Plans for a Sealed CO2 Laser

Built By David Knapp

David Knapp is with the Energy and Minerals Applied Research Center (EMARC) University of Colorado, Boulder - Department of Geological Sciences.

Since these sites have temporarily disappeared or had broken links in the past, I (Sam) have provided the complete paper below and also in PDF Format. (Note: Some of the embedded equations may be a bit buggy or totally broken in the HTML version. I have also made some corrections in wording and to the first equation.)


The Construction of a sealed, CO2 gas discharge laser was undertaken as an independent study project. Glass laser tube design, as well as clear acrylic housing, make it an excellent demonstrational tool. Sealed operation was characterized in mode, power, warm-up and stability of the period of weeks. Novel design approaches were used for expediency, and cost savings and an anomalous turn-on behavior is also discussed.

1. Introduction

The construction of a sealed CO2 laser was undertaken for several reasons. The CO2 laser provides an impressive and graphic demonstration of quantum mechanical processes. The laser is constructed out of glass and is contained in an acrylic case, so that all of its components can be observed. The laser power supply simply plugs into a standard A/C outlet and more than 20 Watts of coherent, monochromatic laser radiation is produced, enough to burn paper and wood. Another reason why this was undertaken was because all of the supplies, equipment and lab space were readily available; it only took time and inclination to exploit these resources to arrive at a useful device. It was also an achievable project as evidenced by the author's demonstration of a smaller, less complicated flowing gas CO2 laser the previous semester in a junior physics lab course. Most importantly, the author simply wanted to have a relatively compact, portable laser, which did not require bulky vacuum equipment or gas cylinders. In this manner, a CO2 laser would be available for future work.

2. The CO2 laser

2.1 The Resonator

The resonator used is confocal-confocal. Since the optics used in the this project were donated from a company, the author did not have a choice of mirror curvature. Since it was higher power and not best quality mode that was desirable, particular attention was not paid to resonator optimization. With the High Reflectance mirror radius of curvature of 2.685 m, and Output Coupler radius of 5.072 m, the resonator is in the stable resonator regime of 0.66, as determined by the stability relation:

where L is the cavity length, here 65 cm, and R is the radius of curvature of the mirror.

2.2 Energy Transfer in the Discharge

The most commonly observed laser transitions in the CO2 molecule, barring the use of any frequency tuning mechanisms, are from the CO2 asymmetric stretch transitions, from the 00*1 to the 10*0 10.6 micron and 02*0 9.6 micron states, using the notation v1v2*v3, where v1 refers to the symmetric stretch quantum number, v2 refers to the asymmetric stretch quantum number and v3 refers to the asymmetric stretch quantum number. There are literally dozens of other lasing transitions [6] which can be easily chosen by employing an intracativity grating. In a CO2 laser, lasing of one vibrational transition precludes the efficient lasing of another, so that lasing lines 'hop' from one to another depending on instantaneous gain medium and resonator conditions.

Any single possible laser line can be forced through the use of an intracavity grating. Rotational structure, having energies clustered very close to one another, may exist at any time. Nonradiative decay to short-lived lower lying states followed by nonradiative decay to the ground state follows. N2 is added to the laser gas to more efficiently transfer energy from electron impact to the CO2 upper vibrational laser level. The glow discharge is a very effective mechanism for vibrational excitation of nitrogen. Since N2 is a homonuclear molecule, dipole radiative de-excitation is forbidden. This allows for long-lived vibrational states which makes excited N2 molecules more readily available for collisional excitation of CO2. De-excitation is only accomplished collisionally with the wall or other gas constituents, the most beneficial of which is the CO2 molecule. The N2 v=2 state is only 18 cm^-1 (2.2 E -3 eV) from the upper laser level of the CO2 molecule. This makes resonant energy transfer between N2 and CO2 more likely. This energy is much smaller than the average kinetic energy of the molecules in the surrounding glow, so vibrational energy can easily be supplied to the CO2 molecules. Energy transfer occurs from vibrational levels up to v=4 in N2, because the ensuing anharmonicity of these states, due to bond stretching, is still well below the average molecular kinetic energy [7]. CO is isoelectronic with N2 and also has vibrational levels easily excited in the glow discharge. Figure 1 Details the more common energy transfer routes in the CO2 laser. Excited N2 and CO transfer vibrational energy through collision to CO2, exciting any of a number of stretch and vibrational modes.

Fig 1 - Energy level diagram of CO2 laser

Thermal poisoning can occur, which is a build up of lower lasing level populations in CO2. This results in a reduction in laser output power due to a clogging of the path from the upper lasing level to the ground state, where the CO2 upper lasing level is most efficiently populated through collisions with N2. These lower levels are cooled by the addition of He to the gas mix. Helium energy levels are much higher than the molecular energies of N2 and CO2, above 20 eV. For typical electron energies in the glow discharge of 1 to 3 eV, the discharge is not significantly affected by the addition of He, other than to raise the electron temperature of the discharge [7]; Since the first ionization level of He is higher than that of the other gas components, high energy impacts (higher "voltage") is required to make it part of the glow conducting path. Only a small amount of energy is lost from the discharge due to inelastic collisions with He and subsequent collisions with the walls. Thermal conductivity in gases is independent of pressure and since thermal conductivity of He is roughly six times that of CO2 and N2, He makes an efficient transporter of waste heat to the walls of the discharge tube. The efficiency of heat transfer resulting from the addition of He to the mixture allows for a higher discharge current before radiation saturation [7]. CO may also be added to the laser mix to improve efficiency, but it does not transfer vibrational energy as efficiently as N2, due to a difference between the CO v=1 level and the CO2 upper lasing level of 170 cm^-1. CO also has a dipole moment which creates a radiative decay channel to depopulate the electron impact excited CO, thus making CO less available for the job of CO2 excitation. CO is also a component in the dissociation equilibrium of CO2, so when using added CO with CO oxidation catalysis, larger concentrations of CO affect the CO2 concentrations, not always in a predicatable manner. With these drawbacks, CO still adds to more efficient CO2 vibrational excitation than electron impact alone. H2O can be added as a heat transfer enhancer but is less efficient at cooling than He. H2O, in small concentrations, also has the beneficial side effect of homogeneous catalytic recombination of the dissociated CO2 products, CO and O [1]. H2O in larger concentrations overwhelms the beneficial catalytic effects and effectively depopulates the upper lasing levels of CO2. The optimum concentration of H2O in the laser gas has been shown to be a function of the laser bore diameter [7]. Xe may also be added to a laser gas mix to effectively cool the electron temperature of the discharge for a given current, thereby reducing the amount of electron impact dissociation of CO2. The prohibitive cost of laboratory grade Xe prevented this investigator from utilizing it.

3. Construction

3.1 The Laser Tube

The laser is a sealed, DC discharge type with attached ballast tank for long gas life. The silver-copper cathode design was used to reduce the amount of gas consumption by sputter pumping through chemisorption and physisorption. Construction of the laser was kept simple to reduce expense and excessive consumption of time. The laser is constructed of Pyrex, fabricated on the Boulder Campus by the chemistry department's Master Glass Blower. See Figure 2.

Fig 2 - Schematic of a sealed CO2 laser

The design incorporates a laser bore nested in a water cooling jacket, with a feed through the water jacket so that the discharge can go to an external cathode and anode. Having the external electrodes lower than the axis of the laser bore reduces the possibility that sputtering or oxidation products at the electrodes will contaminate the optics. This becomes an important consideration when working with CO2 lasers as intracavity power densities, even in a resonator of this design, can easily exceed 100 W/cm^2, quickly causing thermal damage on the surface of an optic, should a small piece of contaminant land on it. A glass vacuum valve is attached for evacuation and filling of the laser mix. Water inlet and outlets are positioned so that air bubbles which form inside the water jacket are ejected as they rise to the top of the tube. Without this design, air buildup inside the jacket would create a radial temperature differential perpendicular to the optical axis, detuning the resonator and/or thermally poisoning the gain medium.

3.2 Mirror Mounts

Mirror mounts were turned out of scrap rolled aluminum from the Physics Instrument Shop. The mounts are of a simple yet versatile design, incorporating an O-ring which provides both the vacuum seal and the restoring force for the mirror adjustment. The finished pieces were sanded with a very fine grained sand paper, then polished, to reduce high surface area materials in the low pressure environment. This reduces the potential of laser gas poisoning due to adsorbed contaminants. See Figure 3. Since the travel on the mirror for optical adjustments is very small, motion of the mirror on the o-ring for mirror adjustment doesn't break the vacuum seal. Both the high reflectance (HR) mirror, and the output coupler (OC) have mirror mount backings with a depression in them to hold the mirror near the center of the optical axis during assembly. With extra room between the glass end of the laser and the mirror mount shoulder, the mount can be aligned coaxially to the laser bore during assembly and the glass blowing need not obey strict tolerances. The mirror mounts need not be mounted exactly perpendicular to the laser bore either, since even relatively gross alignments can be accomplished with the o-ring backing plate.

3.3 Assembly

The mirror mounts were attached to the glass laser tube with the aid of a mandrel, which aligned the mirror mount coaxially with the optical axis. The mandrel consisted of a piece of Delrin turned so that one end fit snugly into the center of the mirror mount, and the other end fit snugly into the laser bore. This alignment procedure was necessary as tight tolerances were not requested for the glass work in order to avoid excessive cost. The laser tube was tipped on end and the mirror mount brought up from beneath with a small lab jack, which then met the laser tube end. After the mandrel had aligned the mirror mount to the axis correctly, it was removed for the next step. Apiezon W vacuum sealing wax was broken into small chips and dropped along the inside edge of the mirror mount, between the glass and the aluminum shoulder. See Fig 3. A heat gun was used to warm the mirror mount so that the sealing wax melted and flowed to form a positive vacuum seal between the mount and the tube. While Apiezon W makes a suitably clean vacuum seal, it does not have adequate structural strength for this application, so five minute epoxy was added on top of the wax, to enhance the mechanical integrity of the mount. Alignment of the laser using a Helium Neon laser required the mounting of the laser on an optical bench. Since the laser itself had been built into an acrylic box to protect it from breakage, and the user from electrical hazards, the entire box was mounted on an optical rail. The laser was secured in an acrylic box, on acrylic mounts lined with Sorbethane shock absorbing material. Aluminum strips were used to secure the ballast tank to its mounts while acrylic pieces screwed in the laser mounts secure the laser tube. For alignment of the laser resonator, two alignment mandrels were again made, each fitting snugly in the ends of the laser bore, but easily passing through the mirror mount.

Fig. 3 - Details of mirror mounts

Each had a 1 mm hole drilled in the center so that coaxial alignment of the HeNe alignment laser with the CO2 resonator was assured. After the HeNe beam was aligned with the CO2 laser bore, the mandrels were removed and the rear reflector was mounted and aligned so that the beam was reflected back into the HeNe aperture. The output coupler was then mounted then aligned to the HR by eye, since trying to make sense of the multiple reflection of the HeNe beam was futile. A bright, quartz halogen light was directed into the OC so that the image of the light was visible on the HR when looking down the bore through the OC. The OC was then adjusted so that the light source created "hall of mirrors" effect between the HR and OC. With a concave-concave resonator and such a high gain medium as CO2, this type of alignment procedure is more than adequate. A simple evacuation and fill system was assembled using Polyflo tubing and Swagelock fittings which used polyethylene ferrules. Industry standard CO2 laser gas mix consisting of 4.5% CO2, 14% N2 and balance He was used for the fill.

3.4 The Cathode

Cathode material selection in a sealed DC laser is crucial. The abundance of free, ionized oxygen in the discharge rules out any materials which readily oxidize, like tungsten, nickel, aluminum or stainless steels. The abundance of CO in the discharge also limits many otherwise suitable materials since many materials form gaseous arbonyls, removing CO2, and exposing fresh cathode surface for oxidation. These carbonyls can then be transported to the optics and deposited, forming a strong IR absorber. The "Hochuli cathode" was found through a literature search on cathode materials conducted earlier by the author. Professor Urs Hochuli generously donated cathodes for this project. The Hochuli cathode was designed specifically for long life sealed DC CO2 lasers, but was only tested at much lower currents, of about 5ma. [3] The Hochuli cathode is made of Ag and Cu in a matrix which is internally oxidized. This oxygen equilibrium within the cathode allows the Ag and Cu to not consume oxyge n from the discharge, thereby eliminating the cathode as a chemical sink of oxygen. The materials in the cathode are not readily transported in the discharge, so mirror contamination is not a problem. The cathode is in the familiar hollow cathode design, so sputter pumping of gas through physisorption is less of a problem, although this may be the ultimate source of gas consumption by the cathode. A glass plasma limiting sheath is incorporated in the cathode design to prevent the glow from forming on the outside of the cathode. Should that happen, it would rend the hollow cathode design useless, and sputter pump the heavier gas constituents quite efficiently. The discharge is physically channeled to the inside of the cathode, as the sheath acts as a physical plasma limiter. A small rod of tungsten is used as the anode since it is rather easily fed through the glass using a graded Uranium glass to metal seal. The anode doesn't suffer from the high momentum impacting ions as the cathode does, so it does not act as a sputtering loss element. The cathode is the only electrode which suffers from sputtering, since it is the source of electrons, which requires a large number of ions crashing into its surface to form secondary electron emission. This is the mechanism whereby the cold cathode provides electrons [4].

3.5 Power Supply

The power supply for the laser is of the simplest design. A 15 kV, 60 mA neon sign transformer is used to obtain the high voltage. The output of this is connected to a high voltage bridge rectifier. Since high voltage bridges are expensive and often difficult to find, the bridge was assembled out of several HV diodes in series, one set of three diodes on each leg of the bridge. The output of the b ridge is connected to a HV filtering capacitor, which has a 30M ohm bleeder resistor across its terminals. Since the output of the transformer is center tap grounded, neither end can be tied to ground. consequently, while in operation, all components of the HV end of the power supply are floating with respect to building ground. 200k Ohms of high power ballast resistance is placed in series with the discharge to limit tube current. The tube current is varied rather crudely, by simply using a Variac to adjust the input voltage to the transformer.

4. Laser Operation

After the laser was assembled and aligned, it was connected to a sink and drain for cooling water. A closed circuit heat exchanger will be built to make the laser self contained. The laser head and power supply had been mounted in separate acrylic boxes for safety purposes, and to keep the glass tube away from potential mechanical stresses. An open-ended mercury manometer was used to measure the pressure of the gas fill. Due to daily fluctuations of atmospheric pressure, the manometer had to be set to the day's pressure. The laser and fill system was roughed out until it settled to its lowest value. This lowest value was taken to be "0" Torr, since a good mechanical pump can pump down to at least .1 Torr [5]. Due to this approximation, the laser tube pressure is taken to be within one Torr of measured value. After evacuation, a leak-proof system was verified to first order by simply closing off the system to the pump, then watching the pressure gauge. After several days, no appreciable leaks were detected. The system was filled to a nominal pressure of 30 Torr to attempt lasing. Lasing was obtained but lasted for less than a second. This anomalous behavior was a setback, but after careful consideration revealed itself to be a chemical imbalance, as will be discussed later in this paper. The laser eventually gave consistent performance and output power data were taken at varying pressures and currents. Mode burns were taken in several materials, and in several alignments of the resonator. A burn at the best power alignment was also taken. Due to laser powers exceeding the capabilities of the commercial power meter available, power was measured by a simple thermocouple pair. Laser light is absorbed by a small piece of black anodized aluminum which absorbs 10 micron laser light very well. this small piece is mounted against a larger heat sink to dissipate the laser power. One junction of a thermocouple pair was placed against the target piece, and another series reference junction was placed against the heat sink. The millivolt output of the junctions was monitored as an indicator of input laser power. This meter was then calibrated using a commercial laser power meter.

4.1 Performance

The data in Table 1 were taken over tube current and gas pressure before the laser tube had come to complete equilibrium from the turn-on phenomenon described earlier, and without optimizing the resonator each time. Therefor, the peak power obtained in the data does not correspond to the highest power obtained, of 22 Watts. It is quite clear form the data, however, that an optimum pressure and current does exist for maximum power and efficiency. The optimum pressure was arrived at empirically to be 30 Torr. The optimum current was 9.2 mA. Table 1.

Current (ma) Voltage (kV) Power (W) Efficiency

20 Torr 5.3 9.76 1.1 .021
        5.7 9.59 .93 .017
        6.1 9.55 .22 .004
        6.5 9.36 .99 .016 

24 Torr 6.9 10.46 1.42 .020 6.5 10.44 1.75 .026 7.0 10.29 2.07 .029 7.6 10.15 1.26 .016 7.8 10.12 .99 .012 8.1 10.03 .17 .002

26 Torr 7.4 10.57 2.94 .037 7.7 10.78 3.49 .044 8.2 10.45 2.62 .031 8.6 10.35 2.18 .024 8.9 10.71 1.75 .018

28 Torr 8.4 11.13 9.80 .015 8.8 11.03 7.62 .079

30 Torr 9.2 11.40 11.43 .109 8.9 11.37 10.34 .102 8.8 11.28 9.53 .096 9.1 11.25 5.45 .053 9.4 11.08 .99 .009

32 Torr 9.0 11.41 3.75 .037 9.1 11.31 3.43 .033 9.3 11.17 1.64 .016 9.7 11.08 .55 .005

After cycling the laser power and allowing for mixing as described earlier, output power was measured as a function of time. Power fluctuations at turn on were found to be due to mode changes. It appears that as the laser warms up, the resonator alignment changes. This can be verified by simply tweaking the mirrors for the lowest order mode while it is warming up, then watching that mode be detuned. The mode can be brought back by simply adjusting the resonator again. Thermal detuning appears to be a plausible explanation since the glass tubes which feed the discharge out the laser bore go through the walls of the tube which make up the laser resonator. As these feeds heat up on the outside due to no water cooling over them, they expand and create a wedging in the resonator structure. The laser can be tuned for best mode after stabilizing. If the laser is allowed to cool, then is turned back and allowed to restabilize, the good mode that was achieved previously will come back. The sup ports the idea of thermal detuning. Power performance as a function of time after turn on is given in Figure 4.

Fig 4 - Power as a function of time

Monitoring the mode periodically during the first few minutes at turn on indicates that the power fluctuations are due to the cavity tuning through different modes. In the data shown, the power fluctuations settle down after about 25 minutes and settle to something other than the best power mode. Further measurement has shown this settling behavior to consistently be the case. The spikes in the data are artifacts of a noisy chart recorder. The thermal inertia of the meter does not allow for such a fast response so the spikes cannot be real. The ordinate is one minute per division. Initial variation of 22% of the peak power is seen as the cavity tunes through different modes and the highest power attained is 16.5 Watts. The resonator does not settle to the highest power mode. Vertical sensitivity is 1.17 Watts/division. Typical best power donut or bulls eye laser modes, (TEM 01* or TEM 10) are shown as contrast enhanced burns in thermally sensitive fax paper and plain white cardboard. Other mode burns in cardboard, taken during cycling of the laser during warm up, over the period of about one minutes, shown evolving symmetries corresponding to lower power.

5. Discussion

The anomaly mentioned earlier, whereupon a fresh fill, measurable laser power lasted less than a second, is believed to be due to oxygen specie migration between the laser bore and the ballast tank. This mechanism is believed to occur as follows: Upon initial filling, the gas in the laser bore and ballast tank is uniformly a laser mix of 18.7 :2.4:1 (He:N2:CO2). When the discharge is turned on, many other species are formed due to dissociation of CO2 and CO and O and the subsequent formation of NOx and other compounds. At this point, the laser bore has an abundance of NOx and other species and the ballast tank does not. Diffusive processes drive these byproducts from the laser bore to the ballast tank. Indeed, the characteristic green glow of N+O combination fluorescence can be seen streaming out of the cathode section towards the ballast tank. Also, a characteristic white discharge in the laser bore indicates a CO rich medium, which would be the case if free atomic oxygen were leavin g the bore. In low current discharges of CO2 laser mix, the pinkish emission bands of the N2 fourth positive system are seen which are due to nitrogen first-positive transitions [2]. If a white discharge is seen, dissociation is dominating the discharge. The length and brightness of the exiting fluorescence stream is demonstrated to be proportional to the discharge current. A rapid migration of oxygen containing species out of the laser bore makes for a sink of CO2 in the ballast tank and the laser then becomes lean in its lasing medium. If this migration hypothesis is correct, we would expect that the laser power should last longer upon initial turn on if dissociation products are allowed to come to an equilibrium between laser bore and ballast tank. This is indeed the case. After repeatedly turning on the laser then allowing diffusion time, performed over a period of a couple days, power does begin to remain consistent at turn on. The maximum power attainable does improve after several days of turning the laser on and off and allowing it to sit. Ultimately, a laser power of 22 Watts (44 Watts/meter) has been achieved this way, with relatively stable output power lasting for over and hours worth of operation.

6. Conclusions

The original objective of building a sealed, DC CO2 laser was accomplished with satisfactory laser performance. The anomalous turn-on effect delayed progress since many other possible problems were investigated first such as a leaky system, water contamination, dirty optics etc. but is now believed to be understood. The laser has yielded powers up to 22 Watts and has repeatable power performance, on a single fill, on the order of several weeks. The best mode attainable for this resonator design is a donut mode or a TEM 01* mode.

7. Further Investigations

It is recommended that, should further research be possible on this laser design, the oxygen transportation phenomenon be more thoroughly studied. More detailed analysis will most likely require the use of a mass spectrometer for time resolved characterization of the gases in the laser bore and the ballast tank after initial turn-on. Improvements on the resonator design with respect to thermal stability and warm up cycles is desirable. Recent work in Au intracavity catalysts has been said to improve laser performance to powers in excess of 100W/m. Further studies in intracavity catalysts such as Pt/Pd on SnO2 or Rh on Sn O2 may prove fruitful.


The author wishes to acknowledge the contributions and support of Professors D. Bartlett, B. Ridley and S. Robertson of the University of Colorado at Boulder Physics Department and Professor U. Hochuli as well as the technical assistance and support of E. Lutter and R. Tyler.


1. Campbell, I. M., Catalysis at Surfaces. Chapman & Hall Ltd., New York, 1988.

2. Gaydon, A. G., & Pearse, R.W.B, The Identification of Molecular Spectra. Chapman and Hall, New York, 1976.

3. Hochuli, U., et al. IEEE JQE., 10, 139-244, 1974.

4. Llewelln-Jones, F., The Glow Discharge. edited by Worsnop, B. L., John Wiley & Sons Inc. New York, 1966.

5. O'Halnon, J. F., A Users Guide to Vacuum Technology. John Wiley & Sons, New York, 1980.