This section discusses whether or not the second condition is met. It begins by clarifying the meaning of the reliability of standard gas. Next, the conditions to meet the reliability are identified. After discussing whether these conditions were observed, a suggestion to improve reliability is made. Finally the zero point error in the calibration are estimated and the assumptions are stated.
For the standard gas to be reliable, two conditions should be fulfilled:
1 ) the calibration of the gas monitor using the same standard gas for the second time should give the same calibration setting as the first calibration. In this case the same gas means that the second gas sample is taken from the same gas bottle.
2) the calibration of the gas monitor using the standard gas from the second gas bottle, of the same purity, should give the same calibration setting as the gas taken from the first gas bottle.
The first condition for standard gas is achieved if the gas bottle and the connecting tubes to the gas monitor are made of material inert to the standard gas. In this research, the gas bottle is made of steel and the connecting tube is made of Teflon. Teflon is a type of tetrafluorethylene polymer (Fachinformationszentrum Chemie GmbH, 1992). The properties of tetrafluoroethylene which are relevant to this condition are: non-stick to airborne pollutants, melting point of 327 degrees Celsius, and exceptional resistance to chemical attack (Saunders, 1988).
The second condition for standard gas is achieved if the standard gas of the two gas bottles have the same purity. Purity refers to the types and relative concentration of inteferent gases in the gas bottles. Interfèrent means any substances, which if exist in the sample cell, contribute to the current detected in the microphone of the gas monitor. Ideally the standard gas should not contain any interférants. However, the standard gas is normally supplied with a known type and concentration of interférants. Thus, to meet the second condition, the standard gas in the two bottles should have the same type and concentration of interférants.
To improve the reliability, as required in the second condition, the two gas bottles should contain the standard gas manufactured in the same batch. In this research the standard gas from more than one bottle was used. It is not known whether the standard gas was manufactured from the same batch. However the uncertainty in the reliability of the standard gas may be estimated if the manufacturer specifies the type and concentration of the interferents.
The uncertainty In the reliability of the standard gas may be estimated in terms of zero point and span calibration errors.
Zero point error, in this case, means the maximum expected uncertainty in determining zero point due to the unreliability of the zero gas used in the calibration. The zero gas used in this research was the BOC pure nitrogen of grade N5.5. Besides nitrogen, the zero gas contained carbon dioxide, carbon monoxide, and total hydrocarbon. The concentration of each of the three impurities was not more than 0.5 ppm. For this reason the zero point error for optical filter UA 0983 specific for carbon dioxide was 0.5 ppm. Similarly the zero point error for optical filter UA 0984 specific for carbon monoxide was 0.5 ppm.
The zero point error for optical filter UA 0987 was assumed to be 0.5 ppm. The assumption in this case was that n- hexane, 1-decene, ethylene, 2-butanone, and n- buty I acetate, if existed in the zero gas tested in determining its impurity, would not give a concentration larger than 0.5 ppm if it were to be measured using the gas monitor.
In the above estimation of zero point error, it was assumed that the zero gas was perfectly dry. The concentration of water, which is an interfèrent for the three filters. In the zero gas is not stated by the manufacturer. Therefore the uncertainty in the reliability of the zero gas due to water cannot be estimated. This unreliability can be avoided by using the zero gas from the same gas bottle or gas bottles from the same manufacturing batch.
Span error in this case means the maximum expected uncertainty in determining two points calibration span. The span error may be estimated as follows:
a) TVOCs
Since the standard gas used was of 100 ppm and an error up to 0.5 ppm may be introduced at zero point by zero gas, the error due to span calibration is 0.5 percent (or 0.5 of 100)
b) carbon dioxide
Since the standard gas used was of 540 ppm and an error up to 0.5 ppm may be Introduced at zero point by zero gas, the error due to span calibration is 0.09 percent (or 0.5 of 540)
c) carbon monoxide
Since the standard gas used was of 10 ppm and an error up to 0.5 ppm may be introduced at zero point by zero gas, the error due to span calibration is 5 percent (or 0.5 of 10)