EP1727924A1

EP1727924A1 - Deposition by magnetron cathodic pulverization in a pulsed mode with preionization

Info

Publication number: EP1727924A1
Application number: EP05728585A
Authority: EP
Inventors: Mihai Ganciu-Petcu; Michel Hecq; Jean-Pierre Dauchot; Stephanos Konstantinidis; Jean Bretagne; Ludovic De Poucques; Michel Touzeau
Original assignee: Materia Nova ASBL; Centre National de la Recherche Scientifique CNRS; Universite Paris Sud Paris 11
Current assignee: Materia Nova ASBL; Centre National de la Recherche Scientifique CNRS; Universite Paris Sud Paris 11
Priority date: 2004-03-22
Filing date: 2005-03-22
Publication date: 2006-12-06
Also published as: RU2006133826A; RU2364661C2; CN100587107C; US20070034498A1; CN1985022A; WO2005090632A1; JP2007529632A; US7927466B2; EP1580298A1; KR20070040747A

Abstract

The invention relates to the deposition, in a magnetron reactor (1) fitted with a magnetron cathode (CM), of at least one material on a substrate (11a), said material being vaporized, by means of magnetron cathodic pulverization, with the aid of a gas which is ionized in a pulsed mode. In order to promote the formation of high current pulses of a short duration, while avoiding the formation of arcs and enabling efficient ionization of the pulverized vapor, the gas is preonized prior to application of the main voltage pulse to the magnetron cathode (CM) such that it is possible to generate current pulses (IC) whose duration of decline (Td) is less than 5µs after interruption of the main voltage pulse (IT).

Description

MAGNETRON CATHODE SPRAY DEPOSIT IN PULSE SYSTEM WITH PREIONIZATION

OBJECT OF THE INVENTION The present invention relates generally to the deposition on a substrate of a material vaporized by sputtering in a magnetron. It relates to a new method of depositing at least one material on a substrate by magnetron sputtering in pulsed regime, as well as a new pulse electrical supply for magnetron, and a magnetron equipped with said pulse supply.

STATE OF THE ART The technique of depositing material on a substrate by magnetron sputtering is known and has developed rapidly since the 1970s. This deposition technique and its main improvements known to date are by example described in the article: "Magnetron sputtering: a revie of recent develop ents and applications" PJ Kelly, RD Arnell,

VACUUM 56 (2000) pages 159-172.

In summary, the technique of depositing a material on a substrate by magnetron sputtering generally consists in bombarding a target, which forms the cathode of a magnetron reactor and which is produced in the material to be deposited. , with ions from an electrical discharge (plasma). This bombing ionic causes the target to sputter in the form of a "vapor" of atoms or molecules which deposit, in the form of a thin layer, on the substrate placed near the magnetron target, the substrate possibly being fixed or mobile.

In a non-reactive regime, the gas intended to form the plasma is an inert gas, for example argon. In the so-called “reactive” regime, a gas is used, generally diluted in an inert gas, to generate a compound which may or may not be electrically conductive; with a titanium target cathode, for example, a mixture of argon and nitrogen is used as reactive gas, which results in the formation of an electrically conductive titanium nitride (TiN) compound, or a mixture of argon and oxygen, which results in the formation of a titanium dioxide compound (Ti0 ₂ ) which is an electrical insulator. Whatever gas is used, the atoms are ionized essentially by collisions with the electrons produced in the electric discharge. In the vicinity of the target is also created a magnetic field which makes it possible to trap the electrons formed in the gas, thus forming a more strongly ionized plasma near the surface of the target. The magnetron sputtering methods are implemented at low pressure (typically between 0.1 Pa and a few Pa). As a result, the ions undergo little or no collisions during their path towards the target cathode, which increases the effectiveness of the sputtering. Also, this low pressure makes it possible to facilitate the transport of pulverized material to the substrate, by reducing the collisions of the atoms or molecules pulverized, and therefore reducing the importance of the processes of deflection and / or loss of kinetic energy. particles (atoms or molecules) sprayed.

The aforementioned technique of depositing material on a substrate by magnetron sputtering is particularly suitable for depositing, essentially in thin layers, a large variety of materials in very varied fields of application such as microelectronics ( metallic interconnection and dielectric deposits for MOSFET structures), optoelectronics (manufacture of piezoelectric substrates), mechanics or connectors (anti-wear, anti-corrosion coating deposits, etc.), the glass industry (functional layers). The materials deposited can be metallic materials or electrically conductive compounds, insulating ceramics such as nitrides, oxides, carbides, oxynitrides, etc. Compared to other known techniques of evaporation deposition (evaporation and thermal deposition process, in particular by inductive heating, or deposition process by evaporation under electron beam or by arc), magnetron sputtering presents the advantage of being more directional, due to the existence of an emission lobe, more or less open, perpendicular to the target cathode, with ejection speeds of the sprayed particles which can advantageously be greater. Originally, in the aforementioned technique of magnetron sputtering deposition, the electrical discharges in the plasma were carried out in steady state by applying a continuous voltage to the cathode or by radio frequency (RF) excitation of the gas. . In general, the limitations of the magnetron sputtering technique in steady state are linked to a generally insufficient quality of the deposited layers, particularly in that which relates to porosity, and to the difficulty of obtaining homogeneous deposits on substrates having a deposition surface with complex geometry, and at deposition rates which remain relatively low (typically of the order of μm / h). More particularly, with regard to the quality of the layers deposited, there is often observed in steady state (DC or RF) the untimely formation of electric arcs, which cause the ejection of material from the target in the form of droplets, which droplets are deposited on the substrate, detrimentally creating defects in the coating.

In addition, in the case of the deposition of insulating materials by magnetron sputtering, the use of RF excitation alone remains relatively complex and it is difficult to control the process. In addition, spray rates are significantly reduced. To improve the deposition by magnetron sputtering of insulating materials, and in particular of oxides such as alumina, it has already been proposed to draw the magnetron discharge (see the aforementioned article by PJ Kelly, RD Arnell, pages 166 to 168 / section 7 entitled “Pulsed magnetron sputtering”). Thus, according to this technique in pulse mode, voltage pulses are applied to the magnetron cathode, with the aim of creating in the plasma gas a pulse discharge current (current pulses) making it possible to ionize the gas and form a plasma ( later said "main" plasma) strongly ionized. In particular, in this article, it is indicated that by drawing the magnetron discharge at frequencies above 20 kHz, and in particular in the frequency range 20 kHz - 100 kHz, it would be possible to avoid the formation of electric arcs, caused by a target poisoning by an insulating layer. The deposition rates are also improved (of the order of ten μm / h). It is also specified in this article that the duration of the "On" pulse must be limited (adapted) to limit the surface charge due to the poisoning of the target at the origin of the electric arcs, and that during the "Off" pulse, the surface electrical charge is neutralized. More particularly, it is specified that the most effective suppression of electric arcs is observed when the duration of the pulses "Off" approaches or is equal to the duration of the pulses "On". In the exemplary embodiment given (FIG. 12), the duration of the “Off” pulses represents approximately 10% of the total cycle, and is equal to 5 μs, the duration of the “On” pulses being equal to 45 μs.

Under these conditions, the average current is only slightly lower (by 10%) than the pulse current.

More recently, it has been proposed, in particular in US Pat. No. 6,296,742 (Kousnetsov) to improve the deposition technique by magnetron sputtering in pulsed regime allowing efficient ionization of the sprayed vapor, using pulses having a power very high instantaneous (0.1k - IMW) and generated so that the gas (plasma) located near the cathode quickly reaches a state of strong ionization (operating range S8 of FIG. 1 of US Pat. No. 6,296,742). It is explained in US patent 6,296,742 that this solution would apply a high voltage to the cathode while avoiding the formation of electric arc. In practice, it can be seen that the implementation of this solution is accompanied by the formation of electric arcs detrimental to the quality of the deposits. These untimely appearance of electric arcs can be explained by the fact that the gas, before reaching a state of strong ionization must pass through an arc discharge region (region S7 of FIG. 1 of US Pat. No. 6,296,742). The plaintiffs are not, however, bound by this explanation. With the magnetron sputtering deposition technique in pulsed mode, if one wishes to increase the instantaneous power of the electric discharges produced in the gas, it is necessary to apply to the cathode voltage pulses which are the shortest possible (typically less than 50μs, more particularly less than 20μs, and preferably less than 5μs). It is found in fact that by reducing the duration of the voltage pulses, the risks of untimely formation of electric arcs are reduced. The inventors have however demonstrated that the reduction in the duration of the voltage pulses applied to the cathode generates two disadvantages detrimental to the formation of current pulses in the gas. These two drawbacks are related to the delay time of the formation of the current pulse with respect to the voltage pulse, that is to say the time which elapses between the application to the magnetron cathode of the voltage pulse, and the start of the formation of the current pulse in the gas. This delay time is linked to the development time of the electronic avalanche.

1 ^st disadvantage: too long delay time In the worst case, it sometimes happens that the aforementioned delay time is greater than the duration of the voltage pulse. In this case, the current pulse does not form. If this delay time is shorter than the duration of the voltage pulse, but is such that the pulse current does not have time to develop properly before the end of the voltage pulse, we result in the formation of a current pulse of low amplitude, and thereby, detrimentally, in the formation of '' a low power electrical discharge. In summary, the greater the delay time relative to the duration of the voltage pulse, the smaller the amplitude of the current pulse (and is even almost zero in the event of a delay time greater than the duration of the voltage pulse).

^2nd disadvantage fluctuation of the delay time [0022] ^'The second disadvantage is a large fluctuation of the delay time, which results in instability and poor reproducibility in the time of formation of the current pulses. This drawback makes the deposit process random and not reproducible.

Aims of the Invention The present invention aims to improve the known methods of depositing a material by magnetron sputtering in pulsed regime, by proposing a new solution which makes it possible in particular to overcome the aforementioned drawbacks, while allowing ionization effective steam spray.

More particularly, the invention aims to propose a method of depositing a material by magnetron sputtering in pulsed regime, which makes it possible to generate electric discharges, preferably of high power, with good stability and good reproducibility over time, while reducing the risk of arcing, and reducing the risk of redeposition on the cathode of the sprayed material. Summary of the invention The above-mentioned objectives are achieved by the process of claim 1. This process is known in that at least one material is deposited on a substrate in a magnetron reactor equipped with a magnetron cathode. , by vaporizing said material by magnetron sputtering, and using a gas which is ionized (main plasma) in pulse mode by applying to the magnetron cathode main voltage pulses. Typically according to the invention, prior to each main voltage pulse, a preionization of said gas is carried out so as to generate current pulses whose decay time, after the main voltage pulse is cut off, is less than 5μs, and preferably less than lμs.

Prior to the invention, there has already been proposed in the international patent application WO02 / 103078 a method for generating in a magnetron reactor a plasma in pulsed regime, with a prior regime of preionization by means of a direct current. , prior to the voltage pulse.

However, on the one hand this international patent application WO 02/103078 does not address the problem of stability and reproducibility over time of the current pulses generated.

On the other hand, the power supply, used in the international patent application WO02 / 103078 (see Figure 9), implements a circuit comprising inductors (L1, L2) in series with the magnetron cathode and allowing limit discharge currents (page 22, line 2). More particularly, according to the teaching of this publication (page 22, lines 2 and 3), the value of the inductance (L1) must be as high than necessary to produce the preionization current. However, the implementation of this inductance in series (L1) is detrimental for the current pulse. Indeed, it introduces a time constant, which prejudicially increases the cut-off time of the current pulse (that is to say the duration of decay of the current pulse after the cut-off of the current pulse voltage). In practice, with this type of power supply, the cut-off time of the current pulse is important and is greater than 10 μs. Thus, and unlike the invention, with the technical solution described in international patent application WO02 / 103078, it is not possible to obtain very short cut-off times for the current pulse, it is ie cut-off times less than 5μs, and preferably less than lμs. The inventors have been able to demonstrate that it was essential, in order to obtain optimal conditions for the deposition of the sprayed material on the substrate, to generate current pulses with a very short breaking time. In fact, it has been demonstrated that with significant cut-off times (> 10 μs), such as those obtained for example with the technical solution described in the international patent application WO02 / 103078, the risks are detrimentally increased of redeposition on the cathode of the pulverized material which are associated with the self-spraying, which decreases the rate of deposition on the substrate. Thanks to the invention, it is therefore advantageously possible to very significantly reduce, or even avoid, the redeposition phenomena on the cathode which are associated with self-spraying.

Also, the preionization of the gas according to the invention before each main voltage pulse results in the creation, in the magnetron reactor, of an initial plasma which advantageously makes it possible to reduce the delay time of the current pulse relative to the main voltage pulse, and make this delay time less fluctuating. Compared to a process without preionization, this generates impulse electrical discharges of higher instantaneous power, and with an average power relatively constant over time. Thus, thanks to the process of the invention, the formation of the current pulse is promoted (stabilized and / or accelerated). This preionization is particularly advantageous in a pulse operating regime with short main voltage pulses, which also advantageously makes it possible to reduce the probability of untimely formation of electric arcs. Another role of the preionization is to create in the magnetron, between the cathode and the substrate, a sufficient density of free electrons which promotes the transport towards the substrate of the vapors of ionized material. Another important advantage of the process of the invention is the obtaining of a high ionization of the vapors arriving on the substrate. Thanks to preionization, an ionization rate of the vapors arriving on the substrate is obtained which is greater than 10% and preferably greater than 70%. This ionization rate is measured in the vicinity of the substrate by an absorption spectroscopy method of the type described in one or other of the two publications below: S. Konstantinidis, A. Ricard, M. Ganciu, J.-P. Dauchot, M. Wautelet, and M. Hecq, "A study of an ionized magnetron source by pulsed absorption spectroscopy" Proceedings of the 46th Annual Technical Conference of the Society of Vacuum Coaters, published by the Society of Vacuum Coaters , Albuquerque, NM, USA (2003) 452; O.Leroy, L.de Poucques, C.Boisse-Laporte, M. Ganciu, L.Teulé-Gay, M.Touzeau "Determination of titanium temperature and density in a magnetron vapor sputtering device assisted by two microwave coaxial excitation Systems" J of Vacuum Science & Technology A: Vacuum, Surfaces, and Films, 22 (2004) 192 [0035] Obtaining, thanks to the invention, a high rate of ionization of the vapors arriving on the substrate advantageously allows '' improve the quality of the layers deposited on the substrate (in particular better adhesion and better compactness of the layers deposited). Another advantage of implementing a preionization according to the invention is to provide conditioning of the surface of the substrate (pre deposition, ablation, polarization) which allows the modulation of the properties of the deposited layers (adhesion, structure, uniformity) depending on the applications envisaged. This advantage is for example particularly advantageous in the field of microelectronics. According to the invention, the preionization of the gas before each main voltage pulse can be obtained in several different ways.

In a first alternative embodiment, the preionization of the gas is obtained by applying a preionization voltage to the magnetron cathode. This preionization voltage is preferably continuous, but can also be drawn, the main thing being that the preionization of the gas (formation of the initial plasma) is effective before the application of the main voltage pulse for the formation of the main plasma. In a second embodiment, the preionization of the gas is obtained by RF excitation of the gas. In a third alternative embodiment, the preionization of the gas is obtained by microwave excitation of the gas or by any other means making it possible to create a sufficient free charge density in the space between the cathode and the anode . For example, for the magnetized zone which faces the cathode, this density (estimated using the relationship between the electronic density and the density of the ion current extracted from the plasma) is preferably greater than 10 ⁸ cπf ³ and more preferably still greater than 10 ⁹ cm ^"3. In a fourth variant, the preionization can be ensured by a pulse repetition frequency high enough to form a stable current pulse at each voltage pulse, so as to use the density Residual electronics between two successive pulses In the context of the invention, the main voltage pulses for the formation of the main plasma can be applied to the magnetron cathode in a regime of mono-imupulsion type or in regime of type of multi-pulse, that is to say by being generated by successive trains of at least two close pulses.

Preferably, whatever the manner of carrying out the preionization of the gas, and whatever the pulse regime (mono or pluri-pulse) the duration of the main voltage pulses is less than 50 μs, more particularly less than 20 μs , and preferably between lμs and lOμs. More particularly, the frequency of the pulses (single pulse regime) or of the close pulse trains is less than or equal to

100 kHz, and is preferably greater than 50 Hz.

The invention also has other objects as a new impulse power supply for magnetron reactor referred to in claim 13, as well as a magnetron reactor equipped with this impulse power supply.

Brief description of the figures

Other features and advantages of the invention will appear more clearly on reading the description below of several alternative embodiments of the method of the invention, which description is given by way of non-exhaustive example and not limiting of the invention, and with reference to the appended figures in which: FIG. 1 schematically represents a magnetron reactor equipped with a pulse supply of the invention. Figure 2 is an electrical diagram of an exemplary embodiment of a pulse power supply of the invention. FIG. 3 schematically represents oscillograms respectively of the voltage (U) measured on the magnetron cathode and of the current (I) passing through the magnetron cathode, in an operating regime of the single-impulse type.

Figures 4 to 17 are real oscillograms of the control signal (S), the voltage (U) and the current (I) above, obtained by screenshot of an oscilloscope under different operating conditions described in detail later. Figure 18 shows schematically another alternative embodiment of a magnetron reactor of the invention, with RF loop for preionization. Detailed description of the invention

Block diagram of the magnetron reactor (1) / Figure 1

There is shown in Figure 1, the block diagram of a magnetron reactor 1 equipped with a pulse electrical supply 2 which is according to the invention. The magnetron reactor 1 is known per se. Its structure and general functioning are briefly recalled. The magnetron reactor 1 essentially comprises, and in a manner known per se, a deposition chamber 10 inside which is mounted a cathode CM, hereinafter called the magnetron cathode. Right and at a distance from this magnetron cathode CM is mounted a substrate holder 11 intended to receive a substrate 11a. The enclosure 10 is further equipped with an intake system 12 for the introduction of a gas or gas mixture, which once ionized will form a plasma. The magnetron cathode consists of a sample of the material which it is desired to deposit in the form of thin layers on the substrate 11a. In general, and in a manner known per se, the power supply 2 allows in operation to apply to the magnetron cathode CM a negative voltage which generates within the gas of the deposition chamber 10 of the electric discharges resulting in the formation of a discharge current passing through the magnetron cathode CM, and by a strong ionization of the gas (formation of a main plasma). The ions of this main plasma bombard the surface (target) of the magnetron CM cathode with sufficiently high speeds to pulverize the magnetron CM cathode on the surface. It is formed thus within the enclosure 10 a vapor of the material constituting the cathode CM, which vapor of material is deposited in thin layers on the surface of the substrate 11a. At the underside of the cathode CM are further provided magnets 13 which create a permanent magnetic field having the main function of capturing and confining the electrons of the main plasma in the vicinity of the magnetron cathode. This magnetic field thus makes it possible to form and maintain a more strongly ionized magnetized plasma in the vicinity of the magnetron cathode CM.

Power supply structure (2) / Figure 2

FIG. 2 represents the electrical diagram of an exemplary embodiment of a pulse electrical supply 2 according to the invention. The power supply used in the invention is designed to allow the advantages of continuous or pulsed preionization to be used. The power supply 2 includes an output (Out) which is connected (in a manner known per se) to the cathode CM of the magnetron reactor, first means (S0, GI, TI, T, SI, Cl, RI, Dl , D3) for generating on the output (Out) main voltage pulses, and second means (S2, R, D) for generating a preionization voltage on the output (Out) simultaneously with the main voltage pulses. More particularly, in the embodiment illustrated in FIG. 2, the first aforementioned means for generating the main voltage pulses comprise: a DC voltage source SI (delivering in negative voltage output HT), means (S0, Gl, Tl) for generating control pulses, switching means T which are mounted between the DC voltage source SI and the output (Out), a resistor RI and a separation diode Dl which are connected in series between the switching means Tl and the output (Out). The junction J between the switching means T and the resistor RI is also connected to ground, via a separation diode D3. The output of the voltage source SI is also connected to ground by means of a capacitor Cl. The means for generating control pulses more particularly comprise: a supply S0 delivering a control signal S, of the type rectangular, adjustable frequency and duty cycle, - a pulse generator (GI) triggered by the control signal (S), a pulse transformer for controlling the switching means T. [0062] More particularly, in the example illustrated, the switching means T is a bipolar junction transistor of the IGBT type, the gate of which is coupled to the pulse generator GI, by means of the pulse transformer TI.

The DC voltage source SI allows, only when the transistor T is on, to apply a negative voltage on the magnetron cathode CM, via the resistor RI and the separation diode Dl connected in series with the cathode CM. The aforementioned second means for generating the preionization voltage comprise a DC voltage source S2 connected to the output (Out) via a resistor R and a separation diode D in series. The voltage source S2 delivers a negative negative preionization voltage (HTP) at the output. A switch SW is further mounted in series with the output of the voltage source S2. When the switch SW is closed, the negative preionization voltage (HTP) is applied to the magnetron cathode CM, via the resistor R and the separation diode D. The aforementioned switch SW is optional and is used essentially to be able to carry out comparative tests (see below the description of FIGS. 4 and following) between: [switch (SW) open] / operation of the magnetron reactor 1 in an impulse regime alone (conventional operation) and, [switch (SW) closed] / an operation in accordance with the invention, with superposition, on the impulse regime, of a permanent regime allowing the required preionization of the gas to be obtained (in in this case, application of a continuous negative preionization voltage on the magnetron cathode (CM)). To achieve a power supply according to the invention, we can replace the transistor T which in the example of Figure 2 is an IGBT transistor, by any equivalent means known to those skilled in the art, that is ie by any rapid switching means controlled by an electrical signal. The transistor T can, for example, and in a non-exhaustive manner, be replaced by a Behlke type switch. Power supply operation (2) / Figures 2 and 3

The operation of the power supply 2 will now be explained with reference to Figure 2 and also with reference to Figure 3. Figure 3 shows schematically an example of oscillograms of the voltage (U) measured between the magnetron cathode CM and the ground, and of the discharge current (I) passing through the magnetron cathode CM. For the description of the operating phases (I to IV) below, it is considered that the switch (SW) of the supply 2 is closed.

Phase I / Pre-ionization (FIG. 3): The transistor T is blocked. Only the negative preionization voltage (HTP) is applied to the magnetron cathode CM, via the current limiting resistor R. This negative DC voltage generates within the gas a continuous electrical discharge resulting in a continuous preionization current Ip which passes by the magnetron cathode CM, the separation diode D and the resistance R.

Phase II:

At the end of the first preionization phase (I), the gate of transistor T is controlled by the pulse generator GI (transistor T passing), which makes it possible to temporarily apply the negative voltage (HT) on the magnetron cathode CM via the current limiting resistor RI and the separation diode Dl. A pulse of main voltage IT, of duration ti, is thus applied to the cathode CM, which results in a main current pulse IC passing through the cathode. magnetron CM. This main current pulse IC is generated with a low delay time Δ, relative to the start of the voltage pulse IT.

Phase III: The transistor (T) is again blocked (reverse switching of the transistor by the pulse generator GI). The negative high voltage HT is no longer applied to the magnetron cathode CM (end of the main voltage pulse). The current (I) remains supported only by the inductive energy accumulated in the equivalent inductance (L) of the magnetron 1 reactor. The current (I) decreases with a time constant substantially equal to [L / (RE)] where RE is an equivalent series resistance limited lower by the value of the resistance RI. The difference between RE and RI is related to the nonlinear impedance of the plasma. The diode D3 (FIG. 2) ensures the passage of the current (I) after the switching of the transistor T, at the same time avoiding overvoltages on this transistor.

Phase IV: After the impulse current has stopped, the impedance of the magnetron 1 reactor remains low due to the state of strong ionization of the magnetized plasma. Consequently, for the same direct preionization current (Ip), the voltage (U) on the magnetron cathode remains low for some time. As the ionization state of the magnetized plasma decreases, the impedance of the magnetron reactor 1 increases, and we gradually return to the initial conditions of the aforementioned preionization phase (I).

The cycle is resumed with a repetition frequency (f) which is fixed according to the application considered.

Choice of RI, Cl and R.

The choice of the resistance RI is important for the proper functioning of the power supply 2. This choice is made so as to comply with the following two opposite constraints: - The duration (Td) of decrease of the current pulse IC during phase (III) (see Figure 3) must be as low as possible (compared to the duration ti of the main voltage pulses IT). More particularly, the resistance RI is chosen so that the duration (Td) of decrease (current cut-off time) is less than 5 μs, and preferably less than or equal to l μs, which corresponds substantially to the choice of a ratio ( L / Rl) less than 2.5 μs, and preferably less than or equal to 0.5 μs. This constraint thus makes it possible to set the minimum value of the resistance RI appropriate for a given equivalent inductance L of the magnetron reactor 1 and of the connection conductors. - The resistance (RI) must not be too high in order to limit the maximum current (Imax / figure 3) generated during phase II; in practice we will choose (RI) so that this current (Imax) is in all cases less, for example, twice the maximum operating current of the magnetron reactor, in order to avoid any risk of abnormal operation of the magnetron reactor . Preferably, the supply is adjusted so that one and / or the other of the following conditions are met: the maximum density of preionization current (at target level, i.e. magnetron cathode) is between 0.05 mA / cm ² and 100 mA / cm ² ; the maximum current density (at the target) is greater than 0.1 lA / cm ² , and preferably greater than 1A / cm ² ; the current rise time is less than 20μs and more particularly less than lμs. The capacitance (Cl), of low series inductance, is chosen so as to obtain, during the aforementioned phase II, a pulse current while maintaining a suitable high voltage (U) on the magnetron cathode (CM) for l applied pulse. The resistance (R) is chosen so as to define and limit the initial preionization current. In a specific embodiment given as an indication and not limiting of the invention, the voltage (HT) was worth at most -1100 V. The voltage (HTP) was worth at most -1100V Equivalent inductance (L) of the magnetron 1 reactor was worth approximately 0.5μH. The resistance RI was worth 5Ω, so that (Td) was worth about 0.1 μs. Resistor R was worth 300kΩ. The capacitance Cl was worth 10 μF.

Comparative tests; pulse regime without preionization pulse regime with preionization (Figures 4 to 17)

Comparative tests were carried out with the magnetron reactor 1 and the supply 2 previously described with reference to Figures 1 to 3, in order to highlight the advantageous effects of preionization on the formation of the pulse of current IC during phase II, in particular on the delay time (Δ) of the current pulse, on its intensity maximum I _m χ. ^and ^has the stability over time of the voltage pulses IT and current IC. The main conditions common to all the tests were as follows: the gas used to form the plasma was argon; the cathode (target) was made of titanium. The pressure inside the enclosure (10) of the magnetron reactor was equal to lOmTorr (ie approximately 1.33 Pa). For each test, the following three signals were captured simultaneously by means of an oscilloscope (see FIGS. 4 to 17) (S): control pulse delivered by the generator S0 (I): current passing through the cathode magnetron (CM) (U): voltage measured between the magnetron cathode (CM) and earth.

The tests were carried out each time in a comparative manner [see column (2) of the table below]: without applying a preionization voltage [switch (SW) open) / Figures 4, 6, 8, 10, 12, 14, 16], and by applying a DC preionization voltage (HTP) (switch (SW) closed) / Figures 5, 7, 9, 11, 13, 15, 17]. For all the tests with preionization, the DC voltage (HTP) of preionization was equal to -1000V, and the direct current (Ip) of polarization was worth approximately 3.3mA. It should be noted that because of the importance of the scale used to measure the current (I), the preionization current (Ip) of very low value is not visible on the oscillograms of Figures 5, 7, 9 , 11, 13, 15, 17. The comparative tests were carried out in single-pulse mode) (Figures 4 to 7 and Figures 12 and 13), and in double-pulse mode (Figures 8 to 11 and Figures 14 to 17), that is to say by generating successive trains of two pulses IT voltage close together. The main variable parameters and results of these tests are summarized in the table below. In this table, the duration (ti) [column (4)] corresponds to the width of each main voltage pulse IT; the duration (Ti) [column (5)] corresponds, in the case of a double pulse regime (that is to say a succession of trains of two closely spaced voltage pulses), to the duration separating the two pulses a train of pulses; the frequency (f) [column (6)] corresponds to the frequency of repetition of the IT voltage pulses (in the case of a single-pulse regime) or of repetition of the trains of two closely spaced IT pulses in the case of a double regime impulse ; the delay time Δ [column (7)] measures the time separating the start of the current pulse IC with respect to the start of the voltage pulse IT. The values Imax and Umax reported in the table below (columns (8) and (9)) correspond to the maximum amplitude of the current pulse and the voltage pulse respectively; in the case of tests in double pulse mode, these amplitudes (Imax and Umax) are measured on the first pulse of the train of two pulses. Table / Comparative results: (D (2) (3) (4) (5) (6) (7) (8) (9)

(*) Oscilloscope in envelope mode (**) No first current pulse

The analysis of Figures 4 and 5 clearly shows that in the absence of preionization (Figure 4), the current pulse has a significant delay time compared to the voltage pulse, which results in a current pulse of very low amplitude, whereas under the same operating conditions with preionization (FIG. 5), a current pulse is obtained much more quickly (Δ very weak), and therefore with a large amplitude. The power of the electric discharges generated within the gas is therefore, advantageously, significantly greater in the case of operation in pulse mode with preionization. In certain cases, in the absence of preionization, it is even possible to achieve a delay time (Δ) greater than the width (ti) of the voltage pulse, which results in the almost total absence of current pulse. Referring to Figure 6 (f = 100Hz), in the absence of preionization the current pulse develops more significantly than for the case of Figure 4 (f = 50Hz), however compared to operation with the same repetition frequency (f = 50Hz) and with preionization (figure 7), the current pulse is made later (Figure 6: Δ = 3, βμs / Figure 7: Δ = 1.2μs). In the absence of preionization, the density of free charges is very low and the time necessary for the formation of a plasma of sufficient density to give rise to a current usable for spraying is too long. Preionization makes it possible to reach the saturation phase of the magnetron current much more quickly. Referring to Figure 8, in double pulse without preionization, the first current pulse occurs late and therefore has a very low amplitude (situation comparable to the figure

4 in single pulse mode). In the case of FIG. 14 (f = 100 Hz, Ti = 23.8 μs), we even note the total absence of the first current pulse in the absence of preionization. Comparatively, by implementing a preionization (FIG. 9 and FIG. 15), there is the formation of a first current pulse which forms very early and which has a large amplitude. Of moreover, with preionization, the second current pulse advantageously has a greater amplitude than the second current pulse generated in the absence of preionization. Figures 10 and 16 (double pulse regime without preionization / oscilloscope in envelope mode) show a very high instability in time of the current pulses (IC) and voltage (IT), which is detrimental to the quality and the reproducibility over time of the deposits of material on the substrate. Comparatively, with preionization (FIGS. 11 and 17), there is a very good stability over time of the current and voltage pulses (IC and IT), which advantageously makes it possible to obtain better reproducibility and better stability of the deposit in time.

The invention is not limited to the above examples of implementation, and in particular to the electrical parameters described above for the power supply and the magnetron. It is up to a person skilled in the art to size and adjust the power supply on a case-by-case basis for a given magnetron, for example by observing the current (I) and voltage (U) curves by means of an oscilloscope, and by modifying the preionization so as to obtain the desired effects in particular in terms of delay time and / or amplitude and / or stability of the current pulses or even so as to obtain the desired ionization rate for the vapors arriving on the substrate. Also, the preionization can be adjusted so as to limit the effects of poisoning of the target in the case of operation in reactive gas.

More generally, the pre-ionization of the gas prior to the application of each IT voltage pulse can be obtained by any known means of the skilled person. More particularly, the preionization can be ensured by a sufficiently high repetition frequency of the pulses so as to use the residual electronic density between two successive pulses. By way of nonlimiting example, the preionization can also be obtained by creating a plasma by RF excitation or secondary microwave or by any other means making it possible to obtain a sufficient electronic density of preionization (corona discharge, UV preionization ) in addition to the main pulse plasma. For example, with reference to the variant illustrated in FIG. 18, an RF excitation of the gas is carried out by means of a copper loop 14 placed in the deposit enclosure near the magnetron cathode and supplied by means of a generator 15 by an RF current of predefined frequency. In FIG. 18, the block 16 shows diagrammatically an impedance matching network which is interposed in a manner known per se between the generator 15 and the loop 14. [0093] Results comparable to those previously described could be obtained with a RF excitation at a frequency of 13.56 MHz, the distance between the magnetron cathode CM and the copper loop 14 being approximately 4 cm. The coupling of the RF excitation is not necessarily of the inductive type, but can also be of the capacitive type. In another alternative embodiment (not shown), the pre-ionization can also be obtained by means of microwaves applied in the deposition enclosure, for example at a frequency of 2.45 GHz. In the examples of Figures 1 to 17, the preionization was obtained by means of a direct preionization current. This is not limitative of the invention. In another variant, the preionization current could be a pulsed current, the essential being that the preionization of the gas precedes the application of the voltage pulse. In this case, the preionization current pulses must precede the main voltage pulses (IT). This remark is also valid in the event of RF or microwave excitation or any other preionization excitation, these can equally be carried out in continuous mode or in pulsed mode.

Claims

CLAIMS 1. Method of depositing, in a magnetron reactor (1) equipped with a magnetron cathode (CM), at least one material on a substrate (lia), method according to which said material is vaporized, by magnetron sputtering, using a gas which is ionized in pulse mode by applying to the magnetron cathode (CM) main voltage pulses (IT), characterized in that prior to each main voltage pulse (IT), one performs a preionization of said gas, so as to generate current pulses (IC) whose duration (Td) of decay, after breaking of the main voltage pulse (IT), is less than 5 μs. 2. Method according to claim 1 characterized in that the duration (Td) of decrease of a current pulse (IC), after breaking of the main voltage pulse (IT), is less than lμs. 3. Method according to claim 1 or 2 characterized in that the ionization rate of the vapors measured in the vicinity of the substrate is greater than 10% and preferably greater than 70%. 4. Method according to one of claims 1 to 3 characterized in that the preionization of the gas is obtained by applying to the magnetron cathode (CM) a preionization voltage. 5. Method according to claim 4 characterized in that the preionization voltage is continuous. 6. Method according to one of claims 1 to 5 characterized in that the density of the preionization current at the magnetron cathode is between 0.05 mA / cm ² and 100 mA / cm ² . 7. Method according to one of claims 1 to 3 characterized in that the preionization is ensured by a pulse repetition frequency high enough to form a stable current pulse with each voltage pulse. 8. Method according to claim 1 to 3 characterized in that the preionization of the gas is obtained by RF excitation of the gas. 9. Method according to claim 1 to 3 characterized in that the preionization of the gas is obtained by microwave excitation of the gas. 10. Method according to any one of claims 1 to 3 characterized in that the main voltage pulses are generated by successive trains of at least two close pulses (IT). 11. Method according to any one of claims 1 to 10 characterized in that the duration (ti) of the main voltage pulses (IT) is less than 50 μs, more particularly less than 20 μs, and preferably between lμs and lOμs . 12. Method according to any one of claims 1 to 11 characterized in that the frequency (f) of the pulses (single pulse regime) or of close pulse trains is less than or equal to 100 kHz, and preferably greater than 50 Hz 13. Impulse power supply (2) for magnetron reactor (1) and comprising an output (Out) intended to be connected to the cathode (CM) of the magnetron reactor, characterized in that it comprises first means making it possible to generate on the output (Out) of the main voltage pulses (IT), and second means making it possible to generate a preionization voltage on the output (Out) simultaneously with the main voltage pulses (IT), and in that the first means for generating main voltage pulses include: - a DC voltage source (1), - means (S0, GI, Tl) for generating control pulses,

- switching means (T) which are mounted between the DC voltage source (SI) and the output (Out), and which are controlled by the means (S0, GI, Tl) for generating control pulses,

- a resistor (RI) which is connected in series between the switching means (T) and the output (Out), the junction (J) between the switching means (T) and the resistor (RI) being connected to ground . 14. Supply according to claim 13 for magnetron reactor (1) having an equivalent inductance (L), characterized in that the resistance (RI) is chosen so that (L / Rl) is less than 2.5 μs and preferably less than 0.5μs. 15. Power supply according to claim 13 or 14 characterized in that it comprises a diode (D3) between the junction (J) and the ground, allowing the passage of a discharge current between the output (Out) and the ground, by protecting the switching means (T) against an overvoltage. 16. Power supply according to one of claims 13 to 15 characterized in that the second means for generating the preionization voltage comprise a DC voltage source (S2) connected to the output (Out) via a resistor (R) and a separation diode (D) in series. 17. magnetron reactor (1) equipped with an impulse electrical supply referred to any one of claims 13 to 16, the output (Out) of the supply (2) being connected to the cathode (CM) of the magnetron.