pressure values and established baseline
specifications.
As seen in Figure 2 the flow curves for the
MFC in N2 is different than the flow curve for
the same MFC in Cl2 gas.
Figure 3 shows the range of flow errors
for the 200sccm and 500sccm MFCs as a
percentage of set point flow in Cl2. At 10%
the error ranges from is 6.34% to - 4.09%
and at 100% the error ranges from 1.43 to
-1.2%. While these errors may seem small,
they become accretive to the process recipe
and can significantly impact low flow steps
in a process at less than 20% set point. While
this is an example of one MFC manufacturer,
the lack of full characterization with real gas
flow will result in MFC—MFC and tool—tool
performance variability.
Figure 4 shows the typical gas cal factor
for various process gases. While approximating a gas cal factor with a best linear
fit might meet generic flow conditions, it
is not sufficiently accurate for emerging
process requirements. Current gas flow
requirements dictate accuracy of <1% at
operational set points and will not be able
to tolerate gas flow variations to the process
tool caused by a “close fit” gas cal factor as
shown in Figure 4.
Figure 5 shows a comparison between a
commonly used gas cal factor for Cl2 and the
contrast with the cal factor derived from this
study. The ROR system cal factor is calculated
by the comparison of the measured N2 flow
and Cl2 flow in the ROR system in this study.
This is not a single number or a linear correlation between flow and set point for this MFC,
however, it follows a polynomial curve, as a
function of the set point. In this case using a
single number as the gas cal factor leads to
larger flow errors in the real gas (Cl2)
The data in Figure 5 was further corroborated when testing the devices using
another independent sonic flow ROR device.
Again the flow curves for the MFC in N2 and
Cl2 deviate, and the calculated gas calibration factor is not a straight line as seen in
Figure 4, but a polynomial curve over the
range of test set points from 2-100% FS.
This preliminary data shows the need for
calibration and certification of mass flow controllers in the actual gas being used. While
several advances have been made with digital
MFCs utilizing various algorithms, including
polynomial curve fits, to compensate for cal
Figure 4. Example of commonly used linear cal factor applied to real gaes
Figure 5. MfC cal factor variations with Cl2
