Gas carburizing is a well-understood process. The most important parameter aside from temperature is the carbon potential (C-potential).

 Fig. 1. Principle of the O2 probe

Gas carburizing is a well-understood process.[1] The most important parameter aside from temperature is the carbon potential (C-potential). The C-potential of a furnace atmosphere is defined as the carbon content (expressed in weight percent) of unalloyed austenite that is in equilibrium with the corresponding atmosphere. For example, an atmosphere with 0.7% C-potential would be in equilibrium with an austenite with 0.7% C. An austenite with more than 0.7% C would be decarburized to 0.7% C, whereas an austenite with lower C content would be carburized up to this value. It should also be mentioned that the C-potential of a certain atmosphere depends on temperature. In order to get a definite carburization depth in the parts, C-potential of the furnace atmosphere has to be measured and controlled during the heat-treatment process.

Indirect Determination of C-Potential

Basically, the C-potential can be determined directly or indirectly. The direct-measurement methods are not suitable for a continuous measurement of C-potential and a control based on this measurement. However, the direct methods are used to check and adjust the indirect measurement methods if necessary. The following equations are the basis for the indirect determination of the C-potential:

C + 0.5O2 « CO (1)
C + CO2 « 2CO (2)
C + H2O « CO + H2 (3)

These reactions take place in the furnace atmosphere as well as on the surface of the workpieces. It follows from these reactions that the CO gives off the carbon and O2, CO2 and H2O absorbs it. If the C-potential of the atmosphere is higher than the carbon content on the surface of the workpieces, CO transfers the carbon to the workpieces, whereas O2, CO2 and H2O remove the carbon from the furnace atmosphere. If the C-potential is lower than the carbon content on the surface of the workpieces, CO transfers the carbon to the atmosphere whereas O2, CO2 and H2O remove the carbon from the workpieces. In both cases, these reactions cause an equalization of the carbon content on the surface of the workpieces and the C-potential of the atmosphere.

For the calculation of the C-potential, the activity also becomes important. The activity depends on temperature and content of carbon that is dissolved in austenite. For a given temperature, each C-potential equates to a certain value of the activity ac(%C,T). The reactions 1-3 can be linked to the activity via the following equations:

The equilibrium constants of equations 1-3 – KO2, KCO2 and KH2O – are functions of temperature. These functions are well known in literature. Therefore, in order to calculate the C-potential, it is enough to measure the temperature and the partial pressures appearing in one of the equations 4-6.

In certain atmospheres (e.g., endogas or mixtures of nitrogen and methanol), the partial pressures p(O2), p(H2) are much higher than the partial pressures p(O2), p(CO2) p(H2O). When the C-potential is changing, the partial pressures p(CO) and p(H2) vary by percentage only a little compared with the percentage variation of p(O2), p(CO2) and p(H2O). This is the reason why for the calculation and control of the C-potential, in most cases, the partial pressures p(CO) and p(H2) are assumed to be constant and only the partial pressure or volume fraction of O2, CO2, or H2O is measured.

Continuous measurement is essential for the control of a process. The market-available dew-point sensors, which could be used for continuous determination of the partial pressure of H2O, are not robust enough for measurements in carburizing atmospheres.

The measurement of CO2 is done continuously with infrared sensors and is frequently used for comparison. The CO2 measurement has little correlation with the O2 measurement. In addition, the CO2 sensor has higher maintenance requirements because of the drift of zero point. Therefore, the zero point has to be corrected daily with nitrogen if the CO2 measurement is used for C-potential control. These are the main reasons why the measurement of oxygen partial pressure with the oxygen probe (O2-probe) has become widely accepted for the control of C-potential.

 Fig. 2. Types of O2 probes

O2-Probe Construction & Functionality

Figure 1 shows the construction of the O2-probe. The measuring cell of an O2-probe consists of a one-sided, closed-probe ceramic. The material is zirconium dioxide dotted with yttrium trioxide, for example. These intentionally implemented lattice defects enable the needed oxygen-ion conductivity. Furthermore, these defects also stabilize the ceramic against thermal and mechanical stress. The ceramic is connected with platinum inside and outside. These are the electrodes of the measuring cell. On the inside of the probe, reference gas is provided, mostly air with the known 20.9% O2. The outside is exposed to the furnace atmosphere. The O2-probe is applicable for in-furnace measurements only at temperatures above 700°C (1292°F) since the oxygen ions’ conductivity is temperature-dependent.

At the inside platinum electrode, oxygen is reduced to oxygen ions. These ions move via ceramic defects to the outer electrode in order to compensate the low oxygen concentration there. The accumulation of electric charges results in a voltage (Nernst equation) that can be measured between the electrodes. The voltage increases when the atmosphere oxygen content decreases. In the process of carburizing, this implies the following: Higher voltage corresponds to a higher C-potential, and conversely lower voltage corresponds to a lower C-potential.

There are primarily three different models of O2-probes on the market (Fig. 2):

• The first model has a one-sided closed zirconia element that is glued or welded to a ceramic pipe made from alumina. Apart from that, the probe works as described: reference air inside, furnace gas outside, oxygen ions move through the ceramic.
• The second model has a zirconia ball.
• The third model has a continuous, closed and dense zirconia pipe.

Probes with a zirconia element or ball are less costly than the probes with continuous zirconia pipe, but these probes have lower leak tightness. The difference of the expansion coefficients of the two different ceramics causes fine hairline cracks to occur at the contact point of the two ceramics. Through these cracks, the furnace atmosphere leaks into the inside of the probe ceramics and modifies the reference air. The probe voltage decreases, and the calculated C-potential is lower than the actual value in the furnace. This error can be minimized by increasing the flow rate of the reference air.

For calculation of the C-potential via O2 measurement and the equation of chemical equilibrium, the temperature is important for two reasons:

• Firstly, the temperature is needed for the determination of the O2 content (eq. 7).
• Secondly, the temperature is needed to determine the equilibrium condition from which the C-potential is calculated (eq. 4).

Therefore, more often than not the furnace-atmosphere probe temperature is measured with a thermocouple that is fitted inside the probe ceramic. It is recommended to use oxygen probes with thermocouples of type S or with no thermocouple. In the latter case, a separate furnace thermocouple, which is mounted close to the probe, has to be used for the calculation of the C-potential. There are also low-cost probes – mainly with zirconia element or ball – equipped with type-K thermocouples. Because of the ceramic design, it is possible to use light-gauge thermocouples. Through ceramic leakages, furnace atmosphere reaches the thermocouple. This gives rise to lower accuracy and a reduced lifetime. For this reason, probes with K-type thermocouples are only conditionally recommendable.

 Fig. 3. Lambda (L) probe

Uncertainty of Measurement of the O2-Probe in the Field

When handling the O2-probe in the field, the following points should receive attention:

• The O2-probe is exposed to mechanical and thermal stress. Since the ceramic is not breakproof, the probe is potentially at risk. In particular, the unbalance of circulating fans can decrease the probe lifetime. Therefore, mounting of the probe near such equipment should be avoided. Furthermore, due to thermal sensitivity, it is time consuming to exchange the O2-probe under process conditions.
• Probe leakages increase over the course of time. The increase of the porous area and hairline cracks in the ceramic result in a growing flow rate of reference air. Aside from the already-described effect on the probe voltage, the temperature of the O2-probe thermocouple is also affected.
• Materials such as washing agents, oils and covering paste evaporate in the furnace and precipitate on the outer platinum electrode. This yields a falsified measurement and a reduced probe lifetime. To remove residues from the probe element, an extended purging of the probe is necessary. One of the most known problems is sooting of the outer electrode, particularly when the process is controlled near to the soot limit. The sooting of the electrode yields a falsified measurement too. In this case, the only help is a cyclic purging with air. Keep in mind that the valve for air purging has to be mounted near the probe. Long pipes produce long dwell time after the purging. The valve must close well and should not jam.
• At temperatures above 1100°C (2012°F), the electron conduction in the zirconia is increasing and cannot be neglected anymore.[2][3] Decarburizing processes at high temperatures (e.g., decarburizing of cast iron) can be controlled with the oxygen probe if, and only if, the electron conduction effect is compensated.
• If chromium-nickel steels are used as a protection tube, microscopic analysis indicates that, near the zirconia tip, the chromium in the tube dissolves away in the furnace. Soot particles and oxygen cling to the porous surface. The soot on the probe causes faulty measurements. The oxygen induces increased probe reaction time and long relaxation time after probe purging.

 Fig. 4. Principle of the L-probe

L-Probe Construction & Functionality

Due to the aforementioned problems, attempts were made to place a smaller oxygen-measurement cell outside the furnace. The use of oxygen-conductive solid electrolytes (e.g., zirconia ceramic) for combustion control in the automotive industry has led to miniaturized oxygen probes, namely the Lambda probe (L-probe). As a result of mass production, a robust and economically priced sensor is available today (Fig. 3).

The functionality of the L-probe and O2-probe is basically the same. However, there is an essential difference in construction and measuring setup. The form of the L-probe is similar to the O2-probe – one-sided closed pipe (Fig. 4). The ceramic is also the same. The probe ceramic surface is covered with a microporous platinum layer inside and outside. These two platinum layers are the electrodes of the measuring cell. The outside platinum layer is protected with a highly porous ceramic layer. In order to enable the movement of oxygen ions through the ceramic, a nickel heating element heats the ceramic inside the L-probe. A rapid heating of the probe is possible because of the PTC-characteristic.

 Fig. 5. Temperature profile of a controlled and an uncontrolled L-probe

A few points should be considered for an accurate handling of the L-probe. If one arranges measurements with L-probes in air, a voltage of 8-15 mV instead of 0 mV is indicated. The reason for this is an external thermocouple effect. Each L-probe is heated differently because of minor differences of the heating elements. For the heating of the probes, a voltage of 12 V is provided, and the temperature of the ceramic takes on values from 500-600°C. In order to accurately measure oxygen and accurately calculate C-potential based on this measurement, the aforementioned characteristics have to be incorporated as correction factors in the calculation.

For an accurate calculation, the control of the L-probe ceramic temperature is very important. Therewith, different perturbations such as changing of the flow rate, ambient temperature or gas composition are eliminated. Figure 5 shows the temperature profile of a controlled and an uncontrolled L-probe by equal change of process conditions. In this test, the modification of gas composition and flow rate has resulted in a 40°C temperature change of the uncontrolled probe. In contrast, the temperature of the controlled probe has hardly changed, and the adjustment to the primary value occurred after a short time.

The power supply NTV44P (Fig. 6) was designed by MESA Electronic GmbH for the control of the L-probe ceramic temperature at a constant value. Flow rates higher than 50 l/h are now also possible by use of this power supply. As a result, the dead time (time which the furnace gas needs to pass the probe) can be significantly reduced. For a gas-sampling pipe with nominal length of 720 mm and a gas flow rate of 20 l/h, the dead time is 17 seconds. It decreases to less than 6 seconds if the flow rate is 60 l/h. Both times are harmless in the process of carburizing. Furthermore, the working temperature is reached much faster if the L-probe ceramic temperature is controlled instead of heating the L-probe with constant voltage.

It should also be noted that for certain oxygen partial pressures in the furnace, the O2-probe and the L-probe give different voltages. To control the C-potential with the L-probe you need either a controller that can calculate and control the C-potential using the L-probe voltage and correction factors (Fig. 7) or a measurement transducer that is able to do the same calculation or to transform the L-probe signal to the O2-probe signal (Fig. 8). One of the many options of the power supply NTV44P is the transformation of the L-probe signal to the O2-probe signal.

 Fig. 6. MESA Electronic power supply NTV44P

Gas-Sampling Fitting with L-Probe for C-Potential Control

One of the most important points connected to the correct handling of the L-probe for measurements of furnace atmosphere is the appropriate construction of the gas-sampling fitting (Fig. 9). The fitting is constructed such that it can be mounted without mechanical work in the entry of an O2-probe. A ceramic pipe guides the furnace gas up to the outer furnace wall. On the steel protection pipe of the gas-sampling fitting, a mark shows the end of the inner ceramic tube. The remaining part of the steel protection pipe serves as the cooling section. The following is achieved by this construction:

• The hot reactive gas and the steel parts of the gas-sampling fitting are not in contact. Temperature decreases and with it sooting of the fitting is avoided.
• A back reaction in another equilibrium state is impeded because of the sufficient thermal isolation of the gas up to the cooling section.
• The gas-flow velocity is increased by reduction of the width in the ceramic tube.
• The gas is thermal isolated up to the furnace outer wall only.
• At the furnace outer wall, the flow velocity is reduced severely because of the larger width.
• When the gas leaves the furnace isolation and enters the cooling section, a fast cooling and a refreezing of the gas components occurs. A back reaction in another equilibrium state is thereby impeded. Another equilibrium state of the gas would falsify the C-potential calculation and produce sooting of the sampling pipe.

 Fig. 7. Controller to correct for differences between L-probe and O2-probe

The measurement with the L-probe takes place outside the furnace. The internal temperature of most L-probes is between 550 and 600°C. These temperatures are still high enough to sustain the conductivity of the zirconia oxygen ions. In contrast to the O2-probe, a lower temperature is sufficient due to the construction difference. The temperature in the measurement chamber is low enough that the back reactions of the measuring gas are excluded.

 Fig. 8. Transducer used to transform the L-probe signal to O2-probe signal

Both probes, O2-probe and L-probe, qualify very well for control of the carburizing process. In addition, compared to the O2-probe, the L-probe possesses some noteworthy advantages:

• The L-probe is very robust and was designed for operating conditions with changing temperatures and mechanical impacts (e.g., the exhaust pipe in cars). The L-probe is insensitive to mechanical vibrations. Therefore, it can be mounted closer to the fans than the O2-probe.
• Since the L-probe is outside the furnace, it can be replaced during the process in only few minutes.
• Contrary to the O2-probe, the L-probe can be used in processes with temperatures up to 1600°C (2912°F) without signal falsification. For these types of processes, however, the cooling section has to be designed properly.
• Because of the design of the gas-sampling fitting, the L-probe is not in contact with the high temperature of the furnace. Moreover, contrary to the O2-probe, the ceramic of the L-probe is not affected by the temperature variation of the furnace. These are the essential reasons for the higher lifetime of the L-probe compared with the O2-probe. Based on L-probe carburizing applications over the last 20 years, its lifetime is double that of the O2-probe in the same process. In processes in which the O2-probe is exposed to a high thermal stress, the lifetime of the L-probe is many times that of the O2-probe.
• One important advantage of the L-probe is the price for the initial purchase as well as for the repeat order. When replacing an O2-probe with an L-probe, the initial price is higher because an NTV44P power supply is needed to transform the signal. Taking into account the longer life and lower follow-up costs for the replacement and repairing of the L-Probe respectively, however, a return of investment is achieved after the first breakdown of the O2-probe.

 Fig. 9. Appropriate construction of the gas-sampling fitting is shown

Conclusion

L-probe experience over the last 20 years shows that these probes are very well qualified for the control of C-potential. The same can be said for the O2-probes. There are pros and cons for both. The evident benefits of the L-probe are the lower price and higher lifetime. Consequently, cost savings without loss of quality is achieved using the L-probe. IH

For more information: Contact Dr. Džo Mikulovic, Director, MESA Electronic GmbH, Leitenstr. 26 82538 Geretsried – Germany; tel.: +49 (0) 8171-76930; e-mail: dmikulovic@mesa-international.de