Flatness is an important surface tolerance requirement during the manufacturing of a part of an ASML machine. Flatness can be measured by either measuring the variation in height of the surface in one measurement (a plane-wise measurement) or by measuring the height on multiple xy-coordinates on this surface (point-wise measurement). Point-wise distance measurement sensors tend to have small physical dimensions, while having good specifications on resolution perpendicular to the measured surface. During a separate research, a gap is found in performing a point-wise flatness measurement while accounting for undesired stage deviations. The stage deviations are in the order of
micrometers, while the sensor accuracy is in the order of nanometers. The performance of a point-wise flatness measurement is therefore in a negative sense dominated by the stage inaccuracies rather than sensor performance. During this research the gap is filled by designing, building and testing a feasibility demonstrator which uses reference sensors to account for stage inaccuracies. By performing a live correction with either 1 or 3 reference sensors, the stage inaccuracy (in z, Rx and Ry) can be compensated for. The measured stage wobble for this specific setup is 3.2 µm, which is reduced to 0.13 µm when using three reference sensors. The method of using reference sensors for flatness measurements is shown to work, mainly when using 3 reference sensors.

This report describes the design, build and verification of a setup to measure and adjust the Rx, Ry, and Z coordinates of an element within a microscope at ASML to improve the precision of a microscope. The required tolerance budget for the measurement and adjustment of these coordinates are $\pm 1~mrad$ in Rx and Ry and $0.1~mm$ in Z. The design is done by first finding different solutions for the sensor, the calibration approach and the adjustment. These different solutions are then combined to create different concepts for the setup. Subsequently, these concepts are traded off against each other using different selection criteria. The selected concept uses a lasertriangulator on an X and a Y directed stage that measures the height of a selection of preselected locations on a coplanar reference, which has the desired Rx, Ry, and Z coordinates. This set of data points is then used to make a plane fit from which the Rx, Ry, and Z coordinates of the reference can be retrieved. Subsequently, the same process is used to determine the Rx, Ry and Z coordinates of the element. The required adjustment of the element can then be calculated from the difference between the two sets of coordinates. After a 3D design is made for the setup, the different subsystems of the setup are elaborated and their tolerances are budgeted. To verify the setup, the setup is build in the cleanroom and the different subsystems are tested for their functionality and their performance. These results are then compared with the calculated tolerances. Since most of the tolerances are approximately the same as budgeted and the setup works as expected, it can be concluded that the selected design is suitable. This is however with the exception the tolerance due the stage wobble, which could not be measured as accurately as desired, and with the exception of the tolerance due to the heat generated by the stages, which was unexpected. In the end, recommendations are given on how to improve the setup. From these recommendations, changing the lasertriangulator for a sensor which preforms better on a reflective surface and implementing a solution for the tolerance due to the heat generated by the stages are the most important to alter the setup and make it a production ready design. If a sensor would be implemented that can measure reflective surfaces, the stage wobble, which is the last unknown tolerance contributor, could also be measured after which it can be fully concluded whether the setup works.

Holistic lithography uses feedback from a chip to improve the production process for the following batches of chips. However, as the features on a chip are on a nanometer scale, acquiring feedback is not simple. The features are too small for conventional light based metrology systems, therefore, scanning electron microscopes (SEM’s) are used.

As a SEM uses one single beam it is too slow to implement in the lithography production process. To improve the speed multibeam SEM’s have been designed. These microscopes use the same operating principle but with multiple beams simultaneously.

Each beam has its own focal point, which together form the focal plane. To acquire a sharp image the focal plane needs to very precisely align with the sample that has to be measured. To achieve this the focal plane has to be very flat, which requires a very precise electrostatic lens.

In this thesis the flatness tolerances are evaluated and determined whether they are feasible. This is evaluated by calculating the errors with FEM software and by experimentally testing the remaining errors.