Open Access
Add to BibSonomyAbstract
The direct replication of W/Si multilayers and the effect of separating layer thickness on the performance of the multilayer before and after replication are investigated systematically. Platinum separating layers with different layer thicknesses were first deposited onto different supersmooth mandrels and then W/Si multilayers with the similar structure were deposited onto these Pt-coated mandrels by using a high vacuum DC magnetron sputtering system. After the deposition, these multilayers were replicated onto the commercially available float glass substrates by epoxy replication technique. These multilayers before and after replication are characterized by grazing-incident X-ray reflectance measurement and atomic force microscope. The measured results show that before and after replication, the reflectivity curves are much similar to those calculated and the surface roughness of each sample is close to that of the mandrel, when the separating layer thickness is larger than 1.5 nm. These results reveal that the W/Si multilayer with the separating layer thickness larger than 1.5 nm can be successfully replicated onto a substrate without modification of the structure, significant increase of surface roughness or apparent change of reflectivity.
©2011 Optical Society of America
1. Introduction
Replication technique has been investigated widely in many countries and used successfully by many X-ray telescopes for its advantages [1–7]. One advantage of the replication technique is that the optical figure and roughness of the final replicated mirror is strongly influenced by the figure and polish of the mandrel. Thin and floppy mirror shells with high accuracy profiles and low microroughness may be produced. Another advantage is the possibility of making many identical shells from one mandrel, which is very attractive in the case of multi-nested telescopes.
Replication techniques include two widely used methods: one is nickel electroforming replication [8–11] and the other one is epoxy replication [12–15]. The main advantage of Ni electroforming replication method is high imaging resolution, combined with high throughput of the telescope design which is formed by many confocal thin mirror shells. However, the density of pure Ni is very high (8.9 g/cm3), which increases the mass-to-collecting area ratio and the weight of the system. Multilayer shells based on epoxy replication can be fabricated by using two different approaches: a) the metal-coated foil substrate was first produced by epoxy replication from a supersmooth mandrel, and then was deposited with the multilayer, and b) the multilayer was first deposited on the supersmooth mandrel and then was replicated to the foil substrate. For the former, because of the small heat capacity of the foil mirrors, the temperature rises sharply during the multilayer deposition. To avoid damaging the underlying epoxy layer, the temperature has to be kept below 323.15K by idling the chamber for a period in each cycle to allow the foil to cool, which makes the process time consuming and unsuitable for mass production. The direct replication of a multilayer can solve the problem. By direct replication, damage from heat can be avoided since the multilayer is deposited directly on the mandrel, and the process of coating a metal foil is not needed. The technique of multilayer replication is thus of interest for fabricating multilayer mirrors which require special substrates: i) a foil substrate with very small thickness which can't be polished, ii) a substrate which could be deformed or damaged by the heat during the multilayer deposition, iii) a nonplanar substrate with special profile which is difficult to coat uniformly.
As part of a program to develop grazing incident multilayer mirrors for Chinese future hard X-ray telescopes, replication technique is investigated by our research group. As an initial study, the replication process of periodic multilayers and the effect of different separating layer thickness on the performance of these multilayers before and after replication were investigated systematically. Flat supersmooth borofloat glasses were used as mandrels and commercially available float glasses as substrates. Tungsten and silicon were selected as the material combination. The measured results demonstrate that before and after replication, the W/Si multilayers with the separating layer thickness larger than 1.5 nm can be successfully replicated without modification of the structure or increase of the interface roughness.
2. Experiments
2.1 Multilayer Deposition
Tungsten/silicon multilayers are widely used to fabricate X-ray mirrors because of their high theoretical reflectivity, excellent thermal and temporal stability, and low interface roughness [16,17]. In this study, we also chose W and Si as the material pairs. The supersmooth borofloat glasses (3.8 mm in thickness, 30.0 mm in diameter) were used as the mandrels. These mandrels were first coated with a platinum layer and then were deposited with a W/Si multilayer using a high vacuum DC magnetron sputtering system. The Pt layer was used as a separating agent between the mandrel and the multilayer to reduce the adhesive force between them. The W/Si multilayers deposited on different Pt-coated mandrels have the similar structure. The number of periods is 40. The difference among these multilayer samples lies in the different Pt thicknesses which are set to be 50.0, 25.0, 12.0, 6.0, 3.0, 2.0, 1.5, 1.0, 0.84 and 0.72 nm, respectively.
The base pressure was lower than 2.0 × 10−4 Pa before deposition. Argon was used as the working gas with purity better than 99.999%. The working pressure was 0.133 Pa during deposition. The deposition rates of W and Si were about 0.06 and 0.09 nm/s, respectively.
2.2 Multilayer Replication
The aim of the replication is to transfer these multilayers onto the substrates. In this study, ordinary flat float glasses (2.0 mm in thickness, 45.0 mm in diameter) were used as the substrates. In order to create a replica with the best performance, bare spots, surface blemishes, or loss of surface smoothness should be minimized.
Before replication, a two-component epoxy resin was mixed and then diluted with toluene. The surface of the substrate was sprayed with the fluid epoxy and attached with the mandrel. This process was performed in an evacuated chamber, which can prevent air from being injected into the epoxy resin when combining the two objects. Air in the epoxy can hinder accurate duplication of the mandrel surface and result in degradation of the surface quality of the replicated mirror. The mated mandrel and substrate then were heated to cure the epoxy at 323 K for about 14 hours. Finally, the mandrel was detached from the substrate at the boundary between Pt and the mandrel with minimum force.
3. Measurements
3.1 GIXR Measurements
The W/Si multilayers with Pt thickness of 50.0, 25.0, 12.0, 6.0, 3.0, 2.0, 1.5, 1.0, 0.8 and 0.7 nm before replication are designated as ML1, ML2, ML3, ML4, ML5, ML6, ML7, ML8, ML9 and ML10, respectively, while the corresponding multilayers after replication are denoted by ML1’, ML2’, ML3′, ML4’, ML5′, ML6’, ML7’, ML8’, ML9’ and ML10’, respectively. All of these multilayers are characterized by grazing incident X-ray reflection (GIXR) at the energy of Cu Kα line (E = 8.04 keV). The period thickness (d) of each W/Si multilayer before and after replication is obtained according to the modified Bragg’s formula. Other structural parameters, such as the ratio (Г) of W layer thickness to the period, the interface roughness (σ) of W-on-Si and Si-on-W, are obtained by fitting the reflectivity curves. Figure 1(a) shows the typically measured results of ML6-ML10 and their corresponding fitted curves. Figure 1(b) shows the correspondingly measured results of ML6’-ML8’ and their fitted curves. All the measured and fitted parameters were listed in Table 1 .
Fig. 1 The measured and fitted reflectivity curves of W/Si multilayers with different Pt thickness: (a) before replication, (b) after replication.
Table 1. Structural Parameters of Each W/Si Multilayer Sample Before (Labeled as ML1-ML10) and After (Labeled as ML1’-ML8’) the Replication, Respectively
From these measured and fitted results, we can see that when the separating layer thickness is larger than 1.5 nm, the Bragg peaks’ positions before and after replication are matched well before and after replication and that the structural parameters of the replicated multilayer keep almost the same as these deposited. This indicates that the multilayer structure is not modified after replication even though the structure is inverted. These measured reflectivity curves are consistent with those calculated, indicating that the reflectivity performance is not degraded. In addition, we can see that before replication, the thickness of the bottom Pt layer has little effect on the measured reflectivity curves. When the separating layer thickness is equal to 1.0 nm, the interface roughness of the replicated multilayer is increased and the difference between the measured and calculated results can be observed. As the separating layer thickness continues to decrease (less than 1.0 nm), it is hard to separate the mandrel and the substrate, probably caused by the discontinuity of the separating layer and the increase of the adhesive force between the mandrel and the sample. Figure 2 shows the measured reflectivity curves of the multilayer before (square line) and after (triangle line) replication and the mandrel after replication (dot line), (a) Pt thickness is 0.84 nm and (b) Pt thickness is 0.72 nm. It is obvious that after replication there are still some multilayer remnants on the mandrel. The intensity of each Bragg peak of the mandrel is higher than that of the replicated multilayer when the separating layer thickness is 0.84 nm, and the result is contrary when the separating layer thickness is 0.72 nm. This indicates that with the separating layer thickness decreasing, the proportion of the remaining multilayer on the mandrel is increasing and the difficulties in separating the mandrel and the substrate is accordingly increasing until this process cannot proceed.
Fig. 2 The measured reflectivity curves of (a) ML9 and (b) ML10 with Pt thicknesses of 0.84 and 0.72 nm, respectively. The square and triangle lines correspond to the reflectivity of the multilayer before and after replication, and the dot line represents the reflectivity of the mandrel after replication.
These results demonstrate that the W/Si multilayers with the separating layer thickness larger than 1.5 nm can be successfully replicated onto the substrates from the mandrels without modification of the structure or degradation of the reflectivity performance.From Fig. 1, we also can see that the total external critical angle of the replicated sample is shifted compared with that before replication and the lower-order Bragg peaks are suppressed with the Pt layer thickness increasing. This is mainly due to that after replication the multilayer structure is turned upside down and the Pt layer becomes the topmost layer. This Pt layer has an enhancing effect of the reflectivity at small angles by total reflection and also has a suppressing effect on the multilayer Bragg peaks. The small peaks between multilayer Bragg peaks are the fringes of the single Pt layer.
3.2 AFM Measurements
To evaluate the surface roughness of these samples quantitatively, atomic force microscope (AFM) is used. All the AFM images are obtained in tapping mode and each covers a typical scan area of 10 µm × 10 µm. To ensure the accuracy of data, several points in different regions are measured for each specimen, and the results show good reproducibility.
Figure 3 presents the AFM images of (a) the mandrel and (b) the float glass substrate, and the root-mean-square roughness (rms) was 0.28 nm and 0.47 nm, respectively. The surface of the mandrel was much smoother than that of the float glass substrate.
Fig. 3 AFM surface images of (a) the mandrel and (b) the float glass substrate with σ = 0.28 and 0.47 nm, respectively. The typical scan area is 10 µm × 10 µm.
Figure 4 shows a typical AFM image of ML8 with Pt thickness of 1.0 nm. Figures 5(a) and 5(b) show the AFM images of ML7’ and ML8’, respectively. The AFM images of other samples have the similar profiles and thus are not presented here. The surface roughness of typical samples before and after replication is listed in Table 2 . The measured results show that the surface roughness of all the samples before replication and these replicated samples with Pt thickness larger than 1.5 nm is close to that of the mandrel. This indicates that the ultrasmooth surface of the mandrel has been finely replicated and that the Pt layer has little effect on the sample’s roughness, which is in good agreement with those results obtained by GIXR measurements. The slight roughness difference between mandrel and the multilayer is mainly due to the impossibility of exactly repeated measurements at the same point. When the separating layer thickness is 1.0 nm, the surface roughness after replication reaches to 0.61 nm. When the separating layer thickness is less than 1.0 nm, the separating process is hard to perform which may be resulted from the discontinuity of the separating layer. Even the mandrel is detached with the substrate by using great force, the multilayer sample is destroyed. Figure 6(a) shows the AFM image of the replicated multilayer with Pt layer thickness of 0.84 nm and Fig. 6(b) shows the picture of the corresponding mandrel after replication. It is evident that the multilayer sample has been destroyed due to the large adhesive force between the mandrel and the sample. The sample with Pt thickness of 0.72 nm experiences the similar case.These results demonstrate that this replication process can produce high quality multilayer mirrors with ordinary substrates which are not superpolished so long as the separating layer thickness is larger than 1.5 nm.
Fig. 5 AFM surface images of W/Si multilayers after replication with Pt thickness of (a) 1.5 nm and (b) 1.0 nm, respectively.
Table 2. The Surface Roughness of W/Si Multilayer Samples Before and After Replication
Fig. 6 (a) AFM image of the replicated multilayer with separating layer thickness of 0.84 nm and (b) the picture of the corresponding mandrel after replication.
4. Conclusion
The replication of W/Si periodic multilayers and the effect of different separating layer thickness on these multilayers are investigated. The results show that when the separating layer thickness is larger than 1.5 nm, the replication process can be successfully carried out. When the separating layer thickness is larger than 1.5 nm, the structure of a replicated multilayer is the same as the as-deposited one, its surface roughness is close to that of the mandrel and its reflectivity is consistent with the theoretical value. However, when the separating layer thickness is lower than 1.0 nm, the mandrel can hardly be removed from the substrate, probably due to the discontinuity of the separating layer and thus the increase of the adhesive force between the mandrel and the sample. These results demonstrate that high quality multilayers can be produced by epoxy replication technique without modification of the structure or increase of surface roughness or degradation of the reflectivity performance when the separating layer thickness is larger than 1.5 nm. Based on this technique, future studies will focus on examining the effect of the replication process on the optical figure of a replicated multilayer and fabricating multilayer mirrors on even thinner substrates of different types and larger apertures.
Acknowledgments
This work was supported by National Basic Research Program of China (No. 2011CB922203), National Natural Science Foundation of China (No. 10825521, 10978002), National International Cooperation Program between China and Japan (No. 2008DFA01920), Shanghai Natural Science Foundation (09XD1404000).
References and links
1. O. Citterio, G. Bonelli, G. Conti, E. Mattaini, E. Santambrogio, B. Sacco, E. Lanzara, H. Brauninger, and W. Burkert, “Optics for the x-ray imaging concentrators aboard the x-ray astronomy satellite SAX,” Appl. Opt. 27(8), 1470–1475 (1988). [CrossRef] [PubMed]
2. G. Pareschi, O. Citterio, M. Ghigo, F. Mazzoleni, A. Mengali, and C. Misiano, “Nickel replicated multilayer optics for soft and hard X-ray telescopes,” Proc. SPIE 4012, 284–293 (2000). [CrossRef]
3. G. Pareschi, O. Citterio, M. Ghigo, F. Mazzoleni, P. Gorenstein, S. Romaine, and G. Parodi, “Replication by Ni electroforming approach to produce the Con-X/HXT hard X-ray mirrors,” Proc. SPIE 4851, 528–537 (2003). [CrossRef]
4. M. P. Ulmer, Y. Matsui, D. K. Bedford, G. M. Simnett, and P. Z. Takacs, “Electroformed grazing incidence x-ray mirrors for a mirror array telescope,” Appl. Opt. 26(18), 3852–3857 (1987). [CrossRef] [PubMed]
5. Y. Tawara, K. Yamashita, H. Kunieda, K. Tamura, A. Furuzawa, K. Haga, N. Nakajo, T. Okajima, H. Takata, P. J. Serlemitsos, J. Tueller, R. Petre, Y. Soong, K. Chan, G. S. Lodha, Y. Namba, and J. Yu, “Development of multilayer supermirror for hard X-ray telescope,” Proc. SPIE 3444, 569–575 (1998). [CrossRef]
6. Y. Tawara, K. Yamashita, Y. Ogasaka, K. Tamura, K. Haga, T. Okajima, S. Ichimaru, S. Takahashi, A. Gotou, H. Kitou, S. Fukuda, Y. Tsusaka, H. Kunieida, J. Tueller, P. J. Serlemitsos, Y. Soong, K. Chan, S. M. Owens, B. Barber, E. Dereniak, and E. E. Young, “A hard X-ray telescope with multilayer supermirrors for balloon observations of cosmic X-ray sources,” Adv. Space Res. 30(5), 1313–1319 (2002). [CrossRef]
7. Y. Ogasaka, K. Tamura, T. Miyazawa, Y. Fukaya, T. Iwahara, N. Sasaki, A. Furuzawa, Y. Haba, Y. Kanou, D. Ueno, H. Kunieda, K. Yamashita, R. Shibata, T. Okajima, J. Tueller, P. Serlemitsos, Y. Soong, K.-W. Chan, E. Miyata, H. Tsunemi, K. Uesugi, Y. Suzuki, and Y. Namba, “Thin-foil multilayer-supermirror hard X-ray telescope for InFOCμS/SUMIT balloon experiments and NeXT satellite program,” Proc. SPIE 6688, 668803, 668803-8 (2007). [CrossRef]
8. G. Pareschi, S. Basso, O. Citterio, M. Ghigo, F. Mazzoleni, D. Spiga, W. Burkert, M. Freyberg, G. D. Hartner, G. Conti, E. Mattaini, G. Grisoni, G. Balsecchi, B. Negri, G. Parodi, and A. Marzorati, “Development of grazing-incidence multilayer mirrors by direct Ni electroforming replication: a status report,” Proc. SPIE 5900, 590008, 590008-12 (2005). [CrossRef]
9. M. P. Ulmer and W. R. Purcell, JrJ. E. A. Loughlin and M. P. Kowalski, “Electroform replication used for multiple x-ray mirror production,” Appl. Opt. 23(23), 4233–4236 (1984). [CrossRef] [PubMed]
10. M. P. Ulmer, R. Altkorn, M. E. Graham, A. Madan, and Y. S. Chu, “Production and performance of multilayer-coated conical x-ray mirrors,” Appl. Opt. 42(34), 6945–6952 (2003). [CrossRef] [PubMed]
11. M. P. Ulmer, R. Altkorn, A. Madan, M. Graham, Y.-W. Chung, A. Krieger, C. Liu, B. Lai, D. C. Mancini, and P. Takacs, “The fabrication of Wolter I multilayer coated optics via electroforming: an update,” Proc. SPIE 3773, 113–121 (1999). [CrossRef]
12. Y. Soong, L. Jalota, and P. J. Serlemitsos, “Conical thin foil X-ray mirror fabrication via surface replication,” Proc. SPIE 2515, 64–69 (1995). [CrossRef]
13. M. A. Jimenez-Garate, W. W. Craig, C. J. Hailey, F. E. Christensen, and A. Hussain, “Fabrication, performance, and figure metrology of epoxy-replicated aluminum foils for hard x-ray focusing mulitlyaer-coated segmented conical optics,” Opt. Eng. 39(11), 2982–2994 (2000). [CrossRef]
14. Y. Soong, K. Chan, and P. J. Serlemitsos, “Recent advance in segmented thin-foil X-ray optics,” Proc. SPIE 4496, 54–61 (2002). [CrossRef]
15. A. Furuzawa, K. Yamashita, H. Kunieda, Y. Tawara, K. Tamura, K. Haga, N. Nakamura, N. Nakajo, H. Takata, T. Okajima, Y. Ogasaka, G. S. Lodha, and Y. Nanba, “Replication of multilayer supermirror,” Proc. SPIE 3444, 576–582 (1998). [CrossRef]
16. D. L. Windt, F. E. Christensen, W. W. Craig, C. Hailey, F. A. Harrison, M. Jimenez-Garate, R. Kalyanaraman, and P. H. Mao, “Growth, structure, and performance of depth-graded W/Si multilayers for hard X-ray optics,” J. Appl. Phys. 88(1), 460–470 (2000). [CrossRef]
17. K. K. Madsen, F. E. Christensen, C. P. Jensen, E. Ziegler, W. W. Craig, K. Gunderson, J. E. Koglin, and K. Pedersen, “X-ray study of W/Si multilayers for the HEFT hard X-ray telescope,” Proc. SPIE 5168, 41–52 (2004). [CrossRef]
