The invention concerns a process for the production of broad band elements, especially broad band mirrors, for the ultra violet to hard x-ray wavelength range, especially the extreme ultra violet wavelength range according to the claims. The invention is also related to optical broad band elements, especially broad band mirrors.
In the ultra violet to hard x-ray wavelength range, especially in the extreme ultra violet range (approx. 10 to 100 nm) multilayer systems are used for optical elements as a rule. To this end layers are arranged successively with their respectively constant thickness out of a low and high absorbent material. Here the thicknesses of the individual layers are selected in such a way that the Bragg condition for a determined wavelength is fulfilled and on the other hand as little radiation as possible is absorbed. Radiation is reflected at each high absorbent layer. The individual part beams build up in a constructive way, so that the reflectivity is high for a determined wavelength or energy. The wavelength or energy range that is reflected or is diffracted in case of high energy radiation in Laue geometry, can be too narrow for many applications.
This is the case when broad band sources are used and either a maximally integrated reflectivity of an optical system is desired (e.g. with microscopy or lithography) or when radiation of differing energy or differing angles of incidence are to be reflected altogether (e.g. collector mirrors or premonochromators) or when the rotation of multilayer systems for reflecting differing wavelengths is not wanted (e.g. optical systems of satellites).
In order to further improve the multilayer systems, especially so as to receive rectangular reflectivity profiles, there has been a move over to using multilayer systems without constant layer thicknesses.
The distribution of layers can be found inter alia by two varying means. In E. Ziegler et al., SPIE Vol. 3737, 386 (1999), there are described between 10 and 30 keV for x-rays, whereby the thicknesses distribution of the layers in the multilayer system can be calculated through analytic recursive procedure. Moreover, it is explained there how it is possible to optimize broad band multilayer mirrors for application with synchrotron radiation, while in particular selecting materials which are heat and radiation resistant to a great extent. Here wolfram and osmium are preferred as highly absorbent materials.
In P. van Loevezijn et al., Appl. Opt; 35, 3619 (1996), the thickness distribution of the layers for a broad band reflector for the soft x-ray range (band width 13-19 nm) is optimized numerically. Iteration is started with an ordered layer thickness distribution. At every step in the iteration a layer is picked out at random and has its thickness altered by a random amount. Moreover, this layer is permutated with the layers of the same material in closest proximity. The thicknesses distribution which leads to the highest reflectivity serves as the initial distribution for the next iteration step.
With the help of the methods described it is possible to design optical broad band elements whose reflectivity is essentially rectangular and whose integrated reflectivity should lie up to four times or more above that of the multilayer systems with a constant layer thicknesses distribution.
On conversion to actual optical broad band elements however there occur a number of problems. For example, in the extreme wavelength range and soft longwave range use is made preferably of molybdenum and silicon as layer material. With these materials it has been known for some time that they interact with each other at their border stratum with Mo and Si diffusing into each other and forming either one of their silicides, Mo5Si3 or MoSi2, or a mixture of these suicides. In some cases the mixture can contain additional pure Mo or Si.
In EP 0 280 299 B1 this problem was solved by the application of an intervening layer of hydrogen on to every layer of molybdenum or silicon, so saturating the surface of the respective layer. As the absorption coefficient for hydrogen is very low, it was assumed that these intervening hydrogen strata would not have an effect on the reflectivity performance of the multilayer system. For mass-production of optical broad band elements e.g. for use in lithography, this approach is less suited, as constant working with hydrogen is connected with an increased risk of explosion. Also system saturated with hydrogen are not very stable and changing their parameters with time.
Another problem arises in that oxide strata or adhesive layers can form on the surface of the broad band element, which also have a negative effect on the reflectivity performance. Interlayers work as actual layers, they introduce additional interfaces in the period at the positions different from that in the ideal multilayer with no reaction between layers. As a result the angular or wavelength reflectivity profile of the broad band element becomes strongly distorted leading to unacceptable deviation from the desired reflectivity response.
These deviations of the actual multilayer systems from the calculated multilayer systems lead to reflectivity losses of several percent. This is particularly detrimental to the use of optical broad band elements in lithography, as a large number of optical elements are successively linked in series in lithography systems. Every individual optical element leads to a certain loss of intensity. The individual losses compound each other in severity. Working on the assumption that there are three optical broad band elements connected successively in series, whose actual reflectivity lies at 5% below the reflectivity of the ideal broad band elements (e.g. 45% instead of 50%), there would emerge across all three broad band elements an intensity loss of 23%.
Given this background, it is the task of the present invention to prepare a process by which optical broad band elements can be produced which differ from the ideally calculated broad band elements as little as possible in their structure and also in their reflectivity performance.
This task is met by a process according to the claims.
According to a first alternative of the invention the materials are chosen in such a way that two successive layers interact with each other as little as possible or controllably.
So as to optimize broad band elements, the materials used for the layers are selected on the basis of whether and how they interact with each other. This interaction can either be in a mixture which leads to a solid solution or consists of a chemical compound. In the invention those materials are chosen which either interact with each other as little as possible or interact with each other in a way that is controllable, so that we may proceed on the basis of a determined thickness of an intervening stratum especially less than 0,3 nm between two layers.
According to a second alternative of the invention the numerical optimization takes into account interlayers of a certain thickness and composition. The resulting interlayer that controllably occurs is also taken into account in the optimization procedure.
The term interlayer means layer with constant depth composition. Insofar interlayers work as actual layers which shift reflecting interfaces in the period so that these interfaces become situated at different positions in the period compared to an ideal multilayer with no interlayers. Interlayer shouldn't be mixed up with non-sharp boundaries which do not change the effective reflecting interfaces compared to the ideal multilayer.
In order to achieve for example a flat angular or wavelength reflectivity response the existence of interlayers is accepted independent of their thickness because at least their thickness is taken into consideration during the optimization procedure.
The interlayer can be a native interlayer produced by the interaction of the used multilayer material or an artificial interlayer that is deposited between the layers of the multilayer system.
In case the thickness of an interlayer is unknown a corresponding layer system made of the selected materials is produced and a measurement of at least the thickness of the interlayer is conducted.
Preferably, the measurement of thickness of the interlayer/s is conducted by grazing incidence x-ray reflectometry, as it is described for example in “Determination of the layered structure in Mo/Si multilayers by grazing incidence x-ray reflectometry”, A. E. Yaksin et al., Physica B Volume 283, pp. 143-148 (April 2000).
As far as a simulation method is discussed in this publication it relates to the design of multilayer mirrors with high reflectance at near normal incidence. These mirrors that are no broad band elements have a periodical layer structure with two materials and two interlayers in the period. Only the thickness of the thickest layer in the period is modified laterally or from sample to sample, to be able to extract information about the layered structure.
Moreover it is intended to have a cap layer on the optical broad band elements, whose material is also selected according to the criterion that it does not interact with the environment or only in a controlled way, and so forms no adhesive stratum that might have an unforeseeable thickness, and that it does not oxidize or oxidizes only in a controlled way, so that we may proceed on the basis of a determined oxide stratum thickness.
Two or more materials are chosen for the formation of the multilayer systems themselves. These are brought together in a determined sequence. A stacked arrangement in which every layer of a determined material occurs at least once in a determined material is called a set or period. Preferably the thicknesses of the layer vary so that the multilayer system is a so-called depth graded multilayer system.
With conventional molybdenum-silicon multilayer systems it would be a matter of a set of a molybdenum-silicon pair of layers. In selecting the order of the layers care is taken that only those layers succeed each other which are made of materials which interact with each other as little as possible or only in a controlled way. The condition that the layer materials should interact as little as possible or only controllably does not have to apply to all materials interchangeably but only by pairs.
Subsequently, on the one hand the number N of sets must be established, where according to every wished for reflectivity profile only one single set is allowed. On the other hand, the layer thicknesses and the cap layer thickness has to be established. Establishing the number of sets and in particular the layer thicknesses and the cap layer thickness can be achieved by the recursive analytical procedure, as described in E. Ziegler et al. If necessary, the cap layer thickness can for the time being be left externally and affixed separately. It is possible to numerically optimize the layer thicknesses as per P. van Loevezijn et al. It is of decisive importance to take into consideration the cap layer thickness both during initial calculations, and, if present, intervening strata between two controllably interacting material layers and potentially an oxide or adhesive stratum on the cap layer should be taken into account as separate layers.
After the layer parameters have been established, the layers are applied to a substrate. For this purpose all known procedures for stratum-coating are suitable, as for example electron radiation vaporization, magnetron sputtering or ion radiation sputtering, etc.
Finally, the cap layer is applied to the multilayer system. The great advantage of the process according to the invention consists in the fact that optical broad band elements can now be produced which on the one hand only differ slightly from the calculated broad band elements. The differences lie below 1% and are essentially attributable to tolerances with the stratum-coating process and surface asperities of the individual layers. On the other hand, the optical broad band elements produced according to the process in the invention show integrated reflectivities, which lie at some percent above the integrated reflectivities of optical broad band elements, where in calculations potential interactions between the individual layers or with the environment were neglected.
In selecting only two materials for layer formation molybdenum carbide and silicon have proved to be particularly suitable. Both these materials do not interact with each other and so form a clearly defined thin border stratum.
Preferably at least three materials A, B, C are selected for the formation of layers, of which at least two materials can interact with each other and at least one material C does not interact with A and B.
The layer formed of material C can be a native interlayer or it is a layer deposited between the layers of materials A and B. In case material C is a native layer it comes into being when e.g. the following materials are used for A, B, C respectively: Mo, Si, MoxSiy; MO, C, MoxCy; W,Si, WxSy.
In case the layer formed of material C is an artificial layer that has an effect as an barrier layer so that no interdiffusion effects between the multilayer material can occur, it is preferred to use Mo5Si3, Si3N4, Rh5Si3, Rh2Si, RhSi, Ru2Si, Ru2Si, RuS1, MO2C, SiC, Nb4Si, Nb5Si3, Y5Si3, YSi, YSi2, diamond-like C, Zr2Si, Zr5Si3, MoB, B4C or B.
In case of artificial layers that are deposited one does not need to measure the deposited thickness, if the thickness can be controlled precisely during the deposition process.
The thickness of these artificial layers can be in the range of 0,5 to 1,5 nm, preferably between 0,8 and 1,2 nm.
Particularly in the production of optical broad band elements for the extreme ultra violet wavelength range, it has proved advantageous to select for layer materials molybdenum and/or silicon and an inert compound on the basis of molybdenum and/or silicon. In the case of the compound, it can be a question of a solid solution or also a chemical compound. Molybdenum carbide and/or silicon carbide or even MoxSiy are especially preferred as a compound. In MoxSiy the values of x and y are not to be specified since some deviation from the nominal stoichiometry may be allowed. Molybdenum carbide and silicon carbide are inert compounds compared with molybdenum and silicon. In the case of MoxSiy it is either one of the silicides, Mo5Si3 or MoSi2, or a mixture of these suicides. In some cases the mixture can contain additional pure Mo or Si. While during the stratum-coating process molybdenum and silicon are applied simultaneously, there forms a stratum of MoxSiy which is already saturated, so that no molybdenum and silicon can diffuse into each other from adjacent pure layers.
On the whole it is preferred to choose layer materials from the group of materials made up by molybdenum, rubidium, rhodium, ruthenium, yttrium, strontium, silicon, silicon carbide, molybdenum carbide, molydenum boride, TiN, C, Si3N4, B4C, BN, rubidium-hafnium or rubidium-sulphide. It should simply be taken care that molybdenum and silicon layers are not arranged in succession to each other.
It has proved especially advantageous for the formation of a cap layer to use the materials: silicon, rhodium, ruthenium, gold, silicon dioxide, silicon carbide, molybdenum carbide, MoB, MoxSiy, C, TiN, Si3N4, B4C and BN.
The object of the invention is also solved by an optical broad band element comprising a substrate, a cap layer system if necessary and a depth-graded multilayer system having periods consisting of at least two different materials. The materials of two successive layers do not interact so that the multilayer system is free of intermediate layers. The already mentioned materials can be used. Preferred materials are MO2C and Si.
According to another embodiment the layer system of the optical broad band element comprises a substrate, a cap layer system if necessary and a multilayer system having periods consisting of at least two different materials and is characterized in that the layer system is designed by incorporation of different materials into the numerical optimization procedure.
According to another embodiment the optical broad band element has a substrate, a cap layer system if necessary and a depth-graded multilayer system having periods consisting of at least three different materials A, B and C. At least two materials A, B can interact with each other and at least one material C does not interact with A and B and the layer formed by material C is a native layer or is deposited between the layers of materials A and B. Preferred materials are already mentioned in connection with the process claims.
The invention is to be explained by means of the following examples and drawings. These indicate:
For producing a broad band reflector for a wavelength of 13.4 nm and an angle band width of 20°, molybdenum carbide and silicon were chosen as materials. Molybdenum carbide and silicon are two materials which do not interact. Silicon was also chosen as cap layer material. No adhesive stratum forms on silicon, but simply a negligibly thin silicon dioxide stratum, should the broad band reflector not be kept in a vacuum. Both silicon and molybdenum carbide, on the basis of their absorption coefficient, are suitable for the production of optical elements in the extreme ultra violet wavelength range.
Given a set or period number N=100 and a layer thickness distribution which was obtained according to E. Ziegler et al., the thickness distribution was optimized as per P. van Loevezijn et al. Their resulting thickness distribution is represented in
Three-material depth-graded multilayer deposited in the sequence: Mo/MoSi2/Si/MoSi2. Design contains variable thicknesses of Mo and Si, with the thickness of MoSi2 being kept constant at 1.0 nm (
A comparison of the reflectivity curve in
In
In general for a real multilayer, the flat reflectivity profile can not be achieved basing on designs done for an ideal stack that does not take into account interactions of materials in the multilayer. These interactions will cause oscillations in reflectivity profile as shown in
Number | Date | Country | Kind |
---|---|---|---|
02006984.5 | Mar 2002 | EP | regional |
This application is a continuation-in-part application of International Application No. PCT/EP03/03200, filed Mar. 27, 2003 and published as WO 03/081187 on Oct. 2, 2003, which claims the priority to European Application No. 02006984.5, filed Mar. 27, 2002.
Number | Date | Country | |
---|---|---|---|
Parent | PCT/EP03/03200 | Mar 2003 | US |
Child | 10950936 | Sep 2004 | US |