/v,) D,,] and E,,. vessel was sealed. (The pressure of the methane
was varied to obtain different concentrations in the
EXPERIMENTAL PROCEDURE cylinders.) Carbon was deposited on the cylinder
A. Counting procedure surface by heating with an induction furnace to
approximately 500°C for 30 min. The temperat’ure
An end window (window thickness 1 mg/cm2)
of the cylinder was t)hen raised to 850°C for about
GM. tube was mounted as shown in Fig. 1. This
90 min.
fixture insured constant geometry for surface counting
After evacuating the gases remaining in the vessel,
sample ends. The counting port in the aluminum
the entire vessel and cylinders were placed in an
end shield was concentric with the samples and had a
annealing furnace at 750°C for 72 hr.(l”) The car-
diameter of 0.120 in. The V block held the samples
burized cylinders, which were bright and shiny in
for counting and grinding. The reproducibility of
appearance, were reweighed to determine the weight
counting geometry and the reliability of the counting
gain and counted to insure a uniform carbon dis-
equipment was checked periodically by means of
tribution. The uniform count rate of a cylinder was
chi-square tests.
between 50 and 200 cpm depending on the carburiza-
B. Sample preparation tion treatment. The diffusion couple was formed by
butt welding, in a helium atmosphere, a pure iron
Cylinders of 0.187 in. dia. and 18 in. long were
cylinder to a carburized cylinder (using a special
carefully machined from 99.98% zone refined iron
fixture and after suitably preparing the weld surfaces).
stock.oO) The cylinders were then etched in a solution
The weld was made by passing 3-5 pulses of 900
of 10 g of oxalic acid, 20 cc of 30 ‘A hydrogen peroxide
amperes through the sample under a small pressure
and 300 ml of H,O at 45°C until approximately
load. The time of a pulse never exceeded 5 sec.
0.020 in. were removed from the diameter and the
Metallographic examination of similar cylinders of
surface was very bright and shiny.
Armco iron welded with the same procedure indicat,ed
The cylinders to be carburized were weighed on a
good welds with no discernible entrapments at the
Mettler Grammatic microbalance and placed in a
interface.
quartz carburizing vessel. After the vessel was
Since this diffusion experiment will be extended to
high pressures, a new technique to encapsulate the
sample was developed.03) Because (1) the solubility
of Carbon in Cu is low,(14) (2) the diffusivity of Cu in
a-Fe is low with respect to the diffusivity of C in a-
Feo5) and (3) Cu is an extremely good “getter” of O,,
it was decided to plate the diffusion couples with
0.003 in. of Cu. After a diffusion anneal, it was possible
to remove part of the copper oxide layer to obtain
a bright Cu color. The solubility of C in Cu was
GM END qualitatively investigated by a counting technique.
.. ..__..
‘WIN”C)ul
COUNTING C. Diflusion anneal
A temperature control over the sample length of
05°C at 725°C during a diffusion anneal was possible
by means of a diffusion furnace and controller. The
GM TUBE
SUPPORT temperature fluctuations during the anneal were
measured by means of a chromel-alumel thermocouple
imbedded in the Cu heat bath of the diffusion furnace.
ENDWINDOW
SHIELD AND The thermocouple potential was measured with a
COUNTING PORT L&N K-3 potentiometer against a cold junction of
V-BLOCK glycerine in a dewar. The chromel-alumel couple
was calibrated before and after a diffusion run with
a Leeds and Northrup standard Pt-Pt 10% Rh
FIG. 1. Schematic of counting assembly. thermocouple using an ice bath cold junction. The
1074 ACTA METALLURGICA, VOL. 12, 1964
temperature of the cold junction was measured sections counted for at least 10,000 counts; and for
with a ~alibrate~l thermometer. sample #6, every section was counted for at least
The sample was inserted into the furnace with a 3,000 counts. Background measurements were made
chrome]-alumel thermocouple welded to the low periodically during the counting procedure and varied
activity side of the couple. Measurements of the between 23.4 and 24.3 over several months of counting.
heating time of this thermocouple as compared to the A summary of experimental data has been tabulated
furnace thermocouple indicated that a maximum in Table #I.
time of 1 minute was needed to heat from 0.8 T, to TABLE 1. Diffusion annealing data
z.-. ._ - -.
TD for diffusion runs lasting from 60 to 90 min.
tn
After the anneal, the sample was quenched within Couple no. &vt.y$J a0
iwm) (%! x 10s sec.
10 set to R.T. The data has not been corrected for 1 0.022 191 721.0 f 0.3 4.56
the heat-up time. 1A 0.022 191 721.0 f 0.3 4.56
2A 0.021 184 732.0 * 0.3 4.20
2B 0.021 184 732.0 * 0.3 4.20
3A 0.009 80 730.9 & 0.4 3.60
5 0.006 55 814.9 + 1.0 3.54
The sample was mounted in the V block and the 6 0.006 53 823.5 & 1.0 4.08
Cu plating was removed from the end with a precision 7 0.006 52 616.0 * 0.3 5.40
10 0.008 209 844.2 i 1.0 3.60
surfacegrinder. The grinder has an accuracy of -
0.0001 in. and produced a #lS finish on the sample
RESULTS
end. Sectioning and counting from the high carbon
(activity) end showed that the original cpm measured A. Measurement of Dexp.
before the anneal could be obtained within 0.005 in. Figure 2 is a typical penetration curve which was
of the end. This result qualitatively confirmed the obtained from the measurements of the activity of
fact of the extremely low solubility of C in Cu. At Sample #l after a diffusion anneal of 76 min at
least three sections were made to within 0.2 in. of 721.0%. If one plots
the weld to be sure t,hat the concentration in the
(erf)-r %
~ - ‘b
carburized portion of the sample was uniform after vs. distance
the anneal. All samples reported in this paper were a0 - %
within 1 “/;; of the original value of a0 determined from the interface on probability paper, the data may
before the anneal in this region. The penetration was be fit with a straight line. Results of this experiment,
measured from this point. For samples #I & #lo, some of which are plotted in Figs. 3-8 indicates that
every section was counted for at least 10,000 counts; in the ranges of carbon concentration and temperature
samples fA, 2A. 3A, 5 and 7 have at least three st,udied the diffusivity is independent of carbon
HOMAN: DIFFUSION OF C IN THE ALPHA Fe 1075
99.99
98 9e 98 9.9
95 95 95 95
ox-(lb
yo 90 ox-Ob 90 so
(lo-(lb
(lo-oh e. so 80 80
70 70 70 70
60 60
60 60
50 50
50 50
40 40 40 40
30 30
30 30
20 20
20 20
10 10
10 10
5 5
5 5
2 2
2 2
-60
-30.0 -15.0 0.0 I5.0 30.0 -40 -20 0.0 t20
DISTANCE FROM WELD (1631N.)
DISTANCE FROM INTERFACE ( lu3 IN.)
FIG. 3. Probability plot of (a, - ab)/(q, - a,) vs. distance FIG. 5. Probability plot of (a, - ab)/(rzO - ub) vs. distance
from weld: sample #lA. from weld: sample #5.
99.99 99.99 99 99
9999
DC# 3A DC* 6
To=730.9+.4”C To =823.5 i I.O°C
to=3.6Xl03sec 1D =4.OBXlO3sec
Cl4 IN w-Fe Ci41N d-Fe
D = Ii 6 X10-’ cm2/sec D = 4 7.1 X 10-7cmz/sec
98 96 98 98
95 95 95 95
90 90 90
-90
80
armgo
a*-Ob 80 80
oa-Ob ‘0 70 70 70
60 60 60
60
50 M 50
50
40 40 ‘la 40
30 30 30 30
20 20 20 20
IO
10 IO 10
5t-+ 5 5
2
5
2
2
-80 -60 -40 -20 0.0 +10 -80 -60 -40 -20 0.0 +10
1
99 99 99 99 99.99
99.99
DC* 7
To =616.0-+0.5°C =644.2f I.O’C
to=5.4x103 set
C’41Noc-Fe
D= 2.36X IO-‘cm hec
D =62.0 X 10-7cm%ec
98 98
98
95
95
90 90
EO
70
O_ 8o
ao-(lb ‘0
60 60
50 50
40 40
30 30
20 20
10 IO
5 5 I I I I 5
I*
2
Fro. 7. Probability plot of (a, - aa)/(ao - a*) vs. distance FIG. 8. Probability plot, of (a, - ab)/(aO - (lb) vs. distance
from weld: sample #7. from weld: sample # 10.
concentration. Table #2 is a summary of all the pre-exponential terms to be 2.2 cm2/sec and the
experimental values of the diffusivities determined activation energy (B - Efv - E,,) to be 29,300
in this experiment. Cal/mole.
However, a plot of the diffusivity of carbon in DISCUSSION
alpha iron measured in this and other(3s4) experiments
A. Comparisovb with the results of other measurements
over the temperature range studied is not of the
Comparison of the D, values obtained in this experi-
simple exponential form as shown in Fig. 9. In order
ment with previously reported D values(314) indicate
to study this effect a subsidiary plot was made of
good agreement in the temperature range 616”C-
the measured diffusivities shown in Fig. 9 minus the
corresponding extrapolated diffusivities D, from low 844°C. The quantitative agreement of this data with
temperature anelastic measurements.@) Figure 10 Smith’s14) work using nonradioactive carbon indicate
indicates that two competing mechanisms are in that an isotope effect, if any, is not detectable. It
operation. Rewriting equation (12) would be expected that an isotope effect would yield
values of diffusivities for C-14 approximately 7 x,
(D, - D,) == 6 ii D,, exp
B - E,, - En%, smaller than for the nonradioactive carbon.
(13)
RT Most of the low temperature diffusion measure-
Hence from Fig. 10, one may find the value of the ments involve a quenching technique in the sample
TABLE 3. Summary of diffusion calculations
to X* c, - cbt 11
Couple no. x 103 SW (in.) erf l#J 4 lo-’ cm2/sec
c0 - cb
the diffusivities of hydrogen in nickel and iron@O) FIG. 10. Plot of De,,, - D, vs. 103/ToK for carbon in
X
= erf I___ (19)
21/tD,,t
or
D exp I D,,t (20)
(15)
under the conditions of this experiment.
The room pressure values are indicated by the sub- This effect has been investigated by redetermining
script 0. Integrating one obtains the diffusivity of sample #2B looking from the low
activity side. In this counting situation
= D, + yoD,, exp -
D,(P. 117)
D exp 2 Dact (21)