In almost all processing for Integrated Cicuits and other devices we follow a path with a long list of layers, masking and etching. In the final circuit we cannot see the performance of single layer or resistor or transistor anymore; it's embedded in the final product and all we can say is it's a pass or failure.  The cause of failure can be more clear if we know something about every single step.One way to do this is to procces simple structures related to the main procces simultaniously. A layer can be monitored by means of vandePauw structures, linewidth with a resistor, contacts from layer to layer with contact string, performance of transistors by series of single transistors with different dimensions, isolators are put between layers and capacitance is measured.  These structures are placed in between circuits 5 in a cross on a wafer, with 300mm wafers 10 structures.

This technique is called PCM-testing, where PCM stands for Procces Control Monitor and is used within Philips as a standard for years in all centra. 
If test-data shows that one or more results are not within the limits, the wafers are not going to final test and are analyzed by SEM and other techniques to reveal the cause, and the wafers are scrapped.

An other use is to make full PCM-wafers, we then call it PEM (Procces Evaluation Monitor). Here the goal is to see the parameters of a step of a new process to study and improve steps. One example is the linewidth study of deep-UV. Kelvin-modules were designed to measure linewidth related to focus, exposure, resists and etching.

The last example is a very dense wired stucture ( displays, memories ). A module was designed to study the feasability of relaxed, normal and very tight dimensions ( module area 25 mm2) The defects from the proccessing , cleanroom, equipment, handling is measured electrically. It classifies small, large and multiple shorts and opens. We call it PYM : Procces Yield Monitor.

We use a Keithley 450 with source/measure units and a switching matrix. This matrix is connected to a probe-ring were we can place different fixed probecards. This ring is part of the Proberstation (Electroglass 4090) which accepts cassettes with max. 25 wafers at a time. If this prober is trained it can measure for days without any help. 

Testprograms are made by TFF. A 1-25 parameter job can be made in a day : a 498 parameter job takes a week. Data can be delivered in any form ; most used is Exel-Input-Friendly type and is most of the time copied to the user or fileserver of the user. A few VBA routines are available for datapresentation ( graphs and datamapping )

Although most testroutines are 'standard' we can still make new ones for special application. For data-evaluation we can program or help with the data-proccessing.