Accepted Manuscript

Effect of Retained Austenite on Subsequent Thermal Processing and Resultant

Mechanical Properties of Selective Laser Melted 17-4 PH Stainless Steel

Tyler LeBrun, Takayuki Nakamoto, Keitaro Horikawa, Hidetoshi Kobayashi

PII: S0261-3069(15)00272-1
DOI: http://dx.doi.org/10.1016/j.matdes.2015.05.026
Reference: JMAD 7257

To appear in: Materials and Design

Received Date: 13 February 2015

Revised Date: 22 April 2015
Accepted Date: 9 May 2015

Please cite this article as: LeBrun, T., Nakamoto, T., Horikawa, K., Kobayashi, H., Effect of Retained Austenite on
Subsequent Thermal Processing and Resultant Mechanical Properties of Selective Laser Melted 17-4 PH Stainless
Steel, Materials and Design (2015), doi: http://dx.doi.org/10.1016/j.matdes.2015.05.026

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Effect of Retained Austenite on Subsequent Thermal Processing and Resultant Mechanical

Properties of Selective Laser Melted 17-4 PH Stainless Steel

Tyler LeBrun1‡, Takayuki Nakamoto2, Keitaro Horikawa1, Hidetoshi Kobayashi1

1: Graduate School of Engineering Science, Osaka University

1-3 Machikaneyama-cho, Toyonaka, Osaka, Japan
Tel. +81 06-6850-6200
Fax +81 06-6850-6204

2: Technology Research Institute of Osaka Prefecture

7-1 Ayumino-2, Izumi, Osaka, Japan
Tel. +81 07 2551-2563
Fax +81 07-2551-2599

‡: Corresponding Author

Abstract

Thermal processing and mechanical characterization of tensile specimens fabricated from bulk 17-4

PH material produced by selective laser melting (SLM) is presented. Prior to specimen evaluation, various

industry standard and non-standard heat treatments were performed. The focus of this thermal processing

was to evaluate the effect of aging on mechanical performance with or without a prior solution heat

treatment. Volumetric fraction of metastable austenite was measured to vary with heat treatments

performed and the initial conditions from which aging was initiated. Material aged following a solution

heat treatment was found to have an austenite reversion from a fully martensitic structure to volumetric

fractions as high as 20.4%. Whereas, initial concentrations of retained austenite in SLM as-fabricated

material increased following a peak-age heat treatment, but decreased with higher temperature, overaged

heat treatments. Specimens with large amounts of austenite demonstrated stress-induced transformation to

martensite during tensile testing. This behavior was reflected in substantially reduced yield strengths,

increased work hardening rates across greater ranges of strain, and delayed onset of localized plastic

deformation. Ultimate tensile strengths for varieties of specimens aged to the peak-aged condition were

similar, but differed greatly at strain levels where this strength was achieved.

Keywords: Additive Manufacturing; Heat Treatment; Stainless Steel; Selective Laser Melting; Retained
Austenite

1. Introduction

1
 Hardware production via additive manufacturing (AM) technologies has proven versatile across a

variety of industries as a means to bypass many aspects of existing, traditional manufacturing processes and

tools. However, by circumventing many established manufacturing methods, consequences of this

relatively new material processing technology may not be fully understood or appreciated. Of the alloys

currently deployed in commercial AM applications, a small number may be further thermally processed to

achieve desired strength and toughness characteristics by way of precipitation hardening (PH) kinetics. The

effectiveness of these subsequent thermal processing steps when performed on bulk material produced by

AM remains largely unstudied. Similarly, some investigators have identified enhanced mechanical strength

as a result of AM [1,2], but consideration for how these properties are altered as a result of additional

thermal processing is lacking. Of the materials available that may take advantage of further PH through

aging, this study has focused on 17-4 PH stainless steel (AISI 630) as processed by selective laser melting

(SLM).

17-4 PH is primarily a lath martensitic (bct, α’) stainless steel at room temperature following the

allotropic transformation from an austenitic structure (fcc, γ-iron) upon cooling. Strength may be enhanced

through subsequent aging as a consequence of the nucleation and growth of Cu-rich precipitates. These

precipitates initially form as coherent spherical bcc structures that may transform into enlarged fcc non-

coherent structures at greatly extended aging times [3]. With aging, a reversion of martensite back to the

parent austenite structure may be noted. This behavior is believed to be a result of localized chemical

stabilization following the diffusion of Ni and Cu coinciding with the formation of precipitates, increasing

localized concentration of austenite stabilizing elements and locally lowering the martensitic transformation

start (Ms) temperature below room temperature [4,5].

AM processing of 17-4 PH has been studied by some regarding initial mechanical and material

properties. Previous work by the authors of this paper [6] outlined the relative increases in mechanical

strength and the strain rate sensitivity associated with the highly refined microstructure as a result of

selective laser melting (SLM). The small, prior-austenite grain sizes observed for the as-fabricated 17-4 PH

material is attributable to the rapid melt and solidification of the working powder during processing.

Several studies by others have investigated the retention of metastable austenite following the formation of

bulk material by SLM and other similar processing techniques. In one case, a concentration ratio as high as

72% austenite to the 28% martensite was reported for 17-4 PH at room temperature [7]. The retention of

2
this metastable phase is largely seen as a consequence of mechanical stabilization due to the residual

thermal stresses within the bulk material. Additionally, the influence on the residual phase fractions and

texturing by processing parameters, initial working powder fabrication methods, and SLM processing

environment have also been identified [8]. How this retained austenite and its evolution with additional

standard and non-standard thermal processing affect mechanical properties for SLM 17-4 PH is the focus of

this study.

2. Experimental Procedures

Water-atomized pre-alloyed 17-4 PH powder was used as the working material to fabricate high-

density bulk primitives. Resultant density was measured using the Archimedes method on polished cubic

specimens. The averaged density, where surface defects were removed, was 7.7 grams per cubic

centimeter. This amounted to a relative density of 98.7% when compared to fully densified material. Figure

1 highlights the particle shape and regularity of the powdered material used. The two dissimilar particle

shapes, irregular and spherical, are a result of the water atomization process used in its fabrication. Table 1

outlines the chemical composition of the powder as compared to the ASTM A564/A564M [9] material

standard maximums and ranges. Table 2 details the powdered material size distribution as percent weight

on a per volume basis.

Figure 1

Table 1

Table 2

Bulk material was fabricated using an EOS M280 SLM machine (EOS, GmbH, Germany). The total

quantity of primitives and subsequent tensile specimens were built across two different operations, but the

same powder source and operational settings with the machine system were used. In an effort to minimize

oxidation, a protective nitrogen atmosphere was maintained by an attached nitrogen generator throughout

the entire SLM process. The M280 system uses a 400 W capable Nd:YAG fiber laser system to scan, melt,

and consolidate the powdered material into bulk form. Laser operational settings such as scanning speed,

power, focus (laser spot diameter), layer thickness and scanning pattern may be adjusted based on the

material and purpose of the build. Settings deployed in this study included a laser power of 190 W,

scanning speed of 787 mm/s, laser spot diameter of 100 μm, and a scan centerline spacing of 110 μm (10

3
μm gap between individual scanning paths). The laser was operated in continuous wave mode. Layer

thickness was controlled by the vertical displacement of the build chamber’s baseplate between laser

scanning. In this study a layer thickness of 40 μm was used. While scanning patterns are a focus of some

studies [10], this investigation used a single pattern of parallel scan lines with each layer’s orientation

rotated by 60 degrees relative to the previous layer.

The focus of this experimental investigation was to characterize the mechanical properties of 17-4

PH produced via SLM. However, these evaluations were intended to be irrespective of the geometries

fabricated by this process (i.e. independent of the shape of the volume submitted to the machine for

building). It is well recognized that resultant surface finishes associated with this form of AM are currently

inferior to what is capable by traditional machining [11,12]. Tensile specimens were therefore machined

from within the bulk material primitives so as to eliminate surface-related effects and artifacts of the SLM

process. Tensile specimens evaluated in this study were all fabricated with their loading axes parallel to the

SLM building plane (i.e. layers are oriented in-line with the load applied during tensile testing). Figure 2

depicts the geometry of the compact-size tensile specimens produced. Tensile specimens of this geometry

were selected because of the requirements associated with additional, dynamic tensile testing performed

outside of this study and the need for congruency in mechanical behavior irrespective of specimen

geometry.

Figure 2

Following initial machining, a subset of the tensile specimens did not undergo additional thermal

processing. These were evaluated with material in a condition hereafter referred to as “as-fabricated”. A

variety of material standards govern heat treatments of 17-4 PH, but for the purpose of this study the

ASTM A564/A564M-10 standard is referenced [9]. The remaining specimens underwent the standardize

H900 (480 °C for 1 hr), H1025 (550 °C for 4 hrs), and H1150 (620 °C for 4 hrs) heat treatments. Half of

those treated to each temper were initially treated with a solution heat treatment to the Condition A state

(1040 °C for 30 min) and quenched prior to the subsequent aging. The reported treatment temperatures and

times for both varieties of specimens are in accordance with the aforementioned material standard;

however, the minimum yield and ultimate tensile strengths, elongations, and hardness values are for

materials aged after an initial solution heat treatment. Some researchers working with this material

processed by SLM have evaluated properties without this initial step and treated the as-fabricated condition

4
as a solution treated state [8]. A solution heat treatment was included as part of the preparatory processing

for half of the specimens evaluated to measure its effect. For shorthand notation and disambiguation, aged

specimens that have had a solution heat treatment prior to further aging will be referred to in this paper with

the shorthand "CA" prefix attached to the name of the aged condition (e.g., CA-H900). All heat treatments

were performed in a quartz chambered electric vacuum furnace. At the end of the designated heat

treatment, all specimens were then exposed to laboratory air and quenched in water.

Material samples not intended for mechanical evaluation underwent the same thermal processing.

These bulk samples were evaluated by X-ray diffraction (XRD) (spectral) analysis. XRD analysis was

performed using a Rigaku RINT Ultima III X-Ray Diffractometer with a Co (Kα) X-ray source (λ = 1.79

Å). Rupture surfaces of tensile specimens were evaluated by standard scanning electron microscopy using a

JEOL JSM-5310LV Scanning Microscope. Transmission electron microscopy (TEM) images were

collected using a JEOL JEM-2100.

Tensile testing was performed using a Shimadzu AG-Xplus, a screw-driven uniaxial tension testing

machine. Cross-head speed was set to 0.5 mm/min for an equivalent 1.04x10-3 s-1 strain rate. Because of the

specimens’ relatively small gage length, specimen strain was evaluated by measuring cross-head

displacement. Elastic compliance of the testing apparatus was corrected in all recorded tests to arrive at the

correct specimen displacement and corresponding elongation with respect to the applied load. All testing

was performed at room temperature (approximately 24 °C).

3. Results and Discussion

3.1 Mechanical Properties

Engineering stress-strain curves typical of samples evaluated for each condition are presented in Fig.

3 (a) and (b). Figure 3 (a) shows curves for specimens following aging heat treatments to the CA-H900,

CA-H1025, and CA-H1150 states. These curves are also plotted alongside a reference curve for a typical

SLM specimen following only a solution heat treatment. Figure 3 (b) shows a similar comparison, but with

specimens treated to the same aging conditions without the intermediate, post-SLM solution heat treatment.

Figure 3 (b) also shows a stress-strain curve for a typical as-fabricated specimen without any additional

thermal processing following machining. Averaged values and ranges for micro-Vickers hardness, proof

stress (0.2% elastic offset), ultimate tensile strength (UTS), uniform elongation (UEL), and hardness for the

5
various tested conditions are summarized in Table 3. Reference minimums per the material standard are

included in parenthesis. In the cases where an upper and lower yield point are observed, the upper yield

point is presented in the table. All averages and ranges were of 3 measurements for each condition.

Figure 3 (a) and (b)

Table 3

Specimens shown in Fig. 3 (a) demonstrate an evolution of mechanical behavior across the aging

conditions evaluated consistent with standard thermal processing for 17-4 PH [9,13,14]. Aging to the CA-

H900 results in the peak-age condition characterized by the high yield and ultimate tensile strengths. The

two other aged conditions, where tempering was performed at higher temperatures and for longer durations,

demonstrated an expected reduction in mechanical strength and accompanied increase in elongation. Total

elongation following uniform plastic strain for all solution heat treated specimens was impaired due to local

stress concentrations at embedded porosity sites. For samples of the peak-aged condition, shortly after the

onset of localized plastic deformation (necking) failure was observed to occur. And for progressively

longer and higher temperature heat treated specimens, both UEL and total elongation were observed to

increase. The increase in UEL also indicates the material’s increasing ability to strain harden after initial

yielding until achieving UTS.

A largely different stress-strain response is observed for specimens following thermal processing

without an initial solution heat treatment. Both the as-fabricated and peak-age specimens yield at reduced

stress levels, but rapidly work harden across early plastic strain. The general shape of the stress-strain curve

observed of the as-fabricated condition is preserved following the peak-age H900 heat treatment, but at

elevated stress levels with noted increases in both YS and UTS. However, the shoulder associated with

initial yielding of the as-fabricated condition shows both an upper and lower yield point. This demonstrates

a gradual relaxation in stress with continued elongation for approximately 1% of the nominal strain before

once again strengthening. Unlike the CA-H1025 and CA-H1150 heat treatments, the H1025 and H1150

show an increase from the initial as-fabricated condition in both yield and ultimate strength with reductions

in total and uniform plastic elongation. This is in contrast with the behavior observed for those two similar

tempers post solution heat treatment outlined in Fig. 3 (a).

Figure 4 (a) and (b) show the work-hardening rates as a function of strain across all conditions

tested. These curves are generated from the individual tests depicted in both Fig. 3 (a) and (b) and are also

6
representative of the behavior for all tests conducted from each condition evaluated. The work hardening

rate was calculated by performing a finite difference of the true stress, divided by the corresponding

increment of true strain. Without in situ measurements of specimen diameter, the approximate work

hardening rates calculated by this method are only reliable under the assumption that uniform plastic

deformation is maintained until the onset of necking. Values calculated here are with units of MPa per unit

strain. The trend for each family of thermal processing conditions, those with a prior solution heat

treatment and those without, illustrate different behavior across the aged conditions. The two curves for the

as-fabricated and H900 conditions in particular depict strain-hardening rates that are not monotonically

decreasing with increasing strain. The reversal in strain-hardening rates explains the sigmoidal shape of the

stress-strain curve. Additionally, the H900 curve is observed to be negative for a short region of strain. This

corresponds to the softening observed post initial yielding outlined in Fig. 3 (b). When compared to the as-

fabricated condition, the H900 specimens show reduced work hardening rates over early stages of plastic

strain following the recovery from initial softening, but remain elevated by comparison after 5% strain and

persist across greater ranges of uniform elongation. Despite the increase in both yield and ultimate tensile

strengths observed for the H900 specimens over the as-fabricated condition, the relative increase in strength

as a result of this work hardening is slightly diminished. Relative differences between averages of yield

strength and UTS for the as-fabricated and H900 specimens are 594 and 472 MPa, respectively. This

illustrates a reduced capacity to work harden for this particular sequence of thermal processing and these

two conditions.

For those tested at the CA-H900 and CA-H1025 treatments, little to no work hardening capacity

over extended regions of strain is demonstrated by the rapid transition from nearly vertical to nearly flat

curves. The CA-H1150 shows a slightly elongated range of strain over which elevated work hardening

occurs. For specimens without an initial solution heat treatment and aged for an extended period (H1025

and H1150) an opposite trend in behavior is observed. Strain-hardening rates more rapidly transition

towards a constant value earlier in strain. The general shape for these two conditions progressively

approach those specimens subjected to an initial solution heat treatment.

Figure 4 (a) and (b)

Other researchers have noted by hardness testing that the absolute and relative strengths attained

post-H900 heat treatment, without a preliminary solution heat treatment, are consistent with those observed

7
for traditionally processed 17-4 PH [8]. However, micro Vickers hardness measurements made and

reported here for the H900 condition were found to approach, but not exceed the minimum required per the

material standard for this heat treatment (average 375 HV, 38.5 HRC equivalent, as compared to the

material standard minimum 40 HRC). Whereas, performing a solution heat treatment prior to the peak age

resulted in hardness measurements that well exceeded the minimum required (CA-H900) (average 417 HV,

42.5 HRC equivalent). These observations are also supported by the dissimilarity in the observed yield

strengths between the two conditions. Whereas, the ultimate tensile strengths were found to be similar

(1417 and 1444 MPa for H900 and CA-H900, respectively). Unlike hardness measurements, tensile testing

has highlighted the different yielding and plastic deformation behavior associated with the two thermal

processing sequences that hardness testing cannot directly capture.

Across all heat treatments performed it is useful to compare the relative increases in strengthening

associated with aging of the material. The relative increase in strength following an aging treatment may be

simply expressed by:

St = (S – So) / So × 100 (1)

Where St may describe the relative increase in strength due to aging over an untreated specimen (either

from the as-fabricated condition or solution heat treated condition, as the respective baseline condition), So,

and the post-treated aged strength, S. For example, the relative increase in strength of a specimen in the

CA-H1025 condition is based on the relative difference to a sample in the Condition A state. And a H1025

specimen is directly compared to the strength measured from specimens of the as-fabricated condition. As

is common with investigations into aging kinetics, previous researchers relied upon measured hardness

values to establish this relationship as a relative change in strength. For the purpose of this study, true stress

at UEL is instead used. Similarly, tempering charts commonly provide hardness as a function of

temperature at which aging is performed. These charts only outline the relationship across a single

temperature with varying exposure times. However, the various standardized heat treatments evaluated

were performed across a variety of different temperatures and durations. Relating the effects of aging

across these two variables is necessary. The Larson–Miller or Hollomon–Jaffe tempering parameter (P)

[14,15] may be used to interrelate these as it is a time-compensated temperature term. The Hollomon-Jaffe

parameter is expressed as follows:

P = T (C + log(t)) × 10-3 (2)

8
Where T is the absolute temperature in degrees Kelvin, t is the tempering time in hours, and C is a material

constant (ranging between 10 and 20, in many applications it is considered to be 20).

Figure 5 shows the relationship between these two expressions for the tensile specimens evaluated.

Both exhibit a downward trend off of the peak-age condition commonly characterized by a tempering

parameter of 15 for 17-4 PH. However, samples without an initial solution heat treatment preceding aging

did not exhibit the expected negative relative strengthening for larger values of P above values of 17.9.

These trends in behavior may be partially explained by a combination of interrelated mechanisms. To help

elucidate those differences, it is useful to examine the phase presence and volumetric fractions of

metastable austenite relative to the primary martensitic matrix.

Figure 5

3.2 Phase Composition

Figures 6 shows the XRD spectrums for the various heat treatment conditions as measured across

the vertical reference plane (SLM layers oriented normal to the surface examined). Mechanical processing

and preparation for XRD analysis may disturb the austenite and initiate a strain induced transformation to

martensite due to the localized stresses associated with grinding and polishing. Therefore, the observed

concentrations of austenite may be artificially low from what is initially produced via SLM. Observations

of these measurements, however, still point to reasons for the observed mechanical behavior. Table 4

details the calculated volume fractions of martensite and austenite observed for all specimen conditions

from the XRD line spectra. Quantitative X-ray diffraction phase analysis and calculated spectra were

performed using the Rietveld method provided by the MAUD software package by a full pattern fitting

procedure on data collected via XRD [16,17].

From Fig. 6, specimens with an initial solution heat treatment show increases in the major and minor

γ-iron peaks, e.g. (111)γ or (200)γ, that progress with increased exposure temperature and time. The initial

state prior to these aging treatments, Condition A, contains no discernable fraction of the metastable

austenite phase post SLM-processing that is otherwise found in the as-fabricates state. Aging from this

initial condition shows a gradual reintroduction of austenite by reverting martensite to its parent austenitic

structure. This observation is consistent for standardized thermal processing of conventional 17-4 PH

[14,18]. This process of austenite reversion has been generally attributed to the chemical stabilization by

9
the diffusion of alloying element constituents, commonly toward lath boundaries. More specifically, as Cu

diffuses to form precipitates, the local M temperature is effectively lowered at the sites of their formation
f

and thereby permits austenite chemical stabilization at room temperature [19]. Viswanathan et al. identified

the presence of austenite grains to be closely accompanied by coarsened Cu-precipitates at lath boundaries

at elongated aging time and elevated temperature.

Figure 6

Table 4

The evolution from the as-fabricated condition is not as straightforward. Following the initial H900

treatment, an increase in austenite is measured. Treatments to H1025 and H1150 conversely see reductions

in the intensity of all γ-iron indices. At their peak, the H900 condition demonstrated an austenite volume

fraction of 40.5%, an increase from the as-fabricated fractional percent of 36.0%. The reduced strain-

hardening capacity observed for both the H1025 and H1150 specimens correlates with the similar reduction

in retained austenite following heat treatment. Work-hardening rates for both of these conditions begin to

approach those observed for the family of heat treatments that included a preliminary solution heat

treatment. The resultant concentrations at these two processing conditions is believed to be a combination

of two competing mechanisms for its retention. The reduction in volume fraction between H900 and H1025

conditions is likely a result of the release of thermal stresses associated with the SLM process when

samples are exposed to increasingly elevated temperature and longer duration heat treatments. It has been

suggested that the retention of austenite within SLM fabricated martensitic steels is a result of mechanical

stabilization by hydrostatic pressures generated during initial formation [20]. With a reduction in

mechanical stabilization, austenite is free to complete its transformation to martensite when cooled below

the expected Mf temperature. And by extension, it is outright eliminated in the case of a solution heat

treatment at 1040 °C for 30 minutes. While, as previously mentioned, the reversion or local stabilization of

austenite by the diffusion of Cu and Ni during aging maintains the observed austenite concentrations above

zero.

Comparative differences in relative strengthening between the two CA-H900 and H900 conditions

highlight the significance (see Fig. 5) of austenite concentrations and the material’s ability to age-harden.

Specimens with large starting volume fractions of austenite and heat treated to the peak-age condition do

10
not demonstrate the same relative strengthening capacity as that demonstrated by those initially solution

heat treated, such as CA-H900. In these cases, the austenite does not age-harden and consequently inhibits

strengthening by the intended growth of Cu-precipitates [5,21]. Precipitate nucleation requires a saturated

state. And in this case, austenite maintains a higher solubility of copper, thereby reducing age hardening

throughout. This is most evident when YS is considered for the two variations of the peak-age. Samples

aged without the solution heat treatment, H900, exhibited a 30.1% lower yield on average to those with

such preliminary treatment, CA-H900. Despite the differences between the relative strengthening for the

these two peak-aged conditions, the H900 specimens showed a capacity to strain harden throughout the

comparatively elongated uniform plastic deformation region. The plastic deformation behavior for both the

as-fabricated and H900 conditions with their high volume fractions of austenite demonstrate the capacity

for transformation induced plasticity (TRIP). This is further supported by two XRD measurements made

after the completion of a tensile test for an as-fabricated specimen.

Figures 7, 8, and 9 show TEM imagery collected of the two overaged states, CA-H1150 and H1150.

Figure 7 (a) shows a bright field image of a CA-H1150 specimen. The dark field reflection highlights the

reverted austenite structure interspersed amongst the martensite lath structure adjacent to lath boundaries.

Figure 8 shows a magnified view of overaged precipitates commonly visible throughout this sample's

microstructure. Circular or oval-shaped precipitates were observed uniformly dispersed outside of austenite

grains, occasionally coinciding with dislocations. The long axis of the oval-shaped precipitates was found

to be limited to approximately 20 nm in length. In contrast to Fig. 7, Fig. 9 shows the same aging treatment

applied without an initial solution treatment. The dark field reflection shows austenite without the same

distribution or frequency as the CA-H1150. The precipitate structure observed in Figure 8 for the CA-

H1150 specimen was not observed with the same regularity. Twinning is observed in the selected area

diffraction patterns for both instances of austenite in the CA-H1150 and H1150 specimens.

Figure 7 (a), (b), and (c)

Figure 8

Figure 9 (a), (b), and (c)

Figure 10 shows an additional two XRD spectra measured from within a sectioned, as-fabricated

tensile specimen post-test. The upper curve demonstrates the combination of both martensite and austenite

as measured from within the threaded grip section of the specimen. No plastic deformation occurred as part

11
of the tensile test in this region of the sample. The lower curve shows the reduced volume fraction of

austenite as measured from within the gauge section adjacent to the fracture surface. The region of the

sample where localized plastic deformation occured (necking region) is a smaller target to successfully

measure for tensile samples of this shape and size. Therefore, these measurements made by XRD are

limited to the remainder of the gauge section. The measured phase presence and volumetric fractions

highlight the austenite to martensite transformation within the region largely deformed by uniform plastic

deformation. The calculated volume fraction of the grips section of the sample was found to be unchanged

from the bulk material also examined by XRD. The gauge section was measured to have fallen from 36.0%

to 5.2% by volume of austenite post-test. This is good evidence of strain-induced transformation of

martensite of the as-fabricated material. Similar behavior is noted in the peak-aged condition without an

initial solution heat treatment.

Figure 10

3.3 Fracture Surfaces

Figures 11 (a) and (b) depict the fracture surfaces at low magnification for specimens in the CA-

H900 and H900 conditions. These images were taken at angle to highlight the macro-scale texturing in their

respective rupture surfaces. Despite the differences in total elongation and work hardening, these two

figures show the similarity in fracture behavior between the peak age conditions. Of note is the planarity

observed of both specimens. Figure 11 (c) shows a similar view, but of the as-fabricated condition. In

contrast to Figs. 11 (a) and (b), the material without further thermal processing is found to have a highly

irregular fracture surface. A greater degree of shearing is observed throughout and surrounding a less clear

fracture plane. Some isolated pores are observed to be atop local mounds of material or in the base of a

corresponding divots in the fracture surface. Figure 12 highlights such an example of a pore isolation with

local shearing of the surrounding material.

Figure 11 (a), (b), and (c)

Figure 12

Figures 13 and 14 show at high magnification the texturing across the surface observed for the CA-

H900 and as-fabricated specimens. The tensile loading direction was normal to the these images. When

viewed at this scale and normal to the fracture surface, the micro-scale dimple structure shown in both

12
figures suggests localized ductility. However, the formation of micro-voids and their coalescence into

larger dimple structures commonly observed in traditionally processed 17-4 PH is not observed here. When

the fracture plane is normal to the applied load, the observed texturing at the scale shown in Figs. 13 and 14

is consistent across all SLM specimens despite heat treatment condition. The shared fracture surface texture

at the smallest size scale suggests that mechanics of deformation achieve commonality between all

specimens despite their differing initial microstructural conditions. For specimens with high volumetric

fractions of austenite, this follows the transformation of that phase to martensite as a result of plastic strain

due to the tensile loading.

Figure 13

Figure 14

3.4 Discussion

High volumetric fractions of austenite are found to persist after the formation of bulk material via

SLM. Application of a solution heat treatment at 1040 °C exceeds the transus temperature of 727 °C where

all martensite reverts to the parent austenite structure (α' ⟶ γ) and residual stresses have been eliminated.

Cooling to room temperature initiates the reverse transformation, γ ⟶ α', starting at ≈ 100 °C (MS) with a

transformation finish temperature of ≈ 32 °C (Mf). Subsequent aging treatments nucleate and grow Cu-rich

precipitates that further coarsen with extended time and temperature. Reversion of martensite to austenite

accompanies the diffusion-controlled segregation of stabilizing, substitutional elements as a part of the

aging treatment. These austenite grains are observed to nucleate on lath and parent austenite grain

boundaries. TEM observations show such formations interspersed within the lath structure of the overaged

CA-H1150 specimens.

The CA-H900 treatment demonstrates a mechanical behavior characteristic of the peak-age

condition with a rapid transition in work hardening and limited ranges of plastic strain before the onset of

incipient necking. Ductility of these specimens is limited due to early fracture from embedded porosity. For

CA-H1025 and CA-H1150 specimens, progressive reversion of the primary martensite to austenite, along

with the coarsening of the precipitate structure, both reduce yield and ultimate tensile strengths while also

extending the range of UEL and total elongation. Softening of the material via increased fractions of the

weaker phase and a likely transition to Orowan bowing around coarsened precipitates are reasoned for the

13
transition in behavior for overaged specimens.

As-fabricated specimens and those further aged without an initial solution heat treatment

demonstrate different behavior. Temperatures applied for all examined post-processing heat treatments do

exceed the temperature range over which the γ ⟶ α' martensitic transformation is to occur and therefore

maintain austenite stability throughout treatment. The solubility of Cu within these austenite grains is

higher than the surrounding martensite and therefore does not precipitate. Their absence is noted in TEM

observations of H1150 specimens. However, the temperatures associated with the overaged conditions,

H1025 and H1150, do begin to sufficiently relieve the expected residual thermal stresses [19].

Consequently, austenite that remains stable during the heat treatment is then permitted to fully transform

into martensite upon cooling in the absence of mechanical stabilization. This is observed as reductions in

austenite volume fraction, as noted for H1025 and H1150 specimens via XRD spectra analysis.

Additionally, any metastable martensite that may reside in the as-fabricated material is permitted to

decompose to austenite during aging treatments. However, without an accompanying segregation of

austenite stabilizing elements, these grains will revert to a stable lath martensite upon return to room

temperature. The lower temperature, peak-age treatment of H900 does not demonstrate the dramatic

reductions in retained austenite likely because the aging temperature is insufficient to relieve the

aforementioned thermal stresses. As supported by measurements of the CA-H900 specimens, increases in

the austenite volume fractions for H900 are believed to originate from the stabilization that accompanies

the initial diffusion and precipitation of Cu in the fraction of the structure that is initially stable martensite.

With high volume fractions of austenite the H900 tensile specimens closely reproduce the stress-strain

response of the as-fabricated material, but at elevated stress levels due to the limited precipitation

hardening.

Enhanced ductility with rapid transitions in work hardening rates of the as-fabricated and H900

conditions are a result of progressive, stress-induced transformation of metastable austenite grains to

martensite. Due to the lower yielding stresses needed to initiate the martensitic transformation, stresses

within tensile specimens do not attain levels necessary to initiate premature failure by crack formation at

embedded porosity sites early in plastic strain. This behavior is in contrast to what is observed in CA-H900

specimens. For those with higher concentrations of austenite, plastic deformation is permitted to continue

until sufficient transformation to martensite raises localized stress to levels necessary to initiate necking

14
and ultimately failure.

With greater relief of thermal stresses accompanying over-aged treatments and accompanying

reductions in austenite concentration, the now largely martensite structures for H1025 and H1150

specimens do not carry the same ability to transform with applied stress. Work hardening rates for these

specimens rapidly transition across reduced ranges of strain. This is similarly reflected in the reduced UEL

and total elongation observed from the initial condition of the as-fabricated state. Reductions in volumetric

fractions of austenite for the H1025 and H1150 treatments, where martensite remains the primary structure,

do not show the same transformation induced strengthening associated with the as-fabricated and H900

conditions.

4. Summary and Conclusions

Quasi-static tensile testing was performed on specimens machined from bulk 17-4 PH produced via

selective laser melting. Mechanical performance varied greatly as result of the evolution of phases resulting

from the various thermal processing performed. From these evaluations several conclusions regarding the

aging performance and subsequent mechanical response can be made.

1. 17-4 PH produced by SLM and in the as-fabricated state was measured to yield at reduced stress levels.

Onset and continued plastic deformation was dominated by the stress-induced transformation of the

retained austenite to martensite. Plastic deformation and specimen elongation via phase transformation

resulted in significantly increased work hardening rates over lengthened regions of uniform plastic

deformation. The delay of incipient necking was a result of the increased work hardening capacity

driven by martensite formation. Continued plastic deformation beyond necking remained relatively

short due to the presence of embedded porosity, but total elongation was enhanced despite the presence

of porosity.

2. Solution heat treatment of as-fabricated material was sufficient to eliminate the metastable austenite

phase. Subsequent aging of the material from this condition resulted in yield and tensile strengths

exceeding standardized minimums. Austenite reversion to levels as high as 20.4% by volume were

measured at increased aging temperatures and times. An increase in both uniform and total elongation

resulted.

3. Initially high volumetric fractions of austenite in the as-fabricated material inhibit strengthening by

15
 aging. However, increased temperature heat treatments progressively transform SLM retained austenite

to martensite. The likely relief of residual stresses permits the transformation of austenite to martensite

upon cooling post-treatment. The over aging of the H1025 and H1150 heat treatments does not result in

the expected negative relative strengthening observed for material initially treated with a solution heat

treatment and aged from an entirely martensitic structure.

4. Aging efficacy from the as-fabricated condition is limited due to the higher solubility of precipitating

elements within the austenite phase. Relative strengthening post treatment is reduced when compared

against material that is initially solution heat treated prior to aging.

Acknowledgements

This research received financial support from Grants-in-Aid for Scientific Research in an

Innovative Area (24120008) from the Ministry of Education, Culture, Sports, Science, and Technology

(MEXT). The authors would also like to thank Dr. Takahiro Kimura of the Technology Research Institute

of Osaka Prefecture who provided assistance in the fabrication of our bulk material.

References

[1] Zhang, K., Wang, S., Liu, W., & Shang, X. Characterization of stainless steel parts by Laser Metal

Deposition Shaping. Mater Design 2014;55:104–119. DOI: 10.1016/j.matdes.2013.09.006

[2] Amato, K. N., Gaytan, S. M., Murr, L. E., Martinez, E., Shindo, P. W., Hernandez, J., … Medina, F.

Microstructures and mechanical behavior of Inconel 718 fabricated by selective laser melting. Acta Mater

2012;60(5):2229–2239. DOI: 10.1016/j.actamat.2011.12.032

[3] Murayama, M., Hono, K., & Katayama, Y. Microstructural evolution in a 17-4 PH stainless steel

after aging at 400 °C. Metall Mater Trans A 1999;30(2):345–353. DOI: 10.1007/s11661-999-0323-2

[4] Wang, J., Zou, H., Li, C., Qiu, S., & Shen, B. The effect of microstructural evolution on hardening

behavior of type 17-4PH stainless steel in long-term aging at 350 °C. Mater Charact 2006;57:274–280.

DOI: 10.1016/j.matchar.2006.02.004

[5] Hsiao, C. N., Chiou, C. S., & Yang, J. R. (2002). Aging reactions in a 17-4 PH stainless steel. Mater

Chem Phys 2002;74(2):134–142. DOI: 10.1016/S0254-0584(01)00460-6

[6] LeBrun, T., Tanigaki, K., Keitaro, H., & Kobayashi, H. Strain rate sensitivity and mechanical

anisotropy of selective laser melted 17-4 PH stainless steel. Mechanical Engineering Journal, JSME,

16
2014;1(5):1–13. DOI: 10.1299/mej.2014smm0049

[7] Facchini, L., Vicente, N., Lonardelli, I., Magalini, E., Robotti, P. and Molinari, A. Metastable

Austenite in 17–4 Precipitation-Hardening Stainless Steel Produced by Selective Laser Melting. Adv Eng

Mater 2010;12:184–188.

[8] Murr, L. E., Martinez, E., Hernandez, J., Collins, S., Amato, K. N., Gaytan, S. M., & Shindo, P. W.

Microstructures and Properties of 17-4 PH Stainless Steel Fabricated by Selective Laser Melting. J Mater

Res Technol 2012;1(3):167-177. DOI: 10.1016/S2238-7854(12)70029-7

[9] ASTM Standard A564/A564M-13 (2013). Standard Specification for Hot-Rolled and Cold-Finished

Age-Hardening Stainless Steel Bars and Shapes, ASTM International, DOI: 10.1520/A0564_A0564M,

www.astm.org

[10] Li, R., Shi, Y., Wang, Z., Wang, L., Liu, J., & Jiang, W. Densification behavior of gas and water

atomized 316L stainless steel powder during selective laser melting. Appl Surf Sci 2010;256(13):4350–

4356. DOI: 10.1016/j.apsusc.2010.02.030

[11] Syed, W. U. H., Pinkerton, A. J., & Li, L. A comparative study of wire feeding and powder feeding

in direct diode laser deposition for rapid prototyping. Appl Surf Sci 2005;247(1-4): 268–276. DOI:

10.1016/j.apsusc.2005.01.138

[12] Santos, E. C., Shiomi, M., Osakada, K., & Laoui, T. (2006). Rapid manufacturing of metal

components by laser forming. Int J Mach Tool Manu 2006;46(12-13):1459–1468. DOI:

10.1016/j.ijmachtools.2005.09.005

[13] ASM International. Handbook Committee. ASM Handbook. Vol. 1., “Properties and Selection:

Irons, Steels, and High Performance Alloys”, ASM Internationalv, 1993.

[14] Mirzadeh, H., & Najafizadeh, A. Aging kinetics of 17-4 PH stainless steel. Mater Chem Phys,

2009;116(1):119–124. DOI: 10.1016/j.matchemphys.2009.02.049

[15] Hollomon, J. H., & Jaffe, L. D.. Time-temperature relations in tempering steel. Trans. AIME

1945;162:223-249.

[16] Lutterotti, L., Matthies, S., Wenk, H. R., Schultz, A. S., & Richardson Jr, J. W. Combined texture

and structure analysis of deformed limestone from time-of-flight neutron diffraction spectra. J Appl Phys

1997;81(2):594-600. DOI: 10.1063/1.364220

[17] Lonardelli, Ivan, et al. Texture analysis from synchrotron diffraction images with the Rietveld

17
method: dinosaur tendon and salmon scale. J Synchrotron Radiat 2005;12(3):354-360. DOI:

10.1107/S090904950500138

[18] Viswanathan, U. Kinetics of precipitation in 17–4 PH stainless steel. Mater Sci Tech

1989;5(April):346–349. DOI: 10.1179/026708389790222546

[19] Starr, T., Rafi, K., Stucker, B., & Scherzer, C. Controlling phase composition in selective laser

melted stainless steels. In Proceedings of the Solid Freeform Fabrication Symposium 2012;439–446

[20] Colaço, R., & Vilar, R. Stabilisation of retained austenite in laser surface melted tool steels. Mat Sci

Eng A-Struct 2005;385(1-2):123–127. DOI:10.1016/j.msea.2004.06.069

[21] Salje, G., & Feller-Kniepmeier, M. The diffusion and solubility of copper in iron. J Appl Phys

1977;48(5):1833–1839. DOI: 10.1063/1.323934

18
Figure 1: SEM imagery of the pre-alloyed 17-4 PH powder used to fabricate the bulk primitives.

Note the dual shape morphology attributable to the water atomization process.

Figure 2: Tensile specimen shape. All dimensions are in mm.

Figure 3 : Nominal stress-strain curves for (a) samples initially solution treated prior to aging, (b)

samples aged without an initial solution heat treatment. The indicated aged condition is shown

adjacent to the terminus of the stress-strain trace. For (a) the initial condition, Condition A, is

shown as a dashed trace and labeled ‘CA’. For (b) the as-fabricated condition is shown as a

dashed trace and labeled ‘As-Fabricated’.

Figure 4 (a): Variation of work-hardening rate across UEL for heat treated specimens with a

preceding solution heat treatment. (b) Variation of work-hardening rate across UEL for heat

treated specimens without a preceding solution heat treatment. The as-fabricated condition is also

included.

Figure 5: Effect of tempering parameter, P, on aging behavior for the two different sets of tensile

specimens evaluated. The tempering parameter is defined as a time-compensated temperature

term, P=T(C + log(t)) × 10 , where T is the absolute temperature in Kelvin, t is the tempering
-3

time in hours, and C is the materials constant (ranging from 10 to 20, 20 is commonly used in

many applications).

Figure 6: XRD spectra for material samples aged concurrently with tensile specimens.

Figure 7: (a) TEM Bright field image of a material sample aged to CA-H1150. (b) Dark field

image showing fcc reflection of reverted austenite. (c) Corresponding SAED pattern with

19
twinning indicated by the secondary spots in the austenite structure. [1 2] diffraction plane

shown.

Figure 8: TEM bright field image of the copper precipitate structure of an overaged CA-H1150

specimen. Precipitates were of circular or oblong shaped with long axes approximately 20 nm in

size. Some, but not all, precipitates are associated with dislocations. Arrows indicate example

precipitates.

Figure 9: (a) TEM bright field image of a material sample aged to H1150 (without a preceding

solution heat treatment). (b) Dark field image showing fcc reflection of retained austenite

structure. (c) Corresponding SAED pattern with twinning. fcc [011] diffraction plane shown.

Figure 10: XRD spectra as measured from within the both the gauge section and threaded grips

section of an as-fabricated tensile specimen. These measurements were made post-test.

Figure 11: (a) Angled view of a CA-H900 specimen. (b) Angled view of a H900 specimen. (c)

Angled view of an as-fabricated specimen.

Figure 12: Angled view of a planar pore geometry penetrating into the tensile loading direction of

an as-fabricated specimen. The fracture surface is not limited to the pore’s periphery, but found to

locally remove surrounding material to form a divot in the fracture surface. Tensile loading

direction was at angle to image shown.

Figure 13: High magnification view of the rupture surface as imaged from the CA-H900

specimen. The orientation of the applied load was normal to this image.

20
Figure 14: High magnification view of the rupture surface as imaged from the as-fabricated

specimen. The orientation of the applied load was normal to this image.

21
Table 1: Composition of 17-4 PH Powder
Composition (w%)

C Mn P S Si Cr Ni Nb Cu Fe
Powdered Working Material 0.054 0.18 0.011 0.005 0.41 16.51 4.16 0.42 3.97 bal

ASTM A564/A564M Limits 0.07 1.00 0.040 0.030 1.00 15.00-17.50 3.00-5.00 NS 3.00-5.00 bal
Limits are in percent maximum unless shown as a range
NS – Not Specified

Table2: Particle Size Distribution

μm <2.8 2.8-3.9 3.9-5.5 5.5-7.8 7.8-11 11-16 16-21 21-31 31-44 >44

Weight Pct 0 0 0 0.5 2.8 15.2 36.9 31.9 9.5 3.2

Table 3: Summarized Mechanical Properties

Hardness, HMV Proof Stress,
Ultimate Tensile Strength, MPa Total Elongation, %
Condition (Material Std Minimum) 0.2% Elastic Offset, MPa UEL, %
(Material Std Minimum) (Material Std Minimum)
(Material Std Minimum)

As-Fabricated 333±2 661±24 1255±3 9.9±0.2 16.2±2.5

H900 375±3 945±12 1417±6 11.7±0.8 15.5±1.3

H1025 399±8 870±25 1358±8 6.5±0.03 13.3±1.5

H1150 381±3 1005±15 1319±2 4.8±0.0 11.1±0.4

Condition A 330±3 939±9 1188±6 3.5±0.2 9.0±1.5

CA-H900 417±5 1352±18 (1170) 1444±2 (1310) 2.9±0.1 4.6±0.4 (10.0)

CA-H1025 350±4 1121±9 (1000) 1172±2 (1070) 4.8±0.1 9.6±1.7 (12.0)

CA-1150 317±3 859±11 (725) 1017±15 (930) 7.7±0.3 16.6±1.2 (16.0)

ASTM A564/A564M-10 minimum values are shown in parentheses for aging treatments that follow a solution heat treatment.

Table 4: Volumetric fractions of austenite and martensite

Martensite (bct, α’) (%) Austenite (fcc, γ) (%)

Condition A 100 0

CA-H900 96.7 3.3

CA-H1025 95.3 4.7

CA-H1150 79.6 20.4

As-Fabricated 64.0 36.0

H900 59.5 40.5

H1025 89.6 10.4

H1150 94.4 5.6

22


Figure 1


Figure 2
 1600
CA-H900
1400

1200

Nominal Stress (MPa)

CA-H1025
CA
1000
CA-H1150
800

600

400

200

0
0 0.05 0.10 0.15 0.20

Nominal Strain

(a)
1600

1400
H900
H1025
1200
Nominal Stress (MPa)

H1150 As-Fabricated
1000

800

600

400

200

0
0.00 0.05 0.10 0.15 0.20

Nominal Strain

(b)

Figure 3
 3
50x10

Work Hardening Rate (MPa/Unit Strain)

CA-H900
CA-H1025
CA-H1150
40

30

20

10

0 0.02 0.04 0.06 0.08 0.10 0.12

True Strain


(a)

3
50x10
Work Hardening Rate (MPa/Unit Strain)

As-Fabricated
H900
H1150
40 H1025

30

20

10

0 0.02 0.04 0.06 0.08 0.10 0.12

True Strain


(b)

Figure 4
 30

20

Relative Strengthening (%)

10
H900 HT without Sol'n HT

-10 H1025 HT

with Sol'n HT
-20
H1150 HT

-30
14 15 16 17 18 19 20

Tempering Parameter, P


Figure 5
 1000 Counts

(211 α')
(110 α')

(200 α')
(200 γ)
(111 γ)

(220 γ)

(222 γ)
(310 α')
(311 γ)
Condition A
CA-H900
CA-H1025
Counts

CA-H1150

As-Fabricated
H900
H1025
H1150

20 40 60 80 100 120
2θ


Figure 6
 a)

b)

c)

(000)

(220)
(110)

(131)

Figure 7 (a), (b), (c)

Figure 8
 a)

b)

c)

(111)

(000)
(200)

(111)

Figure 9 (a), (b), (c)

 (110) α'
(111) γ
1000 Counts

(200) γ

(211) α'
(200) α'

(220) γ

(311) γ

(310) α'
(222) γ
Bulk Material

Counts
Grips Section

Gauge Section

20 40 60 80 100 120
2θ


Figure 10
 a)
!

Figure 10

b)

c)

Figure 11 (a), (b), and (c)



Figure 12


Figure 13


Figure 14
Highlights

 We examine changes to stress-strain behavior for 17-4 PH produced by SLM.

 We measure evolution of volumetric fraction for phases with heat treatment.
 Retained austenite in SLM material reduces strengthening via aging treatments.
 Stress-induced transformation of austenite to martensite alters yield behavior.

23