KINETICS O F THE FLUORINATION
267 1
O F SILICON AND B O R O X
Kinetics of Reaction of Elemental Fluorine.
111.
Fluorination of
Silicon and Boron'
by A. K. Kuriakose and J. L. Margrave Department of Chemistry, Rice University, Houston, Texas
(Reeeiced M a y 7 , 1964)
The kinetics of fluorination of silicon between 75 and 900" and of boron between 300 and 995" have been investigated a t fluorine partial pressures of 2.8-52.5 torr in helium, using a gravimetric method. Both reactions follow a linear rate law under all the conditions studied although Arrhenius plots of the results indicate rather sharp decreases in activation energies a t 150 and 3-52" for silicon and boron, respectively. The activation energies below the transition points are 12 kcal./inole for silicon and 37.9 kcal./niole for boron and those above are only 1.3 and 2.5 kcal./mole, respectively. Based on these values, it is proposed that an adsorption process controls the reaction rates in the low temperature regions, while gas phase diffusion becomes rate determining in the high temperature region. The reaction rate of silicon with fluorine is proportional to the 0.6-1.0 power of the fluorine partial pressure, while for boron the rate is first order. The transition temperatures are essentially independent of fluorine pressure, when corrected for the accompanying larger surface temperature rises at higher pressures.
Introduction Xot much is known regarding the interaction of fluorine with silicon and boron except for the early observations of Rloissan2 that these materials burn in fluorine a t or slightly above room temperature depending on their physical state, with the production of the corresponding gaseous fluorides. Wise, et al.,3 determined the heats of formation of BFJ(g) and Si4F(g) by direct fluorination of boron and silicon. The amorphous forms react faster than the crystalline ones, and silicon reacts more readily than boron. Although these findings are true, spontaneous ignition of these elements does not occur if the fluorine concentration is limited by reducing the partial pressure appropriately; and under such controlled conditions, the rates of their reactions can be measured by ;suitable means. This paper presents a study of the kinetics of the fluorination of silicon and boron a t teiiiperatures up to about 1000" and fluorine partial pressures up to 52.5 torr, using a thermogravimetric method in a continuous flow system. Experimental
Apparatus. A detailed description of the apparatus used and experimental procedure have been given else-
where.4 The fluorination reactions were carried out in a nickel combustion tube equipped with a quartz spring from which the specimens could be suspended by means of fine nickel wire. Materials. Fluorine gas of 99.8+% purity obtained froni the Allied Chemical Corp. was used after passing it through an H F trap containing NaF pellets. Polycrystalline boron in the form of rods on 12-p diameter tungsten substrate filaments was obtained from Texaco Experiment Inc. Cylindrical specimens, approxiniately 1 cni. in length and 0.2 cin. in diameter, were made and washed with acetone and benzene and mere dried. By calculation froni the geometrical dimensions, the boron samples had a density of 2.24 f 0.06 g . / c n ~ ~ . The silicon samples were prepared in the f o r m of thin slices of approximately 1-nim. thickness cut from
(1) Presented at the 147th National AMeeting of the Ameriran Chemical Society, Philadelphia, Pa., April, 1964. (2) (a) H. Moissan, Compt. Rend., 139, 711 (1904); (b) Ann. C h i m . Phys., [SI 24, 237 (1891). (3) S. Wise, H . Feder, W. Hubbard, and J. Margrave, J . Phys. Chem., 6 5 , 2157 (1961); 66, 381 (1962); 67, 815 (1963). (4) (a) A. K. Kuriakose and J. L. Margrave, ibid., 68, 290 (1964); (b) 68, 2343 (1964).
V o l u m e 68, N u m b e r 9
September, 1968
2672
A. K. KURIAKOSE AND J. L. MARGRAVE
large crystals of silicon of purity 99.9+% supplied by E. I. du Pont de Nemours, lnc. The specimens had a surface area of about 1.5 cni.2 and a calculated density of 2.1 + 0.1 g./cm.3. The surface areas of both boron and silicon were computed from the measured geometrical dimensions. It should be mentioned that the boron surface was obviously rough so that the calculated value could be in error. The silicon had a smooth shiny surface.
Results As stated in the Introduction, since both silicon and boron ignite spontaneously in pure fluorine around room temperature, the kinetic experiments were cdrried out a t very low fluorine partial pressures. With silicon samples of surface area about 1.5 cm.2, no measurable reaction was observed for a period of 1 hr. below a temperature of about 8 5 O , with a fluorine partial pressure of 2.8 torr. Boron (with a surface area of approximately 0.7 cm.z) was even more resistant to fluorine under the same partial pressure conditions and measurable reaction was observed only above 300'. When the fluorine partial pressure is increased, specimens of both silicon and boron break up a t around room temperature a t an experimental fluorine partial pressure of about 200 torr. The lower reactivity of boron than silicon with fluorine is in agreement with the observations of NIoissan. The Eflect of Temperature. The results of the fluorination of silicon and boron are given in Tables I and 11, respectively. The runs in which the reactions were very fast were made only for short periods so that surface area changes did not become appreciable. As ex-
pected, the reactions are linear and the rate constants were calculated from a plot of the wt. loss per cmS2 against time. Table I1 : Rate D a t a for the Fluorination of Boron kl,
kl1
Temp.,
PF,,
OC.
torr
300 310 320 336 336 336 396 396 396 500
2.8 2.8 2.8 2.8 13.2 52.5 2.8 13.2 52.5 2.8
PF,,
mg./om.l/ min.
Temp., OC.
torr
0.029 0.063 0.113 0.271
500 500
13.2 30.6 52.5 2.8 2.8 2.8
0.800 3.900 0.581 1.576 4.940 0,645
500 608 683 765 870 970 995
2.8 2.8 2.8
mg./om.z/ min.
1.630 3.210 5.279 0.975 0.990 1,325 1.237 1.389 1.250
Figures 1 and 2 are Arrhenius plots for the reactions of silicon and boron, respectively; and there is a clear change in the nature of temperature dependence between the low and high temperature regions. For silicon there are two sets of lines corresponding to fluorine partial pressures of 2.8 and 13.2torr. The solid lines are not exactly parallel to each other in the low temperature regions and the calculated activation energies are, respectively, 15.7 and 18.7 kcal./mole a t 2.8 and 13.2 torr although the upper portions of the lines yield a value of 1.3 kcal./mole for both pressures. For boron, corresponding to the two portions of the line, the activation energies are 42.7 and 2.7 kcal./mole. Temperature, *C
Table I : Rate Data for the Reaction
1000800 I
between Silicon and Fluorine kl,
kl,
Temp.,
PF,,
OC.
torr
75 85 85 95 95 104 120 120 120 120 150 170 170 170 170
13 2
2 8 13 2 2 8 13 2 2 8 2 8 13 2 30 6 52 5 2 8 2 8 6 7 13 2 30 6
rng./cm.z/ min.
Temp.,
0,066 0.053 0.117 0.100 0.363 0.177 0,341 0.987 2.475 4.025 0.724 0,844 1.517 2.086 3.718
170 212 272 272 325 430 500 500 600 600 600 692 795 900
The Journal of Physical Chemistry
OC.
PF,, torr
52 13 2 13 13 13 2 13 2 13 52 2 2 2
5 2
8 2 2 2 8 2 8 2 5 8 8
8
8 1 1 2
319 887 982 547 503 561 203 101 515 027 150 827 724 000
200
100
40
I
I
I
I
+ 0.4
mg./cm.z/ min.
5 1 0 2 2 2 1 2 1 3
I
400
0.0 Y CI,
3
-0.4
-0.8
-I .2
Figure 1. Arrhenius plot for the silicon-fluorine reaction from 75 to 900'.
KINETICS OF
THE
Temperature,
'C
1000 800
o1
-0.6 PI:
-I
= 2 . 8 Torr
.o
-Ia4
t
0.6
2673
FLUORINATION OF SILICON AND BORON
I
I .a
I .4
I .8
I
%OK
Figure 2. Arrhenius plot for the boronfluorine reaction from 300 to 995".
On considering the values for silicon one notes an increase in activation energy with increase of fluorine partial pressure in the low temperature region. This must be attributed to the surface effects which are associated with the fluorination processes. Since temperature measurements below 700" were made using a thermocouple, the measured values could be lower than the actual surface temperatures of the samples under the fluorination conditions. This was very clearly seen for runs above 700" where the actual surface teniperature before and during fluorination could be recorded with an optical pyrometer. Temperature rises of the order of 50 and 35 O were observed for silicon and boron, respectively, at furnace temperatures of 745 and 730" at 2.8 torr fluorine pressure. The reaction teniperatures above 700" given in Tables I and I1 are the actual sample surface temperatures while those below 700 " are the furnace temperatures. Since all the specimens of silicon or boron used had nearly equal surface areas and masses, a rough approximation may be niade that the temperature rise in each case was proportional to the rate constant, assuming specific heat and heat loss considerations remain nearly constant. Based on this assumption, the actual surface temperature rise at 85 " would be only 1.5O and that at 120" would be about 9.9Oso that the activation energy after applying the corrections (obtained from the dotted line, A, in Fig. 1 ) woulld be only 12 kcal./mole instead of 15.4. Similarly applying the temperature correction for the reaction at fluorine partial pressure, 13.2 torr, the activation energy becomes 12 kcal./niole (obtained
from dotted line, B, in Fig. 1) which is the same as that at 2.8 torr. For boron, after applying similar corrections for temperature effects, the experimental activation energy of 42.7 kcal./mole conies down to 37.9 kcal./mole (corresponding to the lower portion of the dotted line in Fig. 2). The greater activation energy for the boron-fluorine reaction is in keeping with the observation that silicon reacts faster and a t lower temperatures. From Fig. 1 and 2 it is possible to malie an estimation of the temperature at which the break in the temperature dependence occurs. For silicon, a t 2.8 and 13.2 torr fluorine partial pressures, the change takes place, respectively, at 131 and 118O experimentally. This difference vanishes when the corrections for tempera-. ture rises at the sample surfaces are applied as indicated earlier and the transition temperature becomes 150 5 " at both the fluorine partial pressures, corresponding to the breaks in the dotted lines A and B in Fig. 1. I n the case of boron the experimental transition tern-perature of 344 " becomes 352" after the correction. Above these transition points the activation energies are very low, 1.3 kcal./niole for silicon and 2.5 kca,l./ mole for boron. The abrupt changes in the activation energies indicate a definite change in the inechanisni of fluorination between the two temperature regions. The E$ect oj Pressure. The variation of the fluorination rates with fluorine partial pressure between 2.8 and 52.5 torr was examined at various temperatures for silicon and boron. Figures 3 and 4 illustrate the pressure dependence of the reactions a t various temperatures for these elements. For silicon the reaction is fractional in order, vi?., about 0.6 a t 170 and GOO", but first order at 120' with respect to the fluorine partial pressure, while for boron it is first order at all temperatures studied. Tentative Reaction Mechanisms. The reactions of fluorine with silicon and boron are very similar to the oxidation of carbon which has been the subject of investigation by various w0rkers.5-'~ Temperature effects very similar to those observed in the present investigation have been reported by all these workers. When considering the reaction between a solid and a gas to produce only gaseous products, the various steps
*
(5) ( s t ) C. M.TU,13. Davis, and M . C. Hottel, I n d . Eng. Chem., 26, 749 (1934); (b) G. Blyholder and H. Eyring, J . Phys. Chem., 61, 682 (1957). (6) G. Blyholder and H. Eyring, ibid., 63, 1004 (1959). (7) E. A. Gulbransen,-T(. F. Andrew, and F. A . Brassart, J. Electrochem. Soc., 110, 476 (1063). (8) E. A. Gulbransen and K. F. Andrew, I n d . E n g . Chem., 44, 1034 (1952).
(9) J. M .Kuchta, A. Knnt, and G. H. Damon, ibid., 44, 1559 (1952). (10) M. Levy, I n d . Eng. Chem., Prod. Res. Dewlop., 1, 19 (1962).
V o l u m e 68. Number 9
September, 1964
26 74
A. E(. KURIAKOSE AND J. L. MARGRAVE
.-cE
3
x-
PFz , Torr
Figure 3. Effect of fluorine partial pressure on the silicon-fluorine reaction rate.
which may involve (a) the dissociation of the gas molecules and (b) breaking of bonds in the solid, (4) decomposition of the activated complex to form the primary reaction products, (5) the desorption of the product molecules, and (6) diffusion of the product molecules away from the solid surface. When the reaction rate is small, L e . , in the lower temperature regions, steps 1 and 6 will not be rate determining because of the formation of only small quantities of the products as compared to the amounts of reacting gas. Under these conditions the reaction kinetics will be controlled by one or more of steps 2, 3, 4, and 5 . By consideration of the high affinity of fluorine for silicon and boron, one tends to believe that all of the adsorbed molecules should react, i.e., the activated complexes, once formed, predominantly decompose to yield the products rather than going back to elemental fluorine. This step may be assumed very fast and, hence, not involved in the rate-determining process. Since the reactions of both silicon and boron are pressure dependent, it is reasonable to assume that steps 2 and 3 are rate determining in tbe reactions and not step 4. Thus, the observed activation energies (-12 and -38 kcal./mole) are presumably a combination of the heats of adsorption of fluorine on silicon and boron, the energy to break a bond in these and the dissociation energy of fluorine, 37.7 kcal./ mole.12 If the heats of adsorption on the two metals are essentially the same, one predicts that, as a first approximation, the difference in activation energies (38 - 12 = 26 kcal./mole) arises because of differences in the energy to get free silicon or boron atoms for reaction and this should be approximately equal to the differences in the heats of sublimation (136 - 112 = 24 kcal./mole). An alternative would involve the formation of very different kinds of initial products, like BF(g) and SiF2(g), which are created by different mechanisms. Optical spectra of BF(g) are well known,ls~14and mass spectrometric studies,I5 optical spectra,16’17 and transpiration studies18 are available on SiF,. The final ~
~~~~
(11) D. R. Stull and G. C. Sinke, “Thermodynamic Properties of the Elements,” Advances in Chemistry Series, No. 18, American Chemical Society, Washington, D. C., 1956. P i 2 , Torr
Figure 4. Effect of fluorine partial pressure on t h e rate of the boron-fluorine reaction.
in the reaction may be assumed to be as follows: (1) diffusion of the reacting gas to the surface of the solid across the gaseous products, (2) adsorption of the gas on the surface, (3) formation of an activated complex The Journal of Physical Chemistry
(12) J. G. Stamper and R. F. Barrow, T r a n s . Faraday Soc., 79, 1320 (1957). (13) R. Onaka, J’. Chem. Phys., 27, 374 (1957).
(14) M. Chrbtien, H e h . P h y s . Acta, 23, 259 (1950). (15) G. D. Blue, J. W. Green, T. C. Ehlert, and J. L. Margrave: ASTM Conference on Mass Spectrometry, San Franrisco, Calif.: May 23, 1963, p. 344. (16) J. W. C. Johns, G. W. Chantry, and R. F. Barrow, T r a n s . Faraday SOC.,54, 1589 (1958). (17) D. R. Rao and P. Venkateswarlu, J . Mol. Spectry., 7 , 287 (1961).
KINETICS OF
THE
2675
FLUORINATION OF SILICON AND BORON
products detected in the fluorination of boron and silicon are BF3 and SiF4 and these could be formed by gas phase or surface reactions between BF and Fz or SiFz and Fz. It has not yet been possible, however, to chill the BF and SiFz formed (if any) as soon as the reaction occurred at the surface. Infrared and ultraviolet spectrophotometric and mass spectrometric studies involving elemental fluorine present various problems and, hence, it is hard to ascertain the exact nature of the primary reaction products in these fluorination reactions. It has been shown above that the fluorination reactions of silicon below 150O and of boron below 3.52 O are apparently controlled by an adsorption process. This point is also indicated by the rounding off of sharp edges on the samples which are highly active centers for adsorption, during the reaction. At about the transition temperatures, the reactions have become fast enough for the products to be formed in amounts comparable to that of fluorine and, hence, diffusion of
fluorine lriolecules through the outgoing product inolecules becomes a slow process compared to the other steps of the reaction. The reactions, therefore, are now controlled only by the diffusion of fluorine through the product gas layer, for which the energy of activation must be very low. The observed values are only 1.3 kcal. for silicon and 2.5 kcal. for boron. I t may also be observed from Fig. 1 that the change in mechanism takes place sooner when the rate is higher, as in the case of the reaction of silicon with fluorine a t 13.2 torr conipared with that a t 2.8 torr, which further indicates that diffusion processes play a significant role in determining reaction rates.
Acknowledgment. The authors are pleased to acknowledge the financial support of the project by the U. S. Air Force through contract No. AF33(616)-7472 administered by Dr. Leslie A. McClaine of A. D. Little, Inc. (18) A. S. Kana'an and 6.L. Margrave, Inorg. Chem., 3,1037 (1964).
Volume 68, Number 8
Septembsr, 186Y
