Introduction
Correlation to traceable certified standards is absolutely critical for any metrology tool. In an effort to show how well scanning white light interferometry (SWLI) compares to traditional methods of certifying roughness artifacts, a correlation study was performed which used three Zygo NewView profilers and four traceable roughness standards. The NewView profilers were compared by measuring several standards and calculating correlation coefficients between results and certified values. Correlation coefficients on the order of 0.999 were achieved for almost all comparisons; and differences between tools and standards were either smaller than or very close to the uncertainty of the certification. The high degree of correlation (as measured by the correlation coefficients) and small differences between the average measured values for most of the individual parameters for each tool lead to the conclusion that the NewView 5000, 6300, and 6300 with 1k camera are well correlated to each other and with traceable standards.
Study Description
Several versions of the NewView optical profilers measured the same parts and part locations to determine correlation. This application note will focus on roughness specimens. For additional detail as well as information on step height correlation, please refer to the supporting data for this document.
All roughness measurements used the same 5x SLWD Michelson objective with image zoom set to 1x. The same vertical and lateral standards calibrated the PZT objective scanner and lateral coordinates in the field of view prior to making measurements. Following calibration, the part was aligned using a crosshair and then moved a defined distance (laterally) away from the initial position by using a MetroPro pattern. Software post-processed 20 SWLI (scanning white light interferometry) data sets for each instrument at similar locations for using identical acquisition and analysis settings for each measurement.
Instrumentation
The following NewView Profilers were used for this study:
- NewView 5032 with analog VGA camera
- NewView 6300 with standard digital VGA camera
- NewView 6300 with optional digital 1k camera
Traceable Standards
The following standards were used for this study:
- Halle KNT2058-01 Ra = 216 nm
- Halle KNT2070-03 Ra = 23.8 nm
- Mahr PGN-1 Ra = 566 nm
Roughness (Ra) Correlation
The instruments’ performance with respect to the certified Ra value for the four roughness standards was compared to the Makers’ certified measurements in two ways. The first method compared the horizontal and vertical lay data separately. The second method averaged the results from the horizontal and vertical lay tests. The standards’ certified Ra value – referred to as Maker’s Ra and Stylus Cert in the charts below – is subtracted from the measured Ra – referred to as WLI using FDA in the charts below (WLI refers to White Light Interferometry). This difference should be nominally zero. The maker’s calibration uncertainty for the certified value is shown as the Y-axis error bars around zero in the charts below.
For each standard, there are three measurement points in the plots – one for each of the NewView instruments used in this study.
Figure 1 – Differences with sample in a horizontal orientation
Figure 2 - Differences with sample in vertical orientation
Figure 3 - Differences with horizontal and vertical readings averaged
It can be seen that in the horizontal orientation, all the differences are quite close to the makers’ uncertainties. In the vertical lay orientation, there is an obvious outlier on the high side of the sinusoidal standards. This outlier is explained by the footnote¹ below.
Finally, when the lay data are averaged together and differenced from the certified values, we see very close agreement with the Makers’ uncertainties on all standards.
Observations and Conclusion
For the Halle standards, the parameter of particular interest is the Ra parameter which did measure within the uncertainty of the standard for all tools. The same can be said for Ra and Rz on the Mahr standards.
The high degree of correlation (as measured by the correlation coefficients) and small differences between the average measured values for most of the individual parameters for each tool lead to the conclusion that the NewView 5000, 6300, and 6300 with 1k camera are well correlated to each other and with traceable standards.
footnote¹ This is the NV 5000 data which experiences data dropout due to the use of an analog camera when the lay is vertical. This dropout cannot be corrected. The NV6000 is an all digital system whose data does not suffer this data dropout, and its data is very similar to the data seen in the horizontal lay plot.
