Pulse Power Generating Circuit With Energy Recovery

1. Field of the Invention

The present invention relates generally to pulse power sources and more particularly, to an apparatus and method for providing high voltage pulse power and energy recovery for use in excimer and other pulsed lasers.

2. Description of the Prior Art

In applications where pulsed lasers are operated for extended periods of time, there is a need for energy efficiency, long term reliability and high component lifetimes. In these applications, magnetic compression circuits in conjunction with solid state switches have recently been utilized to supply pulse power because of the reliability and useful life of the magnetic compression circuits and solid state switches.

Such magnetic compression circuits generally utilize a multi-stage LC network to provide the power compression. Various techniques exist for increasing the power output of such devices, including the use of complicated pulse generating networks such as that described in U.S. Pat. No. 5,138,622 to Friede et al. and the use of multi-stage magnetic modulators to achieve a high packing factor. These techniques, however, are complex and expensive to implement and their use provide long term unreliability. Specifically, they are energy inefficient and promote short laser component life-times. For example, after the laser discharges, energy left in the circuits is dissipated in the laser, resulting in reduced electrode life.

In addition, the cost of conventional thyratron-based pulse power lasers is high, while the lifetime of the thyratron is limited by cathode and anode erosion. The most serious effect of cathode erosion is a decrease in thyratron switching efficiency. With age, high operating voltages are required to drive the thyratron, which further stresses the thyratron, until eventual replacement is necessary.

Furthermore, anode erosion in conventional thyratron-based pulse power lasers severely reduces the thyratron's high voltage hold-off ability. This causes the thyratron to exhibit occasional prefiring, which results in a premature and weak laser pulse. In applications where high precision is required, such as the manufacture of silicon wafers, prefiring can result in the complete rejection and disposal of the silicon wafers, which is an expensive undertaking.

Accordingly, there is a need in the technology to provide an apparatus and method of increasing the energy efficiency of magnetic compression circuits which is easy to implement, which does not exhibit prefire behavior and which also minimizes electrode erosion.

A high voltage pulse power generating circuit capable of providing energy recovery is disclosed. The energy efficient pulse power generating circuit comprises a high voltage charge storing element which is periodically discharged into a compressor circuit. An energy recovery circuit coupled to the compressor circuit reverses the polarity of energy recovered from the laser by the pulse power generating circuit and transfers the what would have been waste energy back to the power source. In a preferred embodiment, the pulse power generating circuit of the present invention utilizes an SCR switch for periodically discharging the charge storing element. This embodiment of the present invention provides pulses consistently and will not provide even a single missed pulse.

FIG. 1 is a schematic diagram of a preferred embodiment of the pulse power generating circuit of the present invention.

FIG. 2 is a schematic diagram of an alternate embodiment of the pulse power generating circuit of the present invention.

FIG. 3 is a graphical representation of the relationship between the flux density and field strength in a typical saturable core inductor.

FIGS. 4A-4E are timing diagrams illustrating the energy transfer and recovery cycles in the pulse power generating circuit of the present invention.

FIG. 5 is a perspective view of a laser system utilizing the pulse power generating circuit of the present invention.

The present invention relates to a pulse power generating circuit which is capable of recovering energy that otherwise would be wasted in a system through a transfer of the dissipated energy back into the power source of the system.

FIG. 1 is a schematic diagram of a preferred embodiment of the pulse power generating circuit 10 of the present invention. The pulse power generating circuit 10 comprises a high voltage DC power supply 12, an energy storage capacitor C.sub.0, an inverting circuit 14, a solid state switch S and a pulse compressor circuit 16 for compressing the duration and increasing the voltage of pulses applied. The pulse power source is coupled to a load 18, typically a pulse discharge load such as an excimer laser.

FIG. 2 is a schematic diagram of an alternate embodiment of the pulse power generating circuit 10 illustrated in FIG. 1. In this alternate embodiment, a step up transformer 20 is coupled to the output of the compression circuit 16. Alternatively the transformer 20 may instead be coupled to the power generating circuit at one of several nodes (for example, nodes 1, 2, . . . n) in the pulse compressor circuit 16, depending on design requirements. A transformer 20 coupled to the circuit 10 at a node closer to the power supply 12 will subject downstream components of the pulse compressor circuit 16 to a higher voltage but lower current than the embodiment of FIG. 2, and of course will effect the design of such components because of the different impedance and other operating conditions they will be subjected to. Conversely, a transformer 20 coupled to the circuit 10 at a node closer to the load 18 will allow use of low voltage but high current components supplying the transformer. Further the transformer itself can be somewhat different, depending on its placement, as stepping the voltage up to the desired high level before the final pulse compressor reduces the required peak current capabilities of the transformer.

In a preferred embodiment, the power supply 12 is a 0-1 K volt D.C. power supply, with the pulse power supply providing an energy per pulse of 2 Joules in a pulse width of 100 nanoseconds at approximately a 1 Khz pulse rate. The D.C. supply 12 is used to charge capacitor C.sub.0, shown as a single capacitor, though typically comprising a bank of parallel connected capacitors. For instance, to obtain a 16 microfarad capacitance for capacitor C.sub.0, 160 0.1 microfarad, 1000 volt capacitors may be connected in parallel.

The switch S may be a silicon controlled rectifier (SCR), an insulated-gate bipolar transistor (IGBT), a metal-oxide semiconductor field-effect transistor (MOSFET), a MOSFET controlled thyristor (MCT), or any other suitable switching element as known in the technology. The pulse power generating circuit 10 preferably comprises a diode D.sub.s connected in series with the switch S. The inclusion of D.sub.s in the circuit 10 ensures that the circuit 10 is nonconductive with respect to reverse current, as will be described in detail in the section below. In another preferred embodiment, the pulse power source 10 further comprises an inductor L.sub.0 selected in conjunction with C.sub.1 to provide a time constant for the LC combination to slightly slow the buildup of current flow through the switch S so that the switch may become completely closed prior to high current flow through the switch. This substantially reduces power dissipation in the switch, conserving energy and reducing the heating in the switch and peak current conduction requirement of the switch to reasonably manageable levels. However, it provides a current pulse through the switch which is too low in current and too long in duration to itself provide the intended output for the pulse power source without pulse compressor, provided by the rest of the circuit.

In a preferred embodiment, the switch S is a solid state switch such as an SCR switch. The use of an SCR switch in the pulse power generating circuit 10 of the present invention in a laser offers a distinct advantage. Specifically, it was found that laser systems utilizing the pulse power generating circuit 10 with an SCR switch generate pulses consistently. In fact, the pulse power generating circuit 10 utilizing an SCR switch does not provide even a single missed pulse throughout the life of the laser. This is because SCR switches do not suffer from gradual degradation nor exhibit prefire behavior, as will be discussed in greater detail in the following sections.

The pulse power source 10 also includes bias inductors L.sub.B1 and L.sub.B2, each of which is connected at one end to a current source, I.sub.BIAS - and I.sub.BIAS + respectively, and at the other end, to compressor circuit 16. As shown in FIG. 1, a negative current relative to the desired output pulse is used to bias the compressor circuit 16, as will be discussed in greater detail below. The bias current is supplied through inductors of relatively high inductance to provide relatively high time constants in the bias circuit relative to the pulse rate of the pulse power source to isolate the bias current source from the pulse power source itself. Energy recovery is facilitated in part by inverting circuit 14 which comprises inductor L.sub.ER and energy recovery diode D.sub.ER.

The compressor circuit 16 comprises banks of capacitors C.sub.1, C.sub.2, . . . C.sub.n-1 and C.sub.n and one or more saturable core inductors L.sub.1, L.sub.2, . . . , L.sub.n-1 connected in a ladder network. As shown in FIG. 1, a first saturable core inductor L.sub.1 is connected between C.sub.1 and C.sub.2, a second saturable core inductor L.sub.2 may be connected in series between C.sub.2 and C.sub.3, . . . , and a saturable core inductor L.sub.n-1 is connected in series between C.sub.n-1 and C.sub.n.

The circuit of FIG. 2 is identical to that of FIG. 1 with the exception of the addition of the step up transformer 20 shown between the output of the pulse power source and the load 18, for the preferred embodiment an excimer laser, though the step up transformer may be inserted in circuit at C.sub.1 or any of the capacitors between C.sub.1 and C.sub.n. The presence and location of the transformer does not effect the function of the other components of the pulse power supply, though being an impedance transforming device (transforming voltages and currents), the values, current capacities and voltage capabilities of the components from and including the power supply 12 to the primary of the transformer will differ substantially from the corresponding components of FIG. 1, even though the same will normally be chosen to have substantially the same time constants as the embodiment of FIG. 1.

The method of operation of the present invention will now be discussed. FIG. 3 is a graphical representation of the relationship between the flux density B and magnetizing force per unit length .sup.Ni /.sub.l characteristic of the major hysteresis loop of a saturable core inductor, where N in the number of inductor coil turns on the core, 1 is the length of the core and i is the current in the inductor.

In addition, the total flux .PHI. in the saturable core is given by the equation:

where B is the flux density and A is the cross-sectional area of the core.

As also known in the technology, the relationships between the electromotive force (emf) induced in a coil having N turns, the rate of change of flux in the core about which the coil is wound and the rate of change in current in the coil are: ##EQU1## where L is the inductance of the inductor.

Substituting equations (1) into equation (2), we have ##EQU2##

Thus, the inductance L of a coil is directly proportional to the rate of change of B with respect to i.

Thus, with reference to FIG. 3, the value of ##EQU3## and thus the inductance when the inductor is operating along the line a-b, the line c-d and anywhere along the line c-e of the curve is small since the gradient of the curve is slight. However, the value of ##EQU4## is large along line b-c and line d-a where the gradient of the curve is steep. Thus, starting at point a (set by the negative bias current I.sub.BIAS - see FIGS. 1 and 2), L is relatively small along line a-b. However, once enough current passes through the inductor coil in the positive direction to start to substantially change the magnetization of the core, the inductance L of the inductor suddenly greatly increases, as shown by the change of ##EQU5## at point b.

The inductance remains large until point c is reached, at which time the inductance returns to a low value, allowing the current through the inductor to rapidly rise to point e and beyond, and to later return to point d, all in accordance with the voltage across the inductor. Between point e and point a, the inductance is again relatively high.

The method of the present invention provides a bias current from I.sub.BIAS - and I.sub.BIAS + through L.sub.B1 and L.sub.B2 to the compression circuit 16 to ensure that the saturation inductors L.sub.1, L.sub.2, . . . L.sub.n-1 are biased to operate along line c-e of the curve as shown in FIG. 3. This ensures that the current passing through saturable inductors L.sub.1, L.sub.2, . . . L.sub.n-1 is conducted in one direction for energy recovery. It also ensures a large time constant for each LC stage of the compression circuit 16 so that sufficient time is provided for full energy transfer from one capacitor (e.g., C.sub.1) to another (e.g., C.sub.2) through the inductor (e.g., L.sub.1). At the end of a transfer cycle, the inductors L.sub.1, L.sub.2, . . . L.sub.n-1 are biased to ensure that their corresponding inductance is reduced to that at point a in FIG. 3, so that proper biasing of the inductors L.sub.1, L.sub.2, . . . L.sub.n-1 may be accomplished during the next pulsing cycle. Specifically, the inductance of the inductors L.sub.1, L.sub.2, . . . L.sub.n-1 must increase with the corresponding increase of positive or uni-directional current in the inductor, as will be discussed in detail below.

The operation of the pulse power generating circuit 10 will be described with reference to FIG. 3 and the timing diagrams illustrated in FIGS. 4A-4E. With the D.C. power supply 12 on prior to the time t.sub.0 (when switch S is open), capacitor C.sub.0 is charged by the current supplied by the D.C. power supply 12. When fully charged, the voltage across capacitor C.sub.0 is V.sub.C0 (see FIG. 4A). At t.sub.0, the switch S closes, and the energy stored in capacitor C.sub.0 is transferred to C.sub.1 (see FIG. 4B) by action of the LC circuit comprising L.sub.0, C.sub.0 and C.sub.1 with a time constant of approximately ##EQU6## C.sub.1 becoming nearly fully charged and the current in inductor L.sub.0 reducing substantially but still remaining positive at time t.sub.1. Inductor L.sub.0 is included in the circuit to purposely provide a sufficiently large time constant in this energy transfer to ensure that S is fully switched before high currents are encountered, and to limit the maximum current through the switch for a given energy transfer to a value manageable by a solid state switch by extending the time of the transfer. As capacitor C.sub.0 starts to discharge through switch S and capacitor C.sub.1 begins to charge during this transfer, the current in the coil of saturable inductor L.sub.1 rapidly changes from the negative bias current of point a to the low positive current of point b of FIG. 3, at which point the inductance jumps to a relative high value to limit the rate of further current increase through L.sub.1.

At time t.sub.1, while the current in inductor L.sub.0 is low but still positive, the saturable core of inductor L.sub.1 saturates, at which time the inductance of inductor L.sub.1 jumps to a low value, allowing the current there through to rise to a very high value to transfer the energy from capacitor C.sub.1 to capacitor C.sub.2 with a time constant approximately equal to ##EQU7## where L.sub.ls is the inductance of the inductor L.sub.1 with the core in saturation (line c-e in FIG. 3). Note that this time constant may be substantially shorter than the time constant ##EQU8## because L.sub.ls may be made much less than L.sub.0.

In the manner described above, energy is transferred through each stage of the compressor circuit 16, if more than one stage is used, compressing the pulse in time in one or more stages as desired, and if of low voltage originally, being stepped up in voltage where desired. When C.sub.n-1 is substantially fully charged and the core of inductor L.sub.n-1 saturates, the energy is coupled through the saturated inductor from C.sub.n-1 to C.sub.n in a short time period pulse, with the voltage across C.sub.n being applied to the laser 18.

At time t.sub.L, the voltage across C.sub.n builds to the level required to initiate discharge in the laser, and the laser 18 fires. Energy stored in C.sub.n and still being fed to C.sub.n through the inductor is delivered to the laser discharge in the compressed time period of the pulse power supply (see FIG. 4E). After discharging, C.sub.n is driven negative. When this occurs, L.sub.n-1 is already saturated for current flow from C.sub.n-1 to C.sub.n. Thus, instead of having energy ringing in the laser 18, eroding the electrodes, the saturated state of L.sub.n-1 causes the reverse charge on C.sub.n to be transferred resonantly back into C.sub.n-1 as shown in FIG. 4D. This transfer is accomplished by the continued forward flow of current through L.sub.n-1.

When energy is transferred back into C.sub.n-1, C.sub.n-1 is driven negative with respect to C.sub.n-2. This process repeats back up the L-C ladder network of the pulse compressor circuit until the energy which would have been destructively dissipated in the load 18, is transferred from C.sub.n-1 to C.sub.n-2. This transfer process continues until the energy in C.sub.n is passed to C.sub.0, as shown in FIG. 4A. It should be noted that such reverse energy transfer is possible only if all saturable inductors L.sub.n, . . . L.sub.n-1 remain forward conducting to operate along curve c-e of FIG. 3, since the voltage of C.sub.n with respect to C.sub.n-1 (and similarly, C.sub.n-1 with respect to C.sub.n-2, . . . etc.) must be negative for the energy propagation to be accomplished. Thus, an important factor in the design of the pulse power generating circuit 10 is that all the coils of all saturable inductors L.sub.1, L.sub.2, . . . L.sub.n-1 must be forwardly conducting to operate in the proper state for one way current transfer until waste energy is fully recovered.

When the waste energy is propagated back into C.sub.0, C.sub.0 is driven negative with respect to its initially stored charge, and switch S is opened. The inverting circuit 14, comprising an inductor L.sub.ER and a solid state diode D.sub.ER coupled to ground reverses this polarity through resonant free wheeling--a half cycle of ringing of the LER-C.sub.0 circuit as clamped against reversal of the current in inductor L.sub.ER by diode D.sub.ER, with the net result that the energy is recovered by the partial recharging of capacitor C.sub.0, thereby the energy which otherwise would have contributed to erosion of the laser electrodes and reducing the supply requirements of the D.C. power supply.

The pulse power generating circuit of the present invention may be utilized in lasers which operate on the principle of transverse discharge. These lasers include systems which discharge at high pressure, such as CO.sub.2 lasers, copper vapor lasers and excimer lasers including KrF excimer lasers. FIG. 5 illustrates a laser module 20 which may utilize the pulse power generating circuit 10 (enclosed in chassis 22) of the present invention as a power source. A typical example of such a laser includes the HPL-100 laser module manufactured and sold by Cymer Laser Technologies, assignee of the present invention.

Several advantages exist through the application of the pulse power generating circuit of the present invention. First, by using the pulse power generating circuit, the energy efficiency of a laser system is increased while electrode erosion from after discharge ringing is minimized, since the waste energy which would otherwise be left ringing in the laser 18 is effectively removed from the laser 18. The energy reflected by the laser chamber due to impedance mismatch no longer rings back and forth between the pulse power generating circuit and the laser chamber since this energy is recovered by the pulse power generating circuit of the present invention. As described, the pulse power generating circuit is designed to transmit this reflected energy all the way back through the pulse forming network into C.sub.0. Upon recovery of this energy onto C.sub.0, the switch turns off, ensuring that this captured energy remains in C.sub.0. Thus, regardless of the operating voltage, gas mixture or chamber condition, the voltage waveform across the laser electrodes exhibits the behavior of a well-trained system. This performance is maintained over all laser operating conditions.

As discussed, after-ringing has been found to have a detrimental impact on the laser chamber's electrode lifetime. After the main laser discharge, current flow between the chamber electrodes becomes nonuniform and breaks up into localized areas of high current density called streamers. Elimination of this after-ringing by the solid state pulse power generating circuit starves the streamers of energy and reduces the amount of electrode damage caused by these streamers.

Secondly, by incorporating an SCR switch in a preferred embodiment of the pulse power generating circuit of the present invention, prefiring in the laser system can be eliminated while predictability and reliability of the laser system can be increased. This is because SCR switches, unlike conventional thyratron switches, neither suffer from gradual degradation nor exhibit prefire behavior. Specifically, the use of SCR switches in pulse power generating circuits has been demonstrated to last for greater than 30 billion pulses under typical operating conditions without a single missed pulse or failure of the pulse power generating circuit. In addition, experimental results in a prototype of the pulse power generating circuit of the present invention has thus far accumulated in excess of 6 billion pulses without a single failure or missed pulse. It has also been determined experimentally that the pulse power generating circuit of the present invention exhibits the same efficiency at 6 billion pulses as it did at 300 million pulses.

Thirdly, the use of the solid state pulse power generating circuit of the present invention in a laser has increased the useful chamber life of the laser by 100%, from 3 billion pulses to 6 billion pulses, without any degradation in performance.

Finally, three other factors are significant in the reduction of cost of operation in using a solid state pulse power generating circuit instead of a thyratron-based pulse power system. The most significant reduction in cost is a ten-fold increase in the expected lifetime of a laser system using the pulse power generating circuit of the present invention. The increase in expected lifetime is from 3 billion pulses for a thyratron pulse power laser system to 30 billion pulses for a solid state pulse power laser system. The second factor in overall system cost reduction is the complete elimination of missed laser pulses and thus the reduction in wafer level rework. The third factor involves a reduction in downtime for laser service procedure. A thyratron-based pulse power circuit used on a typical laser system must go through a 10 minute warm-up period to achieve proper thyratron temperature. The solid state pulse power generating circuit of the present invention requires no warm-up time, and thus each laser service procedure can be shortened by approximately 10 minutes.

Although the invention has been described in terms of a certain preferred embodiment, other embodiments apparent to those skilled in the art are also within the scope of this invention. Accordingly, the scope of the invention is intended to be defined only by the claims which follow.