EP2220267A1

EP2220267A1 - Substrate having a coating comprising copper and method for the production thereof by means of atomic layer deposition

Info

Publication number: EP2220267A1
Application number: EP08855886A
Authority: EP
Inventors: Thomas WÄCHTLER; Thomas Gessner; Stefan Schulz; Heinrich Lang; Alexander Jakob
Original assignee: Fraunhofer Gesellschaft zur Forderung der Angewandten Forschung eV; Technische Universitaet Chemnitz
Current assignee: Fraunhofer Gesellschaft zur Forderung der Angewandten Forschung eV; Technische Universitaet Chemnitz
Priority date: 2007-12-05
Filing date: 2008-12-04
Publication date: 2010-08-25
Also published as: DE102007058571A1; DE102007058571B4; US20100301478A1; WO2009071076A1; US8507038B2

Abstract

The invention relates to a method for producing a coated substrate having a coating made of copper or a coating comprising copper by means of atomic layer disposition (ALD). To this end, a fluorine-free copper (I) complex of the formula L2Cu (XnX) is used as a copper precursor, wherein L is a s donor/p acceptor or a s,p donor/p acceptor ligand and wherein XnX is a bidentate ligand, namely a ß-diketonate, a ß-ketoiminate, a ß-diiminate, and amidinate, a carboxylate, or a thiocarboxylate. A smooth ALD coating (2) containing copper is obtained on tantalum nitride (3).

Description

description

Substrate with a copper-containing coating and method for their preparation by means of atomic layer deposition

The invention relates to a method for the deposition of thin copper layers or copper-containing layers by means of atomic layer deposition (ALD). This method is particularly suitable for depositing layers containing copper or copper on semiconductor substrates (for example, for the production of the multilayer interconnect system in highly integrated microelectronic circuits-ULSI circuits).

Up to now, mostly physical methods have been used to produce thin copper layers, especially sputtering. However, this method has the disadvantage that, in particular in the production of copper starting layers for the interconnect system of highly integrated circuits - with increasing reduction of the geometric dimensions - no uniform closed layers are obtained.

As alternative methods to produce such layers, chemical methods are considered, for example, variants of chemical vapor deposition (CVD). In this case, source substances or precursors which contain the desired metal (eg copper) in the form of a chemical compound are supplied in the gaseous state to a vacuum chamber designed as a hot or cold wall reactor, in which layer deposition takes place later. The precursors are transferred for this purpose before the deposition in the gas phase. On The surface of the heated wafer substrate is then subjected to a film formation reaction. This can consist of a deliberately thermally controlled decomposition of the precursor; often also reducing or oxidizing agents for the layer deposition are necessary. However, the CVD methods have the disadvantage that the layer growth is not uniform here and closed layers form only from a thickness of a few 10 nm.

By using Atomic Layer Deposition (ALD) these disadvantages can be avoided. This is a cyclic process, in which usually two reactants are fed in pulses of Reaktionskämmer. These pulses are separated from each other by inert rinsing and / or evacuation steps, so that both reactants never encounter each other in the gas phase and only surface reactions of the second partner with adsorbates of the first partner lead to layer formation. The first partner is first chemisorbed on the substrate surface, so that the substrate is substantially covered with a monolayer of the precursor. Further monolayers, which are formed by physisorption, are removed during the rinsing or evacuation pulses. It is thus necessary that the precursors can chemisorb on the substrate to be coated. By means of the ALD method, it is therefore possible to control the desired layer thickness very precisely by way of the number of ALD cycles.

To prepare copper layers by ALD, two approaches to deposition can generally be chosen. Either directly elemental copper can be generated during the individual ALD cycles; Alternatively, a copper species can first be prepared (eg copper oxide - CuO _x ), which is then reduced to copper. The first variant however, it is usually difficult to produce elemental copper.

US Pat. No. 6,869,876 B2 describes a generic process in which a copper halide layer is first produced on the substrate, which is subsequently used by means of a

Reducing agent is reduced to a copper layer. The precursors used here are copper (I) and copper (II) complexes. Complexes of the type LCu (XnX) are mentioned here in particular as copper (I) complexes. The bidentate ligand XnX represents β-diketonate, explicitly hexafluoroacetylacetonate (hfac) is called. The ligand L is a stabilizing ligand, e.g. an olefin such as trimethylvinylsilane (tmvs). The reduction can be carried out, for example, with diethylsilane.

The ^'US 6,482,740 B2 describes a generic ALD method in which first a copper oxide layer is obtained. Here are used as precursor copper (I) - and copper (II) - compounds. As the copper (I) compound, for example, (PEt ₃ ) is called Cu (hfac). To produce the oxide layer, an oxidation pulse with water, H ₂ O ₂ , O ₂ , O ₃ or similar oxidizing agents is carried out in each case during an ALD cycle. To reduce the copper oxide layer, reducing agents such as ammonia, hydroxylamine, hydrazine, alcohols (eg methanol), aldehydes (eg butyraldehyde), carboxylic acids (eg formic acid or acetic acid) and hydrogen are used. The reduction takes place at temperatures between 310 and 450 ⁰ C.

The above processes have the disadvantage that fluorine-containing precursors are used. Fluorine can get attached accumulate the interface with the substrate material and reduce the adhesion of the copper layer on the substrate.

WO 2004/036624 A2 discloses an ALD process in which a precursor is used which is not fluorine-containing. Homoleptic copper complexes are proposed as precursors, for example copper (II) -β-diketonates and copper (I) -tert-butoxide. For the oxidation pulse ozone, oxygen, water or mixtures thereof are used; the reduction takes place by means of a hydrogen-containing gas.

The present invention has the object to overcome the disadvantages of the prior art and to provide an improved ALD method for producing a copper layer.

This object is achieved by the method according to claim 1. Claims 16 and 17 indicate a substrate coated by this method with a copper layer or a copper-containing layer. Subclaims teach advantageous developments. Claim 21 indicates an advantageous use of the method according to the invention.

The present invention provides a method by which a substrate can be coated with a copper layer or a copper-containing layer. The layer deposition takes place by means of atomic layer deposition (atom layer deposition - ALD), wherein a copper (I) complex which contains no fluorine is used as precursor. The process according to the invention can be carried out in two variants: either a reduction step after the predetermined number of ALD cycles have been completed. Alternatively, the reduction of the deposited copper-containing layer may be accomplished by means of a reduction pulse during an ALD cycle. Finally, the reduction step can be completely dispensed with, if no coating of elemental copper is desired, but a copper oxide layer or a layer of another copper salt.

The precursor used in the process according to the invention is a complex of the formula L ₂ Cu (XnX) in which the two ligands L may be identical or different and the σ-donor π-acceptor ligands and / or σ, π-donor π Acceptor ligands. The ligand XnX is a bidentate ligand selected from the group consisting of β-diketonates, β-ketoiminates, β-diiminates, amidinates, carboxylates, and thiocarboxylates. In principle, these bidentate ligands thus are a compound which is simply negatively charged in an uncoordinated state and has two heteroatoms which are linked to one another via one or three carbon atoms. As a rule, in the case of the precursor of the invention, the bidentate ligand is coordinated to the Cu at room temperature via both heteroatoms. However, it may also happen that at least part of the precursor is coordinated only via one of the heteroatoms of the bidentate ligand. A complex in which a heteroatom of the bidentate ligand is coordinated (at room temperature) to two copper atoms does not belong to the precursors suitable according to the invention. However, it is possible in individual cases that the two heteroatoms in each case coordinate to different Cu atoms, so that a 2-nuclear complex is formed. As a rule, however, the complexes according to the invention will be mononuclear complexes. According to the invention, it was recognized that by using 18 valence electron complexes instead of 16 valence electron complexes, precursors having a much higher stability are used, but that these precursors decompose or undergo copper under milder conditions than is customary in the prior art Oxidation to a copper salt such as copper oxide can be implemented. According to the invention, it was recognized that the complexes according to the invention are better suited for ALD, since they are less prone to self-decomposition (eg by disproportionation), which would result in greater CVD effects. This is accompanied by the fact that the precursors to be used can be stored for a certain time without decomposition phenomena occurring and thus a more economical implementation of the method is possible. According to the invention it was also recognized that when using such precursor at the start of the decomposition reaction, a sufficiently large nucleus density occurs and thus the formation of growth islands occurs very rarely.

During the layer deposition step, the following (ALD) partial steps are preferably run through successively, an oxidation pulse and / or a reduction pulse always taking place:

Adsorption pulse followed by a purge pulse or evacuation pulse

optionally oxidation pulse followed by a rinse pulse or evacuation pulse

optionally, a reduction pulse followed by a purge pulse or evacuation pulse.

The single pass through the specified sub-steps represents an ALD cycle. The reduction pulse and the subsequent rinsing or evacuation pulse are only then required if the reduction is to be performed during each ALD cycle and the reduction does not take place until after the predetermined number of ALD cycles have passed or the reduction is completely dispensed with. If a reduction pulse occurs during the ALD cycles, it is often not necessary to carry out an oxidation pulse (especially when forming layers on metal substrates).

During the adsorption pulse (also called precursor pulse), the precursor of the reaction chambers in which the ALD method is carried out is supplied in such a way that chemisorption and optionally also physisorption of the precursor on the substrate surface take place on the substrate arranged in the reaction chamber. For this purpose, the precursor is supplied in vapor form to the reaction chamber (in particular by evaporation or sublimation of the liquid or solid precursor) or in the form of a precursor / solvent mixture which has been converted into the vapor phase. Usually, for this feed, a carrier gas, in particular an inert gas, e.g. Argon, used.

The adsorption pulse is followed by a flushing pulse or an evacuation pulse. This has the task of removing excess precursor so that ideally only one monolayer of the chemisorbed precursor remains on the substrate surface.

If provided, an oxidizing agent is fed to the reaction space in the subsequent step. This oxidizing agent reacts with the chemisorbed precursor molecules so that, as a rule, copper or a copper oxide (or another copper compound formed by the oxidation, depending on the oxidant used) is formed. This is followed by a rinsing or Evacuation pulse to remove reaction products from the reaction chamber. In general, an inert gas is usually used for the rinsing pulses (for example argon).

If provided, this can be followed by a reduction pulse in which a reducing agent is supplied to the reaction space. As a rule, this takes place at least partially, preferably completely, a reduction of the copper salt (in particular copper oxide) obtained in the oxidation pulse to elemental copper. In order to remove the reaction products obtained from the reduction pulse also from the reaction space, this is followed in turn by a purge or evacuation pulse.

Repeated repetition of this ALD cycle results in growth of the desired layer on the substrate. Typically, an ALD process consists of several 100 ALD cycles to produce layers several nanometers thick.

Preferably, the ALD process is carried out in a temperature range in which the cycle growth varies very little or not at all with the temperature. This has the advantage that on the areas of the substrate to be coated (also in shadow areas) a uniform uniform layer with high uniformity of thickness is obtained. In contrast to CVD processes, in particular in this temperature window, because of the slower and more controlled layer growth, layers with a lower minimum thickness can be obtained, which in many cases also form closed layers even at layer thicknesses that are <5 nm.

With the method according to the invention, even after carrying out the reduction step remains a closed Layer received. In contrast, in the prior art, as a rule, islanding due to agglomeration processes can be observed.

In an advantageous embodiment of the method according to the invention, a precursor complex is used which can be evaporated at moderate temperatures (up to 100 ^° C.) (ie which is liquid or sublimable at these temperatures). The transfer to the vapor phase can be done for example by applying a negative pressure. The height of the vapor pressure of the precursor used is not essential here; it is only important that the precursor molecules can chemisorb on the substrate surface to be coated.

Particularly preferred are precursor complexes which are already liquid under normal conditions (ie at room temperature and normal pressure). These have the advantage of being easier and more controlled to vaporize than solids.

However, the use of solid precursors can also be advantageous. Solids can be dissolved, for example, and can thus also be used in "liquid" form. Such solid or dissolved precursors can be vaporized or sublimated with a corresponding metering or evaporator system even at very low vapor pressures at a high rate.

Under normal conditions, liquid complexes are obtained, in particular, when the ligand L of the complex L ₂ Cu (XnX) is a trialkylphosphane. In particular, trimethylphosphine, triethylphosphine, tri-n-propyl-phosphine and tri-n-butyl-phosphine are mentioned here.

The vapor pressure of the liquid or solid precursor should be preferred at the particular vaporization temperature be at least 0.005 mbar, particularly preferably 0.01 mbar, so that the adsorption of a monolayer precursor molecules on the substrate surface takes place in a technically acceptable time, so that the precursor pulse can be kept as short as possible. The

Evaporation temperature is hereby preferably chosen so that the complex, as long as it is not yet in the reaction chamber, just does not decompose

In general terms, ligand L may be any σ-donor π acceptor or σ, π-donor π acceptor. Particular mention may be made as ligands: isonitriles, alkynes and olefins (whereby also dienes as bidentate ligands in which L _{2 stands} for exactly one diene ligand or an en-in ligand in which both unsaturated groups are coordinated to the central atom ), in particular olefin and / or alkyne complexes in which the olefin or the alkyne acts as σ, π donor π acceptor, and finally phosphine ligands. In the case of the phosphine ligands of the formula PR ^δ ₃ , the radicals R ^{6 may be} identical or different, with complexes usually being used in which the radicals R ^{5 are} identical. In particular, R ⁶ may also be an alkoxy radical OR ⁷ . The radicals R ⁶ or R ⁷ may be in particular alkyl or aryl radicals. Suitable alkyl radicals are branched, unbranched or cyclic alkyls, in particular having 1 to 15 carbon atoms, particularly preferably the alkyl radicals are methyl, ethyl, n-propyl, i-propyl, n-butyl or t-butyl and Cyclohexyl. Suitable aryl radicals are all substituted and unsubstituted aryl compounds; these can also be bound via an alkylene linker to the phosphorus or oxygen atom. A particularly preferred aryl radical is phenyl. All the above alkyl and aryl radicals may also be substituted by heteroatoms or carry functional groups with heteroatoms (eg a bidentate ligand in which a coordinating NR ₂ group is included). As a rule, however, pure hydrocarbon radicals are used for cost reasons. Particularly suitable bidentate phosphine ligands are ligands in which the two phosphorus atoms are linked via an ethylene or methylene linker. The remaining radicals bound to the phosphorus atom correspond to the definition of the radicals R ^δ and R ⁷ . Particularly suitable bidentate phosphine ligands are bis (diphenylphosphino) ethane, bis (diphenylphosphino) methane and bis (diethylphosphino) ethane.

As a bidentate ligand XnX, it is possible to use, inter alia, a β-diketonate, a β-ketoiminate or a β-diiminate. It is therefore a ligand of the general formula RC (X) -CR ⁸ -C (Y) -R ¹ or

Here, X and Y are the same or different and represent O or NR ² .

The radicals R, R ¹ , R ² and R ⁸ are herewith identical or different and branched, unbranched or cyclic alkyl radicals, aryl radicals or trialkylsilyl radicals. R ⁸ can also be a hydrogen atom. Preferably, these radicals have 1 to 15 carbon atoms. The alkyl and aryl radicals may be substituted by heteroatoms, but are usually pure hydrocarbons; likewise, the radicals can carry functional groups. R, R ¹ and R ² are particularly preferably methyl, ethyl, n-propyl, i-propyl, n-butyl, t-butyl or phenyl and R ^{8 is} hydrogen, methyl, ethyl, n-propyl, i-propyl, n- Butyl, t-butyl or phenyl. Both Trialkylsilyl radicals, the individual alkyl radicals may be identical or different; there are branched, unbranched or cyclic alkyls into consideration, particularly preferred are trialkylsilyl radicals having the alkyl radicals methyl, ethyl, n-propyl, i-propyl, n-butyl and t-butyl. Particularly preferred as β-diketonate ligands acetylacetone ("acac"), heptane-3, 5-dione, 2, 6-dimethylhepta- 3,5-dione, 2,2,6,6-tetramethylhepta-3,5- dione, N, N, N'N'-tetramethylmalonamide, N, N, N'N'-tetraethylmalonamide, 1,3-dimorpholin-4-yl-propane-1,3-dione, 1,3-diphenylpropane-1, 3-dione and 1, 3-dicyclohexylpropane-l, 3-dione.

As β-ketoiminate ligands, 4- (methylamino) -3-penten-2-one and 4- (ethylamino) -3-penten-2-one are particularly preferred. As β-diiminate ligands are (2Z ₁ 4S) -JV-isopropyl-4- (methylimino) pent-2-en-2-amine and N- [(IB, 2Z) -l-methyl-2-pyrrolidine-2 -ylidene ethylidene] methanamine is particularly preferred.

Alternatively, the bidentate ligand XnX contained in the precursor complex may be an amidinate, a carboxylate or a thiocarboxylate. In the general formula R ³ -C (X ') - Y' or Therefore, X ¹ and Y ^{1 represent} either two oxygen atoms or one oxygen and one sulfur atom or two NR ⁴ groups.

The radicals R ³ and R ⁴ are identical or different and preferably branched, unbranched or cyclic alkyl radicals or aryl radicals. Preferably, these radicals have 1 to 15 carbon atoms. The radicals can be substituted by heteroatoms, but are usually pure hydrocarbons; likewise, the radicals can carry functional groups. The radicals R ³ and R ⁴ are particularly preferably methyl, ethyl, n-propyl, isopropyl, n-butyl, t-butyl or phenyl. The radical R ³ can furthermore be a trialkylsilyl radical in which the individual alkyl radicals can be identical or different. Branched, unbranched or cyclic alkyls come into consideration, particularly preferably the alkyl radicals are methyl, ethyl, n-propyl, i-propyl or n-butyl and t-butyl.

In an advantageous embodiment, the amidinate, the carboxylate or the thiocarboxylate in the alkyl or aryl radical have a carboxylate group. In particular, here are carboxylates of the formula R ⁵ OC (O) -ZC (O) -O or to call.

Here Z is an alkylene or alkylidene bridge or a

Single bond between the two carbonyl groups C (O).

Preferably, the alkylene group has the formula - (CH ₂ -) _n (where n = 0, 1 or 2); the alkylidene bridge preferably has the

Formula - (CH = CH-) _m (with m = 0, 1 or 2).

The radical R ⁵ is preferably a branched, unbranched or cyclic alkyl radical or an aryl radical, in particular a

Methyl, ethyl, n-propyl, isopropyl, n-butyl, tert-butyl or phenyl radical.

As the amidinate, carboxylate or thiocarboxylate ligand, particular preference is given according to the invention to acetate, benzoate, benzylate, propionate, pivalonate, 2-methylpropionate, silyl-substituted acetates, thioacetate and N, N'-dimethylacetamidines. According to the invention, it has been found that some precursor complexes are particularly preferred. These are synthetically easily accessible complexes, which are also liquid at room temperature and can be easily converted into the gas phase and continue to be storable for a certain time. The following complexes can be mentioned here:

- The acetylacetonate complexes (acac complexes) of the formula (R ⁷ ₃ P) ₂ Cu (acac) with R ⁷ = methyl, ethyl, n-propyl, isopropyl, n-butyl and tert-butyl

the heptane-3,5-dionate complexes of the formula (R ⁷ ₃ P) ₂ Cu (C ₂ H ₅ -C (O) -CH-C (O) -C ₂ H ₅ ) where R ⁷ = methyl, Ethyl, n -propyl, isopropyl, n -butyl and tert -butyl

- The acetate complexes of the formula (R ⁷ ₃ P) ₂ Cu (O ₂ CCH ₃ ) with R ⁷ = methyl, ethyl, n-propyl, isopropyl, n-butyl and tert-butyl. For the oxidation pulse of the ALD process according to the invention, a liquid or gaseous oxidizing agent is usually used. Preferably, the oxidizing agent is gaseous and more preferably the oxidizing agent is selected from the group consisting of oxygen, water, H ₂ O ₂ , ozone and N ₂ O.

These oxidizing agents are preferred, as can be performed using this pulse, an oxidation at relatively mild conditions (in particular at temperatures in the range from 105 ⁰ C to 135 ° C).

The use of wet oxygen (ie a mixture of water vapor and oxygen) is particularly preferred because wet oxygen is a more effective oxidant than oxygen or water vapor alone, and thus the ALD better even at low temperatures, in which the precursor itself just does not decompose works. This is usually accompanied by a lower tendency to agglomeration of the produced copper-containing layer. The reduction agent for the reduction step or the reduction pulse is preferably selected from the group consisting of hydrogen plasma, molecular hydrogen (H ₂ ), carbon monoxide, hydrazine, boranes, silanes and organic reducing agents.

Particularly suitable organic reducing agents are alcohols, aldehydes and carboxylic acids. Usually, these will be low molecular weight compounds having a molecular weight <100 g / mol, preferably <61 g / mol. Namely, compounds having such a low molecular weight are inherent in that their oxidation products are relatively easily volatile and therefore can be relatively easily removed from the reaction chamber (or can be removed from an equilibrium reaction). In general, all organic reducing agents have the advantage that with these a reduction under mild conditions is possible, so that agglomeration and island formation during the reduction process does not occur here or only to a small extent. On the other hand, if hydrogen or a hydrogen plasma is used, no significant reduction occurs at temperatures <200 ^° C. (ie no reduction in the oxygen content in the copper-containing layer-in particular when hydrogen is used). In addition, the ALD should preferably be carried out without the use of a plasma in order to achieve uniform layer growth in all areas, even in deep and geometrically complicated substrates. The use of a plasma can then also lead to a strong tendency to agglomeration occurs.

Particularly suitable organic reducing agents are methanol, isopropanol, formaldehyde, acetaldehyde, formic acid and acetic acid, of which in turn formic acid is very particularly preferred.

These compounds are intrinsic to the fact that under particularly mild conditions, a reduction can be carried out, in particular a reduction at temperatures 5 200 ⁰ C; in the case of formic acid even at temperatures between 105 and 115 ° C.

The ALD method according to the invention is preferably carried out in such a way that the individual pulses (adsorption pulse, oxidation pulse, reduction pulse and rinsing or evacuation pulse) are in each case no more than 15 seconds in each case. However, the pulse duration is also dependent on the volume of the respective reactor. Preferably, the length of the pulses is 3 to 11 seconds. For very compact reactor hammers, the required pulse length can also be in the range of 10 to a few 100 milliseconds.

For the adsorption pulse, the pulse duration is particularly preferably 3 to 6 seconds. Further preferably, the pulse duration of the adsorption pulse is just as long that at least one deposition rate or a cycle growth of 0.08 Ä / cycle and particularly preferably 0.12 Ä / cycle is achieved. This is usually the case when the pulse duration of the adsorption pulse is at least 2 seconds.

The adsorption pulse is preferably carried out at a temperature of 105 to 165 ° C, more preferably at a temperature of 115 to 135 ° C.

In this temperature window, the cycle growth is relatively little dependent on the temperature and thus a more targeted production of a copper layer or copper-containing layer with a certain layer thickness possible. The cycle growth is usually very constant at a temperature of 115 to 135 ° C. Furthermore, the inventive method is preferably carried out so that the reduction step or the reduction pulse at a temperature <250 ⁰ C, preferably <200 ⁰ C, is performed. However, the temperature to be selected also depends on the reducing agent, so that - even if it would be desirable to work at lower temperatures - a free selectability of the reduction temperature exists only within certain limits. For example, a significant reduction in the use of hydrogen (H ₂ ) will take place only from temperatures of 400 to 450 ⁰ C. Relatively high temperatures are therefore required for the reduction of the copper oxide with molecular hydrogen (H ₂ ). Due to the increased agglomeration tendency of the copper at these temperatures, it is then no longer possible to obtain thin and closed layers. A lower process temperature is possible by using hydrogen plasma. However, this has the disadvantage that the plasma acts differently on structured substrates. Free surfaces are preferably attacked while it is difficult to achieve complete reduction of the copper oxide or copper salt on sidewalls of deep trenches, in holes and in shaded areas.

The temperature should therefore be as low as possible, if a small agglomeration or island formation should take place during the reduction step or reduction pulse. Preference is therefore given to carrying out the reduction step or reduction pulse at a temperature which corresponds to that of the adsorption pulse or is below this temperature. This can be accomplished, for example, by means of organic reducing agents; Very particular preference is given here to the use of formic acid, which gives good results even at a temperature of 105 ^° C. delivers.

Furthermore, if possible, plasma processes should be dispensed with in order to ensure a uniform reduction of the layer applied by ALD, even in structured substrates or shaded areas of structured substrates. Preference is therefore given to purely thermally-acting reduction processes.

As a substrate to be coated, a single-layered or multi-layered substrate can be used. In this case, at least one layer of the multilayer substrate (preferably the layer of the multilayer substrate adjoining the copper layer or copper-containing layer) or the single-layer substrate itself is a transition metal, a transition metal salt (in particular a ceramic compound such as a transition metal nitride or transition metal oxide) ), a semiconductor material, an organic polymer and / or an inorganic polymer or contains one or more substances of the aforementioned substance classes.

Furthermore, this material is preferably selected from the group consisting of tantalum, titanium, tungsten, niobium, vanadium, tantalum nitride, titanium nitride, tungsten nitride, niobium nitride and vanadium nitride, platinum, palladium, ruthenium, rhodium, a silicon dioxide, a silicate, zinc oxide, hafnium oxide, Alumina, zirconia, silicon, germanium, gallium arsenide, aluminum gallium arsenide, gallium nitride, aluminum gallium nitride, indium phosphide, indium gallium phosphide and carbonitrides or silicon nitrides of transition metals, particularly tantalum, tungsten and titanium; Alternatively, this material may contain one or more of the aforementioned substances. Surprisingly, it has been observed that on substrates of a transition metal nitride during the reduction step less tendency for formation of islands occurs than when using the pure transition metal as a substrate. This effect is particularly strong when using tantalum nitride (especially in comparison with elemental tantalum).

Furthermore, it has been observed that the nitrogen content of the transition metal nitride is of great importance for the growth of the copper-based ALD layer. On the pure transition metal is a faster decomposition of the copper precursor, so that a more uniform layer growth takes place. However, this also means that the decomposition of the precursor on a transition metal nitride only starts at relatively high temperatures, so that the use of non-stoichiometric transition metal nitrides, in particular nitrides of tantalum, titanium, tungsten, niobium and / or vanadium, is particularly preferred. if a particularly uniform copper layer is to be produced. Here again, a nitrogen content of 75 to 99%, based on the corresponding stoichiometric transition metal nitride compound MN _x , is present. This effect is particularly strong again for tantalum / tantalum nitride.

According to the invention, the substrate, which can be produced by the method described above, at least partially a closed copper coating or copper-containing coating, as a rule, a completely closed copper coating or copper-containing coating. Under a closed layer is understood according to the invention a layer in which in the measurement by means of X-ray photoelectron spectroscopy (X-ray photoelectron spectroscopy, XPS - uses Mg-Kα radiation) no more signals of the substrate can be detected. The exit depth of the photoelectrons should be 1 nm to 3 nm at a quasi-perpendicular acceptance angle. The measurement procedure is always performed according to "S. Hofmann: Depth Profiling in AES and XPS in: Practical Surface Analysis Second Edition Volume 1 - Auger and X-ray Photoelectron Spectroscopy. (Eds. D. Briggs and MP Seah) John Wiley & Sons, Chichester et al., 1990 ".

The copper coating has no fluorine-containing impurities (since no fluorine-containing precursor molecules were used) and has a layer thickness of 2 nm to 100 nm, preferably 2 nm to 30 nm. The layer thickness is always according to the invention by means of spectral ellipsometry in the spectral range 3.3 to 6, 3 eV measured. The measurement is carried out according to H.G. Tompkins and E.A. Irene (ed.): Handbook of Ellipsometry. Springer-Verlag, Berlin 2005. For microelectronic and nanoelectronic components, copper nucleation layers with thicknesses between 2 and 20 nm are preferably used for the production of the interconnect system of integrated circuits.

Further preferably, the substrate according to the invention with copper coating has a roughness Ra according to DIN EN ISO 4287 of 0.2 nm to 2.7 nm, with a roughness Ra of the uncoated substrate between 0 ¹ , 1 nm and 0.2 nm. Ra was determined using Atomic Force Microscopy (AFM) in tapping mode. A silicon tip with a tip radius ≤ 10 nm was used. The difference in roughness Ra between coated and uncoated substrate is therefore usually 0 to 2.5 nm, preferably 0 to 0.2 nm Difference depends on the one hand on the thickness of the coating, on the other hand on the surface of the substrate used.

Furthermore, the copper-coated substrate according to the invention frequently has trenches, holes and / or shadow spaces. Such substrates are not coatable by many prior art methods (e.g., CVD methods), at least not coatable with a uniform layer. The inventive method has the advantage that even on such substrates, a uniform copper coating or copper-containing coating can be applied.

With the method according to the invention, a substrate is again obtainable, once again with appropriate process control, which at least partially has a closed coating containing copper and copper oxide. This substrate can be obtained, for example, if a reduction step or reduction pulses is completely dispensed with. This coating then has no fluorine-containing impurities and has a layer thickness of 2 nm to 100 nm, preferably 2 nm to 30 nm. The layer consists essentially of Cu ₂ O in the region of the surface and points between this surface and the surface facing the substrate the coating on a gradient with decreasing oxygen content.

This gradient can be determined by angle-resolved XPS studies (ARXPS) according to S. Oswald et al. Angle-resolved XPS - a critical evaluation for various applications, Surface and Interface Analysis 38, 2006, 590-594. This shows that at a take-off angle of 25 ° to the sample surface usually only Binding energy values are measured, which are attributable to oxidic copper (in particular Cu ₂ O). In contrast, in measurements with a decrease angle of 60 °, a significant proportion of the measured binding energies of the photoelectrons can be attributed to elemental copper. At a take-off angle of 75 °, the proportion of these photoelectrons is further increased. With regard to the roughness, the same applies to the substrate with a coating containing copper and copper oxide as has been said above for a substrate, in which a reduction step or reduction pulse is not dispensed with.

The inventive method can be used for the production of copper layers, in particular copper starting layers for the subsequent galvanic or electroless deposition of metal layers, especially copper layers, in the production of contact and interconnect systems in microelectronic devices and in the production of thin film solar cells.

To produce a multilayer interconnect system or interconnect system in highly integrated microelectronic circuits (ULSI circuits), copper layers in the range from 10 to 100 nm are applied galvanically to prestructured silicon wafers provided with different functional layers. Before the galvanic copper deposition can take place, it is necessary to apply to these wafers an electrically conductive seed or seed layer (a so-called seed layer). The inventive method is best suited for the production of such starter layers, as forming a layer with uniform thickness, the thickness - as described - is arbitrarily adjustable, and therefore always in geometrically complicated arrangements of vias between adjacent Leitbahnebenen (Vias) and the like for Success leads. If a plurality of wafers or microchips are permanently mechanically connected to one another in the field of so-called 3D or vertical system integration by means of bonding methods, through-connections (so-called through-hole vias - THVs or through-silicon vias - TSVs) are used to produce electrical contacts between the individual chips Commitment. These are very deep, narrow holes (in particular with a diameter of 5 μm to 150 μm and aspect ratios of 1 to 20) through the chip or wafer stack. In order to establish the electrical contact between the components to be connected, these holes are completely or partially filled by means of electrodeposition of copper. Again, an electrically conductive start or seed layer is required. The method according to the invention is also ideally suited for the production of these starting layers, since - even with complicated geometries - copper layers with a uniform thickness can be produced.

The aforementioned starting layers must meet high requirements. They must be grown and grown without defects on the particular substrate present, which are usually substrate materials or diffusion barrier layers of transition metals such as tantalum or tungsten, or transition metal nitrides such as tantalum nitride or tungsten nitride. The starting layers must in this case grow evenly in all areas of the substrates to be coated. Layer thickness differences between the bottoms of trenches or holes, their sidewalls and free surfaces on the wafer substrate are undesirable because otherwise the subsequent electrodeposition of copper would result in uneven layer growth due to an uneven current density distribution. At the same time, the starting layers should have a very good electrical conductivity and the lowest possible surface roughness. Finally, they must also adhere very well to the substrate so that the layer stack has a high mechanical stability for subsequent polishing processes. Reduced adhesion of the copper layers also causes a reduced electrical reliability of the interconnect in the later component, in particular with regard to the electromigration.

The method according to the invention and the copper layer which can be produced herewith are described in more detail below without limiting the generality:

1. Preparation of a copper and / or copper oxide-containing layer on a silicon wafer

1.1 Method with an oxidation pulse during the ALD-performed layer deposition step using the precursor ((CH ₃ CH ₂ CH ₂ CH ₂ ) ₃ P) 2CU (acac)

For the ALD, a silicon wafer comprising (a) a tantalum nitride layer, (b) a tantalum layer, or (c) a combination of a tantalum nitride and a tantalum layer (namely, a layer system having a tantalum layer at the top including a tantalum nitride layer is arranged) or (d) with a silicon dioxide layer or (e) is provided with a ruthenium layer, introduced into a reaction chamber (vacuum chamber) and heated to a temperature between 100 ⁰ C and 150 ⁰ C. The precursor used is ((CH ₃ CH ₂ CH ₂ CH ₂ ) ₃ P) ₂ Cu (acac). The precursor is fed by means of a liquid metering system with evaporator unit of the reaction chamber. The precursor is stored under protective gas (in particular argon) at room temperature in a storage container. For dosing, the liquid Precursor using the protective gas overpressure in the reservoir out of this and conveyed via a flow meter to a mixing unit, in which the precursor is metered by means of a nozzle and with a protective gas carrier gas stream (in particular argon carrier gas stream) is mixed. This mixture is fed to the evaporator unit, where the precursor is evaporated at 85 ° C to 100 ⁰ C. The resulting carrier gas / precursor vapor mixture is then fed to the reaction chamber. Appropriately, this is done via heated lines. The duration of the precursor pulse thus carried out is 5 seconds; the precursor vapor is fed to the process chamber at a rate of 15 mg / min and an argon carrier gas flow of 700 sccm (standard cubic centimeters per minute). This is followed by a 5 second purge pulse in which argon is fed to the reaction chamber at a rate of 145 sccm. For the oxidation pulse, a mixture of oxygen and water vapor is used as the oxidizing agent. To generate the water vapor is passed through water, which was heated to 45 ° C to 50 ⁰ C, argon, which is loaded in this way with water vapor. The oxidation pulse is carried out as well as the adsorption pulse at 120-135 ° C and a pressure of 0.6 to 1.2 mbar and has a duration of 11 seconds. For this purpose, the reaction chamber oxygen with a flow of 90 sccm and 18.5 mg / min of water vapor with an argon carrier gas flow of 210 sccm (the latter preferably via heated lines) the Reaktionskämmer supplied. This is followed again by a flushing pulse as described above.

When 400 ALD cycles are performed as described above, a layer of about 5 to 6 nm in thickness is obtained. The layer is very smooth on all substrates and is characterized by very good adhesion to the substrate. FIG. 1 shows the cycle growth (growth per cycle, GPC) as a function of the precursor pulse duration, when the substrate used is tantalum nitride and the deposition temperature is 135 ^° C.

Fig. 2 shows the cycle growth (GPC) as a function of the deposition temperature among others for the substrates (a) to (e). It can be seen from Figure 2 that the "temperature window" typical of the ALD, in which the rate of deposition (i.e., cycle growth GPC in A) does not or only very slightly varies with temperature, is approximately between 100 and 125 ° C. A particularly low variation of the deposition rate results for Siθ2 ~ substrates between 110 and 120 ° C and for TaN substrates between 115 and 125 ° C and for ruthenium substrates between 100 and 120 ° C.

Fig. 3 shows the gradient copper / copper oxide in the obtained layer. The angle-resolved XPS spectrum (ARXPS spectrum) shows the Cu2p3 signal of a 5 nm thick layer on tantalum nitride. The signals were recorded at angles of 25 °, 35 °, 45 °, 60 ° and 75 °, each measured to the sample surface.

4 shows a scanning electron micrograph (SEM) surface image of the obtained layer (2) on tantalum nitride (1) (the oxidation pulse was carried out at 135 ° C.). The ALD layer (2) was partially removed to reveal the underlying tantalum nitride surface (1). It is a smooth, closed ALD layer recognizable. Ellipsometrically, a layer thickness of 4.9 nm was determined.

5 shows a transmission electron microscopic (TEM) cross-sectional image of the obtained ALD layer (2) Tantalum nitride (3), which was obtained according to Example 1.1 (the oxidation pulse was carried out at 125 ⁰ C). To protect the surface of the ALD layer (2) was applied to this a plastic layer (1).

FIGS. 6 and 7 show atomic force microscopic surface photographs of the obtained ALD layer. Fig. 6 shows the ALD layer on silicon dioxide (d) as a substrate (the oxidation pulse was carried out at 120 ⁰ C). The layer thickness is 2.8 nm; for the roughness Ra, a value of 0.2 nm was determined. An uncoated SiO ₂ comparison substrate likewise showed a roughness Ra of 0.2 nm. FIG. 7 shows the ALD layer on tantalum nitride (a) as substrate (the oxidation pulse was carried out at 125 ° C.). The layer thickness is 3.6 nm; for the roughness Ra, a value of 1.9 nm was determined. An uncoated tantalum nitride comparative substrate showed a roughness Ra of 0.2 nm.

1.2 Method with an oxidation pulse during the layer deposition step carried out by ALD using the precursor ((CH ₃ CH ₂ ) _S P) ₃ Cu (O ₂ CCH ₃ )

Similar layers as in Example 1.1 when the ALD process ((CH ₃ CH ₂₎ ₃ P) ₃ is used Cu (O ₂ CCH ₃₎ as a precursor can be obtained. The precursor is evaporated as in Example 1.1 with a Flüssigdosiersystem at 100 ⁰ C to 125 ° C and a flow of 5 to 10 mg / min and mixed with argon as the carrier gas at a flow of 1000 sccm. The ALD process, which is composed in the same way as in Example 1.1 from adsorption pulse, first flushing pulse, oxidation pulse and second flushing pulse, was carried out at temperatures of 150 ⁰ C to 200 ⁰ C on a TiN-coated silicon wafer. 1.3. Method Using a Reduction Pulse During the Layer Deposition Step Using ALD Using the Precursor ((CH ₃ CH ₂ CH ₂ CH ₂ ) ₃ P) ₂ Cu (acac)

As in Example 1.1, a coated silicon wafer is used for the ALD process. In contrast, the coating is an approximately 100 nm thick ruthenium coating, which is applied by means of a Ti adhesive layer of about 10 nm on the silicon wafer. The application of the Ti adhesive layer and the ruthenium coating can take place, for example, by means of electron beam vapor deposition.

As in Example 1.1, such a coated silicon wafer is introduced into a reaction chamber and heated. The precursor used is ((CH ₃ CH ₂ CH ₂ CH ₂ ) ₃ P) ₂ Cu (acac). The precursor is fed to the reaction chamber with the aid of a liquid metering system with evaporator unit and stored under protective gas (in particular argon) at room temperature in a storage container. For dosing, the liquid precursor with the aid of the protective gas overpressure in the storage vessel is conveyed out of this and conveyed via a flow meter to a mixing unit in which the precursor is metered by means of a nozzle and mixed with a protective gas carrier gas stream (in particular argon carrier gas stream). This mixture is fed to the evaporator unit, where the precursor is evaporated at 85 ° C to 100 ° C. The resulting carrier gas / precursor vapor mixture is then fed to the reaction chamber. Appropriately, this is done via heated lines. The duration of the precursor pulse is 5 seconds; the precursor vapor is fed to the process chamber at a rate of 15 mg / min and an argon carrier gas flow of 700 sccm. This is followed by a purging pulse of 5 seconds, in which Argon is fed to the reaction chamber at a rate of 145 sccm. Unlike in Examples 1.1 and 1.2, a reduction pulse now follows. The reducing agent used is formic acid vapor. For its production, a carrier gas (in particular argon) is passed through heated to 45 ° C - 50 ° C formic acid, which is loaded in this way with the formic acid vapor. The carrier gas / formic acid vapor mixture thus obtained is fed via heated lines to the reaction chamber. The duration of the reduction pulse is 11 seconds; the formic acid vapor is fed to the process chamber at a rate of 50 μl / min and an argon carrier gas flow of 210 sccm. It follows again as described above flushing pulse. As in Examples 1.1 and 1.2, the precursor pulse, the rinsing pulses and the reduction pulse are carried out at a pressure between 0.6 and 1.2 mbar.

When 400 ALD cycles are performed as described above, a layer of about 4 nm in thickness is obtained on the ruthenium substrate.

8 shows the X-ray spectrum of the layer obtained according to Example 1.3 by means of energy-dispersive X-ray analysis (EDX) at a primary energy of the electron beam of 3 keV. The spectrum shows signals that are assigned to the substrate (ruthenium and silicon), a very clear copper signal. Since only a very weak oxygen signal is present, the layer obtained according to Example 1.3 consists essentially of metallic copper. The oxygen signal is probably due to a post-oxidation of the copper layer already produced, which could be due to air contact of the layer produced before the measurement. Silicon substrate and Ti adhesion layer are due to the overlying layer stack difficult or undetectable.

2. Reduction of the copper / copper oxide-containing layer

To reduce the aforementioned copper / copper oxide layer, the substrate carrying this layer is treated in a reaction chamber as follows:

The obtained according to Example 1.1 or 1.2, carrying the layer to be coated substrate is heated in a vacuum chamber to a temperature between 100 and 300 ⁰ C. In the presence of a reducing gas or a liquid reducing agent to be evaporated, the reduction process is carried out at a pressure between 0.1 and 13 mbar, usually between 0.7 and 7 mbar. The duration of the process is between 1 min and 60 min, usually between 10 min and 30 min. If a gaseous reducing agent is used, this can be used in pure form or mixed with an inert gas, for example argon. If liquid reducing agents are used, they are usually vaporized by means of liquid metering system or bubbler and then introduced into the reaction chamber together with an inert gas as the carrier gas (eg argon).

For the treatment of copper / copper oxide-containing layers on tantalum or Tantalnitridsubstraten according to Example 1.1, this formic acid was evaporated at a temperature of 65 ⁰ C in a Flüssigdosiersystem at a flow rate of 80 mg / min and together with argon carrier gas at a Flow of 100 sccm also heated to 65 ⁰ C lines fed to the reaction chamber. The layers were at a pressure of 1.3 mbar and at temperatures between 100 ⁰ C and 300 ⁰ C, preferably at 100 to 150 ⁰ C, subjected for 20 min to 40 min of the reduction treatment.

9 shows the sheet resistances of pure tantalum nitride or the combination of a tantalum nitride and a tantalum layer (referred to as "as dep."), Of the corresponding substrates provided with a 5 nm thick copper oxide layer (referred to as "ALD on"). ") and of the layers treated as above with formic acid as the reducing agent. It turns out that at particularly low reduction temperatures, the sheet resistance decreases compared to the non-reduced system, which is an indication that in addition to the reduction no or only minor islanding occurs.

10 shows the change in the copper or copper oxide content after and before the reduction treatment according to Example 2, measured by X-ray photoelectron spectroscopy (XPS). Spectrum "4" was recorded before the reduction treatment; the coated substrate obtained according to Example 1.1 or 1.2 was here additionally stored for 25 weeks in air, whereby the copper and copper oxide-containing layer has completely converted into a copper oxide layer. The binding energies detected in the spectrum are attributable exclusively to oxidic copper. Spectrum "5" shows the binding energies after a 20 minute reduction treatment with formic acid vapor according to Example 2 at a temperature of 115 ° C. The resulting layer has high levels of metallic copper (the copper oxide signals are largely due to the fact that the sample before recording of the spectrum was stored in air for about 7 weeks). The spectra shown in Figure 10 result from XPS measurements with a take-off angle of 45 °. FIG. 11 shows the ARXPS spectrum of the sample shown in FIG. The signals were recorded at angles of 25 °, 35 °, 45 °, 60 ° and 75 °, respectively measured to the sample surface. It can be seen that the oxidized copper content is found substantially in near-surface layers of the sample, which suggests a post-oxidation due to the air contact of the sample after the reduction treatment. For larger acceptance angles, the areas further away from the sample surface are also analyzed; There are only small signals for copper oxide recorded.

FIG. 12 shows the EDX-oxygen signal (normalized) obtained by means of energy-dispersive X-ray analysis (EDX) after carrying out reduction processes with various reducing agents on tantalum nitride or on the combination of a tantalum nitride and a tantalum layer. FIG. 12 shows that when the reduction is carried out at 155 ^° C., only a formic acid reaction is appreciable. When using isopropanol (IPA) occurs a significant reduction effect only between 155 and 200 ⁰ C.

3. Investigation of the adhesion of the layers produced by ALD

The layers obtained according to Examples 1.1, 1.2, 1.3 and 2 on substrates of tantalum, tantalum nitride, ruthenium and silicon dioxide were examined with regard to their adhesive strength to the respective substrate used. For this purpose, the adhesive tape withdrawal test ("Tapetest") according to J. Baumann, "production, characterization and evaluation of conductive diffusion barriers based on Ta, Ti and W for the Copper metallization of silicon circuits ", Shaker Verlag, Aachen, 2004, page 216. The adhesive tape ^Tesa® 4129 with an adhesive force of 8 N / 25 mm was used according to DIN EN 1939. As a result of this investigation, none of the samples tested was used Chipping or peeling of the ALD layers detected by the respective substrate.

Claims

claims

A method for producing a coated substrate, wherein the coating consists of or contains copper, comprising the following steps

Providing a copper precursor and a substrate, wherein the copper precursor is a copper (I) complex that does not contain fluorine,

Layer deposition step in which, by means of atomic layer deposition (ALD), a copper-containing layer is deposited using the precursor at least on partial regions of the substrate surface,

- optionally carried out reduction step, wherein a reducing agent acts on the substrate obtained in the layer deposition step, characterized in that the precursor is a complex of the formula L ₂ Cu (XnX) wherein

- L are the same or different σ-donor π-acceptor ligands and / or the same or different σ, π-donor π-acceptor ligands and

XnX is a bidentate ligand selected from the group consisting of β-diketonates, β-ketoiminates, β-diiminates, amidinates, carboxylates and thiocarboxylates.

2. Method according to the preceding claim, characterized in that, during the layer deposition step, the following substeps are repeated several times in the order indicated:

- Adsorption, in which the substrate with the precursor is beauf

- Rinse pulse or evacuation pulse

- Oxidation pulse, followed by a rinsing pulse or evacuation pulse, and / or reduction pulse, followed by a rinsing pulse or evacuation pulse, wherein in the oxidation pulse, the substrate with an oxidizing agent and the reduction pulse, the substrate is subjected to a reducing agent.

3. The method according to any one of the preceding claims, characterized in that the precursor is liquid at room temperature.

4. The method according to any one of the preceding claims, characterized in that a complex is used as a precursor in which the bidentate ligand XnX has the formula RC (X) -CR ⁸ -C (Y) -R ¹ , wherein

- X and Y are the same or different and represent O or NR ² and

R, R ¹ , R ^{2 are} identical or different and is a branched, unbranched or cyclic alkyl radical, an aryl radical or a trialkylsilyl radical, in particular a methyl, ethyl, n-propyl, isopropyl, n-butyl radical , tert-butyl or phenyl radical

R ^{8 is} a branched, unbranched or cyclic alkyl radical is an aryl radical or a trialkylsilyl radical, in particular a methyl, ethyl, n-propyl, isopropyl, n-butyl, tert-butyl or phenyl radical.

5. Process according to claims 1 to 3, characterized in that a complex is used as the precursor, in which the bidentate ligand XnX has the formula R ³ -C (X ') - Υ', wherein

X ^{1 represents} O, S or NR ⁴ , Y represents Y ¹ O or NR ⁴ and

R ³ and R ^{4 are} identical or different and are a branched, unbranched or cyclic alkyl radical or an aryl radical which optionally contains a carboxylate group, in particular a methyl, ethyl, n-propyl, isopropyl, n-butyl, tert-butyl or phenyl radical.

6. The method according to the preceding claim, characterized in that the bidentate ligand XnX has the formula R ⁵ OC (O) -ZC (O) -O, wherein

Z is an alkylene or alkylidene bridge or a bond between the two carbonyl groups, and preferably the formula - (CH ₂ -) _n (where n = 0, 1 or 2) or the formula

Has (CH = CH-) _m (with m = 0, 1 or 2) and

R ^{5 is} a branched, unbranched or cyclic alkyl radical or an aryl radical, in particular a methyl, ethyl, n-propyl, isopropyl, n-butyl, tert-butyl or phenyl radical.

7. The method according to any one of the preceding claims, characterized in that the oxidizing agent is liquid or gaseous and is selected from the group consisting of oxygen, water, H ₂ O ₂ , ozone and N ₂ O.

8. The method according to any one of the preceding claims, characterized in that the reducing agent is selected from the group consisting of alcohols, aldehydes and carboxylic acids.

9. The method according to the preceding claim, characterized in that the reducing agent is selected from the group consisting of isopropanol, formaldehyde and formic acid.

10. The method according to any one of the preceding claims, characterized in that the adsorption pulse for a maximum of 11 s, preferably 3 to 6 s lasts.

11. The method according to any one of the preceding claims, characterized in that the adsorption pulse at a temperature of 105 ⁰ C to 165 0C, preferably 115 ⁰ C to 135 ⁰ C is performed.

12. The method according to any one of the preceding claims, characterized in that the reduction step or the reduction pulse at a temperature <250 ⁰ C, preferably <200 ⁰ C, particularly preferably at a temperature which corresponds to that of the adsorption pulse or is below this temperature, is carried out.

13. The method according to any one of the preceding claims, characterized in that the substrate is single-layered or multi-layered, that at least one of the layers of a transition metal, a transition metal salt, a semiconductor material, an organic polymer and / or an inorganic polymer or one or more substances contains from these classes of substances.

14. The method according to the preceding claim, characterized in that the substrate or the substrate layer is selected from the group consisting of Ta, Ti, W, Nb, V, nitrides, carbonitrides or silicon nitrides of Ta, Ti, W, Nb, further comprising from V, Pt, Pd, Ru, Rh, SiO ₂ , silicates, ZnO, HfO ₂ , Al ₂ O ₃ , ZrO ₂ , Si, Ge, GaAs, AlGaAs, GaN, AlGaN, InP and InGaP or that the substrate or the Substrate layer contains one or more of these substances.

15. Method according to the preceding claim, characterized in that the substrate layer to be coated with the copper layer or the substrate to be coated has a nonstoichiometric nitride of Ta, Ti, W, Nb and / or V at least in the region of the surface to be coated of this substrate layer Nitrogen content of 75-99% based on the corresponding stoichiometric compound.

16. Substrate having at least partially a closed copper coating, obtainable by the method according to one of the preceding claims, characterized in that the copper coating has no fluorine-containing impurities that the copper coating has a layer thickness between 2 nm and 30 nm.

17. Substrate having at least partially a closed copper and copper oxide-containing coating, obtainable by the method according to one of the preceding claims 1 to 15, characterized in that the coating does not contain fluorine-containing impurities in that the coating containing copper and copper oxide has a layer thickness between 2 nm and 30 nm and that the

Layer in the region of the surface consists essentially of Cu ₂ O and between this surface and the substrate facing surface of the coating is a gradient with decreasing oxygen content.

18. Substrate according to one of the preceding two claims, characterized in that the layer thickness of the copper coating is between 2 nm and 20 nm.

19. A substrate according to any one of claims 16 to 18, characterized in that the difference in the roughness Ra according to DIN EN ISO 4287 of the coated and the uncoated substrate is 0 nm to 2.5 nm.

20. Substrate according to one of claims 16 to 19, characterized in that the substrate is provided with copper coating trenches, holes and / or shadow spaces.

21. Use of the method according to claims 1 to 15 for the production of the contact and Leitbahnsystems in the production of microelectronic devices or for the production of thin film solar cells.