EP2954372A2

EP2954372A2 - Blank of tio2-sio2 glass for a mirror substrate for use in euv lithography and method for the production thereof

Info

Publication number: EP2954372A2
Application number: EP14702598.5A
Authority: EP
Inventors: Stephan Thomas; Klaus Becker; Stefan Ochs
Original assignee: Heraeus Quarzglas GmbH and Co KG
Current assignee: Heraeus Quarzglas GmbH and Co KG
Priority date: 2013-02-11
Filing date: 2014-02-04
Publication date: 2015-12-16
Also published as: JP2016511211A; CN104995557A; KR101922765B1; JP6328665B2; TWI624435B; US20150376049A1; WO2014122111A2; KR20150117675A; TW201446670A; US9522840B2; DE102013101328B3; CN104995557B; WO2014122111A3

Abstract

To provide a blank of TiO2-SiO2 glass for a mirror substrate for use in EUV lithography, in which the need for adaptation to optimize the progression of the coefficient of thermal expansion, and consequently also the progression of the zero crossing temperature Tzc, is low, the TiO2-SiO2 glass has at a mean value of the fictive temperature Tf in the range between 920°C and 970°C a dependence of its zero crossing temperature Tzc on the fictive temperature Tf which, expressed as the differential quotient dTzc/dTf, is less than 0.3.

Description

Blank made of Ti0 ₂ -Si0 ₂ glass for a mirror substrate for use in EUV lithography and method for its production

Description Technical area

The present invention relates to a blank made of TiO ₂ -SiO ₂ glass for a mirror substrate for use in EUV lithography

Furthermore, the invention relates to a method for producing such a blank or a shaped body as a semi-finished product for its production.

State of the art

In EUV lithography, highly integrated structures with a line width of less than 50 nm are produced by means of microlithographic projection devices. In this case, laser radiation from the EUV range (extreme ultraviolet light, also called soft x-ray radiation) with wavelengths around 13 nm is used. The projection devices are equipped with mirror elements consisting of glass containing high-silica and titanium oxide (also referred to below as "TiO ₂ - SiO ₂ glass"), which are provided with a reflective layer system low linear coefficient of thermal expansion (referred to for short as "CTE", coef- ficient of thermal expansion), which can be set by the concentration of titanium. Usual titanium oxide concentrations are between 6 and 9 wt .-%.

Such a blank made of synthetic, titanium-doped siliceous glass and a production method thereof are known from DE 10 2004 015 766 A1. The TiO ₂ -SiO ₂ glass is produced by flame hydrolysis of titanium and silicon-containing starting substances and contains 6.8% by weight of titanium oxide. It is mentioned that the hydroxyl group content of the glass thus produced is 300% by weight. ppm rarely falls below. To increase the radiation resistance of the glass, it is proposed to lower the concentration of the hydrogen contained in the preparation by heating to values below 10 ¹⁷ molecules / cm ³ . For this purpose, the glass is heated to a temperature in the range between 400 and 800 ° C and held at this temperature for up to 60 hours. One of the plane surfaces of the mirror substrate is mirrored, wherein a plurality of layers is produced one above the other.

During normal use of the mirror substrate whose top is mirrored. The maximum (theoretical) reflectivity of such an EUV mirror element is about 70%, so that at least 30% of the radiant energy in the coating or in the near-surface layer of the mirror substrate is absorbed and converted into heat. This results in the volume of the mirror substrate to an inhomogeneous temperature distribution with temperature differences, which can be up to 50 ° C, according to the literature. For the least possible deformation, it would therefore be desirable if the glass of the mirror substrate blank had a CTE which would be zero over the entire temperature range of the operating temperatures occurring in use. In fact, for Ti-doped silica glasses, the temperature range with a CTE around zero may be very narrow. The temperature at which the thermal expansion coefficient of the glass is equal to zero is also referred to below as the zero-crossing temperature or as T _Z c (temperature zero crossing). The titanium concentration is usually adjusted to give a CTE of zero in the temperature range between 20 ° C and 45 ° C. Volume ranges of the mirror substrate of higher or lower temperature than the preset T _Z c expand or contract so that, despite the overall low CTE of the TiO ₂ -SiO ₂ glass, deformations occur under which the image quality of the mirror suffers.

There is therefore no lack of proposals to counteract the deterioration of the optical imaging by inhomogeneous temperature distribution in the mirror substrate blank. For example, in the case of EP 0 955 565 A2 known Mirror provided a metallic substrate material. Due to the good thermal conductivity of the metal, the heat introduced in the mirror is efficiently dissipated via the rear side of the metal substrate, preferably by means of a cooling device. Although it is possible in this way to reduce thermally induced mirror deformations, it is not possible to avoid aberrations. There are still very significant aberrations.

In DE 103 59 102 A1 (~ US 2005/0185307 A1) homogeneity requirements are defined for a SiO ₂ -TiO ₂ glass, which is to fulfill the glass. To this

Purpose of the glass should have a determined by the titanium content location-dependent thermal expansion coefficient. This should also be as independent as possible of the temperature, defined by the amount of the average slope m of less than 1, 5 x 10 ^"9 K ^{" 2} . However, it is not stated how this low temperature dependence of the CTE can be achieved. WO 201 1/078414 A2 provides, in the case of a blank for a mirror substrate or for a mask plate made of SiO ₂ -TiO ₂ glass, to adjust the concentration of titanium oxide over the thickness of the blank stepwise or continuously to the temperature distribution which occurs during operation. that the condition for the zero-crossing temperature T _Z c are fulfilled at each point, that is, the thermal expansion coefficient for the locally adjusting temperature is substantially equal to zero. A CTE is defined as substantially equal to zero if the remaining linear expansion in operation at each point is 0 +/- 50ppb / ° C. This is to be achieved by varying the concentration of titanium- or silicon-containing starting substances in the production of the glass by flame hydrolysis in such a way that a predetermined concentration profile is established in the blank.

Technical task

The methods for optimizing the T _z c by local variation of the titanium concentration require precise knowledge of the tem- temperature distribution over the volume of the component to be optimized and are associated with an enormous design and adaptation effort for the individual component. It should be noted that a projection lens contains a variety of mirrors of different size and shape, which not only have flat, but from convex or concave curved, mirrored surfaces with adapted to the specific use outer contours. The actually occurring temperature profile over the volume of each component to be optimized during operation depends on the specific operating conditions and the environment and can be determined exactly only in the fully assembled projection objective under real operating conditions. The replacement of individual components of a ready mounted projection lens is technically hardly possible.

To make matters worse, that the CTE and thus scaling T _Z c depend not only on the titanium oxide content but also on the hydroxyl group content and on the fictitious temperature of the glass. The fictive temperature is a glass property that represents the ordered state of the "frozen" glass network.A higher fictitious temperature of the TiO ₂ -SiO ₂ glass is accompanied by a lower order state of the glass structure and a greater deviation from the most favorable structural arrangement.

The fictive temperature is determined by the thermal history of the glass, especially the last cooling process. In the process, different conditions inevitably result for near-surface regions of a glass block than for central regions, so that different volume regions of the mirror substrate blank already have different fictitious temperatures due to their different thermal history. The distribution of the fictive temperature over the blank volume is therefore always inhomogeneous. A certain equalization of the course of the fictitious temperature can be achieved by tempering. Annealing processes, however, are energy and time consuming.

A further complicating factor is that the resulting fictitious temperature also depends on the composition of the TiO ₂ -SiO ₂ glass, and in particular on the hydroxyl group content and the titanium oxide concentration. Even by very careful and long tempering the course of the fictitious temperature on the blank volume can not be homogenized, if the composition is not completely homogeneous. However, this is not exactly the case with the hydroxyl group content, which can be changed by drying measures.

The invention has for its object to provide a blank for a mirror substrate of a TiO ₂ -SiO ₂ glass, in which the need for adaptation to optimize the course of CTE and thus the course of T _Z c is low.

It is a further object of the invention to provide a process for the production of the blank according to the invention.

General description of the invention

With regard to the blank, this object is achieved on the basis of a blank of the aforementioned type in that the TiO ₂ - SiO ₂ glass at a mean value of the fictitious temperature T _f in the range between 920 ° C and 970 ° C, a dependence of its zero crossing temperature T. _Z c of the fictitious temperature T _f , which, expressed as a differential quotient dT _Z c / dT _{f is} less than 0.3.

An inhomogeneous distribution of the fictitious temperature over the volume of the blank makes it difficult to set the most homogeneous possible distribution of the CTE or T _Z c. Instead of an elaborate equalization of the fictitious temperature on the blank volume or a complex adaptation of the CTE to a given profile of the fictitious temperature, a certain decoupling of the dependence of the CTE and thus the zero crossing temperature T _Z c of the fictitious temperature is sought. This measure is neither known nor suggested in the prior art and can already be considered as a first step in the direction of the invention.

Basically, TiO ₂ -SiO ₂ glasses show a decrease in CTE and an increase in T _z c at the fictitious temperature. The decoupling of the dependence of T _Z c of the fictitious temperature is shown in a diagram in which T _Z c is plotted as a function of the fictitious temperature, thus in a flat slope of the function T _Z c = f (T _f ). This tangential slope according to the invention is less than 0.3, preferably less than 0.25, at any point within the fictitious temperature interval of 920 to 970 ° C.

It is a material-specific property. This is in the blank according to the invention irrespective of whether the mean fictitious temperature is actually in the said temperature interval. If, for example, a higher fictitious temperature is set, slopes of the function Tzc = f (T _f ) may result where the differential quotient is higher than 0.3. The blank according to the invention can be recognized by the fact that the desired decoupling is ensured if it has an average fictitious temperature in the range between 920 ° C and 970 ° C.

The graph of Figure 2 shows a comparison of the function T _Z c = f (T _f ) in a commercial TiO ₂ -SiO ₂ glass (curve A) and a TiO ₂ -SiO ₂ glass according to the invention (curve B). Curve B shows, within the temperature interval from 920 to about 990 ° C at each point, a tangent slope of less than 0.3 expressed by the differential quotient dT _Z c / dT _f , whereas curve A does not show such a low point at any single point within that interval Slope shows.

This proves for the TiO ₂ -SiO ₂ glass of the blank according to the invention a certain insensitivity of the T _Z c from the fictitious temperature, which is also referred to herein as "decoupling" of the dependence of the zero crossing temperature on the fictitious temperature despite a profile of the fictive temperature that would otherwise either be no longer acceptable or that would have to be compensated by adjusting the course of the CTE (for example, by changing the concentrations of titanium oxide or hydroxyl groups) used to produce a low-deformation mirror substrate Requirements for annealing at the setting of a homogeneous as possible course of the fictitious temperature less or it results in the same effort a more homogeneous distribution of T _Z c (as far as it is due to the fictitious temperature).

A common measuring method for determining the fictitious temperature by means of a measurement of the Raman scattering intensity at a wave number of about 606 cm ^"1 is in" Ch. Pfleiderer et al., The UV-induced 210 nm absorption band in fused silica with different thermal history and Stoichiometry; Journal of Non-Cryst., Solids 159: 143-145 (1993).

The desired decoupling is achieved by a specific method for producing the TiO ₂ -SiO ₂ glass, which will be explained in more detail below. The degree of decoupling of the dependence of the T _Z c on the fictive temperature is to some extent dependent on the absolute level of the fictitious temperature itself. At low fictitious temperature, the desired decoupling easier than at high fictitious temperature succeed. The requirements are therefore higher, and the decoupling achieved is particularly noticeable when the notional temperatures of the blank are in the upper range of the temperature interval, for example, above 940 ° C.

The fictitious temperature again depends significantly on the same annealing treatment from the hydroxyl group content. The higher the hydroxyl group content, the lower the fictitious temperature that arises with the same tempering treatment. If only aiming at setting the lowest possible fictitious temperature, a high hydroxyl group content would be preferable in and of itself. On the other hand, hydroxyl groups to achieve other, in particular optical or mechanical properties to some extent undesirable. As a suitable compromise between these other properties and a low fictive temperature, it has proven useful if the TiO ₂ -SiO ₂ glass has an average hydroxyl group content in the range of 200 to 300 ppm by weight.

This is a medium-high hydroxyl group content. The prerequisite for the setting of this average hydroxyl group content is the production of the TiO ₂ -SiO ₂ glass according to the so-called "soot method" obtained as an intermediate a porous soot body containing hydroxyl groups due to production. These can be removed by reactive chemical treatment by means of halogens to the desired extent. Preferably, however, the drying is carried out by thermal treatment of the soot body under vacuum.

The hydroxyl group content (OH content) is determined by measuring the IR absorption by the method of D. M. Dodd et al. ("Optical Determinations of OH in Fused Silica", (1966), p. 391 1).

Moreover, it has proved to be advantageous if the TiO ₂ -SiO ₂ glass has an average hydrogen concentration of less than 5 × ¹⁰ ¹⁶ molecules / cm ³ , preferably an average hydrogen concentration of less than 1 × 10 ¹⁶ molecules / cm ³ .

The mean hydrogen concentration is determined by Raman measurements. The measuring method used is described in: Khotimchenko et al., "De- termining the Content of Hydrogen Dissolved in Quartz Glass Using the Methods of Raman Scattering and Mass Spectrometry" Zhurnal Prikladnoi Spectroscopy, Vol. 46, No. 6 (June 1987), Pp. 987-991.

Because of the above-described decoupling of T _Z c and fictitious temperature, the mirror substrate blank according to the invention made of TiO ₂ -SiO ₂ glass is relatively insensitive to an inhomogeneous distribution of the fictive temperature over the volume of the blank. The overall low thermal and spatial dependence therefore also facilitates the adaptation of the T _Z c to an inhomogeneous temperature profile occurring in practice.

A further adaptation is provided in a preferred embodiment, in which the blank is delimited by a top side and a bottom side, wherein the TiO ₂ -SiO ₂ glass has a non-homogeneous course of the titanium oxide concentration between top side and bottom side.

In this embodiment of the blank according to the invention, in addition to the inhomogeneous course of the fictive temperature, the titanium oxide concentration of the glass is varied and thereby, for example, the T _zc adapted to the operating temperature setting.

The blank is in this case designed as a composite body which comprises a first shaped body of TiO ₂ -SiO ₂ glass with a first titanium oxide concentration and a second shaped body of TiO ₂ -SiO ₂ glass with a second titanium oxide concentration. which is connected to the first molded body.

In the simplest case, two shaped bodies suffice to be able to adapt the titanium dioxide concentration and the T _Z c to the inhomogeneous temperature profile during operation with sufficient accuracy. The moldings used as semifinished product consist of TiO ₂ -SiO ₂ glass according to the present invention, but with different titania concentration. To complete the mirror substrate blank (or a part thereof), the moldings are joined together using known methods. A higher fictitious temperature acts on the T _Z c in the same manner as a higher titanium oxide concentration, that is, increasing T _Z c for the respective TiO ₂ -SiO ₂ glass.

For this reason, in an advantageous embodiment of the blank _according to the invention, the fictitious temperature is also used to adapt the T _zc to a predetermined temperature profile. In this case, the first molded body has a first mean fictitious temperature and the second molded body has a second mean fictitious temperature, wherein the first and second fictitious temperature differ from each other. Apart from the fictitious temperature and the titanium oxide concentration, the hydroxyl group _content has an effect on the T _zc and can be used as an additional parameter for adaptation to the given temperature profile.

To set the respective fictitious temperature, the first and second shaped bodies are subjected to an annealing process prior to their connection, such that the fictitious temperatures that arise in the process differ from one another. The moldings are plate-shaped with a maximum thickness of 60 mm.

In the determination of CTE and titanium oxide concentration on plate-like shaped bodies, a measured value averaged over the thickness is more meaningful in comparison to a measurement over the greater thickness of the complete mirror substrate blank. The thinner the shaped body is (more specifically, the shorter the measuring distance), the more meaningful and accurate is the measured mean value.

According to the invention, the method for producing the blank according to the invention comprises the following method steps: (a) producing a first porous soot body of SiO ₂ with a first concentration of titanium oxide by flame hydrolysis of silicon and titanium-containing starting substances,

(b) drying and sintering the first soot body such that a first TiO ₂ -SiO ₂ glass having the first titanium oxide concentration is obtained, wherein the average hydroxyl group content is set to less than 300 ppm by weight,

(c) homogenizing the first TiO ₂ -SiO ₂ glass in a homogenization process in which the TiO ₂ -SiO ₂ glass is heated in an oxidative atmosphere to a temperature of more than 2000 ° C., thereby softened and reshaped, so that an average hydrogen concentration of less than 5 × ^{10 16} molecules / cm ^{3 is} established,

(d) forming the first TiO ₂ -SiO ₂ glass having an average hydroxyl group content in the range of 200 to 300 ppm by weight and an average hydrogen concentration of less than 5 × 10 ¹⁶ molecules / cm ³ to give a molded article, and

(e) annealing the shaped body in such a way that the TiO ₂ -SiO ₂ glass has a dependence of its zero crossing temperature T _Z c on the fictitious temperature T _f at a mean fictitious temperature T _f in the range between 920 ° C. and 970 ° C. expressed as the differential quotient dT _zc / dT _{f is} less than 0.3. The shaped body of TiO ₂ -SiO ₂ glass thus obtained can either be used directly as a mirror substrate blank after mechanical further processing, such as grinding and polishing, or it serves as a preliminary product for further processing into the blank. Due to its structure and chemical composition, the TiO ₂ -SiO ₂ glass produced in this way exhibits a CTE and a zero-crossing temperature Tzc, which are relatively insensitive to an inhomogeneous course of the notional temperature over the volume of the blank and therefore a relatively simple design to allow adaptation of the thermal expansion to a temperature profile which occurs in practical use over the thickness of the blank. This will be explained in more detail below:

Effect of chemical composition and glass structure on T ₇ c

The thermal expansion coefficient CTE and the zero-crossing temperature Tzc of TiO ₂ -SiO ₂ glass depend on the titanium concentration, the hydroxyl group content and the fictitious temperature. Since the fictive temperature is characterized by the thermal history of the glass, it is always inhomogeneous over the volume of the mirror substrate blank and can basically only be more or less aligned by energy-consuming and time-consuming tempering processes. The method according to the invention makes it possible to produce a TiO ₂ -SiO ₂ glass which has a low dependency of CTE and T _Z c on the fictitious temperature, so that a certain decoupling is achieved to that extent.

The mean hydroxyl group content of the TiO ₂ -SiO ₂ glass in the range from 200 to 300 ppm by weight can be adjusted during the production of the glass by the so-called "soot method". As a result, a porous soot body is obtained, the hydroxyl groups in large quantities These can be removed by reactive chemical treatment by means of halogens, but preferably the drying is carried out by thermal treatment of the soot body under vacuum. However, with regard to the chemical composition of the TiO ₂ -SiO ₂ glass and its effects on the dependence of the CTE on the temperature, the soot method turns out to be disadvantageous in other respects. For the vitrification of the soot body, relatively low temperatures of around 1500 ° C suffice. It has been shown that in the TiO ₂ -SiO ₂ glass micro-crystals of rutile (TiO ₂ ) can form, so that areas with high titanium dioxide concentration arise, which have an effect on the thermal expansion of the glass. These microcrystals have a melting point of 1855 ° C.

According to the invention, the TiO ₂ -SiO ₂ glass obtained after vitrification is heated to a temperature at which the rutile microcrystals melt. At the same time, the glass is deformed and homogenized - for example, by twisting - to effect a more homogeneous distribution of TiO ₂ -rich areas. For this purpose, the TiO ₂ -SiO ₂ glass is subjected to a homogenization process in which it is heated to a temperature of more than 2000 ° C. and thereby softened and reshaped.

As a result, a more homogeneous distribution of titanium dioxide is achieved via the volume of the TiO ₂ -SiO ₂ glass and the spatial dependence of the CTE as a result of rutile microcrystal concentrations is reduced.

On the other hand, the high temperatures during the homogenization can cause a partial reduction of Ti ⁴⁺ in Ti ³⁺ . It has been shown that the oxidation state of titanium oxide affects the coordination of the ion within the network structure, and that this change adversely affects the distribution of the titanium oxide - similar to the rutile formation. Therefore, according to the invention, during the homogenization at least at times an oxidatively acting atmosphere is set. In the area of the softened glass mass, an oxidizing gas, such as oxygen, is provided in excess and thus a partial reduction of Ti ⁴⁺ in Ti ^{3+ is} counteracted.

In this way, it is ensured that the shaped bodies produced by process step (d) from the respective TiO ₂ -SiO ₂ glass exhibit a largely homogeneous distribution of titanium oxide with titanium in the tetravalent oxidation state. At the same time, because of the high temperature and the oxidatively acting atmosphere, the hydrogen contained in the preparation is reduced, so that in the TiO ₂ -SiO ₂ glass on average a very low hydrogen concentration of less than 5 × 10 ¹⁶ molecules / cm ³ , preferably less than 1 adjusts x10 ¹⁶ molecules / cm ^3.

In this homogenization, the TiO ₂ -SiO ₂ glass can be brought into a near-net shape, for example in plate form. Often, however, this shaping takes place in a separate shaping process.

As a rule, the shaped body obtained by shaping exhibits a strongly inhomogeneous distribution of the fictitious temperature over its volume. In order to achieve a certain standardization of the fictitious temperature and to set values in the range of 920 ° C to 970 ° C, the shaped body is annealed. Annealing methods that are suitable for setting a fictitious temperature in this temperature range are to be developed by means of fewer and simpler tests. Even after the annealing process, the resulting fictive temperatures in the near-surface region of the top and bottom and the fictitious temperature in the volume (in the middle of the molding) differ from each other, the difference depending on the volume and thickness of the blank and at thicknesses 150 mm in the range of less degrees, typically around 5 ° C.

The TiO ₂ -SiO ₂ glass produced and reworked by means of the above-mentioned measures is characterized by a hydroxyl group content and a hydrogen concentration as specified above, and in particular by a zero crossing temperature Tzc, which lies in the interval between 920 ° C. and 970 ° C. To such a small extent depends on the fictitious temperature, as it was not previously known. This small dependence is expressed as the differential quotient dT _Z c / dT _f , which is less than 0.3. In the homogenization process according to method step (c), the first TiO ₂ -SiO ₂ glass is preferably at least partially heated with a burner flame which contains fuel gas and at least one oxidizing gas. rende component are supplied in an excess amount for complete combustion of the fuel gas.

The burner flame, a gas mixture of fuel gas and a fuel gas oxidizing component, in particular oxygen, burned. On the one hand, the excess of oxidizing component in the gas mixture ensures that the fuel gas burns completely and that an excess remains which counteracts a partial reduction of titanium ⁴⁺ oxide.

The resulting decoupling of the T _zc from the fictive temperature largely mitigates the problem of having to take into account the fictitious temperature as an influencing variable when adapting to an inhomogeneous course of the operating temperature which is set in use, and therefore simplifies this adaptation.

If the adaptation to an inhomogeneous course of the operating temperature on the composition of the TiO ₂ -SiO ₂ glass, such as the Ti concentration or the hydroxyl group content, so inevitably accompanied by changes in the fictive temperature of the glass blank, both in terms the absolute value as well as their distribution over the volume. These changes in fictitious temperature in turn affect T _Z c complicating the adaptation and correct adjustment of T _Z c over the volume. This interference of the fictive temperature is mitigated by the use of the blank according to the invention. This is because changes in the fictive temperature have less of an effect on the course of the T _Z c, so that the glass composition can be adjusted to the desired T _zc more precisely - ie with a reduced _{disruptive effect} due to the fictitious temperature. The adaptation to an inhomogeneous course of the operating temperature is preferably carried out by glass layers which differ in their titanium oxide concentration. The glass layers are obtained by prefabricated, in particular plate-shaped moldings are joined together. The prefabricated moldings are to be more accurately characterized by the usual measuring technique than the finished mirror substrate. So can the titania concentration and the CTE can be relatively easily measured optically or by ultrasonic measurement. However, a value averaged over the measuring path is obtained. In comparison with the measurement on the complete mirror substrate blank, the mean measured values are more meaningful when measured on the plate-like shaped bodies present in the intermediate stage. In the TiO ₂ -SiO ₂ glass of the present invention, the changes in the titanium dioxide concentration cause changes in the fictitious temperature; but these affect less than usual on the setting of T _Z c.

In the following, some particularly preferred process modifications are explained in more detail:

The drying of the soot body is preferably carried out by heating the respective soot body under vacuum to a temperature of at least 1 150 ° C, preferably of at least 1200 ° C.

A high temperature shortens the treatment time required to remove the hydroxyl groups to a level in the range from 200 to 300 ppm by weight.

It has proven useful if the homogenization according to method step (c) comprises a twisting, in which a cylindrical starting body made of the respective TiO ₂ -SiO ₂ glass held between two holders zonewise

Melting temperature brought while the heated zone is worked through by relative movement of the two brackets to each other to form a homogenized in three directions drill body.

For this purpose, the starting body is clamped in a equipped with one or more heating burners glass lathe and homogenized by means of a forming process, as described in EP 673 888 A1 for the purpose of complete removal of layers. In the modification of this method according to the invention, the starting body is locally heated to over 2000 ° C by means of a heating burner with an oxidizing burner flame and thereby softened. The starting body is by relative movement of the two brackets twisted to each other around its longitudinal axis in several directions, the softened glass mass is thoroughly mixed. As a result of this homogenization process, rutile microcrystals present in the starting body are melted, a reduction of Ti ⁴⁺ to trivalent Ti ³⁺ and a concomitant change in coordination are avoided, so that an overall uniform distribution of the titanium oxide concentration is established.

If two or more moldings are joined together, preferably a planar contact surface of the first mold body and a planar contact surface of the second mold body are assembled by wringing and welded together.

This is a "cold bonding process" in which at most the immediate area of the contact surface undergoes considerable heating, since this joining process does not require a hot step, the preset properties, in particular with regard to adjustment of CTE and fictitious temperature, are no longer changed.

Alternatively, the bonding may comprise a joining step in which the upper mold resting on the second mold body is softened in an oven and deformed together therewith.

This is a "hot bonding process" in which the individual shaped bodies, such as plates or rods, are joined together by welding, whereby an adjuvant can be introduced between the contact surfaces in order to facilitate cohesive joining, for example a slip with particles TiO ₂ -SiO ₂ glass, however, may cause a pre-set fictitious temperature to change as a result of the high temperature joining process, and if necessary, the finished mirror substrate will be treated in a subsequent annealing step, depending on the chemical composition of the original Shaped bodies usually have different fictitious temperatures. embodiment

The invention will be explained in more detail with reference to embodiments and a drawing. In detail shows:

FIG. 1 shows a diagram for illustrating the temperature stratification in a mirror substrate with curved surfaces,

FIG. 2 shows a diagram of the dependence of the zero-crossing temperature T _zc on the fictitious temperature T _f for different TiO ₂ -SiO ₂ glasses, that is to say their specific function T _Z c = f (T _f ), and FIG

Figure 3 is a diagram with derivatives (tangent slopes) of the specific

Functions T _Z c = f (T _f ) from the diagram of Figure 2.

The diagram of Figure 1 shows the temperature distribution in a circular mirror substrate with curved surfaces, as in the thermal

Equilibrium when the surface is assumed to be at 50 ° C. "X" denotes the radius, and the y-axis indicates the thickness, in each case in m. The maximum temperature difference is only about 5 degrees Celsius. but it does not arise only between the top and bottom of the substrate, but also between the top and side edge.

This temperature distribution makes it clear how laboriously the titanium oxide concentration within a substrate has to be varied if an exact adaptation of the T _Z c to the temperature distribution is to take place only via the titanium oxide concentration. To make matters worse, due to their production, mirror glass substrate blanks have a certain variation in their fictive temperature over their volume, which has an effect on the CTE.

The mirror substrate blank according to the invention shows a lower temperature dependency of the CTE on the fictive temperature, so that this adaptation effort is completely eliminated or at least less. This will be explained below by way of examples. Production of moldings with different titania concentration and fictitious temperature

Sample 1 a: plate of TiO _? -SiO _? -Glass

By flame hydrolysis of octamethylcyclotetrasiloxane (OMCTS) and titanium isopropoxide [Ti (OPr ') ₄ ] as a feedstock for the formation of SiO ₂ -TiO ₂ particles, a soot body made of synthetic TiO ₂ -SiO ₂ is prepared by the known OVD process ₂ glass, which is doped with about 8 wt .-% TiO ₂ .

The soot body is (cknen T _Tro) at a temperature of 1 150 ° C in a heating furnace with a heating element of graphite in vacuum dehydrated. The graphite in the furnace causes the setting of reducing conditions. The dehydration treatment ends after 2 hours (t _Tro ckn _e n) -

Thereafter, the dried soot body is ( ^"2 mbar 10) vitrified in a sintering furnace at a temperature of about 1500 ° C under vacuum to a transparent blank of TiO ₂ -SiO ₂ glass. The mean hydroxyl group content of the glass is at about 250 wt. ppm.

The glass is then homogenized by thermal mechanical homogenization (twisting) and formation of a cylinder of TiO ₂ -SiO ₂ glass. For this purpose, a rod-shaped starting body is clamped in a glass lathe equipped with a oxyhydrogen burner and homogenized by means of a forming process, as described in EP 673 888 A1 for the purpose of complete removal of layers. The starting material is locally heated to over 2000 ° C by means of the oxyhydrogen gas burner and thereby softened. In this case, the oxyhydrogen gas burner to 1 mole of oxygen, 1, 8 mol of hydrogen supplied and thus produces an oxidizing oxyhydrogen flame. By relative movement of the two brackets to each other, the starting body is twisted about its longitudinal axis, wherein the softened glass mass is intensively mixed to form a drill body in the radial direction. An elongate drill body with a diameter of about 90 mm and a length of about 635 mm is obtained. In another forming process, the drill body compressed to a ball-shaped mass, and the starting points of the brackets on the ball-shaped mass are displaced by 90 degrees. By pulling the brackets apart and twisting each other another drill body is obtained. This forming process is repeated until a blank homogenized in all dimensions is obtained. The thus homogenized TiO ₂ -SiO ₂ glass is free of streaks in three directions, it contains no rutile microcrystals and shows a homogeneous titanium oxide concentration.

From the blank so a round plate TiO ₂ -SiO ₂ glass is formed with a diameter of 30 cm and a thickness of 5.7 cm. To reduce mechanical stresses and to set a given fictitious temperature, the glass plate is subjected to an annealing treatment. Here, the glass plate 8 hours during a holding time (t1 _Te m em) under air and under atmospheric pressure at 1080 ° C (T1 _Te m em) is heated and then (at a cooling rate of 4 ° C / h to a temperature of 950 ° C T2 _Tem pern) and kept at this temperature for 4 hours (t2 _Tem pern) long. Thereafter, the TiO ₂ -SiO ₂ glass plate is cooled at a higher cooling rate of 50 ° C / h to a temperature of 300 ° C, whereupon the furnace is turned off and the glass plate is left to free cooling of the furnace.

For further processing, the damaged surface layer of the glass plate is removed and a plan side is polished to give a diameter of 29.4 cm and a thickness d of 5.1 cm.

The plate thus obtained (sample 1a) consists of a particularly high-quality, homogenized, TiO ₂ -SiO ₂ glass which contains 7.7% by weight of titanium oxide. The hydroxyl group content is 250 ppm by weight and for the hydrogen concentration an average value of 1 × 10 ¹⁶ molecules / cm ^{3 is} determined. The mean fictive temperature measured over the entire thickness is 968 ° C.

From the same TiO ₂ -SiO ₂ glass and the same manufacturing method, two more glass plates (samples 1 b and 1 c) are produced. The only difference lies in the tempering process. For sample 1b, t2 _tem pern is shorter and for sample 1 c t2 _Tem pern little longer than Probel. The samples show the following average thickness measured fictitious temperatures:

Sample 1 a: 968 +/- 2.5 ° C

Sample 1 b: 993 +/- 5.1 ° C

Sample 1c: 938 +/- 4.2 ° C

For these samples the mean thermal expansion coefficient is determined interferometrically by the method described in: "R. Schödel, Ultra-high accuracy thermal expansion measurements with PTB's precision interferometer "Meas. Sei. Technol. 19 (2008) 084003 (1 1 pp)". From the CTE values measured in this way, the respective T _Z c of the samples is computationally obtained in a known manner.

From commercially available TiO ₂ -SiO ₂ glass with a titanium oxide content of 7.4 wt .-% and with a hydroxyl group content of 880 ppm by weight were also cut and measured samples. Possibly due to the higher hydroxyl group content, the fictitious temperature is consistently slightly lower for these samples. After annealing, it varies between about 902 and 957 ° C for three samples taken.

The diagram of Figure 2 shows the comparison of the two series of measurements. The curve A combines the measured values of the samples of the commercially available TiO ₂ -SiO ₂ glass and the curve B the measured values of the samples 1 a, 1 b and 1 c for the TiO ₂ -SiO ₂ glass according to the invention. On the ordinate, the zero-crossing temperature T _Z c (in ° C) is plotted against the measured fictitious temperature T _f (in ° C).

Curve B shows a comparatively flat course. Both the slope of the function and T _Z c increase slightly with the fictitious temperature and, at a fictitious temperature of 993 ° C, reach a maximum with T _Z c = 36 ° C.

The diagram of Figure 3 shows the derivatives of the curves A and B of Figure 2. The differential quotient dT _zc / dT _f is plotted against the fictitious temperature T _f . Curve A 'shows the tangent slope of the function of curve A and curve B' the tangent slope of the curve. B1. It can be seen from this that the tangent Curve B at the fictitious temperature reading of 993 ° C is about 0.35 (differential quotient dT _zc / dT _f = 0.35). At fictitious temperatures of less than 980 ° C, however, the differential quotient dT _zc / dT _f is below 0.3. In comparison, the trace A shows a steeper course of T _C z with the fictitious temperature. As can be seen from FIG. 3, the differential quotient dT _zc / dT _f at any point in the T _r interval of 920 ° C. to 970 ° C. is less than 0.3 and only reaches this value at fictitious temperatures of less than 915 ° C. This shows the higher dependence of the T _Z c on the fictive temperature in commercial TiO ₂ -SiO ₂ glass.

In conjunction with the elimination of rutile microcrystals, the reduction or avoidance of Ti ³⁺ formation and the concomitant homogenization of the titanium oxide concentration, this lower sensitivity of the TiO ₂ -SiO ₂ glass prepared according to the invention not only allows a more precise, simple re and more homogeneous adjustment of the coefficient of thermal expansion within the molding, but also a structurally particularly simple adaptation of the mirror substrate blank to the T _Z c-

Samples 2 and 3: further plates of TiO ₂ -SiO ₂ -glass As explained on the basis of sample 1 a, soot bodies of synthetic TiO ₂ -SiO ₂ glass are obtained by flame hydrolysis of OMCTS and titanium isopropoxide [Ti (OPr ') ₄ ] produced different concentrations of TiO ₂ . The concentrations are given in Table 1.

The soot bodies are dehydrated as sample 1 a each at a temperature of 1 150 ° C in a heating furnace with a heating element made of graphite under vacuum.

The dried soot body is vitrified into transparent blanks of TiO ₂ -SiO ₂ glass at about 1500 ° C under vacuum (10 ^"2 mbar). The mean hydroxyl group content of the titanium-doped silica glass is in each case about 250 ppm by weight. The glasses thus obtained are then further processed by thermal mechanical homogenization (twisting) under an oxidizing atmosphere. Sample 2 is thereby homogenized in three directions (as explained with reference to sample 1 a, and this type of homogenization is referred to as "3D" in Table 1.) Sample 3 was homogenized in the same way as samples 1 a and 2 under an oxidizing atmosphere, however in one direction only (in Table 1, this type of homogenization is referred to as 1D).

The round TiO ₂ -SiO ₂ glass plates formed from the blanks have a diameter of 30 cm and a thickness d of 5.7 cm (Sample 2) and 5.1 cm (Sample 3), respectively.

These are subjected to the setting of a predetermined fictitious temperature of an annealing treatment. For sample 2, this corresponds approximately to that of sample 1 a (T2 _Tempem for sample 2, however, is 930 ° C.).

In Sample 3, the TiO ₂ -SiO ₂ glass plate is heated to 1080 ° C during a holding time of 8 hours under air and atmospheric pressure and then cooled at a cooling rate of 4 ° C / h to a temperature of 980 ° C and at this temperature Held for 4 hours. Thereafter, the TiO ₂ -SiO ₂ glass plate is cooled at a higher cooling rate of 50 ° C / h to a temperature of 300 ° C, whereupon the furnace is turned off and the plate is left to free cooling of the furnace. The TiO ₂ -SiO ₂ glass of Sample 3 has an average fictitious temperature of 980 ° C.

From the end faces and the cylindrical surface of the cylindrical samples, a layer is removed before the next treatment step, so that each result in a diameter of 29.4 cm and a thickness of 5.1 cm. From sample 2 both sides of the plan are polished and from sample 3 one of the two plan sides. Samples 4 and 5: comparative samples

As illustrated by sample 1 a, soot bodies of synthetic TiO ₂ -SiO ₂ glass with different concentrations of TiO ₂ are prepared by flame hydrolysis of OMCTS and titanium isopropoxide [Ti (OPr ') ₄ ]. The concentrations are given in Table 1.

The soot body of Sample 4 is dehydrated like Samples 1 to 3. In the soot body of Sample 5, dehydration treatment is omitted.

The soot body be at about 1500 ° C under vacuum (10 ^"2 mbar) vitrified into transparent blanks of TiO ₂ -SiO ₂ glass. The mean hydroxyl group content of TiO ₂ -SiO ₂ glass of Sample 4 is percent at about 250. ppm of sample 5 at 350 ppm by weight.

Sample 5 is then further processed by thermal mechanical homogenization (twisting). The oxyhydrogen burner is operated during the entire process with stoichiometrically neutral flame, ie with a molar ratio of oxygen / hydrogen of 1: 2. Otherwise, the homogenization of sample 5 is carried out as described on the basis of sample 1a. For sample 4, homogenization is dispensed with.

To set a given fictitious temperature, both blanks are subjected to an annealing treatment, as described by sample 1 a. Thereafter, the TiO ₂ -SiO ₂ glass of Sample 4 has an average fictitious temperature of 967 ° C and that of Sample 5 has an average fictitious temperature of 952 ° C due to its higher hydroxyl group content.

From the end faces and the cylindrical surface of the cylindrical samples, a layer is removed before the next treatment step, so that each result in a diameter of 29.4 cm and a thickness of 5.1 cm.

Of both samples 4 and 5, a plan side is polished in each case. Samples were taken from all samples to determine the dependence of T _Z c on the fictitious temperature, as explained in Example 1, and the tangent slope in the range of mean fictive temperature was determined.

The respective production parameters and properties of samples 1a and 2 to 5 are summarized in Table 1. The sample 6 corresponds to the commercially available TiO ₂ -SiO ₂ glass on which the measurement curve A of FIG. 2 is based. The values of the "?" Provided manufacturing parameters are unknown for this glass.

Table 1

Sample 1a 2 3 4 5 6

Ti0 ₂ (wt%) 7.7 7.9 8.2 8.0 8.1 7.1

Dry (C) / dry 1050/50 1050/50 1050/50 1050/50 -?

(H)

OH (ppm by weight) 248 251 247 252 351 880

T1 homogenization (C) /

Scope of Homoge> 2000 /> 2000 /> 2000 / -> 2000 / -

3D / 3D / 1D / 3D /

nization /

Oxidie Oxidie Oxidie nitrogen

Type of burner flame

T1 annealing (° C) / 1080/8 1080/8 1080/8 1080/8 1080/8? t1 annealing (b)

T2 _Te mpern (° C) / 950/4 930/4 980/4 950/4 950/4 ? t2 annealing (b)

Average of Tf (° C) 968 945 980 967 952 940

Delta Tf (±) 4 4 4 4 5 6 dTzc / dTf 0.19 0.08 0.27 0.33 0.42 0.62

^ dTzc 1, 52 0.64 2.16 2.64 4.2 7.44 In the line designated as Delta Tf (±), the fluctuation range of the fictitious temperature from the measured average over the sample thickness is indicated. The line labeled dTzc / dTf indicates the differential quotient of curves A and B, respectively, at the measured average of the fictitious temperature, and the last line gives the maximum difference of T _C z over the entire sample thickness, as determined from the data from the three lines above ,

Each of Samples 1 to 3 can be used by itself as a mirror substrate blank with little deformation at an inhomogeneous temperature course readily.

A further adaptation to the temperature profile results when TiO ₂ - SiO ₂ glasses of different composition are combined together to form a mirror substrate blank. This will be explained in more detail below.

Production of a Mirror Substrate Blank by Joining Several Moldings EXAMPLE 1 In order to adapt the T _Z c to a temperature profile as shown in FIG. 1, a mirror substrate blank is built up from only two layers. Sample 1 and Sample 2 are optically blasted with their polished plano sides, resulting in an attraction-based, bubble-free joint. This joining compound is heated in an oven to a temperature of 1650 ° C for a period of 15 min. As a result, a melt-bonded body is obtained with low-bubble contact surface, which consists of two equal zones, which differ in their titanium oxide concentration, the mean fictive temperature but about the same.

The melted composite is tempered to remove mechanical stress. The temperature profile when annealing the melt-bonded body is as follows: heating to a temperature of 1080 ° C, holding at that temperature for a holding time of 10 hours; Cooling at a cooling rate of 4 ° C / h to a temperature of 950 ° C and held at that temperature for a period of 12 hours, followed by free cooling to room temperature.

The mirror substrate blank produced in this way is composed of only two components of different chemical composition, namely the upper shaped body of sample 1 and the lower shaped body of sample 2. These are connected to one another via a substantially flat and planar contact surface.

This slight adaptation of the T _Z c to the temperature profile of FIG. 1 is sufficient to ensure a small deformation of the mirror substrate as a whole at the given temperature profile. The remaining linear expansion in operation is anywhere in the range of 0 +/- 10ppb / ° C.

The mirror substrate blank is used to produce a mirror substrate made of titanium-doped glass for use in EUV lithography. To produce the mirror substrate, the top surface of the mirror substrate blank formed by sample 1, which faces the EUV radiation when it is used as intended, is subjected to a mechanical treatment which comprises grinding and polishing. The contour of the mirror is generated as a concavely curved surface area.

Comparative Example 1 In order to adapt the T _Z c to a temperature profile as shown in FIG. 1, a mirror substrate blank of two layers is built up using the procedure described in Example 1. In contrast, instead of Sample 1, Sample 4 is used. The method of preparation of sample 4 is similar to that of sample 1, but it is not homogenized. This adaptation of the T _Z c is not sufficient to ensure a small deformation of the mirror substrate as a whole at the predetermined temperature profile. The remaining length expansion in operation is in places above 0 +/- 20ppb / ° C.

Claims

claims

Blank of TiO ₂ -SiO ₂ glass for a mirror substrate for use in EUV lithography, characterized in that the TiO ₂ -SiO ₂ glass at an average value of the fictitious temperature T _f in the range between 920 ° C and 970 ° C has a dependence of its zero crossing temperature T _Z c of the fictitious temperature T _f , which, expressed as a differential quotient dT _Z c / dT _{f is} less than 0.3.

Blank according to claim 1, characterized in that the differential quotient dT _Z c / dT _{f is} less than 0.25.

Blank according to claim 1 or 2, characterized in that the TiO ₂ - SiO ₂ glass has an average hydroxyl group content in the range of 200 to 300 ppm by weight.

Blank according to one of the preceding claims, characterized in that the TiO ₂ -SiO ₂ glass has an average hydrogen concentration of less than 5x10 ¹⁶ molecules / cm ³ , preferably an average hydrogen concentration of less than 1 x 10 ¹⁶ molecules / cm ³ .

Blank according to one of the preceding claims, characterized in that it is delimited by an upper side and a lower side, and that the TiO ₂ -SiO ₂ glass has a non-homogeneous course of the titanium oxide concentration between the upper side and lower side.

Blank according to claim 5, characterized in that it is formed as a composite body having a first molded body of TiO ₂ -SiO ₂ glass with a first titanium oxide concentration and a second molded body of TiO ₂ -SiO ₂ glass with a second titanium oxide Concentration, which is connected to the first molded body.

Blank according to claim 6, characterized in that the first molded body has a first mean fictitious temperature and the second molded body has a second average fictitious temperature, wherein the first and second fictitious temperature differ from each other. Blank according to one of claims 5 to 7, characterized in that the shaped bodies are plate-shaped with a maximum thickness of 60 mm.

A process for the production of a blank according to one of the preceding claims or of a shaped body as precursor for such a blank, comprising the following method steps:

(a) producing a first porous soot body of SiO ₂ having a first concentration of titanium oxide by flame hydrolysis of starting materials containing silicon and titanium,

(c) homogenizing the first TiO ₂ -SiO ₂ glass in a homogenization process in which the TiO ₂ -SiO ₂ glass is heated in an oxidative atmosphere to a temperature of more than 2000 ° C., thereby softened and reshaped, so that a mean hydrogen concentration of less than 5 × ^{10 16} molecules / cm ^{3 is} established,

(d) forming the first TiO ₂ -SiO ₂ glass having an average hydroxyl group content of less than 300 ppm by weight and an average hydrogen concentration of less than 5 × 10 ¹⁶ molecules / cm ³ to give a shaped article, and

(E) annealing the shaped body such that the TiO ₂ -SiO ₂ glass at a mean fictitious temperature T _f in the range between 920 ° C and 970 ° C has a dependence of its zero crossing temperature T _Z c of the fictitious temperature T _f , the , expressed as the differential quotient dT _zc / dT _{f is} less than 0.3. A method according to claim 9, characterized in that the first TiO ₂ - SiO ₂ glass is heated at least partially in the homogenization process according to step (c) with a burner flame, which fuel gas and at least one oxidizing component supplied in an excess amount for complete combustion of the fuel gas become.