The pore-filling process and the effect of entrapped gas in liquid phase sintering of Fe-Cu alloy have been investigated. First, in order to observe the pore-filling process, large spherical pores were created by sintering in a $H_2$ atmosphere at 1125 degrees C a power mixture containing 82 pct. (by weight) of 5 micron Fe power, 10 pct. of Cu powder finer than 22 microns, and 8 pct. of spherical Cu particles of size between 100 and 120 mesh (diameters between 125 and 150 microns, approximately). The heating rate to the sintering temperature was about 60 degrees/min. Upon reaching the sintering temperature, large Cu particles flowed completely into the capillaries between Fe powders, leaving spherical pores of nearly the same size at their sites. During liquid phase sintering, fine natural pores located between solid grains were filled first. After an incubation period, spherical pores were filled with liquid flowing from the surrounding dense region. But spherical pores of smaller size always remained at the top of the liquid pockets. After prolonged sintering, most spherical pores were partially filled with liquid, and the size of the smaller pores in the liquid pockets remained almost unchanged.
The pore-filling behavior described above agreed qualitatively with theoretical predictions based on models by Park and Yoon which showed that solid grains must grow to a certain critical size before pore-filling could take place. But there existed a considerable difference between the critical grain radius calculated based on the theoretical model and the average grain radius measured experimentally at the moment when spherical pores were first filled. The measured average grain radius was larger than the calculated critical grain radius by a factor of about five. It has been proposed that this discrepancy might be attributable to the difference between the non-equilibrium grain shape due to non-closest packing of solid grains in real sintered structures, and the local equilibrium grain shape of a closest-packed structure assumed in the model.
The smaller pores remaining in the liquid pockets were confirmed to be gas bubbles by dipping the specimens into molten liquid of matrix composition. When the specimen was dipped into the molten matrix, all spherical pores were filled quickly, but the size of smaller pores remained almost unchanged even after a long time. Therefore, the bubbles appear to contain a non-diffusing gas (NDG), most likely $H_2O$ formed by reduction of oxides.
The effect of entrapped NDG on pore-filling has been investigated by entrapping Ar in the pores. To entrap the Ar, the specimens were first sintered in a $H_2$-Ar atmosphere. The partial pressure of Ar was controlled by adjusting the ratio of the flow rate between $H_2$ and ar. After sintering the specimens long enough for pores to be sealed from the specimen surface, they were resintered or dipped into a liquid bath for various times in a $H_2$ atmosphere. During resintering, $H_2$ outside the specimen diffused into the pores so quickly that the $H_2$ pressures between the inside of the pore and the outside of the specimen rapidly became equal. Therefore, the total pressure in the pore became higher than that outside the specimen by an amount equal to that of Ar. When the pressure of entrapped Ar was small ($Pp^o < 2\gamma/Rp^o$, where $Pp^o, Rp^o, \gamma$ are entrapped Ar pressure, pore radius, liquid-vapor surface tension, respectively), pores were filled during resintering, but more resintering time was required before the first pore was filled. Also, the pores shrunk only to a very small extent. The limit of the size of the pore was determined by dipping the specimen into a liquid bath. It is worthwhile to note that when $Pp^o$ was higher than $2\gamma/3 RP^o$, pores shrunk at a very slow rate to the minimum size due to grain growth only.
When the entrapped Ar pressure was higher than $2\gamma/Rp^o$, pores were not filled. In the 82Fe-8Cu (100×120 mesh)-10Cu specimens, pores were not filled even after prolonged periods. But, in the 60Fe-8Cu (100×120 mesh)-32Cu specimens with larger amounts of liquid, pores became rather expanded. Therefore, it is proposed that pore expansion for $Pp^o$ being higher than $2\gamma/Rp^o$ should be dependent only on the strength of the skeleton between the solid grains. When dipped into liquid bath for a long time, pores expanded abruptly due to the penetration of liquid from the surroundings of the specimen into the interspaces between solid grains even in 82Fe-8Cu (100×120 mesh)-10Cu specimens.
Reversely, pore-filling behavior was investigated by changing the sintering atmosphere from $H_2$ to Ar. Specimens were first sintered in a $H_2$ atmosphere, then resintered in a $1H_2$-9Ar (volume ratio) atmosphere. During resintering, $H_2$ in the pores was quickly degassed; the total pressure outside the specimen (1 atm) rapidly became higher than that inside the pore (0.1 atm). And the pores were filled faster than when the sintering atmosphere was not changed. The results described above about the effect of NDG and change of atmosphere on pore-filling was in good qualitative agreement with theoretical predictions.
The pore-filling behavior during sintering in a $N_2$ atmosphere was also investigated. Eighty two Fe-8Cu (100×120 mesh)-10Cu specimens were sintered in a $1H_2-9N_2$ atmosphere at 1125 degrees C. During sintering pores were filled, but the rate of pore shrinkage was small. And many pores started to shrink at the same time. Therefore, it is proposed that the rate of pore shrinkage should be controlled by the diffusion of $N_2$ entrapped in the pore through the liquid matrix. By dipping the specimen into a liquid bath, it was confirmed that entrapped $N_2$ in the pores was a slowly-diffusing gas. After a prolonged period, most of the pores were filled, but spherical pores of smaller size also remained in the liquid pocket.