Carbon spark evaporation

Abstract

The invention relates to a method for operating a pulsed discontinuous spark discharge. The spark is fed via a capacitor. Between the pulses there are switched-off time intervals during which no spark current flows. Within the pulses, that is to say during the switched-on time intervals, the supply of charge is stopped upon a current threshold being reached and is restarted, with the result that subpulses occur within the pulses. The time intervals and subpulses are chosen according to the invention such that when the capacitor is switched on again, the spark discharge readily ignites again.

Claims

1. Method for depositing a layer on a substrate, comprising the steps of: providing substrates; placing the substrates in a vacuum chamber; pumping out the vacuum chamber; providing two separate current supplies, namely a direct current supply and a pulse current supply that discharge to achieve a pulsed discontinuous spark discharge, with the pulse current supply comprising at least one capacitor and where the corresponding discharge current of the pulse current supply is controlled by periodical switching on and off of the capacitor to form switched-on time intervals and switched-off time intervals, wherein the capacitor is switched off during the switched-off time intervals, during which no current is supplied from the direct current supply and no current is supplied from the pulse current supply, wherein within the switched-on time intervals, no current is supplied from the direct current supply and the discharge current of the pulse current supply is momentarily interrupted upon reaching a previously determined threshold, with the result that subpulses occur within a current pulse from the pulse current supply; and depositing the layer on the substrate by performing a cathodic spark evaporation by means of the pulsed discontinuous spark discharge, wherein the layer comprises carbon.

2. Method according to claim 1, wherein at least a first plurality of the switched-off time intervals is chosen so to be short enough that after expiration of each one of the plurality of the switched-off time intervals, the switching on of the capacitor automatically causes the renewed ignition of the spark discharge.

3. Method according to claim 1, wherein the first ignition of the spark discharge is achieved by means of an ignition device, and after the first ignition of the spark the capacitor is switched on for the first time and has a first charge state at the time of the first ignition, and the pulsed discontinuous spark discharge method is subsequently further operated with the direct current supply switched off.

4. Method according to claim 3, wherein a length of a second plurality of switched-off time intervals is chosen depending on the charging means of the at least one capacitor, such that after expiration of each of the second plurality of switched-off time intervals, the capacitor has essentially reached the first charge state.

5. Method according to claim 1, wherein the layer is one or more carbon layers that are hydrogen-free and/or contain hydrogen and/or metal-free and/or contain metal and is coated on the substrate.

6. Method for depositing a layer on a substrate, comprising the steps of: providing substrates; placing the substrates in a vacuum chamber; pumping out the vacuum chamber; providing two separate current supplies, namely a direct current supply and a pulse current supply that discharge to achieve the pulsed discontinuous spark discharge, with the pulse current supply comprising only one discharging capacitor and where the corresponding discharge current of the pulse current supply is controlled by periodical switching on and off of the capacitor to form switched-on time intervals and switched-off time intervals, wherein the capacitor is switched off during the switched-off time intervals, during which no current is supplied from the direct current supply and no current is supplied from the pulse current supply, wherein within the switched-on time intervals, no current is supplied from the direct current supply and the discharge current of the pulse current supply is momentarily interrupted upon reaching a previously determined threshold, with the result that subpulses occur within the a current pulse from the pulse current supply; and depositing the layer on the substrate by performing a cathodic spark evaporation by means of the pulsed discontinuous spark discharge, wherein the layer comprises carbon.

7. Method according to claim 3, wherein the ignition device that causes the first ignition of the spark discharge is a DC direct voltage source.

8. Method according to claim 1, wherein the direct current supply provides current for the first ignition of the spark discharge during a first switched-on time interval.

9. Method according to claim 8, wherein during switched-on time intervals after the first switched-on time interval, no current is supplied from the direct current supply.

10. Method according to claim 1, wherein the switched-on time intervals and switched-off time intervals alternate with each other.

Description

(1) The figures show:

(2) FIG. 1: diagrammatic representation of method A (state of the art)

(3) FIG. 2: diagrammatic representation of method B (state of the art)

(4) FIG. 3: vacuum coating facility and circuitry with electric supplies

(5) FIG. 4: temporal progression of U.sub.KA and I.sub.KP when operating the spark discharge as per method A according to the state of the art with a pulse frequency of 500 Hz. A time of 100 s was specified as pulse length.

(6) FIG. 5: temporal progression of U.sub.KA and I.sub.KP for a individual pulse with better resolution when operating the spark discharge as per method A according to the state of the art with a pulse frequency of 500 Hz. The trigger signal for the pulse is 100 s. The current increase takes place during the predetermined pulse time, wherein the current reaches the freely selectable current limitation of approx. 720 A already before 100 s, which results in the voltage being switched off after approx. 40 s.

(7) FIG. 6: representation of an operation with pulsed spark discharge (improved method A). The direct current supply (5) supplies an I.sub.KD of 30 A (not shown). Pulses with the pulse current supply (8) were superimposed over this direct current discharge. The pulse current I.sub.KP is represented in the figure. The pulse length was set in this case at 800 s (trigger signal indicated). Between the pulses, a pulse pause was set at 1200 s. The current was switched off respectively at approx. 700 A. In these conditions, it is possible to observe the formation of a cluster of pulses over the length of the specified pulse, which is composed respectively of 6 subpulses.

(8) FIG. 7: improved temporal resolution of the curves of FIG. 6. The specified pulse length is 800 s with a current limitation at approx. 700 A. In this pulse length, 6 subpulses are generated by the current limitation, wherein in each subpulse the voltage above approx. 40 s is stabilized at a higher level.

(9) FIG. 8: inventive operation of the pulsed spark discharge. The FIG. shows the temporal progression of I.sub.KP and U.sub.KA for the case of the switched off direct current supply (5) (according to an operation with merely discontinuous pulses). The specification was for pulse lengths of 1000 s and pulse pauses also of 1000 s.

(10) FIG. 9: improved temporal resolution of FIG. 8. A single pulse with 7 subpulses is shown. The curve shows that ignition voltages around 60V are sufficient to reignite the subpulses within the single pulses without additional ignition devices being required. Surprisingly, these voltages are also sufficient for reigniting the individual pulses after 1000 s.

(11) FIG. 10: The FIG. shows the temporal progression for U.sub.KA and I.sub.KP (according to the invention without being superimposed over direct current) for a predetermined pulse duration of 900 s, in the course of which 6 subpulses are generated. A time of 1100 s was chosen as pulse pause. For this case, the ignition operation could only just be achieved with the purely discontinuous pulses. For smaller pulse lengths and longer pulse pauses in this experimental configuration, the ignition could no longer be ensured.

(12) In FIGS. 4 to 6, 3 curves are respectively represented. The uppermost curve corresponds to the current pattern. Accordingly, the right axis is relevant. The middle curve corresponds to the voltage pattern. Accordingly, the left axis is relevant. The lowest curve should merely show the pulse duration. Its amplitude is chosen in random units.

(13) The tests were performed on a vacuum coating facility of the type Innova of the company OC Oerlikon Balzers AG. The description will make the transferability of the method to other vacuum coating facilities apparent to the one skilled in the art.

(14) The vacuum coating facility is represented diagrammatically in FIG. 3 and includes a receptacle 1 that can be evacuated, in which is located a rotatable substrate holder 4 for receiving the substrates to be coated. In all tests, argon gas was admitted at a flow of 200 sccm into the evacuated receptacle, with a total pressure in the receptacle of 0.6 Pa being reached. The pressure range of the argon gas can be varied to a large extent and other gases, for example nitrogen, can also be used.

(15) A spark evaporation source containing the carbon target 2 to be evaporated is furthermore located in this evacuable receptacle. The target 2 is mounted accordingly, being held mechanically and being insulated from the receptacle 1 which is connected to ground. It is switched as cathode for the spark discharge. Additionally, an anode 3 is located in the receptacle and is mounted preferably in the vicinity of the target 2 operated as cathode. The spatial proximity of the anode 3 to the target 2 makes the spark operation easier, especially the ignition process in the case of the discontinuous pulse sequences with self ignition or without additional external ignition. The central arrangement of the anode 3 also contributes to the stability of the spark, because it encourages the ignition process especially in the target center and prevents the spark from running to the target edge. The pulsing of the spark current also contributes to stabilizing the spark discharge. First, because it prevents the spark from persistently remaining on one spot of the target but also, and this applies especially in the case of discontinuous spark operation, because, helped by the central placing of the anode 3 relative to the target 2, it encourages the spark ignition in the middle of the target and the spark current can be interrupted (by setting the pulse length) before the spark can reach the target edge. Two electric supplies are used for the cathodic spark evaporation of the target 2, whose functions can of course be fulfilled also within one feed. The direct current supply 5 represents a current supply with an open-circuit voltage of typically 100V. In spark operation, due to the low impedance of the spark discharge (less than 1 Ohm), this voltage will settle to voltages below 50V, in the case of a carbon target typically to voltages around 20V. In the case of such current supplies, typical currents up to approx, 1000 A can be set. In the experiments described here, however, only the current range around 100 A and lower is preferably used since coating rates that are too high, as mentioned above, will indeed reduce the hardness of the carbon layers. The pulsed electric supply 6, which essentially switches a capacitor discharge, is designed with its electronic component elements in such a manner that it can switch high discharge currents. It can therefore make sense to size the electronic component elements (e.g. IGBT) in such a manner that an optimum can be sought between the required discharge current and the switching frequency. The size of the capacitor is chosen so that for a given charging voltage on the capacitor, sufficient energy (q=CU) can be stored in order to feed the cathodic spark or, in other words, to evaporate the corresponding cathode material (here carbon). The design of the pulsed electric supply is thus a compromise between pulse current, pulse frequency, charging voltage and available capacitor capacity. The switches 7 and 8 enable the electric supplies for operating the spark discharge to be connected to the target.

(16) It is also possible, and this is new and part of the invention, to operate the spark discharge after an initial ignition fed alone through the pulsed electric supply, namely only with the pulsed current I.sub.KP. The entire spark current I.sub.KA is then identical with I.sub.KP. The discharge voltage of the spark discharge U.sub.KA is measured between the cathode (target 2) and the anode 3, as represented in FIG. 3. If the spark discharge were to be operated alone with the pulsed supply, it would be required in the state of the art to reignite again and again at each pulse. With the inventive method, this continuously renewed ignition is unnecessary.

(17) The supplies are each fused with diodes 10 which prevent the supplied current from being fed from one supply into the other and thus damages from occurring. Usually, the positive outputs of the supplies are connected to earth, i.e. to the potential on which the receptacle is also placed. It can however also be advantageous to operate the spark discharge floatingly (with zero potential). For this purpose, the switch 9 is opened. Depending on the switch position, the anode is thus operated either with zero potential or on mass potential (same potential as the receptacle). In the tests, for the initial ignition of the spark discharge a mechanical ignition contact 11 was used which momentarily establishes via a low-ohmic resistance a contact with the anode and thus results in an initial evaporation of the target material, which then renders conductive the path from the cathode to the anode so that the electric supplies 5 and/or 6 take over the spark feed for evaporating the target material, i.e. they are operated at lower voltages. After the initial ignition, according to the invention the electric direct voltage supply 5 is switched off for example via the switch 7. The spark is henceforth fed exclusively by the pulsed electric supply 8 without direct current portion.

(18) It has been shown in the literature for discontinuous pulse operation (method 8) that an excessive increase of the current at the beginning of the pulse contributes to the increase of the proportion of multi-charged ionized ions of the evaporated target material (Paperny et al., Ion acceleration at different stages of a pulsed vacuum arc, J. Phys. D: Appl. Phys. 42 (2009) 155201). Unfortunately, the voltage characteristic was not examined during these tests. The excessive increase of the current at the beginning of the pulse is probably caused by the momentary application of a higher voltage, i.e. the excessive increase in the voltage is possibly to be considered the cause itself for the increased ionization and the current increase represents merely a secondary effect.

(19) In method B, an increased spark voltage can be achieved in that the spark is ignited under the condition that a high open-circuit voltage is applied between cathode and anode. This can occur for example in that the capacitor that is used for the discharge in the pulse is charged to a high voltage and the spark is ignited based on this high open-circuit voltage of the capacitor through a further momentary evaporation of material. After ignition, the capacitor then discharges as long as its charge lasts in order to maintain the spark. Immediately after the ignition (typically within the first 50 s), the spark discharge occurs at increased voltage and subsequently falls back again to a lower voltage value which is typically given by the impedance of the respective direct current spark discharge.

(20) According to the present state of knowledge (Paperny et al.), the generation of the ions having higher charge states occurs at the beginning of the current pulse and is then lost after the latter, so that for a pulse length of 200 s, after the first 50 s, only charge states are generated that are typical for direct current spark discharge. Therefore, in order to generate higher charge states, it is important to generate short current pulses in the order of magnitude below 10 s, preferably below 50 s. At typical pulse frequencies of 1 kHz, this would mean a significant reduction of the duty cycle and result in very low coating rates. A faster external ignition through material evaporation is coupled to the charging behavior of the capacitor to be used (the capacity of the capacitor must be sufficient great to generate material plasma to ignite) and in good cases is achieved also at 1 ms charge time (see Anders et al.), i.e. at approx. the frequency of 1 kHz. It would be conceivable as a possibility, albeit an expensive one, to couple several ignition capacitors temporally to one another in order to obtain pulse sequences with higher frequencies. This is however laborious and expensive and has so far, to the inventors' knowledge, also not been pursued to date. The increase of the pulse frequency, especially for short pulses and in discontinuous operation, is a further aspect that is solved by the invention.

(21) Whilst a current increase at the beginning of the pulse in method B it is thus comparatively easy to achieve and can be complex merely because of the high voltage at the ignition mechanism, high spark voltages are more difficult to achieve with method A. The one skilled in the art knows that a gradual voltage increase in the direct current spark discharge results in a limitless current increase without a noticeable overvoltage occurring. It is only the superimposition of the direct current with very steep pulses that increases the voltage of the spark discharge in method A. It is difficult to stabilize this voltage increase over longer periods of time, since the current rises quickly and the capacitor discharges quickly, with the breakdown associated therewith of the voltage at the capacitor. This applies in particular for the case of the underlying direct current, since a direct current spark discharge has small impedances in the range of only approx. 0.1 Ohm.

(22) The inventive methods are also based on a further aspect of the cathodic spark evaporation of carbon. From the investigations of Horikoshi at al., Vacuum Are Deposition of Homogeneous Amorphous Carbon Films at High Growth Rates, New Diamond and Frontier Carbon Technology, Vol. 18, No. 5 (2006), 267-277, it is possible to conclude that the discharge voltage at the capacitor and the size of the charge during the capacitor discharge influence the frequency of splatters during cathodic spark evaporation, in the sense that the higher the discharge voltage is and the greater the charge quantity during discharge is, the smaller the number of splatters is. This was shown for individual current pulses according to method B. The pulse shapes recorded in that connection (FIG. 4 in Horikoshi et al.) represent the typical current-time progression for a capacitor discharge: at the beginning a steep current rise up to a maximum, that is typically reached within the first 100 s, and then slower current drop. The pulse length depends on the capacity and the impedance of the discharge. Discharge voltage and charge also influence the pulse shape.

(23) The experiments described here were carried out with a pulsed electric supply 6 based on a controlled capacitor discharge and which in this special case allowed pulse frequencies up to 2 kHz at maximum discharge currents of 1000 A. The capacitor had a capacity of 8 mF and voltages up to 1000V could be used for charging the capacitor.

(24) The amplification of the capacitor readily makes it possible to also use currents of 2000 A or more and charging voltages of up to 2000V can also be used. The charging time of the capacitor is proportional to the product of R and C, wherein R is the ohmic resistance in the load circuit and C is the capacity of the capacitor. If one thus wishes to charge capacitors with large capacities C, one is forced either to take into account long charging times or to use high charging voltages in order to achieve the same charge quantity of the capacitor with shorter charging times. Higher charging voltages reach the acceptable limits of the electronic component elements and longer charging times limit the pulse frequency (Anders et al.). The use of higher charging voltages is also further limited by the current limitation of the direct current supply that is used for charging the capacitor (the latter is either integrated in the supply 6 or is connected to 6 as an external supply and is not represented separately in the figure). All these facts have so far prevented cathodic spark discharges at increased spark voltages over longer pulse times and at higher frequencies to be performed for the purpose of material evaporation, especially for the purpose of the evaporation of carbon.

(25) First, a method is to be described that results in an improvement of method A. In this method already described in the document EP:00888335_A1 by Grimm, a direct current spark discharge is ignited via a momentary contact of the mechanical ignition finger (anode) with the target, which is then further maintained via the supply 5. Additionally, current pulses from a capacitor discharge (supply 6) are superimposed over this direct current discharge. The temporal current pattern of these pulses does not differ essentially from those that have also been indicated in Horikoshi et al.: increase of the current pulses at the beginning and then drop caused by the discharge of the capacitor and depending on the respective impedance. With an underlying direct current, the latter is on the order of magnitude of 0.1 Ohm, i.e. the capacitor is quickly discharged (time constant RC).

(26) The temporal progression of the voltage U.sub.KA and of the current I.sub.KP for such an operation is represented in FIG. 4. The current I.sub.KD (not represented) was pre-specified at 30 A with supply 5. Usually, this current collapses during the pulse from supply 6 since the diode then blocks the I.sub.KD current. Pulses with a frequency of 500 Hz were generated and a pulse length of 100 s and a pause between the pulses of 1900 s were specified. During the pulse progression, there is an excess increase in both voltage as well as current, which with an improved time resolution is clearer in FIG. 5 which illustrates a pulse specified at 100 s. Consequently, the voltage at the begin of the pulse changes from 20V, i.e. the voltage set in the direct current case with the electric supply (5), to approx. 50V and it then stabilizes at around 40V. The result of this voltage increase in absolute value is a strong current increase. If one were now to proceed according to the previous operation mode for method A (document EP_00666335_A1 of Grimm) and allow the capacitor to freely discharge, the current would increase further, the capacitor would quickly discharge and the voltage would fall with the current increase (charge drain from the capacitor). The voltage would thus vary widely in time.

(27) In order to stabilize higher discharge currents, the magnitude of the spark current is thus limited, even before the voltage at the capacitor has been reduced too much. This is explained on the basis of FIG. 5. At the supply 6, a pulse length can be prescribed that limits the discharge time and for which in this case 100 s were chosen. For the sake of a better comprehension, the trigger signal for the pulse has also been illustrated in FIG. 5 (lowest curve). For the pulse current, an upper threshold of 720 A was set (in the figure, 1.4V correspond to approx. 1000 A). The current was limited in such a way that the voltage of the spark discharge could be maintained stable in time within certain limits or, in other words: the current was switched off at a current value for which the voltage had not yet collapsed to any substantial degree. This depends on the respective experimental conditions (such as for example also the cable inductivities) and on the technical data of the supply 6 used and on the properties of the direct current spark discharge. This current limitation can however easily be determined experimentally by means of an oscilloscope showing the temporal current and voltage patterns.

(28) FIG. 5 shows in particular that the current (uppermost curve in FIG. 5) has reached this threshold value already before the pulse duration of 100 s had been reached and that it is possible with this method of switching off of the current earlier before the voltage collapse to stabilize the higher voltage U.sub.KA over approx. 40 s, i.e. in the range in which the higher charging states of the cathode material are generated.

(29) It has been shown that the excessive voltage increase depends on the charging voltage of the capacitor (300V were used here) but also on the size of the capacitor that has to be charged (here 8 mF). By using in this experiment a direct voltage supply for charging the capacitor in which the maximum current was limited to 30 A and operation was with a pulse frequency of 500 Hz, the capacity between the pulses cannot be fully charged again. This is only mentioned because it has to be shown that the method also works with such limitations.

(30) FIG. 6 now illustrates an operation with pulsed spark discharge. In this case, the direct current supply 5 used again supplies an I.sub.KD of 30 A. Pulses with the pulse current supply 6 are then superimposed over this direct current discharge. The pulse current I.sub.KP is represented in FIG. 6. The pulse length in this case was specified at 800 s (trigger signal indicated). Between the pulses, a pulse pause of 1200 s was set. For switching off the current, the maximum current was again limited to approx. 700 A. Under these conditions, it is possible to observe the apparition of a cluster of pulses over the length of the specified pulse, which is composed each of 8 subpulses. By way of explanation, a better temporal resolution was chosen in FIG. 7. It shows that, if working again with the current limitation, the generation of the subpulses can determine the discharge voltage in a similar manner as for the conditions in FIG. 5 to values of approx. 40V, again over approx, 40 s, i.e. for the duration of a subpulse. The time constant for the renewed switching on of the subpulses is approx. 100 s. This is caused by the commutation behavior of the protection diodes and can be shorted by faster diodes. In FIG. 7 it is also possible to see that in a predetermined pulse of 800 s and a current limitation at 700 A, 6 subpulses can be generated with a sequence of the subpulses that is greater than that possible from the original supply (here greater by a factor of about 7.5). Simultaneously, in the pulse with a length of 800 s, generating 6 subpulses will allow the voltage to be stabilized over a time of 640 s, i.e. overall for 240 s.

(31) For method A, it is thus possible with this method to both increase the pulse frequency as well as stabilize the spark discharge at a higher voltage. However, even more important is the possibility with this method of specifying to a large extent the ratio of direct current portion to pulse current portion in favor of the pulse current portion. In FIG. 4, an I.sub.KD of 30 A is used and a temporal mean of I.sub.KP of 12 A. By contrast, in FIG. 6 a temporally averaged I.sub.KP of 70 A is used for the same ratio of direct current.

(32) It would now be desirable to generate a cluster of subpulses in method B. The direct current portion could thus be completely avoided. In this method, which is based on discontinuous pulses, one is confronted with the problem of ignition at each pulse but even more with the problem of the ignition of individual subpulses that has to take place at the much higher sequence. This problem cannot be solved with the present state of the art ignition mechanisms.

(33) It has therefore been tested whether the combination of spark current limitation, increased discharge voltage and higher sequence of the subpulses thus generated can generate sufficient conductive plasma and for sufficiently long for an external ignition to be discarded, except merely for a single initial ignition via 5 in connection with the mechanical ignition finger 11.

(34) For this purpose, the following procedure was followed. In a first step, the direct current spark discharge with a current I.sub.KD of 40 A was ignited by the momentary contact between the ignition finger 11 and the target 2. Then, in a similar manner to method A, the pulsed supply was switched on and a pulse current I.sub.KP was superimposed to the direct current. As a further step, the direct current supply 5 was switched off. In FIG. 8, the temporal progression of I.sub.KP and U.sub.KA is now represented for the case of the switched-off direct current supply 5 (according to an operation with merely discontinuous pulses). Pulse lengths of 1000 s and pulse pauses of also 1000 s were specified. Surprisingly, it was possible with such a mode of operation to ensure that no renewed ignition between the subpulses within the cluster of pulses and, even more surprisingly, also between the individual pulses was required. FIG. 9, which illustrates with a better temporal resolution an individual pulse with 7 subpulses, shows that ignition voltages with a value around 60V are sufficient for igniting again the subpulses within the individual pulse, without additional ignition devices being necessary. Surprisingly, these voltages are also sufficient for reigniting the individual pulses after 1000 s. The method thus provides a way with which a discontinuous pulse operation can be achieved without additional ignition device. Furthermore, the sequence of the subpulses within the individual pulse can be increased, in the present case to approx. 7 kHz.

(35) The reason why this quick and easy method for igniting sparks works has not been fully explained. It is presumed that the mixture of argon gas and evaporated carbon in connection with the intensive ionization via the pulsed operation (high electron streams) are sufficient for forming a highly conductive plasma sufficiently long, in which the spark discharge can be ignited again at low voltages. In order to estimate the limit for the only just achievable ignition for the device used in these experiments and the special process parameters, the individual pulse was made shorter and the pulse pause longer. In FIG. 10, a pulse with a duration of 900 s is represented, during the course of which 6 subpulses are generated. A time of 1100 s was chosen as pulse pause. For this case, the ignition operation with the purely discontinuous pulses was just barely achievable. For smaller pulse lengths and longer pulse pauses, the ignition could no longer be safely guaranteed. It is however obvious to the one skilled in the art that modified capacities, other current switch-off levels, different (or no) gas flows, other voltages on the capacitor etc. will influence these thresholds of self ignition and can easily be set for the respective experimental arrangements.

(36) The inventive method was described for the cathodic spark evaporation of carbon because it is particularly difficult with this material to operate the cathodic spark, i.e. to prevent the seizing up on one target position. The method can however be used advantageously also for other materials. These can be elemental metallic targets but also targets consisting of several materials such as for example powder-metallurgically produced targets. The method is particularly advantageous for materials with a low melting point, because in this case too the pulsed operation reduces the formation of splatters. This method can also be used advantageously in reactive processes in gases that contain oxygen, nitrogen or carbon to increase the ionization and reduce the formation of splatters.

(37) By having the anode arranged centrally with respect to the target and by using a pulsed spark current, especially in discontinuous operation, it is possible to ensure a stable spark discharge.

(38) The discontinuous pulsed operation also enables the cathodic evaporation of carbon targets with metallic parts up to 50% at.

(39) The method of distributing the individual pulses over clusters of subpulses of higher sequence and over longer times enables a discontinuous pulsed operation without additional external ignition, between the subpulses lying closer to one another as well as between the individual pulses.

(40) The voltage at the capacitor (supply 6) can be changed to a large extent for the purpose of generating higher charged ions of the material evaporated in the cathodic spark discharge.

(41) The higher discharge voltage in combination with a fast but current-limited discharge reduces the incidence of splatters in the synthesized carbon layers or layers containing carbon.

(42) The method represents a discontinuous pulse operation of a spark discharge, without having to be re ignited.

(43) On the basis of the method, no DC maintenance current is required for the spark discharge.

(44) In the frame of this description, a method has been disclosed for operating a pulsed discontinuous spark discharge that is achieved by means of the discharge of at least one capacitor and where the corresponding discharge current is controlled by periodical switching on and off of the capacitor, wherein when the capacitor is switched off during the switched-off time intervals, no current is supplied from the supply, and within the switched-on time intervals the discharge current is momentarily interrupted upon reaching a previously determined threshold, with the result that subpulses occur within the pulses.

(45) Preferably, a first plurality of the switched-off time intervals is chosen so short that after their respective expiration, the switching on of the capacitor readily causes the renewed ignition of the spark discharge.

(46) The first ignition of the spark discharge is preferably achieved by means of an ignition device, even more preferred by means of a DC direct voltage source, and after ignition of the spark the capacitor is switched on for the first time. The capacitor preferably has at the time of the first ignition by means of charging means a first charge state and the spark discharge is subsequently further operated with the DC direct voltage source switched off.

(47) There can be a second plurality of switched-off time intervals depending on the charging means of the at least one capacitor that are chosen so long that the capacitor after expiration of the second plurality of switched-off time intervals has essentially reached the first charge state.

(48) The spark discharge described above can be used for operating a cathodic spark evaporation.

(49) A method for coating substrates has been described that comprises the following steps: providing substrates placing the substrates in a vacuum chamber pumping out the vacuum chamber performing a cathodic spark evaporation of the kind described above.

(50) By means of this coating method, it is possible to produce carbon layers that are hydrogen-free and/or contain hydrogen and/or metal-free and/or contain metal.

LIST OF REFERENCES

(51) 1 receptacle 2 target (cathode) 3 anode (preferably but not necessarily centered relative to the target) 4 substrate holder (rotating) 5 direct current (spark) supply 6 pulse current (spark) supply 7 switch for switching I.sub.KD 8 switch for switching I.sub.KP 9 switch for switching the anode to earth or to floating potential 10 protection diodes 11 mechanical ignition finger 12 resistance for current limitation between ignition finger and anode