Chiang Mai J. Sci. 2013; 40(5) 909

Chiang Mai J. Sci. 2013; 40(5) : 909-922http://it.science.cmu.ac.th/ejournal/Contributed Paper

Application of Flow Analysis in Improving GasNitriding Conditions in a Vacuum FurnaceThanaporn Korad*, Mana Ponboon, Niphon Chumchery and John T.H. PearceNational Metal and Materials Technology Center, Pathumthani 12120, Thailand*Author for correspondence; e-mail: thanapk@mtec.or.th

Received: 25 December 2012Accepted: 27 February 2013

ABSTRACTThis study of gas nitriding conditions in a vacuum furnace was the final part of

a project in collaboration with an aluminium extrusion manufacturer. The project wasaimed at furnace condition optimization to solve the problem of too-thin-nitrided layersexperienced in a vertical top-loading vacuum gas nitriding furnace. Small samples cutfrom AISI H13 (JIS SKD61) tool steel dies were placed at 20 different positions torepresent the different nitriding conditions in the furnace. Four different stacking patternsin loading extrusion dies into the furnace were simulated using COSMOSFloWork® todetermine the most effective loading. Microstructural checks, depths of diffusion layerand case-hardness profiles were carried out. The results of the study enabled thedetermination of the optimum positions within the furnace for dies requiring differenttreatment conditions i.e. new dies receiving first nitriding, and used or repaired diesneeding re-nitriding. Production management was then able to arrange furnace loadingto meet these conditions so that for each batch full utilization could be made of thevacuum nitriding furnace leading to savings in energy consumption.

Keywords: heat treatment, aluminium extrusion, gas nitriding and vacuum gas nitriding.

1. INTRODUCTIONIn one medium-sized aluminium

profiles manufacturer, more than 15 tonsof AISI H13 or JIS SKD61 steel extrusiondies per year require low-temperaturesurface hardening treatment such asnitriding to provide 1,000 HV surfacemicro hardness without distortion. If theextrusion dies are incorrectly treated, theiruseful service life is reduced resulting inproduction delays and waste in time andenergy in re-nitriding. Surface hardening

chemical heat treatment of machine partsand tools in nitrogen gas media introducesnitrogen into the surface of steel at atemperature range of 490-565oC. Thetreatment is carried out on steel parts thathave been hardened and tempered toprovide the desired core strength andtoughness. Although new nitridingprocesses such as ion nitriding and lasernitriding have been introduced to industry,cost optimization still is a concern for

910 Chiang Mai J. Sci. 2013; 40(5)

SMEs. Hence, conventional gas nitriding[1] remains the most popular and effectiveway for nitriding steel dies. Continuingstudies on process optimization of gasnitriding and the improvement of furnacesfor gas nitriding are therefore important.

Process temperature for nitriding isoften restricted to a maximum of 540oC,in order to avoid further tempering duringnitriding which would reduce the corestrength of a quenched and tempered part.Application is therefore limited to steelsthat have been tempered at 530oC or,preferably, 560-570oC or higher [2], whilethe relatively low processing temperaturelimits diffusivity of nitrogen such that evenwith long treatment times of up to 100hours case depths greater than 500 microncannot be achieved [3].

In achieving satisfactory hardeningfrom the nitriding process, severalparameters must be tightly controlled. Inparticular, furnace temperature, time [4],gaseous phase dissociation or nitrogenpotential, and gas flow significantlyinfluence both case depth and the shape ofthe hardness profile [5,6]. Recent work hasshown that computational fluid dynamicssimulation can provide useful informationabout the gas flow inside the effectivenitriding zone and the temperatureprofiles generated in dies during treatment[2]. Simulation for steady state conditionsare well recognized as an effective methodto achieve satisfactory data in additionto diffusion models for studyingprocess variables and predicting nitridingresults [3].

The use of computational fluiddynamics can model both gas andtemperature distribution as a startingpoint for the development of proceduresto forecast hardness profiles on the basisof the assumed process parameters and the

chemical composition of steel being treated[3,7].

The objective of the present work isto apply commercially available softwarethat is commonly used for gas flow analysisto simulate heat and gas flow distributionin a vertical top-loading vacuum gas nitridingfurnace. The results from such simulationare compared with the experimental datafor routine industrial control parametersin order to validate the applied techniqueand to form the basis for setting upoptimum conditions for nitriding i.e., todetermine the suitable positions in thefurnace for the normal first time nitrideddies, re-nitrided or repaired nitrided dies.Such optimization should increase furnaceutilization efficiency and result in energysavings.

Problems that can arise if the gasnitriding process is not correctly controlledinclude:

• Inefficient ammonia gas dissociation• Incorrect surface hardness• Distortion• Incorrect hardness profile• Too thick white layer• Cracks and flakingSignificant temperatures for the heat

treatment of AISI H13 (0.4C, 5.3Cr,1.4Mo and 1.0V) that is used for extrusiondies are:

• Austenitising at 1,020-1,060oC• Annealing at 750-850oC• Stress relief at 600-650oC• Gas nitriding reaction typicallybegins at 450oC [5]

2. EXPERIMENTAL WORKIn this study, all conditions were

simulated based on the routine work of in-house nitriding extrusion dies in thealuminium profiles manufacturer. Thestudy focused on simulation flow analysis

Chiang Mai J. Sci. 2013; 40(5) 911

of gas nitriding treatment using a verticaltop-loading furnace of AISI H13 (JISSKD61 or X40CrMoV5-1) tool steelextrusion dies to be used in the productionof aluminium profiles. The flow analysissoftware (COSMOSFloWorks®) wasapplied to determine only the distributionof temperature while gas density, gasvelocity and pressure are kept constant.Convection heating is achieved using theammonia gas which generates the nitridingatmosphere. Under vacuum conditions,the gas velocity is assumed to be uniformand is excluded from experimental

parameters. For comparison with practicalresults, the experiment used examinationof samples from four different dies stackingarrangements in the furnace which had achamber volume of 850 liters. Previouslyhardened and tempered to 49 HRC H13test samples, 1 × 1 × 2 cm in size, werepositioned to represent the various furnaceloading positions of H13 extrusion dies.During the experiment the nitridingprocesses was set up in accordance with theoptimum practice determined from recentin-house projects in the company.

Figure 1. AISI H13 extrusion dies in various size and thickness (diameter of 175-225mm) are prepared for surface heat treatment.

Figure 2. Aluminium profiles produced by extrusion using nitrided alloy steel dies.

912 Chiang Mai J. Sci. 2013; 40(5)

Figure 3. Vacuum gas nitriding, a vertical top-loading furnace used in this study.

Figure 4. Extrusion dies are randomly stacked into the basket, the random pattern wasroutine in use prior to this study. This is an unsuitable arrangement for sufficient nitridingon every part of each die.

Chiang Mai J. Sci. 2013; 40(5) 913

Figures 1 to 3 provide the backgrounddetail of the project, i.e., extrusion dies,typical aluminium profiles and the vacuumnitriding furnace.

Figure 4 shows the top and the bottommodel views of dies stacked randomly inthe loading basket together with theactual dies for loading. With such stacking,selective positioning of dies for nitridingcannot be applied. Hence, the experimentused ordered stacking of dies in the loadingbasket. Six stacking patterns were used andeach was simulated by the computationalflow analysis. Temperature distributionand flow analysis for a normal atmospherenitriding furnace was as in previous work.Figure 5 shows the important parametersfor the normal atmosphere furnace and for

Figure 5. Conceptual parameters used in the simulation of gas nitriding processcomparing between the process under atmospheric controlled furnace and vacuumfurnace.

the vacuum nitriding furnace which is thesubject of the present study. For each ofthe samples taken, microstructural checkswere used to study the effectiveness ofnitriding with respect to surface hardnessand diffusion layer thickness. Thetraditional temperature for gas nitriding isbetween 500 to 590oC but earlier workshowed that the process could be donemore effectively at lower temperature [8].Hence, the experiment was performedusing the double-stage nitriding profile asshown in Figure 6. Small samples cut fromAISI H13 dies (Figure 7) were placed at 20different positions (Figure 8) to representthe different nitriding conditions in thefurnace.

914 Chiang Mai J. Sci. 2013; 40(5)

Figure 6. Nitriding cycle used in this work. Ammonia is purged only after the furnaceis evacuated and subsequent cooling is by nitrogen gas. The process uses double-stagenitriding with a fixed ammonia gas purging rate.

Figure 7. The samples prepared for placement in the furnace.

Figure 8. Left to right, H13 extrusion dies arranged in the basket and illustration viewsof the basket showing positions for 11 samples marked as Sample 1, 2, 3, 5, 6, 7, 9, 10 and11 for the side area in the chamber are as Sample 4 and 8 for the deep area inside the stack.The inlet for the ammonia purge is situated at the position between Sample 2 and 3.

Chiang Mai J. Sci. 2013; 40(5) 915

3. RESULTS AND DISCUSSIONSimulation results obtained are shown

in Figures 10 to 14. The contour plot(Figure 9) was used to examine theeffectiveness of nitriding process. Nitridingresults obtained from microhardness

checks of the cross sectioned specimens, i.e.surface hardness or maximum hardness atthe edge zone, core hardness at the depthunder the surface without diffusion,diffusion depth and the depth of hardnesschange are given in Table 1.

Figure 9. The contour plot (left) of sectional hardness profile on nitrided specimen [8]commonly used as a reference to compare nitriding results and the plot on a micrograph(right, etched by 3% nital) for determining case-hardness properties.

Figure 10. Illustrations of two different stacking patterns having nearly maximum loadin the loading basket. The simulations were done using a soaking temperature of 550oC.The zone closer to the inlet is determined as the higher temperature zone, while thezone far from the inlet (bottom zone) does not achieve the required temperature.Temperature distribution in each stacking pattern directly affects nitriding.

916 Chiang Mai J. Sci. 2013; 40(5)

Figure 11. To stack dies in symmetrical manner and the use of a furnace retort weresuggested to determine the effects of stacking order and more symmetrical placement.

To optimize nitriding, energy savingmust also be considered. The routine workprior to this study could load an average45-50 dies or around 1.2 ton per batch intothe furnace chamber. Figure 9 shows thestack arrangement for increasing the

number of dies such that 55-60 mixed-size dies or up to 1.5 ton per batch can betreated. However temperature distributionneeded to be determined for this suggestednew arrangement.

Figure 12. The illustration of loading with a large gap between the upper and the lowerbasket aimed at optimizing temperature distribution.

Chiang Mai J. Sci. 2013; 40(5) 917

Figure 13. A more symmetrical stacking pattern was also proposed with orderly stackingof H13 dies, aimed more uniform temperature distribution. The illustration showsa symmetric stacking that optimized requirements to increase the number of treateddies and to achieve more uniform temperature distribution.

The temperature distribution in H13dies (Figure 10 and 13) shows that thetemperature is reduced progressively onmoving from the top to the bottom zonesof the furnace chamber. This indicated thatthe different temperatures between diesat eleven reference positions for thetest specimens represent the highesttemperature zone, medium temperaturezone and the lowest temperature zone

respectively. Nitriding at high temperatureincreases diffusion but decreases corehardness, while nitriding at low temperaturedecreases surface hardness and decreasesnitriding depth. A schematic view of atypical nitrided component is shown inFigure 14 and the relationships betweensimulation and hardness checks are givenin Figures 15 to 18.

Figure 14. Schematic of a typical nitrided component macrostructure [5].

918 Chiang Mai J. Sci. 2013; 40(5)

As shown in Table 1, results fromhardness tests and metallographic examina-tion on the samples placed with the diestacking pattern in Figure 10 representroutine stacking without any symmetricalarrangement. Excluding the samples atpositions 4 and 8 (Figure 8) that represent

the deep positions within the stack, theaverage surface hardness of all samples is56.18 HRC while average diffusion depthis 58.25 micron. A smaller diffusion depthis achieved at the deep-into-stack positions,4 and 8.

Table 2. Results of nitriding stacking dies having a more symmetrical pattern and space(Figure 13).

Position Surface hardness Diffusion depth(HRC) (micron)

1 53.17 34.362 53.53 31.793 52.20 34.184 53.45 52.985 53.55 64.156 54.00 77.127 52.33 51.738 53.67 66.349 53.45 72.9210 54.70 70.6711 53.60 81.01

Table 1. Average nitriding results (as referred in Figure 7) obtained from the stackingshown in Figure 10 determined for positions 1 to 11 in the furnace.

Position Surface hardness Diffusion depth(HRC) (micron)

1 55.93 58.572 56.30 48.143 55.85 42.614 55.95 28.375 55.13 48.376 55.75 58.167 56.28 53.908 55.78 28.969 56.55 66.6610 57.28 76.2811 56.55 71.55

Chiang Mai J. Sci. 2013; 40(5) 919

Figure 15. Plot of the relation between hardness data (raw data) from Table 1 andsimulated temperature obtained from the simulated condition in Figure 10.

Figure 16. Average diffusion depth plotted against simulated temperature as obtainedand shown in Figure 15.

Figure 17. Plot from the simulation in Figure 13, showing the relationship betweenaverage surface hardness decreased when temperature increases.

920 Chiang Mai J. Sci. 2013; 40(5)

Figure 18. Data obtained from samples shown in Figure 17, the better diffusion depthsare shown in the group of samples that were nitrided in a low temperature zone.

The difference between zones in thevacuum furnace is believed to have nosignificant effect on nitriding. The nitridingatmosphere in the furnace is controlled atlow pressure to effectively uniform gaspressure that should cause no difference ofnitriding gas pressure. Nitrogen potentialis one major factor that controls nitridingdepending on gas concentration and gastemperature [9]. The low pressure, uniformgas pressure and uniform gas concentrationare used to minimize the variation from heatconvection. However, heat conductionand heat radiation from the loaded dies stillinfluence temperature distribution in thefurnace. Even though the heat sources ofthe furnace basically are designed toprovide uniform temperature, variation intemperature distribution can occur due todifferences in die stacking causing differentnitriding effects between zones, as shownin Figures 10 to 13 for simulated data oftemperature distribution.

The simulation in this work wassimplified to show only effects of controlledparameters [10]. The nitrogen is suppliedby the dissociation of ammonia at the steelsurface in accordance with equation:

NH3 → [N] + 3/2 H

In gas nitriding, the nitrogen activityis controlled by the degree of dissociationand the flow rate of the gas. Nitridingdiffusivity depends on the partial pressuresof the products of ammonia gas dissocia-tion, the rate of which is not steady evenat constant temperature. As the gas ispurged into the chamber the dissociationproceeds. Since the diffusible surface fornitrogen atom is limited (for only clean andless oxidized surface), the dissociationbecomes limited since the ammonia gaspressure increases faster than nitrogenatoms can diffuse into the die surface. Dietemperature and temperature distributionbetween zones of die stackings determinenitrogen diffusivity and hence nitridingeffects.

The differences in diffusion depth inthe samples at the various zones areillustrated in Table 2 for the conditionswhere sufficient space was set up betweendies to encourage uniform gas conditionsbut there was a temperature gradient of70-80oC between top and bottom of thechamber.

Chiang Mai J. Sci. 2013; 40(5) 921

Typically, increasing temperatureincreases nitrogen potential and diffusivitybut decreases core hardness [6]. In addition,higher diffusivity increases the depth of thenitrided layer but decreases maximumsurface hardness. The experiment involvedchecking only 2 outputs, i.e., surfacehardness and diffusion depth (as nitridedlayer thickness); surface hardness tends todecrease with increasing temperature whilediffusion depth increases. Due to highdiffusivity, nitrogen atoms diffuse deeperinto the surface with less concentration atthe near surface zone. The results inFigures 15 to 17 are in agreement withprevious work [6].

The relationship between measurednitrided layer thickness and simulatednitriding temperature plotted in Figure 18indicates low diffusivity for samples at thenear gas-inlet position. One reason for thislow diffusivity is that the dissociatedammonia gas was blown away by freshammonia gas from the inlet. The vacuumfurnace is controlled at low pressure, butuniform low pressure is not possible in thezone near the inlet due to the incomingammonia. To obtain uniform temperaturebetween zones in the vacuum furnace, astacking arrangement with less spacebetween dies is used to give heat transferby heat conduction and radiation. Heatconvection of the flowing gas in the basicgas nitriding furnace without a vacuumsystem is known to affect gas concentrationand nitrogen potential. Accordingly spacebetween stacked dies should be madesufficient to give uniform gas pressure andgas concentration including nitrogen atomconcentration between zones. In previouswork [6], unsatisfactory nitriding resultswere found in the internal zones of diesstacks, due to the reduced gas flow intosuch zones giving lower nitrogen potential.

4. CONCLUSIONSFlow analysis using a simulation

technique can be used to determinetemperature distribution in the gasnitriding process as well as predicting theeffectiveness of nitriding at differentfurnace load positions for the batchprocess.

Temperature is a main parameter in thegas nitriding process increasing nitridingpotential and diffusivity. The low pressureconditions in a vacuum furnace can supportnitriding under essentially uniform gaspressure, however temperature cannoteasily be made uniform as can gas pressure.As determined by using the simulatedtemperature data, temperature is akey parameter in achieving satisfactorycase-hardening in vacuum nitriding. Toeffectively use vacuum gas nitriding, someconcerns from this work can be concludedas:

• Heat from the furnace wall transfersto loaded dies by only radiation, theconvection via atmosphere is minimizedbecause of the vacuum condition.

• Heat radiation between dies and heatconduction from die to die via theloading basket and retort improvesheat transfer in the loaded dies.

• Even though heat transfer is limitedby the vacuum condition increasingfurnace temperature too high decreasescore hardness of loaded dies.

• Increasing space between dies decreasesthe efficiency of heat conductionresulting in variable temperaturedistribution.

• Increasing the number of dies in theloading (or decreasing space betweendies) increases heat conductionbetween dies that decreasestemperature differences betweenzones.

922 Chiang Mai J. Sci. 2013; 40(5)

In the present work, the simulation isrelated to only steady state conditions,hence, in practice, the results cannot beapplied to multi-step soaking temperatures.In industry, both in vacuum and non-vacuum furnaces, double-stage gas nitridingusing multi-step soaking during increasingtemperature between stages has becomenormal practice. A study of the relation-ships between soaking temperatures,nitriding stages and case depth characteris-tics in such multi-step treatments will bethe subject of future work.

REFERENCES

[1] Kula P., Wolowiec E., Pietrasik R.,Dybowski K. and Januszewicz B.,Non-steady state approach to thevacuum nitriding for tools, Vacuum,2013; 88: 1-7.

[2] Ratajski J. and Suszko T., Modellingof the nitriding process, J. Mater.Process. Technol., 2008; 195: 212-217.

[3] Keddam M., Djehlal M.E. andBarrallier L., A diffusion model forsimulation of bilayer growth (ε/γ’) ofnitrided pure iron, Mater. Sci. Eng. A.,2004; 378: 475-478.

[4] Anichkina N.L. , Bogolyubov V.S.,Boiko V.V., Denisov V.E. andDukarevich I.S., Comparison ofmethods of gas, ionic, and vacuumnitriding, Met. Sci. Heat Treat., 1989;31: 170-174.

[5] Pye D., Nitriding and FerriticNitrocarburizing, ASM International,2003.

[6] Lightfoot B.J. and Jack D.H.,Nitriding, Heat treatment’ 73, TheMetals Society, 1975; 163: 39-50.

[7] Elkatany I., Morsi Y., Bilcblau A.S.,Das S. and Doyle E.D., Numericalanalysis and experimental validation ofhigh pressure gas quenching, Int. J.Therm. Sci., 2003; 42: 417-423.

[8] Korad T., Ponboon M., Khunnam P.,Chumchery N. and Pearce J.T.H.,Flow Analysis Simulation Study onImproving Gas Nitriding Conditionfor Heat Treatment Process of H13Extrusion Die, NETSU SHORI, J. Jpn.Soc. Heat Treat., 2009; 49: 53-56.

[9] Korad T., Polboon M., Chumchery N.and Phongsophitanan U., ImprovingHeat Treating Condition of Thin SteelSection by Applying ComputerSimulation Technique, JMMM, 2011;21: 101-105.