Single Sample Cylinder for RGA Correlation

Abstract

A large portion the manufacturers of high reliability microelectronic devices must have those devices tested to be proven “dry”. A mass spectrographic technique, referred to as Residual Gas Analysis (RGA) has been the primary means of demonstrating that the devices contain less than the 5000 parts per million(ppm) of moisture. However, there are no standards for determining moisture levels of this nature when tested in this manner. Too often, the pass/fail criteria for a given production lot is exceeded and recourse becomes time consuming and expensive. A small group of technologists has successfully fabricated and tested a Single Sample Cylinder (SSC) wherein moisture-laden air can be taken from one analytical facility to another for comparison analyses. The need for such a device is explained by a brief review of why “identical” microelectronic devices from the same production line have been found to contain different chemistries when analyzed via Residual Gas Analysis, per Test Method 1018 of the military standards. The SSC is over two liters in volume; hence it can deliver hundreds of “air samples” that have the same chemical composition. It is maintained at 100° C during test but is kept at room ambient during non-analytical periods. Several major problems had to be overcome in order to be able to continuously generate samples of the same chemistry. The primary two were moisture adsorption to all of the various exposed surfaces, and second was hydrogen diffusion from the materials of construction, with subsequent oxide reduction to form more water. Some anomalies were experienced during initial preparation of the SSC and the early hours of equilibration. These again are believed to be adsorption/desorption related. Thereafter it was shown that the SSC could deliver samples that were within 2% of each other, or better, within a sample set. The level of precision between sample sets was +/− 2.5%. This study was developed on a statistical basis, involving over 100 data points and was carried out over 1450 hours. The gas analyses were performed in a Defense Supply Center Columbus (DSCC) certified testing facility. This study was conducted with the support of an additional analytical chemist and a statistician to assure that data generated met the goals of the stated objective.