Many in the thermal-processing industry can recite dew-point conversion tables with closed eyes. Others may have certain phase diagrams committed to memory. Perhaps a select few are even familiar with Adolf Fick and Walther Nernst – pioneers in developing the Laws of Diffusion that we unknowingly, yet instinctively, rely on.

However, the even older Segal’s Law may be new to us all: “A man with a watch knows what time it is. A man with two watches is never sure.”

This proverb is more relevant today than ever. Technology in all industries is developing faster than ever. So how can it be that, equipped with nothing beyond an Alnor Model 7000 dew pointer, we knew exactly what our equipment’s carbon potential was and never questioned it?

Fast-forward 20 years, and we are surrounded by improved dew-point sensors, three-gas analyzers and a variety of other instrumentation. Yet we are less certain what the atmosphere truly is. How can this be? The overwhelming volume of information makes Segal’s Law a reality for many. This article will help you regain that lost confidence by detailing such instrumentation’s assumptions and its best use (maintenance, production and secondary verification).


Dew-Point Sensors

Dew-point sensors are simple in design, flowing an extracted sample across a dielectric ceramic cell that measures relative humidity. Changes in the sensor’s capacitance coupled with a measured temperature result in a calculated dew point.

Dew point can be converted to an atmosphere’s carbon potential (CP) if the furnace temperature (TR) is known. Conversion tables make several assumptions – notably fixed carbon monoxide (CO, of 20%) and hydrogen (H2, of 40%) concentrations consistent with stoichiometric endothermic atmospheres. With CO as the only carbon-carrying gas, a lower CO concentration results in lower CP. Furnace gas compositions vary, causing such tables to be more a close estimate than a science. As such, dew-point sensors are best used for troubleshooting or secondary verification purposes – not as the primary estimation of carbon potential.

A troubleshooting example could be establishing a “baseline dew point.” To perform this test, close all additive gases and introduce endothermic gas to a furnace. Once the atmosphere has stabilized, measure dew point at both the generator and furnace. In theory, these measurements should be identical (though in practice the furnace is always greater since air leaks from cracked burner tubes, door seals, etc. are the only source altering the atmosphere). Maintaining such logs can help you understand when such leaks worsen.

Measurement error is often introduced by poor sampling methods, with loose tube connections introducing air to the sample. Contamination (in the form of water or soot) is another concern and can damage the sensor. Fiber filters should be installed in-line and replaced when the media becomes darkened. Sample valves should be opened before connecting tubing to allow water and soot to “blow out” before connecting the sensor.

A sensor exposed to moisture (resulting in an unusually high and unresponsive dew point) can be “dried out” by sampling a (low-pressure) dry nitrogen source. This process may take 2-48 hours, depending on the amount of moisture. Dew-point sensors should be factory-calibrated on a regular interval to maintain accuracy.


Carbon Probes

Carbon probes paired with some type of control instrument are referred to as a carbon control system. In understanding such systems, we must first understand the probe itself. Probes are constructed of a hollow, sealed zirconia (ceramic) tube inside a protective metal sheath. Reference air is pumped into the probe’s interior, leading to an oxygen-rich atmosphere compared to the furnace’s oxygen-less reducing atmosphere outside. The specialized ceramic allows oxygen ions to flow between the lower and higher concentrations, generating a purely electromechanical millivolt output.

In this description, it is important to notice the omission of the word carbon. Carbon probes do not measure carbon. These probes actually measure oxygen and calculate carbon based on the Nernst equation (Fig. 3).

The Nernst equation demonstrates that the millivolt output is based on furnace temperature (TR), oxygen concentration inside the probe (%O2 ref, or 20.95%) and oxygen concentration outside the probe (%O2 fce, often in parts per billion). Further, this equation is only applicable to heat treating when CO is a known composition (approximately 20%).

The probe’s nonadjustable output begs the question: How can such a system be calibrated? The answer is that a carbon probe cannot be calibrated. Here enters the second part of the carbon control system, the control instrument. Control instruments calculate CP by measuring even more fixed variables: furnace temperature (TR), probe millivolts (E) and CO concentration. While specific algorithms may vary, the equations quickly become complicated (Fig. 4).

As an industry standard, “calibrations” are offered in the form of adjustment factors (COF). Theoretically, this represents changes in CO but in practice compensates for changes in both the probe and furnace. Such changes are slow in nature and should therefore be small and incremental. Exceptions may apply to deep-case, high-CP processes in which in-situ adjustment produces better metallurgy. As an example, a 0.10% CO change can result in a 0.02% atmosphere change, a large impact on calculated CP. As an industry idiom, we must shift away from the term “calibration,” opting instead for “verification checks” – adjusting COF based on metallurgical and other analyzed results.

When problems arise with probes, it is often a result of improper furnace placement (perhaps areas with poor circulation or erroneous gas mixtures not representative of the work area), lack of reference air, soot buildup between the probe’s ceramic and metal sheath (resolved by proper probe burn-offs), a crack in the ceramic (reducing the oxygen differential between the inside and outside) or an aging probe that requires replacement. Though beyond the scope of this article, many documents are available online to isolate the root cause of a failing carbon control system.

Carbon probes remain one of the oldest technologies in the industry and are still the center of production. Other instruments in our repertoire should be used to “dial in” to this system because it proves the most stable over time due to the low risk of sample contamination (compared to analyzers that extract samples), the robust design of the probe and slower “calibration” drift.


Infrared Three-Gas Analyzers

Infrared gas analyzers (IGAs) operate like dew-point sensors since they extract an atmosphere sample across a sensor, and they are also prone to the same contamination concerns. The dew-point sensor’s single capacitive cell is replaced with several nondispersive infrared cells – each uniquely configured for specific gas wavelengths. IGAs are extremely precise, and sampling from ferrous ports may further react the gas, skewing the results. As such, it is recommended to sample from a ceramic-lined port to avoid further reaction.

IGAs are complex in technology and what they measure but simple in operation. IGAs estimate CP by measuring CO, carbon dioxide (CO2) and methane (CH4) with the user entering furnace temperature (TR). Some IGAs include additional sensors to measure both oxygen (O2) and H2, further increasing the accuracy of the calculation.

Compared to these methods, IGAs measure more and make fewer assumptions, making them seem better suited for both production and troubleshooting. In reality, most commercial heat treaters run such a variety of relatively short processes that use of IGAs result in potential “false alarms.” In such processes, IGAs are best utilized as a support tool to a carbon control system, allowing the user to slowly adjust the COF. In certain deeper-case processes, constant COF adjustment may be ideal, resulting in more consistent metallurgy.

IGAs’ most appropriate uses are troubleshooting and maintenance. Similar to the “baseline dew-point” test, each furnace has unique IGA measurements across every temperature and CP combination. Following a dew-point test, open the additive gases. Once the atmosphere has stabilized, record the results along with furnace temperature, CP and COF. Developing such a record history allows maintenance to understand the “norm” for a furnace, quickly distinguishing between a door seal and burner-tube leak. Unfortunately, lack of such records complicates future IGA troubleshooting.


Carbon Resistance Testing

Carbon resistance testing estimates CP by measuring specific properties of a carburized ferrous material. The material is commonly a steel wire cut to a known length, coiled like a spring to reduce the overall size. Note that resistance naturally increases with length. Therefore, adjustments are required in the form of a length factor.

The coil is inserted into the furnace and processed for a predetermined amount of time, allowing the entire spring to obtain carbon equilibrium with the furnace. It is then extracted and cooled under atmosphere to retain metallurgical properties. Changes in the coil’s resistance coupled with a measured temperature result in a calculated CP.

Measurement error is often a result of improper processing or failure to prepare the coil before measurement. Manufacturers establish strict guidelines for processing, with longer processing creating carbides and altering the resistance. Following extraction from the furnace, the sample must be carefully prepped to remove contamination (oils, soot, etc.) from the coil’s contact points. Contamination alters the surface’s resistance, further skewing the results.

Carbon resistance testing’s purpose is solely secondary verification, which proves a carbon control system’s accuracy. Carbon resistance testers can be either factory- or field-calibrated to maintain accuracy.


Shim-Stock Analysis

Shim-stock analysis is similar to carbon resistance testing with a low carbon (~0.10% C) foil-like test coupon inserted inside the furnace. Rather than measuring resistance, carbon is estimated by a weight method or direct chemical analysis. The weight method compares the original and new weight of the coupon to estimate CP. While simple to perform, this approach can be inaccurate.

Chemical analysis uses a form of combustion to oxidize the coupon, measuring the infrared absorption and thermal changes to estimate CP. This method, while extremely accurate, can be expensive, and the results are often unknown until after completion of the process. This can be lengthy because many send samples to certified test laboratories. For all its disadvantages, chemical analysis remains the best option for carbon measurement and is recommended as often as feasible.



Carbon probes remain the most robust, economical solution for atmosphere control. Their results remain most consistent if COF is slowly adjusted using a secondary method such as IGA, carbon resistance (daily/weekly) or shim-stock analysis (weekly/monthly). The results allow the user to adjust COF to reflect the furnace’s true atmosphere. Dew-point sensors and IGAs are best suited as maintenance tools, with both requiring an established furnace baseline.

Perhaps if Segal had been alive today he would reword his law to state: “A heat treater with an instrument knows his carbon potential. A heat treater properly using more instruments understands his process.”