Ultra Purification of Gallium by Zone Refining

Daniel Kim, Purdue University


High electron mobility is essential to improve two-dimensional electron gas (2DEG) quality in MBE grown AlGaAs/GaAs heterostructures. Recently, gallium purity was identified as limit to electron mobility in such structures. A laboratory horizontal zone refiner was constructed and procedures such as cleaning operation and handling/extracting samples were developed to suit needs for producing ultra-high purity gallium. One major problem is that most of the impurities are already below detection limit of current measurement techniques for starting high purity gallium. The 8N gallium (99.999999%, 10 ppb nominal purity) is the most pure commercially available gallium and measuring the purity becomes difficult after zone refining. Thus, a last zone back-calculating approach was developed to characterize and to quantify purity in the purified zone. First, the performance of the zone refiner was evaluated using 5N gallium, intentionally doped with 14 different impurities (Cd, In, Sn, Ni, Cu, Zn, Pb, Bi, Ge, Au, Mn, S, Co, Al). Methods of measuring the effective partition coefficient (keff) values for the first and the last zone were tested. All elements had keff values less than unity, but the values were increased as number of passes increased from n=1 to n=5 for the first zone and n=5 for the last zone. The keff values were assumed to be the same regardless of number of passes, but the zone length stability was a key factor affecting the keff values. One of the promising results from the 5N gallium study is that Mg and Sb, having C0 below GDMS detection limit of 0.1 and 0.5 ppb, were sufficiently raised to a useful detection limit of 20 and 1 ppb in the last zone. The last zone approach used in the 5N gallium study was applied to the 8N gallium experiments and the purity in the first zone was back-calculated to estimate the overall purity. For the 1st 8N gallium experiment, the starting purity of >7N6 Ga was estimated to be >9N2 Ga based on 16 different elements after ten passes, whereas the starting purity of >7N4 was estimated to be >8N5 after nine passes for the 2nd 8N gallium experiment. However, C0 variations for In and Zn were determined from contamination and purity control studies. The C0,In varied from 120 to 0.3 ppb and from 10 to <0.5 ppb for Zn. Thus, there can be possible errors in estimating the purity for In and Zn when high C0 ingots are placed in the front zone. The expected In concentration in the purified zone was much less than GDMS detection limit of 0.1 ppb, but the measured concentration showed 15 ppb for the 1st batch experiment. The same trend was observed for the Zn in the 2nd batch experiment. Assuming high C0 value of 500 ppb for In and Zn in the first 10 cm zone, the required number of passes to reach below GDMS detection limit is 24 for In and 20 for Zn. Practical number of passes is between 30 and 50 passes, for which the starting purity of 6N gallium is expected to be purified to beyond 10N.




Johnson, Purdue University.

Subject Area

Materials science

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server