Linkam Scientific Instruments (Linkam)—a UK-based manufacturer of material characterisation instruments—and Sensofar Metrology (Sensofar), a Spain-based manufacturer of non-contact surface metrology and device inspection systems, recently undertook a study to characterise the temperature-induced evolution of silicon wafer shape and texture.
The study leads—Robert Gurney, marketing and applications specialist at Linkam, and David Páez, sales support specialist at Sensofar—devised an experimental setup for temperature-controlled optical surface texture measurement experiments.
1 of 2
1
(1) The Linkam LTS420 and Sensofar Linnik interferometer in use in the lab, and (2) a close-up of the sample placed in Linkam’s LTS420 chamber before starting the measurements.
2 of 2
2
(1) The Linkam LTS420 and Sensofar Linnik interferometer in use in the lab, and (2) a close-up of the sample placed in Linkam’s LTS420 chamber before starting the measurements.
Historically, this has been a difficult procedure due to imaging issues caused by spherical aberrations. However, the study’s use of Linkam’s LTS420 temperature control chamber and Sensofar’s Linnik objective, shown in figure 1, minimised these problems, thus allowing for the accurate measurement of 3D topographic maps of nanoscale materials.
1 of 2
Figure 1a
(1a) Experimental setup of the Linkam LTS420 temperature control chamber and Sensofar Linnik configuration, and (1b) schematic of the Linnik configuration.
2 of 2
Image 2b
(2a) Experimental setup of the Linkam LTS420 temperature control chamber and Sensofar Linnik configuration, and (2b) schematic of the Linnik configuration.
The study focused on changes in the topography of silicon wafers as they evolved in temperatures from 20 to 380 °C. Rapid thermal processing (RTP) is an important step in the manufacturing process of silicon wafers, in which the wafer is rapidly heated to high temperatures for a short period of time, then slowly cooled in a controlled manner to impart the desired semiconducting properties. However, RTP causes thermal stress, which leads to other problems in photolithography that may affect the performance of the device, such as breakage due to thermal shock or dislocation of the molecular lattice. Understanding the behaviour of a wafer under these conditions can help optimise the process, improving semiconductor properties and wafer durability.
The method
A key method of evaluating the effects of temperature change during wafer manufacturing is to measure the surface texture of the wafer as a function of temperature using an interferometric microscopy technique. The surface texture is observed in a temperature control chamber, as this allows temperature to be raised precisely to values similar to those during the manufacturing process, using an interferometer.
There are several factors that introduce some complexity in obtaining the interferometric microscopy measurements. First, to visualise the sample and obtain the data while accurately controlling the temperature in the chamber, it is necessary to make observations through the chamber’s optical window, which is 0.5 mm in thickness, but in some cases, can be as much as 1 mm, depending on the degree of thermal insulation required. Second, the window, being of a different refractive index to air, introduces optical aberrations and misalignments that, when analysing silicon wafers, should be corrected to obtain reliable data. Third, when the temperature inside the chamber is increased, heat is emitted to the exterior through the window, and this is not ideal for optical microscopy. Fourth, in the air close to the window, the temperature can reach 60 °C, which can lead to deformation of the objective lens, introducing aberrations.
To address the experimental issues of interferometry at varying temperatures, a Linnik interferometer can be used. The Linnik interferometer introduces the use of measurement optics within the reference arm of a classic interferometer. This allows for compensation and correction of the effects of the optical window such as chromatic dispersion and optical aberrations, enabling work with brightfield objectives, which have a greater working distance than traditional interferometric objectives.
Experimental details
The effect of the RTP process on the silicon wafers was studied while accounting for optical aberrations brought about by temperature changes. Two different samples were used, corresponding to different chip designs from silicon wafers. Sample A was 2.8 x 1 mm, whereas sample B was 3.0 by 2.35 mm. Silicon wafers have typical surface texture values on the sub-micron scale, so the ideal optical technology for this application is coherence scanning interferometry (CSI), as outlined in ISO 25178-604 (Geometrical product specifications (GPS)—Surface texture: Areal—Part 604: Nominal characteristics of non-contact (coherence scanning interferometry) instruments). Depending on the environment in which it is used, CSI offers only nanometres of system noise, regardless of the magnification of the lens being used.
For Sensofar’s Linnik objective, two Nikon 10x episcopic illumination (EPI) lenses (Nikon, MUE12100) with a 17.5 mm working distance were used. The same configuration is available with 10x super-long working distance (SLWD) lenses (Nikon, MUE31100), providing a 37 mm working distance. This makes the thermal emissions from the camera almost imperceptible to the lens and will not affect or damage the measurement quality. The Linnik objective was mounted on Sensofar’s S neox 3D optical surface measuring instrument, which combines four optical technologies in the same sensor head, namely confocal, CSI, phase shift interferometry (PSI) and focus variation.
The temperature was controlled using Linkam’s LTS420 temperature control chamber, which allows the temperature to be ramped and controlled between -195 and 420 °C to a precision of 0.01 °C, while the sample surface texture is observed through the chamber window.
The wafer sample was placed in the chamber under the S neox optical surface measuring instrument with the Linnik configuration. The acquisition routine consisted of ramping the temperature from 30 to 380 °C in 50 °C steps, taking eight topographic measurements of the sample at each step, as shown in figure 2. This procedure was repeated for three samples.
Figure 2
Figure 2: A time-temperature graph showing the temperature steps at which optical measurements were taken.
Results
The results were analysed using Sensofar’s SensoMAP software. A template was created and applied to all samples. The template allowed for the extraction of three profiles in each topography (horizontal, diagonal and vertical) and their representation in the plot shown in figure 3, as well as the building of a sequence of topographies to export it as a video and represent it in a 4D plot.
Figure 3
Figure 3: (a) Horizontal, (b) diagonal and (c) vertical profiles extracted from the measurements of the sample A at eight different temperatures.
Two topographic images of the same sample were produced as the two-dimensional height maps shown in figure 4. The three solid lines represent the three different profiles (horizontal, vertical and diagonal) extracted for each topography. The profiles in each direction are shown in figure 3, where the evolution for the different temperatures at which the sample was taken is evident. The images show that when heating the sample, its topography changes.
Figure 4
Figure 4: Two-dimensional height maps showing the topography of sample A at (a) 30 and (b) 80 ºC. The black lines indicate the three directions at which profiles were taken for further studies.
The data was plotted as the areal topographic image shown in figure 5. By stacking the 3D images as a function of temperature, creating a 4D plot, showing the topographical changes at different temperatures using the same height colour scale and showing how the samples bend as temperature changes, it was clear that the higher the temperature, the greater the bending experienced by the samples.
1 of 2
Figure 5 A
Figure 5: A stacked 4D view of the topographies extracted from (a) sample A and (b) sample B for visual comparison of the experimented bow change when samples go from 30 to 380 ºC.
2 of 2
Figure 5B
Figure 5: A stacked 4D view of the topographies extracted from (a) sample A and (b) sample B for visual comparison of the experimented bow change when samples go from 30 to 380 ºC.
To quantify the bow of the samples, two different parameters were used, as shown in figure 6. The first was the texture parameter (Sz), which is for the maximum peak to valley height of a surface according to ISO 25178–2 (Geometrical product specifications (GPS)—Surface texture: Areal—Part 2: Terms, definitions and surface texture parameters). The second was the waviness parameter (Wz), which corresponds to the counterpart of Rz in profile analysis in ISO 4287 (Geometrical Product Specifications (GPS)—Surface texture: Profile method—Terms, definitions and surface texture parameters). Both Sz and Wz were obtained after applying an S-filter (or lambda-s for Wz) to the surface (or profile) with a 0.8 mm nesting index. In this way, only the longer spatial wavelengths remain in the surface, getting rid of roughness and only leaving waviness for bow analysis. The resulting parameters for samples A and B are shown in figure 7. For sample A, an almost-linear relationship was observed between bow and temperature up to 180 ºC, which stabilises from 180 ºC to 380 ºC. On the other hand, sample B did not show any remarkable bow change until it surpassed 230 ºC.
1 of 2
Figure 6A
Figure 6: A bow evolution in (a) sample A and (b) sample B as a function of temperature. The waviness parameter (Wz) was extracted from the horizontal, diagonal and vertical profiles in figure 5. The texture parameter (Sz) was computed from the surface after applying an S-filter of 0.8 mm.
2 of 2
Figure 6B
Figure 6: A bow evolution in (a) sample A and (b) sample B as a function of temperature. The waviness parameter (Wz) was extracted from the horizontal, diagonal and vertical profiles in figure 5. The texture parameter (Sz) was computed from the surface after applying an S-filter of 0.8 mm.
1 of 2
Figure 7a
(a), (b) Filtered topographies of (above) sample A and (below) sample B at 30 and 380 ºC, respectively. S-filter 2.5 μm, L-filter 0.8 mm. (c), (d) Height and hybrid texture parameters of topographies (a) and (b), respectively.
2 of 2
Figure 7b
(a), (b) Filtered topographies of (above) sample A and (below) sample B at 30 and 380 ºC, respectively. S-filter 2.5 μm, L-filter 0.8 mm. (c), (d) Height and hybrid texture parameters of topographies (a) and (b), respectively.
Conclusion
The feasibility of the proposed configuration has been proven to carry out successful roughness and waviness measurements at different temperatures. Two different behaviours of the surface topography were observed depending on the chip design. Sample A showed bending early on when being heated up, whereas sample B showed bending at a later stage.
The S neox 3D optical surface measuring instrument with Linnik objective and Linkam’s LTS420 temperature control chamber proved ideal for performing such experimental measurements. Moreover, there is the added advantage of different brightfield objectives being compatible with the Linnik configuration, allowing working distances up to 37 mm and magnifications up to 100x to be achieved for applications that require high lateral resolutions.
Linkam Scientific Instrumentswww.linkam.co.uk
Sensofar Metrologywww.sensofar.com/metrology