Design of a machine for the universal non-contact measurement of large free-form optics with 30 nm u

Posted chaining

tags:

篇首语:本文由小常识网(cha138.com)小编为大家整理,主要介绍了Design of a machine for the universal non-contact measurement of large free-form optics with 30 nm u相关的知识,希望对你有一定的参考价值。

4.3. Metrology system design

 

The metrology loop should measure the position of the probe relative to the product within the measurement plane. The probe R and Z-positions should be measured with a maximum uncertainty of about 15 nm, the product R and Z-position to about 5 nm and the tilt to about 0.1 μrad. Figure 5.4 shows the setup in which the probe position may be measured. There are several configurations possible, with variations in mirror positions etc. but the principle remains the same.

To measure the probe position, a heterodyne interferometer beam (1) is delivered to a non-polarizing beamsplitter (2), for instance with folding mirrors attached to the stages (not shown). From here two beams travel to two polarizing beamsplitters (3). Here, one polarization direction of the beam travels straight through, the other part is deflected to the reference mirror (5). The beam is reflected and passes a λ/4 plate (4) for the second time, causing it to pass through the beamsplitter. With a lens (7) the beam is now focused on the center of the reflecting ψ-axis rotor (8). After reflection, it passes another λ/4 plate (4) for the second time, which now causes it to join the other beam again, resulting in interference between the two. With mirrors, the beams (10) can now be transported to the receivers. Basically, one part of the beam travels straight through beamsplitter (3), while the other part makes a detour between the reference mirror and the ψ-axis rotor, resulting in OPD between the beams. This way a direct measurement of the probe displacement in R and Z-direction with respect to the metrology frame has been achieved, free of Abbe errors. All critical stage errors are now measured and can thus be corrected for in the data-processing. As mentioned before, the ψ- angle of the probe is a second order error and will be measured with an encoder on the ψ-axis.

The product is assumed to be rigidly attached to the spindle rotor. The BlockHead 10R spindle is specified at <20 nm axial and radial error motion and <0.1 μrad tilt error motion. The synchronous part of this may be calibrated; the asynchronous part can be measured with capacitive probes measuring to a calibrated edge on the rotor. This way, a short metrology loop between probe and product has been created, in which all the critical positioning errors are measured (Figure 5.5). The metrology frame is shown schematically in this figure; it will be closed at the bottom for stability in future designs.

When measuring a free-form, for instance a toroid of which the surface varies between the continuous and the dotted line of Figure 5.5, the only moving parts in the machine are the continuously rotating spindle and the focusing part of the probe. Since this focusing part may be very light, there will be very little dynamical errors in the system, which will enable high measurement speeds. Most of the vibrations that remain will be measured and can thus be corrected for, which will be further explained in the dynamic analysis of section 6.5.

To the above metrology loop design, the beam delivery path and beam shielding from environmental disturbances will be added in the coming period. The spindle measurement method and metrology frame design will also be continued.

以上是关于Design of a machine for the universal non-contact measurement of large free-form optics with 30 nm u的主要内容,如果未能解决你的问题,请参考以下文章

Machine learning system design---Error metrics for skewed(偏斜的) classes

Advice for students of machine learning--转

AUTOML --- Machine Learning for Automated Algorithm Design.

斯坦福大学公开课机器学习:machine learning system design | data for machine learning(数据量很大时,学习算法表现比较好的原理)

机器翻译自动评价之BLEU详解-BLEU: a Method for Automatic Evaluation of Machine Translation

机器翻译自动评价之BLEU详解-BLEU: a Method for Automatic Evaluation of Machine Translation