+44 (0) 1932 242 444

Leveraging dark data to gain better control of production costs

Leveraging dark data to gain better control of production costs

Semiconductor wafer manufacturing uses a wide variety of speciality gases dispensed to tools in the fab, from a variety of locations and in different ways, including from auto-switchover gas cabinets. 

In order to avoid any interruption to supply the pressure or weight of a gas cylinder is monitored, sometimes manually (via rounds & readings) but often via a SCADA type application, with alarms being generated when an auto-switchover has occurred or when pressure or weight reaches a low set point.

In this way gas technicians can be prepared for the frequent cylinder change-outs required to maintain a continuous supply of gas to the fab.

However this data, captured primarily to ensure a continuity of supply, can also be used, with the right algorithms and an appropriate analytical tool, to gain a better insight into the consumption of these gases and thus better control of production costs.

From weigh scale to mass used per day

Where gas use is monitored via the SCADA system using a weigh scale, then clearly consumption over a period, say a day, is simply the start day/time value minus the finish day/time value but also taking into account any cylinder change-outs that have occurred during this timeframe, for example:

Date: Dec 02 2015 Time: 00:01 Value: 10.5 lbs Start Value

Date: Dec 02 2015 Time: 01:30 Value: 10.3 lbs Cylinder Empty Value

Date: Dec 02 2015 Time: 02:15 Value: 40.3 lbs New Cylinder Full Value

Date: Dec 02 2015 Time: 23:59 Value: 39.2 lbs End Value

Therefore 1.3 lbs were consumed on Dec 02 2015

Often in a fab the same gas is dispensed from several locations, so the total consumption of a particular gas in any period is the sum of that dispensed from all the locations and, if of course the company operates several fabs, the total company use of each gas for any period is the sum of all the locations across all the fabs, which can easily be viewed with the right data collection solution and the appropriate analytical tool.

From pressure transmitter to mass used per day

If the gas use is monitored via the SCADA system using a cylinder pressure transmitter, then it’s a little bit more complicated as the mass use needs to be calculated from:

PV = ZnRT, where Z is the compressibility factor for the gas which varies with pressure & temp

and from:

m = Mn, where M is the molar mass of the gas

Once the cylinder pressure is converted to mass as above, the usage for a period is calculated as in the weigh scale example above.


Specific gas use

Armed with the daily mass used, whether it’s from a weigh scale or pressure transmitter, then it’s a simple automated step to generate a daily cost of gas use metric with the cost/unit mass for each speciality gas. If it’s easier for certain groups to view gas use in terms of number of cylinders, then this is also a simple calculation and another metric.

By including measures of production activity e.g. wafer turns, a further ‘specific gas use’ metric may indicate unusual variations in speciality gas consumption.

But how does this help us reduce production costs?

If your monitoring of speciality gas use is confined to counting the number of cylinders you purchase in a given time period, then you’re likely to be missing important variations that are masked by such an imprecise measure. 

With this higher resolution, unusual consumption can be spotted early and corrective action taken if required and the improved visibility, of increasing or decreasing trends, can be used to help minimise inventory holding costs (including shelf life expiry) whilst avoiding stock-outs and ‘hot shots’.

So if the data is already being captured for one purpose, it would seem pragmatic to utilise it for another and particularly if off-the-shelf packages to do this already exist – as they do.