Atmosphere vs. Vacuum Carburizing

Transcription

1 Atmosphere vs. Vacuum Carburizing Report No Research Team: Lei Zhang (508) Mei Yang (508) Yingying Wei Richard D. Sisson Jr. (508) (508) Project Statements Objectives The surfaces and sub surfaces of steels are the most susceptible regions to the initiation of fatigue failures. Hence, strengthening the materials at or near the surface enhances the fatigue life of machined parts. Carburization is a widely used surface hardening processes in industry. Carburization of steel parts can be achieved in several ways including gas carburizing and vacuum carburizing. The objective of this project is to perform a comparative microstructural study of the properties of steel parts that have been carburized using gas as well as vacuum techniques. Strategy To address the goal of this project, the following strategies are adopted: Perform literature review on both carburization techniques Identify materials to be tested Test for performance characterizations o Residual stress measurements o Hardness o Case depth o Microstructures o Retained austenite The project work is divided into following tasks. 1

2 Task 1- Literature Review o Low pressure carburizing (LPC) process o Modeling the LPC process o Atmosphere carburization process o Metallurgy of 8620 mod, 5120 mod, 4320 and 9310 Task 2 Modeling o Model the LPC and GC process and compare with experimental results Process parameters and carbon profile measurements Task 3 Experiments o Carburization Gas and Low pressure Perform carburization at participating member company facilities Conduct Metallographic Characterization - does it meet the specification above? Task 4 Microstructural Characterization o Optical microscopy o Carbon profile measurements o Microhardness profile measurements o Surface hardness measurements o SEM with EDS o Retained austenite measurements by XRD 2

3 2.0 Executive summary At least half of all mechanical failures in machined parts, such as gears and shafts, are due to fatigue [8]. Therefore, the fatigue performance of materials is of high importance for parts which are subject to cyclic loading. The most susceptible region to fatigue failures are surfaces because the fatigue cracks are generally initiated at or near the surface. Strengthening of the material at or very close to the surface is quite important in enhancing fatigue life. This can be achieved by various mechanical and/or thermo- chemical surface treatments. Besides strengthening the surface layers, the creation of residual compressive stresses at the surface will also hinder fatigue crack initiation. Both effects can be realized by the application of suitable case- hardening heat treatments. Among these, carburizing is a widely used industrial practice to improve the fatigue performance of machined parts [3]. In a carburization process, the steel part surfaces are subjected to an atmosphere with a high carbon potential [3]. During carburizing, carbon diffuses into the steel increasing the carbon concentration near the surface also creating compressive residual stresses. During service, those stresses can counteract the applied tensile stresses and therefore improve the fatigue performance. Carburization can be performed using a variety of processes: liquid salt bath, solid carbonaceous materials, gas atmosphere, vacuum, and plasma carburizing. Among these processes, gas carburizing is the most widely used process in the industry. However, vacuum carburizing is gaining acceptance because of its environmental friendliness and its potential capability to create a uniform carburized layer with no intergranular oxidation and minimal distortion in carburized parts due to the gas cooling [4]. Many factors can affect the fatigue resistance of carburized parts. These include case depth, residual stress, surface hardness, microstructure, carbide formation, intergranular oxidation (IGO), microcracking, inclusions and retained austenite. The properties are discussed in the report. 3

4 3.0 Achievements 3.1 Simulation Software Development CarbTool C is a simulation program developed by CHTE for the calculation of the carbon concentration profile during the carburizing processes. The solution algorithm used in CarbTool C is based on the finite difference method (FDM), and the code is developed using Microsoft Visual C++ in Window OS [2]. The output of CarbTool C is the carbon concentration profile in the steel part Diffusivity improvement Diffusivity data of 10XX, 51XX, 86XX, 48XX series steels in the current version of CarbTool were experimentally measured by O. K. Rowan [3]. Diffusivities of other alloys were built in based on the experience reference data. The comparison of different alloys diffusivity is presented in Figure 1. These curves are almost parallel to each other, so the diffusivity of one alloy should be proportional to that of another alloy. For example, 10XX and 51XX can be expressed as k in the following equation; k is dependent to carbon concentration. D 51XX D 10XX = k 51XX Carbon Diffusivity (cm2/s) 2.8E E E E E E Carbon ConcentraHon (wt.%) "10XX" "51XX" "86XX" "48XX" Figure 1 Carbon diffusivities 4

5 From the flux balance condition at the steel interface and the continuity equation of the mass accumulation within the solid, the rate at which the total flux over the carburizing time is!! C(x, t) dx =!!! Jdt!! Where x 0 is the initial condition of the interface between the two components of the diffusion couple, x is the depth beyond which no concentration gradient exists, and t is the diffusion time. Assuming the isotropic media, based on the Fick s first law, by equating the above two equations, the expression of diffusion coefficient from the carbon profile can be derived [2], J(x 0 ) = D(x 0 ) dc dx (x 0,t) D(x 0 ) = dc(x 0,t) dx Based on this equation, two carbon profiles, treated in the same carbon potential and temperature but different time, are required. There are two parts;the negative inverse of the slope of any position on the carbon profile and the differentiation in terms of time integrated area under the corresponding section [2]. To develop the diffusivity of 93XX, 43XX, and Pyrowear 53. Samples of each alloy was treated in carbon potential of 1.1 wt% at three different temperatures, at each temperature, samples were kept for 1 hour, 2 hours and 3 hours separately. The carbon profile was measured by Optical Emission Spectrometer (OES). Each of the points take the average of three or four measurements. Cubic spline interpolation and polynomial fitting algorithms in MATLAB was applied to smooth the plot to get the gradually changed slope and fit the curve. Ultimately, diffusivity equation was developed as a function of carbon concentration and temperature. 1 d dt x 0 x Cdx 5

6 3.1.2 Gas carburizing module improvement In the gas carburizing module, the process is divided into three stages: the boost step, transition step and diffuse step. Each step can be conducted at different temperatures, carbon potentials, and times. Figure 2 (a) presents the original input interface. In this mode, the parameters of the transition step were fixed as temperature and carbon potential gradually decreased. However, in real cases industry might decrease either temperature or carbon potential first and then decrease the other parameter. Thus, to make the modeling process more flexible a new algorithm was developed, shown in the Figure 2 (b). At each transition point, we can input the temperature, carbon potential and time. (a) (b) Figure 3 Gas module of original (a) and new version (b) Additionally, microhardness profiles can be predicted in the latest CarbTool C. Microhardness changes mainly with carbon concentration, a series of experiments have been applied to investigate the correlation between the microhardness and concentration for specific materials. The polynomial fit method has been used to express the correlation. Then the results has been integrated to CarbTool C as presented in Figure 4. 6

7 Figure 4 Latested version CarbTool C simulation results Process control surface carbon saturation Vacuum carburizing is characterized by the rapid saturation in carbon in the steel surface layer [4] due to the high carbon flux. Vacuum carburizing is usually processed as a series of boost and diffuse pulses. Such a partitioning of the carburizing process is more advantageous because the solubility of carbon atoms in austenite is limited and in excess of carbon atoms a solid carbonaceous layer is being created on the surface of the steel [5,6]. During these process, the surface carbon content should not exceed solubility in Austenite. This is more relevant in alloy steel because of the high chemical affinity of carbon to the alloying elements. In plain carbon steel, cementite will start forming at 1.3 wt% carbon (point A), as shown in Figure 4 (a). In alloy steel, the saturation point is reached at less carbon content approximately at 1 wt% carbon (point B), shown in figure 4 (b). 7

8 . (a) (b) Figure 4 Fe- Fe 3 C phase diagram (a) and 93XX isopleth diagram (b) 8

9 CarbTool, like most current commercial market simulation software, regard the carburizing issues based on the non- reactive diffusion assumptions, which do not consider the formation of carbides on the steel surface [4]. CarbTool outputs the surface carbon content as a function of time, shown in figure 5. In the vacuum carburizing model, with increasing boost time, the surface carbon concentration increases uninterrupted. In the real case, this is impossible. Therefore in simulation it is necessary to consider the maximum solubility of carbon in austenite, otherwise, the carbides of alloying elements will appear on the surface. γ + Fe! C γ Figure 5 Output of carbon concentration as the function of time in CarbTool of single boost and diffuse vacuum carburizing The solution of this issue will be adding an austenite solubility limit as a function of temperature for each alloy onto CarbTool, when the surface carbon exceed the maximum solubility, a pop out window will show up and give a warning showing that the limit is reached and the interval time should be shortened. 9

10 To determine the maximum carbon solubility in austenite, ThermoCalc was utilized to draw the diagram of mole fraction of cementite for alloys. The diagram for 93XX, 86XX, 51XX, 43XX and 10XX was drawn and presented in Figure 6. The diagram was calculated at 927 C (1700 F) and the pressure at 10 torr. These conditions are typical of industrial vacuum carburizing. The carbon solubility of 93XX is the least at approximately 1.0 wt % carbon. This value is agreed with point B in 93XX isopleth diagram. 93xx 51xx 43xx 86xx 10xx Figure 6 Mole fraction diagram of AISI 9310, 8620, 5120 and The above diagram can assist with designing the vacuum carburizing process. Due to the austenite carbon saturation, the interval time of each boost stage on 93XX may be less than the other four alloys. 10

11 3.2.1 Carburizing materials and objectives (1) Materials The untreated alloy composition of 9310, and 5120 provided by Timken is showed in Table 1 measured by OES at CHTE. Table 1 Chemical composition (wt %) of raw materials by OES C Cr Fe Mn Mo Ni P S Si Bal Bal Bal Bal (2) Conditions The general heat treating processes for each alloy are listed in Table 2. Some of the detailed process parameters will be designed in cooperation with the member companies and simulated for comparisons by CHTE. Table 2 Process Conditions Carburizing Quenching Tempering Cryo treatment Post process Vacuum (C6H12) Gas Oil Oil All alloys All alloys 9310/ /4320 IGO removing (3) Specification To compare the properties for both gas and vacuum carburizing process, samples should meet the specification well: Case depth: (0.889mm) at C=0.35 wt. % Surface carbon: 0.80 ± 0.05 for 8620 and ± 0.05 for 4320 and 9310 Retained austenite: 20~30% 11

12 To prepare the samples, trials have been run at our member companies. Initially the recipe is calculated by CarbTool. The experimental results may not meet the simulation results well. Based on the experimental results, a series of analyses were made to improve the CarbTool. Then an updated recipe could be provided. Finally, we got the recipe for both gas and vacuum carburizing. They can meet the specification well. 3.3 Gas carburizing processes Gas carburizing In this process, test samples were gas carburized using boost, transition and diffuse method. For 9310, the heat treating process is to heat up to 1700 F,hold 330 minutes in endothermic gas at a carbon potential of 1.0%, and then hold it at 1700 F for 90 minutes in carbon potential of 0.7% as transition process. The diffuse process keeps samples at 1550 F for 30 minutes at a carbon potential of 0.7%, then it is quenched to room temperature. Figure 7 presents the process schematic. Cp=1.0 Cp=0.7 Cp=0.7 Figure 7 the promoted gas carburizing process 12

13 CarbTool was used to simulate the process. For each step of the process, the temperature and carbon potential parameters are set separately. The simulation results are presented in Table 3, Figure 8 and Figure 9. Table 3 Simulation parameters for Cycle Boost (1700F) Transition (1700F) Diffuse(1550F) Steel Carbon potential Time Carbon potential Time Carbon potential Time (%) (min) (%) (min) (%) (min) Figure 3 presents the trend of the surface carbon with temperature and carbon potential. Different color curves show different steps, red for boost at 1700, yellow for transition at 1700, and blue for diffuse at For each alloy, the carbon concentration may differ (refer to Table 4). The surface carbon content is also shown and meets the specification well. Figure 8 The surface carbon concentration during the process 13

14 CarbTool also provides the carbon profile simulation presented in Figure 4. For each step there is a corresponding curve; red for boost, yellow for transition and blue for diffuse. After the boost step, the surface carbon content is high due to the high carbon potential, while the case is shallow, that will be deeper after the transition and diffuse steps. There is no difference after transition and diffuse. Figure 9 The carbon profile for steel in cycle # Carbon concentration profile The comparison between the experimental measurement and the simulation is shown in Figure 10. The 9310 sample does not fit the profile of the prediction result well. Similar with previous cycle, the surface carbon and the case depth value are both too low. Surface carbide formation might be the prime reason for the low surface carbon concentration and case depth. The other three steels meet both the surface carbon and case depth specification. 14

15 Figure 10 The comparison of the experimental and simulation carbon profile As presented in Table 4, for 4320, 8620 and 5120, the surface carbon and case depth all meet the specification well which demonstrate that the prediction of CarbTool is verified and recipes for this round can be used for the further sample preparation. Table 4 the surface carbon and case depth measurement Surface Carbon (wt.%) Effective Case Depth (mm) Target 0.7± ±0.05 Experiment Simulation Target 0.89±0.05 Experiment Simulation

16 3.3.3 Microhardness profile The microhardness (HV) profile, measured for all samples, is presented in Figure 11. With the results, correlation between microhardness and concentration is studied. CarbTool C predicts the microhardness profile with the correlation. Figure 11 The microhardness profile for gas carburizing process 16

17 3.4 Vacuum carburizing processes Vacuum carburizing tentative cycle The vacuum carburizing process was conducted at Surface Combustion using cyclohexane (C 6 H 6 ) as the atmosphere. In each carburizing process, test coupons were used in tentative recipes first. After being revised by simulation, the final recipes will be applied. (1) Measurement results of tentative cycle Figure 12 is the comparison between experimental carbon profile, modeling carbon profile and hardness profile. In each diagram, the experimental carbon profile agreed with simulation results very well. From the hardness profile, the same carburized depth was reached. Figure 12 Experimental vs. simulated carbon concentration profiles 17

18 (2) Effects of process parameters on flux Previously, we used 8620 test coupons in all three recipes designed for these four alloys, and used the flux calculated from 8620 as boundary condition to determine the processes. However, it turns out the flux varies for different alloys in the same process. Flux is affected by four factors: alloy composition, gas partial pressure, temperature and surface condition. In our case, the pressure and temperature were kept constant in the previous tests. All the materials were cleaned with ultrasonic method before test. Therefore, the variation in flux must be due to alloy composition Final vacuum carburizing cycle With the calculated carbon flux, the recipe for the final cycle contains two boost steps, two diffuse steps, at a temperature of 1700, cooling down to 1575, holding for a short time, then quenching as presented in Figure 13. Figure 13 The vacuum carburizing process The recipes details are based on the CarbTool software to meet the specification of case depth and surface carbon concentration presented in Table 5. 18

19 Table 5 Recipe for the next vacuum carburizing cycle Specification Surface carbon: 0.70 ± 0.05 case depth: 0.889mm at C=0.35 wt. % Carbon flux (g/cm2/s) 1st Boost (min) 1st Diffuse (min) 2nd Boost (min) 2nd Diffuse (min) Cooling (min) Hold (min) E E Specification Surface carbon: 0.80 ± 0.05 case depth: 0.889mm at C=0.35 wt. % Carbon flux (g/cm2/s) 1st Boost (min) 1st Diffuse (min) 2nd Boost (min) 2nd Diffuse (min) Cooling (min) Hold (min) E E OES was performed for carburized samples and is presented in Figure 14 and Table 6. In the carbon profile of each alloy, the surface carbon result fit well for 5120 and 4320, while it s a little shallower for Case depth for each alloy is smaller than the specification. For 9310, both the surface carbon and case depth are much smaller than the specification; carbides are the most likely explanation and further investigation is underway. Table 6 the comparison of target and experiment results Surface Carbon (wt.%) Target 0.7± ±0.05 Test result Effective Case Depth (mm) Target 0.889±0.05 Test result

20 (b) Figure 14 Experimental vs. predicted carbon concentration profile 20

21 3.4.3 Microhardness profile The microhardness profile is also measured for vacuum carburizing process. Figure 15 The microhardness profile for gas carburizing process 3.5 Experimental results comparison between vacuum and gas carburizing samples The experimental carbon profiles resulting from the gas carburizing and vacuum carburizing are presented in Figure 16 and Table 10. The surface carbon concentrations measured on the gas carburized samples meet the specification very well, while the surface carbon concentration values measured from vacuum process samples are all lower than specification. 21

22 Figure 16 the comparison between gas and vacuum carburized samples After several cycles, with the exception of 9310, both recipes can produce samples with required case depth value. For 9310, carbon profiles from the two separated cycles match each other well but cannot match the simulation results. Table 7 the comparison of current gas and vacuum carburizing results Surface Carbon (wt.%) Effective Case Depth (mm) Target 0.7± ±0.05 Gas Vacuum Target 0.89±0.05 Gas Vacuum

23 3.6 Retained austenite Retained austenite plays an important role in the fatigue resistance of carburized steels. Also, excessive amounts of retained austenite may substantially lower the hardness of the material and result in a decreased resistance to fatigue crack initiation. The percent retained austenite was experimentally determined using X- ray diffraction (XRD) techniques. The measurement conditions were as follows: Target: Cr, wavelength: λkα =2.291Å. Figure 17 presents the XRD pattern of 4 alloys. Counts α (110) γ (111) γfe2o3 (115) γfe2o3 (044) γ (200) α (200) Gas 9310 LPC 9310 γ (220) Fe3C (110) 2 theta degree (a) 3000 Counts γ (111) α (110) γ (200) γfe2o3 (044) α (200) Gas 5120 LPC 5120 γ (220) Fe3C (110) 2 theta degree (b) 23

24 α (110) Gas 4320 LPC 4320 Counts γ (111) γfe2o3 (115) γ (200) α (200) γ (220) Fe3C (110) 2 theta degree (c) γ (111) α (110) Gas 8620 LPC 8620 Counts γfe2o3 (115) γfe2o3 (115) 1000 γ (200) α (200) γ (220) theta degree Fe3C (110) (d) Figure 17 Retained Austenite measurement by X- RD 24

25 Based on the following equations, the fractions of retained austenite of four alloys are displayed in Table 7. I =!!!!!!"!!!!!!!!!!!! F! P!!!"#!!!!"#!!!"#!!!!!!! (1) R =!!! F! P!!!"#!!!!"#!!!"#! e!!! (2)!!!! =!!!!!!!! (3) c! =!!!!!!! (4) Table 7 Retained Austenite calculation Gas 14.7% 23.5% 15.8% 17.3% LPC 5.6% 31.8% 17.5% 5.9% As Table 7 presents 4320 and 9310 alloy has more retained Austenite after gas carburizing process than LPC process. While 5120 steel has more retained Austenite in the LPC. For all four alloys, there is iron oxide (III) existing on the surface after gas carburizing process. As discussed previously, the IGO will form during gas carburizing process. IGO cannot be found on the LPC carburized samples. 25

26 3.6 Tempering of carburized samples. 1018, 8620 and 4140 samples were selected to investigate the tempering process for carburized alloy. During tempering, tempered Martensite forms with the precipitation of carbides. Both temperature and time determine the evolution of the microstructure. In the present work, the effects of tempering on the hardness profiles in quenched and tempered carburized steels are experimentally investigated. The tempering and process parameters was presented in Table 8. Table 8: The tempering experiment design for steel 1h 2h 2h+2h 4h 9h 300 (149 ) 350 (177 ) 400 (204 ) Holloman- Jaffe equation [7] is used to present the relationship between the hardness of a steel and the tempering temperature and time. In the equation, P is the Holloman- Jaffe parameter, T is the temperature in Kevin, t is the time in hours. And C is a constant that is characteristic of the steel alloy composition. P =!(!"#$!!)!""" (1) The microstructure of the as- quenched AISI 1018 steel is comprised of a Martensite case with a ferrite and pearlite core. In the transition region, there is a mixture of Martensite, ferrite and pearlite as seen in Figure 2. X- Ray Diffraction revealed no retained Austenite for the 1018 steel. Both 8620 and 4140 steel contains approximate 20% retained austenite at the surface of the case. 26

27 The microstructure of the as- quenched AISI 1018 and 8620 steel is comprised of a Martensite case with a ferrite and pearlite core. In the transition region, there is a mixture of Martensite, ferrite and pearlite as seen in Figure 18. (a) Original 1018 (c) Original 8620 (b) Tempered at 177 for 4h 1018 (d) Tempered at 177 for 4h 8620 Figure 18 The microstructure of 1018 and 8620 before and after tempering process 27

28 Figure 19 and Figure 20 present microhardness changes with depth for 1018 and 8620 alloy. In each figure, tempering time is 1h, 2h, 4h and 9h. The carbon concentration was plotted with depth as talked previously. The correlation between microhardness and carbon concentration was plotted as in Figure 21 and Figure MICROHARDNESS HV As quenched 1 hour 2 hour 4 hour 9 hour DEPTH ΜM Figure 19 The temperature effects on the hardness of carburized AISI 1018 samples 800 MICROHARDNESS HV As quenched 1 hour 2 hour 4 hour 9 hour DEPTH ΜM Figure 20 The time effects on the hardness of carburized AISI 8620 samples 28

29 800 Microhardness Hv As quenched 1 hour 2 hour 4 hour 9 hour Carbon Concentra on wt% Figure 21 The time effects on the hardness of carburized AISI 1018 samples MICORHARDNESS HV hour 2 hour 4 hour 9 hour As quenched CARBON CONCENTRATION WT% Figure 22 The time effects on the hardness of carburized AISI 8620 samples 29

30 Figure 23 and Figure 24 present the tempering effects on the carburized AISI 1018 and 8620 from the Holloman Jaffe equation. The data points are selected from Figure 21 and Figure 22 when carbon concentration is 0.7, 0.6, 0.5 and 0.4 wt%. Hv/T as a function of log(t) is plotted then. The constant C in the equation could be obtained for different process. The Holloman- Jaffe approach will aid the prediction of the hardness of carburized, quenched and tempered steel. Hv/T Log(t) C=0.7 C=0.6 C=0.5 C=0.4 Figure 23 Holloman Jaffe plot for 1018 with different carbon concentration Hv/T C=0.7 C=0.6 C=0.5 C= Log(t) Figure 24 Holloman Jaffe for 8620 with different carbon concentration 30

31 The Figure 25 presents the Holloman Jaffe constant changes with carbon concentration linearly. With the constant, the microhardness for steel with different carbon concentration could be obtained combined with the temperature and time input. More experimental work is planned to fully develop these relationship Constatnt C Carbon concentra on wt% Figure 25 Holloman Jaffe constant vs carbon concentration for 8620 and 1018 steel 4.0 Summary CarbTool can be used along with experimental verification to development process parameters to meet the specification. Updated version CarbTool could predict microhardness profile. There is IGO existing on the alloy surface after gas carburizing, while not on the LPC samples. Holloman Jaffe plot can be used to predict the microhardness profile for tempering of carburized steel. 31

32 References [1] D. Glover, A Ball-Rod Rolling Contact Fatigue Tester, 1982, p 107. [2] O. Karabelchtchikova, and R.D. Sisson, Carbon Diffusion in Steels: A Numerical Analysis Based on Direct Integration of Flux, J. Phase Equilib. Diff., 2006, Vol 27 (No. 6), p [3] O. K. Rowan, and Richard D. Sisson Jr, Effect of Alloy Composition on Carburizing Performance of Steel, J. Phase Equilib. Diff., Vol 30 (No. 3) p [4] P. kula, R. Pietrasik, and K. Dybowski, Vacuum carburizing - process optimization, 13th International Scientific Conference on Achievement in Mechanical and Materials Engineering, 2005, Poland. [5] Ya. Baran, and J.L. Gyulihandanov, Vysokotemperaturnaya nitrotsementatsiya spietchenykh staliv vysokoy plotnosti, MITOM, 1986, Vol 9, p 8-10 (in Russian). [6] W. Grafem, and B. Edenhofer, Acetylene low pressure carburizing a novel and superior carburizing technology, Heat Treatment of Metals, 1999, Vol 4, p [7] Hollomon, J. and L. Jaffe, Time-temperature relations in tempering steel. Trans. AIME, : p [8] Yingying, W., Simulation, optimization and development of thermo-chemical diffusion processes, PHD Dissertation, Worcester Polytechnic Institute,