HIGH-POWER INDUSTRIAL CO2 LASERS EXCITED BY A NONSELF-SUSTAINED GLOW DISCHARGE

Nonself-sustained glow discharge assisted by periodic-pulsed discharge (PSD - pulser sustained discharge) is used to produce an active medium in fast-transverse-flow electric-discharge industrial CO₂lasers (power range 1-40 kW CW or time average). This method was proposed by Reilly [1] and Hill [2], and considerable practical results were reported by Shashkov et al. [3], Seguin et al. [4-6], and Generalov et al. [7-9]. 

Two different schemes are used to create pulse assisted DC glow discharge (Figure 1).

Characteristic feature of the discharge schemes used in [1-6] is that both assisting pulse avalanching voltage and steady (or also pulsed) main discharge voltage are applied to the same electrodes, as shown at the top picture of Figure 1. Spatial uniformity of the discharge is provided by multielement electrodes and additional photoionization.

In other scheme developed by Generalov et al. [7-9] (bottom picture on Figure 1) an avalanching periodic-pulsed voltage is applied to additional pair of large area plane electrodes isolated from plasma by dielectric sheets. This type of assisting discharges is called electrodeless (capacitively coupled) periodic-pulsed discharge (EPD). EPD is characterized by high peak power required to produce uniform ionization in a discharge gap between dielectric sheets and relatively low time-average power. Vibrational excitation of the uniformly ionized medium is produced by a steady-state nonself-sustained discharge, called electrodeless pulse sustained discharge (EPSD) or main discharge. DC voltage of EPSD is applied to tubular metal electrodes (cathode and anode) placed at the edges of the discharge chamber. Fast gas flow is directed from cathode to anode, perpendicular to the axis of an optical resonator. The rest two walls of the chamber have holes to let laser radiation pass to cavity mirrors placed outside. Optical axes of the resonator is folded in Z-shape so that laser beam completely fills an active volume.

This scheme was studied in the Institute for Problems in Mechanics, Moscow. On the base of the study a series of industrial CO₂ lasers "Lantan" of output power from 1 to 5 kW was designed [10]. An experimental model of a 10 kW CW CO₂ laser "Tsyklon" [7-8] was also developed and created. These lasers demonstrate high efficiency, enhanced beam quality [11], wide possibilities for power control, low gas consumption, and high reliability.

High performance characteristics of these lasers are achieved because of the advantages of excitation scheme used: discharge optical homogeneity, simple electrode system, low discharge chemical activity, and original power control scheme.

This paper is dedicated to the detailed description of discharge physics and applications of EPSD method in high power industrial CO₂ lasers. Achieved results and prospects for further applications of this method are considered. 1. Electrodeless capacitive periodic-pulsed discharge

Electrodeless capacitively coupled periodic-pulsed discharge (EPD) is classified among relatively well studied volume pulsed discharges applied particularly to pulsed gas lasers. The characteristic feature of EPD is that electrodes to which potential pulse is applied are isolated from discharge gap by dielectric sheets (Figure 2). This unusual kind of pulse discharges has been investigated theoretically and experimentally in connection with CO₂ laser applications [7-9,18,19].

In order to better appreciate the processes in the discharge, let us consider a simple model.

Voltage pulse applied is assumed to be specified in a step form with peak voltage value U₀ and relatively short front width. Initial free electron density N_e0 is taken to be uniform over the discharge volume. In periodic-pulsed discharge considerable electron density remains after preceding pulse. Potential distribution in discharge gap is taken to be uniform, sheaths of spatial charge near dielectric sheets are assumed to be so thin that potential drop is negligible. In these presumptions equivalent circuit for EPD may be defined as shown on Figure 2. R_g plasma column electric resistance depending on free electron density N_e, C_g is discharge gap electric capacity, and C_d is dielectric sheet capacity.

This model is very similar to that analyzed in [7] and called electrotechnical model. Within this model evolution of discharge electric field, current and electron densities are described by a system of non-linear ordinary differential equations, which may be numerically integrated to obtain plasma electric field, current and electron density waveforms as shown on Figure 3.

It is useful to understand general relations and make evaluations on the base of simple physical considerations.

Voltage value U₀ is assumed to be high enough to start avalanche ionization in the discharge gap. One can define characteristic ionization time T_i as

TABLE 1. Character ionization time at different E/p in CO₂/N₂/He mixture with ratio 1/6/12. 

If front width of the applied voltage pulse is short with respect to characteristic ionisation time correspondent to voltage peak value U_0P, electron density begins to rise with characteristic time T_i(U_0P).

When electron density achieve considerable value, electric current begins to flow in plasma column. Positive and negative charges are then separated at the surfaces of dielectric sheets and screen electric field in plasma column.

Ionization process in plasma is practically terminated as potential drop on plasma column falls below certain value due to plasma polarization process described in previous paragraph. Characteristic time of polarization within electrotechnical model is R_gC_d/2 where R_g is electric resistance of plasma column, and C_d is electric capacity of dielectric sheet. Condition of the end of ionization then may be defined as

Free electron density N_e may be roughly estimated as proportional to plasma conductivity, then

This relation is the main result of our simplified consideration. One can see from (1.3) and Table 1 that N_e strongly depends on U_0P.

In practice the condition that voltage front should be short in relation to ionization time generally does not hold, because peak voltage U_0P is usually higher then required (Figure 3). In this case electric field in plasma starts to decrease before applied voltage achieve maximum value. Overvoltage obtained depends on the voltage rate of rise. Hence T_i, current pulse width, and current amplitude in practice are determined by the voltage rate of rise. To obtain more effective ionization one should apply more sharp voltage pulse. Output power of impulse generator must be sufficiently high to ensure required current amplitude. Restrictions associated with output characteristics of impulse generator lead to additional decrease of obtained free electron density.

Pulse repetition rates required to create quasi-stationary electron density depends on free electrons lifetime. In conditions of CO₂ CW lasers free electrons die in processes of electron-ion recombination and attachment (negative ion formation). Correspondent lifetimes are in the limits of 10-100 чsec and repetition rates are 10-100 kHz.

On the Figure 4 one can see bright and dark layers near the dielectric sheets. These layers are similar to those of common glow discharge and are specific to cathode. Top and bottom layers correspond respectively to positive and negative current half waves obtained from trapezoid shape of applied voltage pulse.

The difference between these layers and those of glow discharge is that in the case of pulsed discharge layers are not stationary, but only quasi-stationary if the discharge is periodic-pulsed.

The nature of the layers were investigated in [19]. It was shown that layers formation during not extremely short pulses also is under strong influence of secondary emission processes on the cathode: formation of cathode potential drop sheath, secondary avalanches, and high energy electron beam with increased penetration ability. Dark space with increased electron density near electrode is the result of these processes: observed thickness of about 10-15 cm Torr is close to that calculated in [19].

If steady-state electric field is applied to the plasma, such as in nonself-sustained discharge, layers thickness will decrease. Free electron lifetime between pulses is determined by recombination processes which in turn depend on electron kinetic energy. If there are no external electric field applied, electron energy is low, but recombination coefficient is high. For this reason initial electron density remaining to the beginning of the next pulse (repetition rate 10 kHz) does not exceed 10⁹ cm^-3, as was accepted in [19].

From the other hand, in the presence of steady-state electric field recombination coefficient decreases, and initial electron density becomes higher (up to 10¹⁰ cm^-3). As result, near electrode layers thickness decreases, potential drop becomes smaller, closer to that of stationary glow discharge cathode sheath. Thus in certain conditions we are able to considered near electrode potential drop as negligible, such as in electrotechnical model considered.

When discharge gap value is comparable to near electrode layers thickness, near electrode effects lead to nonuniform distributions of electric field and electron density in the discharge gap [19]. In this case electrotechnical model fails.

Volume homogeneity of the discharge in the direction perpendicular to pulsed current may be strongly affected by processes similar to arcing of a conventional glow discharge. These processes are suppressed with reduction of discharge gap and gas pressure, decrease in dielectric sheets electric capacity or current pulse width. Fast gas flow in the discharge region also suppresses these instabilities.

Distributed ballast capacity of the dielectric sheets primarily prevents contraction of cathode sheath at the late stage of the breakdown.

High homogeneity of EPD in large volume and considerable average free electron density attainable at relatively low average discharge power determine its successful application as ionization source for nonself-sustained discharge. Typical parameters are presented in Table 2.

TABLE 2. Main parameters of EPSD used in "Lantan-5" industrial CO₂ laser. 

Application of EPD to produce uniformly conductive plasma in large volume permits successfully decide two problems encountered in the design of high power CO₂ lasers.

First, EPD technique is able to control DC discharge parameters and suppress instabilities which are the obstacle to increase discharge power. Second, EPD ionization ensure high optical homogeneity of the discharge in large volume which is difficult to attain in self-sustained DC discharges.

Let us consider physical properties of EPSD applied to high power CO₂ laser "Lantan". Main parameters of the discharge are presented in Table 2.

Mutual arrangement of the electrodes in a discharge chamber of "Lantan" type lasers is shown on Figure 5 together with pulse generation circuit. EPD is excited between two plane electrodes spaced 5.5 cm apart, consisted of water cooled metal plates covered by dielectric coatings. Plasma is formed in a plane channel of rectangular cross section 5.5c90 cm² with fast gas flow assigned by the arrow. Two copper tubular main discharge electrodes (cathode and anode) are placed correspondently at the inlet and outlet cross sections of the channel. Electrodes are mounted parallel to each other spaced 26 cm apart.

Constant main discharge voltage V is applied directly to tubular electrodes, without any ballast resistors. Voltage value is chosen optimal for vibrational excitation, and spatial average ionization rate of main discharge electric field is much lower than time average ionization rate of EPD. Thus EPSD is entirely nonself-sustained and have character properties: monotonously increasing voltage-current characteristics (see Figure 6), current termination when EPD is switched off, and so on.

Been created during the discharge pulse, free electrons die in the processes mainly of dissociative electron-ion recombination and three-body attachment, or simply gone with the gas flow. Following electron density and gas conductivity, main discharge current also flows in periodic pulsed manner (Figure 7).

Time average current thus depends on peak electron density (or ionization pulse amplitude) and pulse repetition rate. Discharge power may be controlled either by regulating high-voltage supply of EPD, as shown on Figure 6, or by varying pulse repetition rate.

Output power is thus produced modulated with EPD frequency, but kinetics of laser levels under pressures considered leads to smoothing off light modulations related to current modulations at appropriate pulse repetition rates, as shown on Figure 7.

Mutual arrangement of electrodes is determined by the conditions of discharge homogeneity and convenience for laser beam extraction. EPSD appears to be transverse with respect to optical axes and longitudinal with respect to gas flow.

On the surface of tubular electrodes cathode and anode potential drop sheaths are formed. The sheaths are equivalent to those of conventional glow discharge. The phenomenon of normal current density also takes place.

Figure 8 shows potential distribution in the discharge chamber measured by electric probe. Electric field is distributed very uniformly over all volume, excluding near electrode zones.

Near (in the region of about 1-2 cm radius) tubular electrodes current density and electric field are increased simply because of electrode geometry. Electrode surface area is much smaller than channel cross section area. This geometry is suitable to fix positions of current spots on the electrodes without multielement electrode structure. The increase of the electric strength near electrodes is followed by formation zones, similar to that of dissociative attachment controlled self-sustained discharge with character electric field strength, electron density, and other properties. Stability of these zones determine stability of the discharge as a whole, because arching in this zone leads to discharge failure.

Stability of the discharge is an ultimate problem for successful laser operation. Instabilities occur in near-electrode zones, where both electric field and current density are high. From the other hand, length of this zones along the flow is small, so gas resident time in this zones is also small. Thus fast gas flow improve stability to a large extent. Furthermore, the most part of the discharge volume, where electric field strength is lower, may be treated as a distributed ballast resistor, which stabilises the discharge in near electrode zones in wide range of deposited power.

Critical power of EPSD arcing in considered range of parameters is increased almost linearly with plasma density obtained in pulsed discharge. It provides the possibility of further increase of discharge specific power up to that of obtained in steady-state E-beam controlled discharges.

The inhomogeneities of the main discharge are bounded in the vicinities of metal tube electrodes. The most part of volume is uniformly excited by non-avalanching electric field and non-self sustained current. Near-electrode zones are relatively small, so the efficiency of active volume excitation and optical homogeneity of the active medium are high. The active medium thus have almost perfect parallelepiped shape, which is favourable to avoid phase distortions of extracted beam. Thus it is possible to produce perfect laser beams with advanced optical resonators [11].

Primarily due to non-avalanching value of the main discharge electric field strength, chemical activity of the discharge is relatively low. Plasmachemical processes are initiated in small volumes near the electrodes, and in short periods of EPD pulses.

Chemical activity of EPSD in typical CO₂ laser conditions is characterized by CO₂ dissociation coefficient less then 20% and nitric oxides relative concentration less then 0.01%. So the discharge can keep high power characteristics for a long time without working gas mixture renewing. In practice small gas consumption (less than 100 standard litres per hour) is required to remove the impurities appearing as the result of dielectric materials outgasing, mainly hydrogen and water vapour.

An important property of EPSD is perfect power control possibilities. For example, main discharge power supply is designed unregulated. Laser power is controlled through ionization degree by means of regulated EPD power supply, much less powerful then that of the main discharge. Normal laser pulses are also realized by EPD amplitude or frequency modulations.

Super-enhanced periodic-pulse laser operation was demonstrated in "Lantan-1" prototype industrial laser model designed in 1980 [9]. Main discharge in this model was designed also self-sustained, switched by thyratron. To prevent main discharge arcing, EPD was used as a very uniform preionization. Several tens kW peak power and 1.5 kW average power at pulse repetition rates up to 200 Hz were achieved.

We can summarize main advantages of the discharge under consideration as follows: high optical homogeneity, high efficiency for vibrational excitation, simple electrode system, low chemical activity, perfect power control.

The following chapters describe design and characteristics features of 5 kW fast-transverse-flow industrial laser "Lantan-5" developed on the base of EPSD.

3. High power industrial CO₂ laser "Lantan-5": design and operational characteristics

Industrial CO₂ lasers "Lantan-5" of 5 kW CW output power based on EPSD technique was designed for metal cutting and welding applications.

"Lantan-5" is fast-transverse-flow (FTF) industrial CO₂ laser. It is more powerful (up to 5 kW) modification of the previous model "Lantan-3M" described in [10]. These two models are based on the same principles and differ from each other with discharge, pulser, and blower power. General specifications of Lantan-5 model are represented in the Table 3, overall view on Figure 9.

Original discharge technique permits "Lantan" lasers to demonstrate high beam quality, fast power control capability, high efficiency, and low gas consumption in reliable and simple transverse flow discharge geometry.

Output power may be modulated in periodic rectangular pulses, or in trapezoid single pulse. All power regimes are automatically controlled, so that this model may be used in flexible processing complexes.

TABLE 3. General specifications of "Lantan-5" industrial CO₂ laser. 

A discharge technique used for excitation is described in preceding chapter. Nonself-sustained discharge is assisted by electrodeless capacitive periodic-pulsed discharge. Main discharge power supply is designed operating at constant nonavalanching voltage, optimal for vibrational excitation. Time average discharge power is controlled by varying of EPD pulse repetition rate. Time average output power in CW or pulse operation is stabilized by means of fast backfeed circuit.

Because of FTF scheme, two parallel axial fans driven by built in water cooled electro-shafts operating at 9,000 rpm are enough for effective gas recirculation. Low operating speed and relative pressure drop less then 1.1 at absolute pressure 6 kPa make the fans very reliable. They ensure uniform gas flow of 100 m/s in the discharge area.

Small signal gain of the active medium produced is easily varied from very small value to 0.5 m^-1 at maximum power.

"Lantan-5" model was primarily designed with a folded stable resonator, schematically shown on Figure 10. Resonator optical length is 7.5 m. Optical axis is folded to obtain 5 passes through active medium, each of 0.9 m. Thus resonator optical scheme includes 4 plane folding mirrors, concave rear mirror and plane ZnSe output coupler with 30% reflectivity.

With a rear mirror of 15 m radius of curvature (semiconfocal configuration) the resonator is operating in multimode regime with output beam divergence of about 4.5-5 mrad. Figure 11 represents high efficiency of beam extraction obtained from "Lantan" with semiconfocal stable resonator. To improve beam quality, radius of curvature was increased up to 30 m. The mixed mode order was reduced to approximately 3 so that the divergence was decreased to 2.5-3 mrad. Extraction efficiency was also reduced to approximately 17% at nominal power. Further increase of the rear mirror radius makes the resonator too sensitive to the thermal distortions of the optical elements and resonator structure and substantial laser efficiency drop.

Multimode operation advantages are high beam extraction efficiency (up to 20%) and relatively uniform near- and far-field intensity distributions. Beam divergence in this case is 5-10 times higher then diffraction limited one. This beam quality meets the requirements of some welding and heat treatment applications, but is not enough for cutting and welding of non-ferrous metals.

To obtain TEM00 operation with the same resonator structure one can decrease intercavity aperture diaphragm to approximately 20 mm. Restricted mode volume leads nominal power to decrease to less then 2 kW in this case. The quality of this beam usually does not exceed M2 = 1.3 because of distortions generated by diaphragm or the next order mode.

High homogeneity and optical quality of the active medium obtained with EPSD technique enables to achieve high beams quality together with high efficiency just by designing of appropriate resonator scheme. The next chapter describes the experiments aimed at extraction of 5 kW diffraction limited output beam from "Lantan-5" laser with a help of unstable resonator with GRM output coupler [11].

Variable reflectivity was obtained by deposition of reflected layer with variable thickness on the anti-reflection coated ZnSe substrate. Total absorption of the mirror at nominal power did not exceed 0.2%.

The result of geometrical-optics calculations of the near-field output beam intensity profile in the case of empty resonator is presented on Figure 12 by solid line together with experimental dots (rectangles) obtained at the output power 4 kW. Equivalent Gaussian beam near field intensity profile is shown by dashed line. Ideal Gaussian beam and considered real beam are treated equivalent if they have equal cross section areas carrying 86% of total power. Ideal beam is presented as the reference point of beam quality.

One can see that near-field intensity profile obtained from GRM resonator appears to be quite smooth and uniform. The difference between theoretical and experimental profiles lays in the limits of used theoretical approximation. Beam diameter was slightly spread when output power was increased which may be connected to variations of the active medium characteristics.

Solid line on Figure 13 represents the result of theoretical calculation of far-field angular intensity distribution from given near-field intensity profile with regard for phase distortion in semi-transparent layer of variable thickness on the output coupler. Rectangles show far field angular intensity distribution obtained in the focal point of 14 m lens. These two profiles are normalized to compare the ratio of intensities at the central lobe and at the first ring (or wings). If they were normalized to equal power, the intensity of the experimental one would be lower, because experimental curve demonstrate wider central lobe then the theoretical one. Nevertheless, far-field measurements with calibrated diaphragms revealed that about 80% of total power is radiated within 0.3 mrad (half angle). It corresponds to that of theoretically predicted profile.

Taking into account the experimental error, intensity in the central lobe observed for GRM resonator may be estimated as 1.7-1.9 times lower then that of equivalent Gaussian beam (dashed line on Figures 12,13).

Figure 14 shows the experimental dependence of the output power against discharge power. The efficiency 13% is achieved at nominal power. Differential efficiency is about 17%, threshold discharge power is 7.5 kW. The last value reveals (in comparison with results represented on Figure 11) that output coupling in this case is about 0.8 and backfeed ratio 0.2. The last value is close to that obtained from geometrical-optics approximation.

Mirror tilting sensitivity of the resonator was found to be moderate, intensity profile remains smooth without hot-spot formation. Heat wedge effects in FTF active medium were negligible in these experiments.

As one can see, extraction efficiency obtained with GRM unstable resonator is relatively low, particularly with respect to that for semiconfocal stable multimode resonator, obtained in similar conditions (Figure 11). It could be explained by low order of super-Gaussian mode excited in resonator. The difference in extraction efficiency between high and low order super-Gaussian modes was experimentally observed by Serri et al. [24].

In spite of high extraction efficiency, high order super-Gaussian output couplers seems to obtain worse beam quality then low order ones. If one compare present results for n = 4 with results for n = 8 from [24] for far field intensity pattern, one can see that the last has more intensive wings that the other.

Experiments with GRMs reported in literature (for example see [24,26]), as well as the present one, show that measured far field intensity pattern exhibits more intensive wings then obtained from geometrical optics calculations regardless of phase distortion effects. The discrepancy is obviously caused by phase distortion of the near-field wavefront, and the question is what is the source of these phase distortions.

In our case geometrical optics approach was used to obtain intercavity intensity distribution with plane wavefront at the output coupler. Output beam intensity profile was obtained from intercavity beam intensity profile and output coupler reflectivity profile. Output beam phase profile was determined as phase shift in a variable thickness layer responsible for variable reflectivity. As one can see from Figure 13, far field intensity profile calculated on the base of this approach is close to experimentally observed: both exhibit equal ratio between intensities of central lobe and of the wings. This demonstrate the feasibility of used theoretical approach.

As follows, present experiments allow to make conclusion that distortion in variable thickness layer is the main limitation for beam quality improvement with a help of GRMs.

The experiment demonstrates the possibility of near diffraction limit operation of 5-kW transverse-flow industrial laser with super-Gaussian output coupler. The results obtained are close to that obtained by Takenaka et al. [17] with phase-unifying output coupler and active medium excited by medium frequency electrodeless AC discharge, and are in good agreement with theoretical prediction on the base of simple geometrical-optics calculations with regard to phase distortion in variable thickness layer. Good agreement between theory and experiment demonstrate high optical quality of the EPSD based active medium used in the experiments.

5. Conclusion

Experimental results and developments presented in this review demonstrate the feasibility of EPSD technique in high power and high beam quality industrial lasers.

In conclusion, let us compare some different excitation methods used in powerful fast-transverse-flow CO₂ lasers, to which EPSD technique may be considered as alternative.

1. Electron beam sustained discharge [12]. In spite of evident advantages of this method, it has not received wide acceptance for industrial use because of the problems related to E-beam injection in gas and X-rays shielding.

2. Electrodeless (capacitively coupled) AC discharge (10 kHz) has been applied in ML-105,-108 5 and 9 kW lasers produced by MLI Lasers Ltd., Israel [13]. Now these models are not produced because of the problems with discharge arcing. Instead of these models ML-5000 5 kW laser with RF-excitation is produced [14].

3. RF-discharge (1-100 MHz) characteristics are close to that of E-beam sustained discharge [15]. Now there are experimental installations (for example, [15,16]) as well as some small-lot production (such as ML-5000) of multikilowatt CO₂ lasers exited by RF-discharges in fast-transverse-flow.

Wide application of this method in transverse flow lasers appears to be limited by technical and economical problems related to powerful RF- technique. From this standpoint more acceptable is capacitively-coupled medium-frequency AC discharge (100-200 kHz) [17] to be used in up to 10 kW lasers. This method is treated to be compromise between AC and RF capacitively coupled discharges.

4. In the case of EPSD most of problems pointed above do not take place: main power is deposited by safe, reliable, and not expensive DC discharge, time average load on dielectric plates is low, biological shielding is not required. Unfortunately in practice specific deposited power achieved in EPSD is lower than that of E-beam sustained and RF-discharges. Nevertheless, EPSD parameters may be considerably enhanced by application of pulse generators of higher peak power. Very significant is also an optimal choice of dielectric plates characteristics (see relationship (1.3)).

It is not excluded that EPSD may be successfully applied to planar waveguide (or slab) diffusively cooled high power lasers instead of RF-discharge. Geometry of the discharge may be similar to that of presented on Figure 5, where dielectric sheets should be moved together to form a narrow gap required for diffusive cooling.

There are some problems encountered on this way. One of these problems is substantially increased electron and current densities required for effective laser excitation in narrow gap. In typical conditions of 1 kW planar diffusively cooled laser [27] specific input power is up to 60 W/cm² - ten times higher then in the case of fast-flow laser (compare to Table 2). To achieve that power density in DC discharge, current density of up to 250 mA/cm² is required. From comparison with data from Table 2 one can conclude that 600 mA/cm² pulse discharge current can produce required ionization degree. Total pulse current may achieve up to 500 A through the area 9.5c77 cm², typical for 1 kW laser [27].

Nevertheless, pulsed discharge ionization in narrow gap may be more effective because of near cathode effects responsible for Faraday dark space formation, and so on. Layer thickness of about 10-15 cm Torr, typical for EPD (see Chapter 1), almost equals to gap width in slab geometry (0.2c70= 14 cm Torr [27]). The effect that prevents application of Y-form of RF-discharge in planar waveguide lasers may be feasible for EPSD. In any case, the possibility of this interesting application require additional studies.

6. References