Over the years, instances of carbon pickup resulting in subtle hardness variations in 0.30% and 0.40% carbon steels after neutral hardening in an endothermic gas atmosphere have been reported and often result in rework or scrapped loads. It’s time to discuss what causes this and what steps can be taken to prevent it. Let’s learn more.


A Real-World Example

Take a recent case of the heat treatment of extremely thin parts, approximately 1 mm (0.040 inch) thick, manufactured from 4130 strip stock. The load consisted of 740 pieces, which were well spread out in a basket and covered with a screen to retain heat during transfer to quench. The net load weight was only 1 kg (2.2 pounds). The parts were run in an integral-quench furnace at 870°C (1600°F), held for two hours at temperature with an oxygen probe setpoint of 0.30% C and then oil quenched. 

    During microstructural inspection for decarburization, the parts were found to have picked up carbon at the surface. A thin “case” was developed, which was confirmed by microhardness testing (Table 1) and metallurgical analysis (Fig. 1). Optical microscopy revealed subtle differences in the martensitic microstructure between the near surface, within the first 0.20 mm (0.008 inch), and the core.

    The most likely explanation advanced was that the component part picked up additional carbon from the furnace atmosphere. A review of the furnace charts, however, confirmed that the carbon setpoint was maintained throughout the cycle and that the austenitizing time and temperature were correct. The temperature uniformity of the furnace was checked and found to be within ±5.5°C (±10°F). The endothermic gas atmosphere had remained consistent as well with an incoming dew point of +6°C (+43°F).

    While the calculation of carbon from the millivolt signal of an oxygen probe depends on a multitude of factors, including the age and condition of the probe and burn-off history, a variance of 5 mV (a conservative estimate) equates to only a 0.03% carbon change given the above process parameters.

    If we consider hardness variation as a function of carbon content in the material, we find that a nominal 0.30% C steel quenched to 99.9% martensite (Fig. 2) would have a resulting hardness of 50 HRC, while a 0.32% C would be 51 HRC and a 0.34% C would be 52 HRC. Hence, a subtle variation in the carbon content of the steel itself could explain this phenomenon.


Sources of Hardness Variation

In order to fully investigate a hardness variation such as the one reported, we must consider all of the factors that may have been involved.


Chemical Composition/Variations in Carbon Content

Variations in steel composition (and resultant hardenability) can occur in component parts. Each steel type has an acceptable range for each of the elements it contains. In the case of 4130 material, the range of allowable carbon is 0.28-0.33%. Variations in carbon can arise from micro-chemical segregation effects within a given heat, heat-to-heat variation in carbon content and from carbon pickup or decarburization experienced during the heat-treatment cycles. Relying solely on the material certification sheet may be misleading. It is always best to analyze the material in question so that the actual carbon content can be used in predicting results.



This can be due to insufficient temperature and/or insufficient time at temperature. The result is that the component or some region within a component never exceeds the upper critical temperature on heating for a sufficient enough time during the austenitizing portion of the hardening cycle. For example, the lower critical temperature on heating (Ac1) for 4130 steel is approximately 749°C (1380°F), while the upper critical temperature (Ac3) is approximately 802°C (1475°F). The normal austenitizing temperature specified for this grade is 870°C (1600°F). If the austenitizing is insufficient, portions or all of the cross section of the part will not be martensitic upon quenching.


Furnace Loading

In an effort to maximize productivity, we run the risk of overloading the furnace or, in the case of certain continuous furnaces, piling parts atop one another in large clumps. In these instances, there is a high probability that some of the parts will be underheated. One way to know this is happening is to have a database of microstructures (and corresponding hardness values) that represent properly austenitized and underheated product based on empirical data gathered over time or designed experiments.



Exceeding the recommended austenitizing temperature for a given steel is never a good practice, especially for steels that have carbon levels where significant amounts of retained austenite may form in the hardened microstructures. Retained austenite is softer than martensite, and increasing amounts will result in lower hardness values. Excessively high temperatures can also promote grain growth. If this occurs prior to quenching, the hardenability will increase and the resultant microstructures and hardness levels will vary accordingly. This is particularly important because of current emphasis on increased throughput by reducing cycle times through the use of higher austenitizing temperatures.


Transfer Time from Heat to Quench

We need to consider the upper critical temperature on cooling (Ar3) and the lower critical temperature on cooling (Ar1) as well as the TTT (Time-Temperature-Transformation) and/or CCT (Continuous Cooling Transformation) diagrams. These will provide guidelines on the temperatures that must be maintained prior to quenching. The cooling rate determines if a fully martensitic microstructure will develop and the corresponding hardness levels that will occur.

    If the components are held in the Ar3 – Ar1 inter-critical range or cooled insufficiently through it, pro-eutectoid ferrite will form. If the cooling rates are insufficient, various amounts and combinations of non-martensitic transformation products (NMTPs) with correspondingly lower hardness levels will result. Figure 2 is an invaluable tool for estimating the relative hardness that can result, and one sees that for a given carbon value, say 0.30% C, the hardness can vary from 50 HRC (99.9% martensite) to 37 HRC (50% martensite).


Summing Up

The challenge in talking about subtle changes in hardness is that the variability can be induced by so many factors. For every instance discussed here, the reader can likely think of examples from their experience that were caused by other factors. It is the discussion that is important, however. It brings to light the fact that, whether large or small, changes in hardness need to be explored and the root cause determined so that our processes are kept under control. IH



1. Mr. Craig Darragh, AgFox LLC, technical contributions and private correspondence.

2. Mr. James Oakes, Super Systems, Inc., technical contributions and private correspondence.

3. Practical Data for Metallurgists, 17th Edition, The Timken Company.