Introduction

Of all gases, oxygen plays a unique part in humanexistence. Without it, life as we know it could not exist.The nature and measurement of oxygen has,therefore, always been an important challenge toscientists.

Modern technological advances both in industry andmedicine have resulted in an increased demand foraccuracy in oxygen measurements. The application ofoxygen analyzers ranges from the measurement of theatmosphere inside an incubator containing apremature baby to the monitoring of the oxygencontent of process and combustion streams in giantchemical plants and refineries to insure safe andeconomical operation.

It is the intent here to discuss in detail TAI's Micro-fuelCell, an electrochemical oxygen transducer aroundwhich an entire line of analyzers has been developedand successfully marketed. At the end of thisdiscussion, the principle of operation of other oxygenanalyzers will be reviewed and comparisons of eachanalysis technique, relative to that of the Micro-fuelCell, will be made.

The Micro-fuel Cell

TAI's Micro-fuel Cell is an electrochemical transducer.A transducer is a device that converts one form ofenergy to another. For example, a loudspeaker is atransducer that converts electrical energy intomechanical energy (sound). A microphone is atransducer that is the reciprocal of this; mechanicalenergy is converted to electrical energy.

Sensors for Gas Phase Oxygen Analysisby Jay Lauer, Development Engineer

AbstractA discussion of gas phase oxygen analysis is presented. Particular emphasis is given to the construction, principle ofoperation, and performance characteristics of the various types of Micro-fuel Cells manufactured by TeledyneAnalytical instruments (TAI). Micro-fuel Cell is the trade name for TAI's disposable electrochemical oxygentransducer.

A second, non-disposable trace level oxygen sensor is discussed. The additional requirements and specialcalibration techniques related to analyzers utilizing this type of oxygen sensor are described. Finally, other analysistechniques are summarized and comparisons drawn.

The most common electrochemical transducer is abattery which transforms chemical energy intoelectrical energy. Although less well known, a fuel cellperforms the same transformation. The primarydifference between a battery and a fuel cell is theplace where the chemical energy is stored. In theinstance of batteries, the chemical energy is storedinside the device itself. With fuel cells the chemicalenergy is stored externally; the rate at which thechemicals are fed into the fuel cell determines theamount of energy or power that is obtained.

Principle of Operation

The Micro-fuel Cell then is a fuel cell (strictly speaking,a hybrid fuel cell) where part of the chemical energy isstored within and the other chemical reactant (oxygen)comes from outside the device. Most fuel cells arepower devices, i.e. the hydrogen-oxygen fuel cells thatare used to power the manned space vehicles used byNASA in the space program. The oxygen sensing cellunder discussion here provides power normally in the1 to 200 microwatt range, hence the name Micro-fuelCell.

All types of electrochemical transducers have threemajor component parts; an anode, a cathode, and anelectrolyte. In the Micro-fuel Cell, the cathode is thesensing electrode or the site where oxygen is reduced.

4e- + O2 + 2H2O -------> 4OH- (1)

In the above reaction, four electrons combine with oneoxygen molecule (in the presence of water from theelectrolyte) to produce four hydroxyl ions. This

cathodic half-reaction occurs simultaneously with thefollowing anodic half reaction:

Pb + 2OH- -------> PbO + H2O + 2e- (2)

The anode (lead) is oxidized (in a basic media) to leadoxide and in the process two electrons are transferredfor each atom of lead that is oxidized.

The sum of half-reactions (1) and (2),

O2 + 2Pb -------> 2PbO (3)

results in the overall reaction (3). From this reaction itcan be seen that the Micro-fuel Cell should be veryspecific for oxygen providing there are no gaseouscomponents in the sample stream capable of oxidizinglead. The only likely compounds that meet thisrequirement are the halogens (iodine, bromine,chlorine, and fluorine.)

Physical Construction

The Micro-fuel Cell (Figure 1) is designed as adisposable unit and when the lead anode is consumeddue to the sensing of oxygen, the cell is simplydiscarded. The cell is, therefore, maintenance freerequiring no replacement of membranes or electrolyteand no cleaning ofelectrodes. It is about 1.25"in diameter and .75" thick.The rear of the cell is fittedwith a contact platecomposed of two concentricmetal foils which areelectrically common to theanode and cathode andmate to spring loadedcontacts in the various cellholding devices usedthroughout the product line.

The anode is composed oflead and is configured insuch a manner (proprietary)so as to maximize theamount of metal availablefor reaction. The cathode isa convex metal discapproximately .75" indiameter with numerousperforations to facilitate continued wetting of the uppersurface with electrolyte and assure minimum internalresistance during the oxygen sensing reaction. Theexternal surfaces of the cathode are plated with aninert or noble metal, normally gold. Silver, rhodium,and platinum are also used in special cases.

The rear of the cell (inside the contact plate) is fittedwith a flexible membrane that is designed toaccommodate internal volume changes that occurthroughout the life of the cell. If it were not for thisvolume compensating membrane, the membranecovering the cathode would move to accommodate theinternal volume changes resulting in changes in outputnot related to oxygen concentration.

The sensing membrane (covering the cathode) ismade of Teflon whose thickness is very accuratelycontrolled. The entire space between these twomembranes is filled with electrolyte (an aqueoussolution typically of potassium hydroxide). This resultsin all surfaces of the anode and cathode being"bathed" in a common pool of electrolyte.

The main body of the cell is fabricated from highdensity polyethylene. This results in an oxygen sensorthat can be placed in virtually any atmosphere orsample stream without any concern that thecomponents of the sampled media will react with thecell assembly.

Oxygen Measurements

In reaction (1) four electrons are transferred for eachoxygen molecule undergoing reaction. In order to bereacted, an oxygen molecule must diffuse through

both the sensing membrane and the thin film ofelectrolyte maintained between the sensing membraneand the upper surface of the cathode. The rate atwhich oxygen molecules reach the surface of thecathode determines the electrical outlet. This rate isdirectly proportional to the concentration of oxygen in

the gaseous mixture surrounding the cell.

As in Figure 2, this proportionality is linear with thesensor exhibiting an absolute zero, i.e. in the absenceof oxygen the cell produces no output.

Calibration: Given these two conditions (linearity andabsolute zero), single point calibration is possible. Theoxygen concentration in air (20.945%) is a very

convenient reference gas to use, and it is accurate tothree or four significant figures throughout the entireearth's surface.

In calibrating a given oxygen analyzer, it is best to usea calibration or reference gas whose oxygen content isnear or equal to the full scale reading. In Figure 2,both air and 100% oxygen are shown as potentialcalibration gases. In theory, any known concentrationof oxygen could be used. Air and 100% oxygen werechosen because they are the easiest and mosteconomical to obtain. In this case, however, air is notthe better of the two gases to use for calibration. Anyerror in reading the readout meter or an error in themeter reading during the setting of the span controlwould be multiplied by a factor of 5 for subsequentreadings at or near full scale. This results fromattempting calibration at a point only 1/5 of full scale.In this case, 100% oxygen would be the calibrationgas of choice.

Since air is such a convenient reference and theMicro-fuel Cell's readings are independent of thesample gas' flow rate (they can give accurate readingin static gas samples such as ambient air), multipleranges are generally provided in the more

sophisticated instruments, one of which is always 0 –25%. 20.94% is about 84% of full scale which willprovide good calibration accuracy which is degradedonly slightly when switching to the other rangesincluded.

Concentration Units: Although most of the oxygenanalyzers manufactured by TAI are set up to readoutthe concentration of oxygen in percentage units, the

Micro-fuel Cell (and allother analysis techniques)are actually sensitive tothe partial pressure ofoxygen in the sample gasmixture. Readouts inpercent are permissibleonly when the totalpressure of the gas beinganalyzed does not change.

To illustrate, assume thatan analyzer with a percentreadout was calibrated inair at sea level. The spancontrol would be adjusteduntil the analyzer readingwas 20.94% (assumingthat degree of readabilitywas available). If theanalyzer were thenelevated to higher andhigher altitudes, the

reading would begin to drop and continue to do so aslong as the analyzer is raised. The percent of oxygenin the atmosphere is the same regardless of thealtitude, i.e. the ratio of oxygen molecules does notchange; the molecules merely get farther and fartherapart. The Micro-fuel Cell's output is a function of thenumber of molecules of oxygen per unit volume whichis decreasing as the analyzer is raised in altitude.

Figure 3 illustrates how the output of the cellincreases when the total pressure is increased (theopposite of the previous illustration). In this case aswith the previous one, the percent oxygen does notchange.

Dalton's Law of Partial Pressures states that the totalpressure exerted by a mixture of gases is the same asthe sum of the individual pressures exerted by theconstituent gases. Or, stated another way, in a mixtureof gases, every gas exerts the same pressure that itwould if it alone were confined in the same volume.

P(total) = P(1) + P(2) + P(3) + …

For air this becomes:

P(total) = P(O2) + P(N2) + P(Ar) + P(Ne) + …

The amount of argon, neon, and other trace gasespresent in air is less than 1%. Ignoring theircontribution and assuming a dry gas mixture (nowater), the expression becomes:

P(total) = P(O2) + P(N2)

At sea level, P(total) would be 1.00 atmospheres, 760millimeters of mercury, or 14.7 pounds per square inch(absolute). Other partial pressure units could be used.In each of the three units above, the concentration ofoxygen and nitrogen would be: 0.209 atm, 159 mm-Hg, and 3.07 psia for oxygen; and 0.791 atm, 601mm-Hg, and 11.6 psia for nitrogen.

Returning to Figure 3, it is evident that when the totalpressure of the gas mixture (air) is doubled, the partialpressure of all the constituent gases will double aswell. The oxygen reading doubles also since thenumber of oxygen molecules per unit volume is twicewhat it was before the total pressure was changed.Therefore, when applying an oxygen analyzer wherethe total pressure will vary, a readout in partialpressure units should be employed.

Temperature Coefficient: In order to obtain accuratereadouts that vary only with changes in oxygenconcentration, one final parameter must be taken intoaccount – temperature. The rate at which oxygen

molecules hit the cathode isgoverned by diffusionthrough the sensingmembrane and the film ofelectrolyte between themembrane and cathode.Since all diffusion processesare temperature sensitive, itis to be expected that theMicro-fuel Cell's electricaloutput will vary withtemperature (at a constantlevel of oxygen). Thisvariation amounts to about+2.5% per degreeCentigrade. This means thatfor every degree Centigradeof temperature rise, the celloutput will increase 2.5%.The cell, therefore, has apositive temperaturecoefficient of 2.5% perdegree Centigrade.

Negative temperaturecoefficient thermistors are

used to compensate for this effect. Figure 4 showsthe manner in which this temperature compensation iseffected. In most instruments, the compensation curveis matched with an accuracy of ±5% or better. Theresultant output function then is independent oftemperature (within the above stated accuracy) andvaries only with changes in oxygen concentration.Figure 4 also shows two simple circuit diagrams wherethe Micro-fuel Cell's output is temperaturecompensated; one involving no amplifier and the othershowing a single stage of amplification.

It should be noted that Micro-fuel Cells do haveenough output power to drive readout meters directlywithout amplification or external power supplies. Thisusually results, however, in single range capability.

Amplification: In most of the oxygen analyzers usingMicro-fuel Cells (with the exception of the portablemedical and safety instruments), a dual stageamplification circuit is used as shown in Figure 5.

The first stage is an I to E (current to voltage)transducer whose negative feedback loop is used tofacilitate range changes. The I to E configuration offersthree distinct advantages.

1) The Micro-fuel Cell can be located at a distance(500 feet) from the remainder of the analyzer withoutpreamplification or noise pickup problems.

2) The cell operates into a short (zeroohm load) which

is the ideal for best linearity and rapid response.

3) The effects of amplifier voltage offset are minimized.

An N-channel field effect transistor (FET) is locatedacross the inputs of the I to E transducer to provide alow resistance current path for the cell when theanalyzer is turned off (no power). When the gate of theFET is powered, the source to drain resistance isapproximately 10 to the 12th power ohms. Whennegative power is removed, this resistance drops toapproximately 30 ohms.

As shown in Figure 5, the output of the first stage ofamplification is as follows:

E(out) = -iR(f) (1)

Where E(out) is the outputvoltage, i is the input current(the current output of theMicro-fuel Cell) and R(f) isthe negative feedbackresistance. This feedbackresistance can be changed(as shown in the diagram) toeffect range changes. Thecell is connected backwards(+ to – and – to +) becausethe polarity of transductionis inverted when theamplifier is operated in thisconfiguration. Hence, theminus sign in equation (1).

The second stage ofamplification is used toeffect temperaturecompensation. Thethermistor must be locatedremotely or at the samelocation as the Micro-fuelCell, and so a total of fourconductors (two for the cellsignal and two for thethermistor) plus a shield,are required when thesensor assembly is locatedremotely.

The output of the secondstage of amplification isthen:

E(out) = E(in) x (R1 + R2)R2

(2)

Where E(out) is the output voltage, E(in) is the inputvoltage (output of the first stage), R1 is the thermistorresistance, and R2 is the reference resistor. Theexpression:

(R1 + R2)-------------

R2

Is the gain of the second stage of amplification.

To complete the amplification circuit of the analyzer, itis only necessary to provide span (calibration) andreadout capability. Figure 5 shows these in schematicform.

Accuracy: It is now possible to discuss the concept ofoverall accuracy since the parameters affectingaccuracy have been delineated. There are two basictypes of errors affecting overall accuracy of a givenanalyzer:

1) Those that produce a percent of readingcontribution, and

2) Those that are a fixed value (usually expressed asa percentage of full scale).

These two error types are expressed graphically inFigure 6.

The most significant percent of reading error is thatcontributed by the matching of the thermistor curveto the output versus temperature curve of the Micro-fuel Cell. As mentioned earlier, this is normallyaccomplished with an accuracy of better than ±5%.A second percent of reading error, although muchless significant, is that imposed during rangechanging. Typically resistors having a tolerance of±1% are used. However, in the more sophisticatedmulti-ranged analyzers, resistors of closertolerances or selected resistors are used todecrease this error contribution.

The major percent of scale type error is thatcontributed by the readout device. Meters can becounted on for linearity errors from ±0.5% to ±2%depending on the quality of the meter. In the case oftaut band meters, linearity errors are the onlysignificant error involved. (The others are eitherinsignificant or are "spanned out" during calibration.)Pivot and jewel type meter movements can addrepeatability errors due to the increased friction of themovement bearings. This can add another ±1% or so.If recorders or digital meters are used to readout theanalog signals, then errors of ±0.25% and ±0.1%respectively can be achieved.

A typical overall accuracy statement for analyzersincorporating a meter readout would be ±2% of fullscale at constant temperature or ±5% of readingthroughout the operating temperature range,whichever is greater. Figure 6 shows this specificationgraphically.

It is important to keep in mind that these accuracyspecifications are worst case and, for example, ifmeasurements are made at the same temperature atwhich calibration was accomplished, there will be notemperature compensation error. In this case, onlyreadout errors would be operative. Small temperaturevariations (i.e. 10 - 20°F) will produce a maximum of±1% of reading error.

Cell Warranty and Classification

TAI manufactures a number of different types of Micro-fuel Cells. The differences are due to differingapplication requirements. Varying the thickness of thesensing membrane, for example, offers a trade off inspeed of response and long life. Thinner membranesproduce relatively rapid response and shorter lifewhereas thicker membranes produce slower responsecharacteristics and longer life.

Cell Life: The life of a cell is generally limited by theamount of lead anode material available inside the celland the rate at which it is consumed. This rate isgoverned by the average or mean concentration ofoxygen that the cell is exposed to throughout its life. Ifa given cell would last for 12 months in air, it would beexpected to last only 21/100 or approximately 1/5 aslong as in 100% oxygen.

If cells are used to monitor the oxygen content of drygas streams, their life will be limited generally by theloss of water which ultimately results in sluggishresponse characteristics and failure to give accurate,low level readings. The water from the electrolyte, inthese cases, diffuses slowly through the sensingmembrane and life limitation is primarily a function ofmembrane thickness.

Good performance can be expected until theelectrolyte volume is reduced to approximately half itsoriginal volume (2 – 3 millimeters). With the thickersensing membranes, the time required to accomplishthis reduction in electrolyte volume is about 12months. In the case of thinner membranes, this time isreduced to about 6 months.

Warranty: The various cell classes are warrantedusually for a 6 or 12 month period based upon theabove considerations, and in the event of failure due

to manufacturing or material defects, the cell will bereplaced. All warranty replacements are on a proratedbasis.

For information on the complete line of Teledyne’sMicro-fuel Cells, please visit our website at:

www.teledyne-ai.com

Packaging and Handling

All Micro-fuel Cells are packaged in special gas barrierbags. Most cells are originally packaged in air, butafter the bag is sealed the cell reacts with the oxygeninside the bag. Eventually oxygen in the bag isconsumed and the cell is in a relatively inertenvironment inside the bag. The barrier bag transmitsnegligible amounts of oxygen and almost no moisture.The cell can thus be stored for periods in excess ofone year and still provide 90 – 95% of expected lifewhen finally placed in use. It is not suggested,however, that cells be stock piled nor stored anylonger than logistics necessitate. TAI maintains a stockof all Micro-fuel Cells and normally cells are shippedthe same week that orders are received.

After the cell is removed from its barrier bag, it shouldcontinue to have a current path at all times. Thismeans the cell should remain installed in the analyzeror, if removed for any reason, the shorting clip shouldbe reinstalled. (Be careful not to puncture the sensingmembrane.)

A current path is required so the cell will be ready togive accurate, reliable readings at all times with aminimal waiting period. If the cell is left unshorted forany significant period, oxygen from the air will diffuseinto the cell without being reacted. Depending uponthe length of time the cell is left unshorted, when thecell is again installed in the analyzer a period of a fewminutes to overnight is required for the cell toconsume all the oxygen allowed to diffuse inside thecell. The end of this period is recognizable as the pointin time when the reading of the analyzer no longerdrops but rather stabilizes to some low value(assuming that the oxygen concentration at the cell isnot varying.)

In general, it is best not to remove the cell from theanalyzer unless there is some indication the cell is notperforming properly. The Micro-fuel Cell is rugged andcan survive dropping and general rough handling. Theonly precaution that should be taken is to assure thesensing membrane is not punctured with a sharpobject. This is especially critical with cells having thinsensing membranes such as the A-1, B-1, and B-2.

If a cell is contaminated with liquids that are

erroneously introduced into the sampling system of theanalyzer, the cell can be removed from its holder andcleaned with water or isoprophyl alcohol and dried witha tissue such as Kleenex or Kimwipe. This procedurewill in no way impair future performance of the cell. Itis important to remove any liquids that condense orotherwise find themselves on the sensing membraneof the cell. Films of liquid impair diffusion through thenormal diffusion path and result in erroneously lowreadings.

Occasionally poor contact between the cell contactplate and the spring loaded contacts inside the cellholder may develop. This is accompanied by erraticand unstable readings. If this occurs, the cell shouldbe removed and the contact plate and holder contactscleaned with water and / or alcohol. Sometimes it maybe necessary to clean the contact plate with cleanser,fine emery cloth, or sand paper.

OPEN TYPE ELECTROCHEMICAL TRACELEVEL OXYGEN SENSOR

TAI manufactures a series of trace oxygen analyzerswhich utilize an unsealed (open) configurationelectrochemical sensor.

Chemistry

There is no difference in basic chemistry between thiscell and the Micro-fuel Cell. The differences are mainlyin physical construction. To facilitate the measurementof oxygen in the very low ppm region, a cathode with alarge surface area is employed.

The cathode assembly is a radial array of 1/2 inchwide, 2 – 1/2 inch long, 80 mesh silver screenelements (see Figure 8). This assembly is mounted inan acrylic cell block so that its lower edge(approximately 1/32 inch) is under the surface of theelectrolyte. The surfaces of the silver screen wires aretreated in such a way so as to hold, by capillaryaction, a very thin film of electrolyte.

When an external current path is completed betweenthe cathode assembly and the lead anode (mountedconcentrically below the cathode), a current isproduced proportional to the concentration of oxygenin the gas phase surrounding the cathode of the cell.

Humidification

In a functioning analyzer, the sample gas is made toflow over the cathode and, unless the sample gas issaturated with water vapor (100% relative humidity),the water-based electrolyte would eventually dry out. It

is therefore necessary tohumidify the sample gasprior to its entering the cell.

Figure 9 shows the flowschematic for the Series 306of trace oxygen analyzers.Humidification water isstored in the reservoir. In theabsence of sample gas flow,the level of water will be thesame in the reservoir andthe humidifier column. Withthe sample bubbling throughthe humidifier, the leveldrops in the humidifiercolumn to a level equal tothe point in the reservoirtank where the samplebubbles through the tankjust prior to venting. Thesample is allowed to"blanket" the tank prior to exhausting to theatmosphere so all oxygen dissolved in thehumidification water will be expelled rather than addedto the sample stream.

Calibration

The Series 306 is not referenced to air as is theirMicro-fuel Cell counterpart. A Faraday type calibrationis utilized which relies upon the electrolysis of water ata known current rate into a known flow rate of samplegas. Faraday's Law states (in this case) that 96,500coulombs of electricity will electrolize 8 grams (onegram equivalent weight) of oxygen from water. Bymeasuring the current flowing through the calibrator(coulombs / second), a direct measure of theconcentration of oxygen added to a sample stream ismade possible. Figure 10 summarizes this process.

Without going into details of the mathematicalconversions, the following relationship holds:

C(O2) = 25i ppm O2 added

Where C(O2) is the concentration of oxygen added tothe sample stream; 25 is the constant ofproportionality; and i is the current in milliamperes thatflows through the calibrator electrodes. In Figure 10,f(i) refers to the voltage that is capable of being readout on the recorder, which is required for readout withthe oxygen analyzer.

Other Oxygen Analysis Techniques

The following section summarizes other oxygenanalysis techniques. Comparisons to the Micro-fuelCell method of analysis will be made at the end ofeach summary. At the conclusion of this section, atabular comparison of the basic electrochemicaltechniques will be made.

Paramagnetic Analyzers

In an oxygen molecule, two of the electrons in theouter shell are unpaired. Because of this, the magneticmovement of the molecule is not neutralized, thusmaking the oxygen molecule strongly paramagnetic(attached by a magnetic field). This property of theoxygen molecule allows an analysis cell to beconstructed wherein two diamagnetic spheres filledwith nitrogen are mounted at the ends of a connectingbar to form a dumbbell shaped assembly. Thisassembly is mounted horizontally on a vertical torsionsuspension. A quartz fiber or platinum ribbon isnormally use for this purpose. Electromagnetic orelectrostatic feedback is used to maintain thedumbbell at a reference or zero position. The spheresof the dumbbell assembly are coated with a thinmetallic coating when electrostatic feedback isemployed.

Figure 11 shows the basic configuration of anelectromagnetic feedback paramagnetic oxygenanalyzer. The dumbbell is mounted inside a stainlesssteel housing having a glass window. At the center ofthe dumbbell assembly is a small mirror upon which isfocused a light beam. The differential output from the

twin photocell is amplified and fed back to a coilaround the dumbbell. With an oxygen free gas in thecell, the torsion suspension is rotated (mechanicalzeroing) until there is a zero signal output from thephotocells. With a calibration gas in the cell, the gainof the amplifier is adjusted so the feedback currentreading corresponds to the concentration of oxygen inthe calibration gas. The currentrequired to keep the dumbbell atthe zero position is a measure ofthe magnetic susceptibility of thesample gas.

Paramagnetic analyzers are, ingeneral, best suited for laboratoryapplications and in instanceswhere suppressed ranges (90 –100%, 95 – 100%, etc.) arerequired. The major disadvantagein industrial applications are:

1. Sample gases must bescrupulously cleaned andrendered free of all particulatematter and condensibles.

2. The dumbbell assembly isfragile and expensive to replace.Also, this maintenance must beperformed at the factory.

3. The analyzer cannot read accurately underconditions of shock and vibration and is position andtilt sensitive.

4. The analyzer cannottolerate wide variations issample flow rates due tomovements of thedumbbell assembly causedby turbulences resultingfrom too high a flow rate.

The Micro-fuel Cellrequires less stringentsample conditioning andcan tolerate someparticulate andcondensible vapors in thesample stream. In theevent that the cell iscontaminated to the pointwhere accurate readingsare no longer possible, thecell can be removed,cleaned, and reinstalled ina matter of minutes. This is

not the case with the paramagnetic analyzer.

Magnetic Wind Analyzers

This type of analyzer (also called thermal magnetic)uses a combination of techniques – thermalconductivity and paramagnetic. The analysis cell

consists of two side tubes with an interconnecting tubepassing through a strong magnetic field (Figure 12).Inside the tube there is a heated filament which formsone arm of a Wheatstone bridge. Sample gas passesup the two side tubes and into the cross tube. Oxygen

present in the left hand tube is attracted into themagnetic field. Upon entering the field, the oxygen isheated by the filament and its magnetic susceptibilityis reduced. The heated gas is pushed across thecross tube by other cold gases entering at the left. Theeffect is as though there were a continuous flow of gasthrough the tube. This gas flow cools the filament, thuschanging its resistance. This change in resistanceunbalances the bridge andproduces a signal proportionalto the oxygen content of thesample gas.

Thermal magnetic analyzerssuffer from the same basicproblems as the paramagneticanalyzer without being asfragile or shock sensitive.However, these analyzers areplagued with several inherentsources of error that areunique:

1) The filament temperature isaffected by changes in thethermal conductivity of thecarrier gas, causing thecalibration to be correct foronly one gas mixture. If thatmixture is changed, a newcalibration mixture must be used. When the samplemixture varies during the required sampling interval,the resultant errors can be substantial.

2) Hydrocarbon vapors in the sample stream react onthe hot filament causing changes in temperature andsignificantly large errors. These hydrocarbons must be

removed if their concentration level is high enough. Inthose instances, oxygen reading errors will result dueto the reduction in sample volume.

3) The cross tube must be perfectly horizontal to avoiderrors due to gravitational flow effects.

Other Electrochemical Techniques

In addition to the Micro-fuel Cell type ofelectrochemical sensor, there are two other similardevices often encountered. These are thepolarographic sensor and the high temperatureceramic sensor (zirconium oxide).

The Polarographic Sensor:The main difference betweenthis type sensor and theMicro-fuel Cell is the sourceof the driving potential. In theMicro-fuel Cell, the drivingpotential is obtained throughthe proper choice of anodematerial, copper, cadmium,lead, tin, zinc, etc. Any ofthese anode metals wouldexhibit a driving potentialsufficient to cause a currentto flow in the presence ofoxygen. In the polarographiccell, the anode is usuallysilver and the cathode isgold. It is necessary that anexternal potential of about0.7 volts be applied between

the electrodes (minus the cathode) before thefollowing half-reaction can occur:

Cathodic reaction:

4e- + O2 + 2H2O ----- > 4OH- (1)

Anodic reaction:

Ag+ + Cl- ----- > AgCl + e- (2)

Overall reaction:

4Ag+ + O2 + 2H2O + 4Cl- ----- > 4AgCl +4OH- (3)

It should be noted that thecathodic reaction (1) isidentical to the cathodicreaction in the Micro-fuelCell. The reaction at thesilver anode occurs in aneutral potassium chlorideelectrolyte. This is thesource of chloride ions (Cl-)in reaction (2).

Unlike the Micro-fuel Cellwhere the composition ofthe electrolyte does NOTchange as the oxygensensing reaction proceeds,the polarographic cellelectrolyte is constantlychanging. Hydroxyl ions(OH-) are being substitutedfor chloride ions (Cl-).When the electrolyte becomes depleted of Cl-, it mustbe replenished. This involves taking the cell apart andconsequently the membrane must be replaced as well(Figure 13). Alternatively, the entire sensor may bereplaced.

Depending upon the design of the sensing assembly,the maintenance cycle can vary from a few hours toseveral months. If the sample gas is dry, this will tendto shorten the cycle since polarographic sensorsgenerally utilize small electrolyte volumes and thewater in the electrolyte evaporates relatively quickly.

Unlike the Micro-fuel Cell, polarographic sensors areunable to measure at the trace oxygen level. This isbecause of the residual of background current that isproduced due to the externally applied potential whichresults in the electrolysis of water. This residualcurrent is normally insignificant if the range ofmeasurement is, for example, 0-25%. However, a

range of 0-1% (0 – 10,000 ppm) will have an offset offrom 0.01 to 0.05% oxygen equivalent. Often thisbecomes greater in cells that are used and haveundergone several maintenance cycles.

And finally, polarographic cells always requireamplification because of their relatively low output andtheir need for an external polarizing voltage.

High Temperature Ceramic Sensor. Figure 14 showsthe arrangement of the major components of the hightemperature ceramic sensor. The sensing cell consistsof a calcium stabilized zirconium oxide solid electrolytewith porous platinum electrodes on its inner and outersurfaces. A temperature controlled furnace maintains

the sensing cell at approximately 800 degrees C (1472F). At this temperature the ceramic electrolyte willconduct oxygen ions (O=) and the half-reaction ateach electrode is:

Reference: O2 + 4e- ----- > 2O =

Sample: 2O = ----- > O2 + 4e-

The above equations hold as long as the referencegas is higher in concentration than the sample gas. Ifthe sample oxygen level exceeds the reference, thanthe O= and electron flow reverse and the output willbe of opposite polarity. The open circuit voltage for thetwo half reactions is:

E = RT in P(O2) referencenF P(O2) sample

Converting from natural log to log (10) this becomes:

E = 0.050 log P (O2) referenceP(O2) sample

As can be seen from these equations, the relationshipbetween the output voltage and the sample oxygenlevel is logarithmic. For every decade drop in sampleoxygen concentration, the output voltage increases 50millivolts.

The manufacturers of the high temperature ceramicsensor analyzers state their instruments can handlehot, wet, even dirty and corrosive sample gaseswithout a sample conditioning system. They also claimability to sense ppm quantities of oxygen under thesame conditions.

There is no question they can operate at hightemperatures and handle samples with high watercontent since their sensor is operating at almost 1500°F. However, this by itself is no real advantage if theoxygen content of sample gases cannot be measured

reliably or accurately.

The main objection to this type of analysis technique isthat at this elevated temperature, any and alloxidizable materials in the sample will combine withoxygen stoichemetrically and consequently willsubtract from the oxygen available. The reading is, ineffect, the net oxygen content of the monitoredsample. This phenomenon almost completelyeliminates the possibility of using the high tempeatureceramic analyzer for trace oxygen analysis since thereis always some hydrocarbons and/or oxidizableinorganic gases that contaminate almost any samplestream.

Also the output of this analyzer is logarithmic andreversed from normal readouts (i.e., the highconcentration is at the low end of the scale).Moreover, there is not true zero level. As a result,these output signals do not lend themselves to directlydriving conventional recorders or data acquisitionsystems.

Comparison of ElectrochemicalTechniques

Figure 15 is a comparison of the four basic types ofelectrochemical sensors, the Micro-fuel Cell, thepolarographic sensor, the open cathode sensor, andthe high temperature, ceramic sensor. Thiscomparison is made on the basis of versatility, cost,and maintenance. §