1.1

Thickness uniformity of single layers

Excellent uniformity in thickness of the required three layers is essential for FEL applications. The mean coating thickness is specified to be in the range from 35-45 nm, although it must be taken into consideration that the tolerances and variation in thickness should be below 4% (rms) over the whole mirror aperture of about 800 mm. This tolerance corresponds to an acceptable thickness variation of less than 1.6 nm rms at a typical thickness of 40 nm. The experimental layer thicknesses of Ni, C, and B₄C in the tangential direction over the whole mirror length of 1500 mm are shown in Figure 1. All measurements were performed on small calibration samples of silicon wafers. The mean, rms and peak-to-valley (PV) values are tabulated in the table 1. All PV values are below 1 nm and all rms values are below 0.3 nm in the tangential direction. Since the mean layer thicknesses are higher than 45 nm for Ni and B₄C and lower than 35 nm for C, it has to be changed in the final deposition according to the specified thickness range. The measured high level of uniformity confirms former results using the HZG magnetron sputtering facility^{5, 9}.Previous investigations of tungsten and carbon coatings reached similar uniformity in thickness. Here, the goal was to coat a very large aperture of 1500 mm x 120 mm. Additional mask technology was developed and optimized in order to achieve a high uniformity over the whole optical area, especially in the sagittal direction of a mirror.

Figure 1.

Film thickness of single layers of Ni, C, and B₄C as a function of the substrate position, i.e. the tangential direction of the mirror, measured by X-ray reflectometry using Cu radiation.

Table 1.

Measured film thickness (in nm) and roughness values (in nm rms) of Ni, C, and B4C layers; all measurements were performed on small calibration samples (roughness values of uncoated silicon substrates are added for comparison).

The current results are achieved after a three-step deposition of Ni, C, and B₄C. Before each step, two covers were adjusted at the carrier in such a way that each stripe was coated in a distinct height above the bottom of the carrier and with a certain width. After optimization and precise adjustment, the first of two mirrors were coated with three stripes side-by-side. The coating process of each material is slightly different and required certain conditions of Ar gas pressure, generator power and carrier velocity. Parameters used for the deposition of Ni, C, and B₄C are mentioned in the second paragraph. It is important to note that this procedure is interrupted twice by a vacuum break. Therefore, a sequence was chosen in order to avoid the coverage of a heavier element onto a lighter element during deposition. A wrong material on top of a layer would lead to a reduction of X-ray reflectivity. The chosen path was as follows: the first stripe was nickel, second boron carbide and the last was carbon. After coating the first mirror, the second one was coated using the same deposition parameters and vacuum conditions to achieve the same film properties. The achieved final results on two 820 mm long mirrors are verified using short witness samples during both processes.

1.2

Stripe pattern of three layers

The use of two covers, fixed at the carrier and parallel to the tangential direction of the mirror, provides the opportunity to coat a stripe (i.e. a defined aperture section) on to a substrate. Typical distributions of the thickness of single layers of B4C around their center are shown in Figure 2. Four certain settings of covers are used for improvement. All thickness measurements were performed with a rectangular slit of 2 mm x 2 mm at various y-positions near the center of each stripe. The first setting employed two covers of 2 mm thick stainless steel (open squares). A rounded thickness profile symmetrically around the maximum thickness in the middle was obtained. At the bottom the profile is expanded to a wide foot, which can be explained by the very broad deposition profile of the rectangular sputtering source. The length of the sputtering source parallel to y is about 30 cm, which is quite large in comparison to the stripe width of 16 mm. The thickness of the covers with 2 mm causes shadowing. Moreover, the distance between cover and substrate of about 2 mm is most likely responsible for the foot at the edge of the stripe. There is also a small amount of coating material coated under the cover. The broadening is due to collisions of target and argon atoms along the way to the substrate. In order to reduce the uncoated area between the stripes, so called dead zone, some improvements are required. The edge of the 1.6 m long covers was tapered in a second setting, so that a long chamfer ends at a 0.5 mm thick edge. Therefore, the uniform area in the middle of the stripe was distinctly broadened. The normalized thickness is distinctly above 90 % over the whole width of the stripe by the use of the improved covers (solid squares). A trapezoidal cross-section of the stripe is determined with a measuring step width of 2 mm. After reduction of the distance between cover and substrate to approximately 1.5 mm, the foot at the edge of the stripe vanished in a third setting (red squares). The last optimization was to adjust the width of the covers to the specified width of the stripe for the final deposition. A perfect positioning of the mask resulted in a stripe width of about 8 mm (solid circles). Finally, the two FLASH mirrors were coated under these optimized conditions in order to achieve a nearly rectangular-shaped thickness distribution in the sagittal direction of the mirror. Future improvements are expected due to the use of foils that reduce the distance to zero without damaging the mirror surface. In case of contact it should be possible to achieve a very small transition zone and high uniformity.

Figure 2.

Normalized film thickness as a function of the y position located around the center of the coating stripe, four settings of various covers and distances to the mirror surface are shown.

Figure 3 shows a typical distribution of the film thickness of three single layers of Ni, C, and B₄C as a function of the y position, which means from side to side in the sagittal direction of the mirror. The optimization resulted in a perfect stripe pattern of three materials. The Ni coating extends from 52 mm to 64 mm and its center is at 58 mm. The center of the a-C coating in the middle is 75 mm, which fulfills the specified difference precisely. The center of the B₄C coating is again shifted by 17 mm to a y position of 92 mm. It can be stated that each material is distinctly separated on the optical area. Therefore, a contamination of a former coated layer of a latter one during the deposition can most likely be excluded. A change of the coating material is possible by a side-step of about 17 mm. An exchange of a mirror in the tunnel section is a time consuming task due to a vacuum break, which would concern all beamlines at the FEL.

Figure 3.

Thickness variation of three stripes of Ni, C and B₄C layers along the sagittal direction of the mirror. Stripe pattern is prepared in three coating runs one after another. Each coating run required a distinct position of the two covers.

For each coating a distinct width between the covers was adjusted before deposition. In the case of Ni, C, and B₄C the cover width was chosen to be 10 mm, 21 mm and 9 mm, respectively. This last stage of pattern optimization before final deposition is shown in Figure 3. Additionally, an intermediate step was done in order to evaluate the experimental result of the achieved stripe pattern on a more realistic surface. A short silicon blank called “little brother” was therefore coated in three runs one after another. Figure 4 shows an image of the prototypical silicon blank with the same cross-section as the long final mirrors and a reduced length of 120 mm. This additional experiment distinctly confirmed the whole three-step process. For the final deposition of the first of two 820 mm long mirrors, the width between the covers was again slightly increased by a tenth of a millimeter in order to extend the coated stripe at the expense of the uncoated area. Thus, the dead zone between two adjacent stripes was about 2 mm or less along the whole mirror length of 1 meter. Then, the experimentally achieved stripe widths are wider than the specified values for Ni, C, and B4C of 8 mm, 20 mm and 8 mm, respectively, and their stripe centers are well located relative to each other with a distance of approximately 17 mm to the center of the central stripe of the a-C coating. X-ray mirror fabrication is based on the assumption that the coating properties of the calibration steps on small Si-wafer samples can be transferred to and repeated in the final coating on very long substrates. The stability and reproducibility of the sputtering facility are therefore very important and essential for the current success and future development projects. It is clear that it is demanding to investigate the coating properties on the final X-ray optics due to their length and weight. Therefore, small witness samples are used to control the achieved layer properties with respect to their X-ray optical properties such as thickness, density and roughness.

Figure 4.

Image of a silicon blank coated as a prototype before the final deposition. The insert shows the perfect positioning of the three stripes to each other.

Reproducibility, stability and precision of the sputtering process were studied intensively over the last years. Former and current experimental results of carbon coatings are compared in Figure 5 that demonstrates three 1.5 m long coatings of amorphous carbon prepared during the last 7 years. The measured layer thickness, normalized by the mean value as a function of the substrate position, depicted the level of the uniformity of the coating process. The density of measuring points was different due to various requirements over the last years. However, the important result is that there is no important change during time. The mean thickness was 45.4 nm, 44.2 nm and 32.7 nm. For all three cases a peak-to-valley thickness variation is about 1 nm and even better, the corresponding rms values are of 0.3 nm and below. It can be stated that the HZG sputtering facility has provided a good stability and extraordinary reproducibility in X-ray optical properties since it started its operation in 2007.

Figure 5.

Normalized thickness variation of three carbon coatings prepared in recent years. The mean thickness of these layers was 30-45 nm. The peak-to-valley value is less than 1 nm and the rms value is less than 0.3 nm.

Micro-roughness of single layers of Ni, C, and B4C were investigated on final calibration samples using atomic force microscopy (AFM) and white-light interferometry (WLI). In Figure 6 a-d, four representative AFM images over a measuring area of 1 μm x 1 μm of uncoated and coated silicon are shown. Additional AFM measurements were performed on a larger area of 10 μm x 10 μm. All data is listed in Table 1. All roughness values are below 0.9 nm rms. The uncoated state of Si exhibits roughness values of 0.2 nm rms and 0.5-0.6 nm rms measured by AFM and WLI, respectively. It is important to note that 820 mm long uncoated mirrors exhibit a significant better quality of micro-roughness than the calibration sample used here to check the coating process. In the AFM image of B4C (Fig. 6 top right), it is distinctly visible that the surface morphology of the film is similar to the one of the underlying silicon substrate (Fig. 6 top left). The B4C coating seems to maintain the morphology of Si. By coating of a-C and Ni (Fig. 6 bottom left and right), a perfect coverage of the Si surface is obtained. Here, the film surface seems to be more flat and totally wet the underlying substrate. This is indicated by the measured roughness on a small scale, since it resulted in very low values of less than 0.2 nm. Especially in the case of a-C coatings, the film roughness can be improved drastically. The roughness values measured by white-light interferometry were not so evident. In general, it can be stated that the deposition process replicates the substrate roughness. In fact, it rather tends to a slight improvement of roughness.

Figure 6.

Micro-roughness measurements of uncoated silicon (top left) and single layers of B₄C (top right), a-C (bottom left) and Ni (bottom right), measured with an atomic force microscope on a measuring area of 1 μm x 1 μm.