Process Physics and Parameter–Structure–Property Relationships

Introduction

Additive manufacturing (AM) converts digital geometry into a physical object by depositing, selectively joining, or solidifying material in successive layers. Although the visible process appears geometrical, the quality of an AM component is governed by coupled thermal, mechanical, fluid-flow, photochemical, and solidification phenomena. A small change in laser power, scan speed, layer thickness, powder size, resin exposure, or build orientation can alter melt-pool behavior, microstructure, porosity, residual stress, surface roughness, dimensional accuracy, and service performance.

The central engineering problem is therefore not simply to produce a shape. It is to control a chain of relationships:

process parameters → thermal and mechanical history → microstructure and defects → properties and performance

This chapter develops that chain for the principal AM process families: powder-bed fusion, directed energy deposition, material extrusion, and vat photopolymerization. The discussion emphasizes physical mechanisms and the interpretation of process–structure–property relationships.

Learning Objectives

After studying this chapter, students should be able to:

1. Explain how heat transfer, fluid flow, solidification, shrinkage, and photochemistry influence layer formation.
2. Relate major process parameters to melt-pool geometry, bonding, porosity, surface formation, and dimensional accuracy.
3. Explain the origins of residual stress, distortion, anisotropy, and microstructural variation.
4. Distinguish lack-of-fusion, keyhole, gas, and process-induced defects.
5. Apply dimensionless process indicators and energy estimates within their limitations.
6. Analyze an AM process window and recommend parameter or design changes.

1. The Physical Sequence of Layer Formation

A typical powder-bed fusion layer follows this sequence:

1. A layer of powder is spread at a specified thickness.
2. An energy source selectively melts or sinters regions defined by toolpaths.
3. The molten region absorbs energy, flows, and exchanges heat with surrounding powder, solid material, and atmosphere.
4. The melt pool solidifies, producing a track bonded to the previous layer.
5. The platform is lowered and the cycle repeats.

In material extrusion, a thermoplastic filament or pellet is heated, pressurized, and deposited through a nozzle. In vat photopolymerization, liquid resin is selectively polymerized by light. In directed energy deposition, powder or wire is introduced into an energy source and fused onto a substrate. Despite their differences, all processes require adequate bonding, controlled geometry, and management of thermal or chemical strain.

A useful abstraction is the process history of a material point. It includes temperature as a function of time, heating and cooling rates, peak temperature, time above a transformation or melting temperature, applied stress, exposure to atmosphere, and the sequence of neighboring deposits. Two points with identical final geometry may therefore possess different microstructures and properties because their histories differ.

2. Energy Input and Its Limitations

For a moving heat source, a commonly used first-order linear energy-density estimate is

where E_l is nominal line energy in J/m, η is effective absorptivity or process efficiency, P is source power in W, and v is travel speed in m/s. If absorptivity is omitted, the quantity P/v is the incident line energy rather than the absorbed energy.

For a raster process, a volumetric energy-density estimate is

where h is hatch spacing in m and t is layer thickness in m. This estimate is dimensionally J/m³. It is useful for comparing parameter sets, but it does not uniquely determine melt-pool dimensions or part quality. It ignores beam diameter, temporal modulation, absorptivity changes, heat conduction, scan strategy, powder packing, evaporation, and the thermal state of the part.

A worked comparison illustrates the point. Suppose a laser operates at 200 W, scans at 800 mm/s, uses a hatch spacing of 0.10 mm, and deposits a 0.04 mm layer. With an assumed effective efficiency of 0.35,

The conversion 800 mm/s = 0.8 m/s, 0.10 mm = 0.00010 m, and 0.04 mm = 0.00004 m is essential. The result is an estimate, not a universal process-quality criterion. Different parameter sets with the same E_v can produce different results because their power, speed, spot size, and scan timing differ.

Other useful indicators include the ratio of beam diameter to layer thickness, the ratio of melt-pool depth to layer thickness, and dimensionless thermal groups. The Fourier number is

where α is thermal diffusivity in m²/s, t is time in s, and L is a characteristic length in m. It compares the time available for heat diffusion with the time required for diffusion over distance L. The Peclet number for moving heat transfer is

where v is source velocity. Large Pe indicates that source motion strongly competes with heat diffusion. These dimensionless quantities support scaling and interpretation, but real AM processes often require transient numerical simulation.

3. Heat Transfer and Melt-Pool Formation

Heat enters a material by conduction, convection within the melt pool, radiation, and—in some processes—absorption of light followed by photochemical reaction. In metal powder-bed fusion, conduction through the consolidated part is usually important, while the loose powder bed has much lower effective thermal conductivity than solid metal. Consequently, the same laser exposure can create different melt-pool behavior in powder and in a dense region.

The general heat-conduction equation is

where ρ is density in kg/m³, c_p is specific heat in J/(kg·K), T is temperature in K, k is thermal conductivity in W/(m·K), and Q̇ is volumetric heat generation in W/m³. The equation assumes a continuum description; material properties may depend on temperature, and phase change requires additional latent-heat treatment.

A moving heat source creates a melt pool whose width, depth, and length depend on power, speed, beam diameter, absorptivity, material properties, surrounding temperature, and previously deposited material. Within the molten region, surface-tension gradients can drive Marangoni flow. If surface tension decreases with temperature, hot liquid generally flows toward cooler regions along the surface. This flow redistributes heat and affects penetration, mixing, and solidification morphology.

At insufficient energy input, the melt pool fails to penetrate the full layer or to remelt enough of the previous track. The result is lack-of-fusion porosity, often characterized by irregular, crack-like voids between tracks or layers. Increasing energy may improve bonding, but excessive energy can cause evaporation, recoil pressure, unstable flow, and a deep narrow keyhole. Keyhole pores form when vapor depression instability or collapse traps gas beneath the surface. Thus, process optimization seeks a stable conduction-mode melt pool rather than maximum energy.

The scan strategy also changes heat accumulation. Short vectors, island scanning, contour passes, rotation between layers, and deliberate pauses alter local temperature and cooling rate. Regions near the build plate or supports generally cool differently from regions near the top or center. Thermal boundary conditions are therefore spatially nonuniform.

4. Solidification and Microstructure

Solidification begins when liquid metal loses sufficient heat to fall below its liquidus temperature. The solidification front is governed mainly by the thermal gradient G and the interface growth rate R. A large ratio G/R generally favors more stable, directionally oriented growth, while a high product G R is associated with finer microstructural scale. These are qualitative solidification guidelines, not universal predictive equations.

The local thermal gradient is the spatial rate of temperature change, with units K/m. The growth rate is the velocity of the solid–liquid interface, with units m/s. Their values vary throughout a melt pool: the bottom and sides may experience different gradients and growth directions from the pool center. Existing grains can grow epitaxially from the underlying layer when the material is not fully remelted. This can produce columnar grains aligned approximately with the build direction, especially in alloys with strong directional heat extraction.

High cooling rates commonly produce fine cellular or dendritic structures and supersaturated solid solutions. Subsequent layers may reheat earlier material, causing recovery, precipitation, grain coarsening, phase transformation, or stress relief. Therefore, the final microstructure is a cumulative result of repeated thermal cycles rather than a single melt-and-freeze event.

In polymer extrusion, cooling controls crystallinity, shrinkage, and interlayer diffusion. A deposited polymer must remain sufficiently mobile for molecular interdiffusion across the interface, but excessive temperature can cause sagging or dimensional loss. In vat photopolymerization, exposure dose controls the extent of polymer conversion and cure depth. A simplified exposure dose is

where H is radiant exposure in J/m², I is irradiance in W/m², and t_e is exposure time in s. For a resin with working-curve parameters, cure depth is often represented approximately by

where C_d is cure depth, D_p is penetration depth, H_c is critical exposure, and the model assumes a homogeneous resin and a suitable optical regime. Overexposure can cause unintended curing beyond the intended boundary, while underexposure produces weak or incompletely bonded layers.

5. Shrinkage, Residual Stress, and Distortion

Thermal contraction is a primary source of residual stress in metal AM. A free body undergoing a uniform temperature change would contract by approximately

where ΔL is length change in m, α is the coefficient of thermal expansion in K⁻¹, L₀ is original length in m, and ΔT is temperature change in K. This linear relation assumes small deformation and approximately constant α over the temperature range.

In AM, newly molten material attempts to contract during cooling but is restrained by cooler surrounding material, earlier layers, the substrate, and supports. Elastic strain develops first; if the local stress exceeds the temperature-dependent yield strength, plastic strain remains after cooling. The final residual-stress field may include tensile stress near surfaces, compressive balancing regions, and strong gradients around supports or geometric transitions.

Distortion results when residual stresses are released by cutting the component from the build plate, removing supports, or machining. Thin walls, overhangs, large flat sections, and abrupt changes in cross-section are particularly sensitive. Mitigation includes preheating, optimized scan paths, reduced thermal gradients, suitable support design, stress-relief heat treatment, and orientation changes.

A simplified elastic relation is

where σ is normal stress in Pa, E is Young’s modulus in Pa, and ε is elastic strain. This relation applies only within the linear elastic range and does not by itself predict AM residual stress because plasticity, temperature dependence, phase transformation, and constraint must also be considered.

Polymer AM introduces thermal contraction and, for semicrystalline polymers, crystallization shrinkage. Vat-cured polymers can contract during polymerization and post-cure. These mechanisms may cause warpage, delamination, dimensional error, and internal stress.

6. Surface Formation and Fluid Flow

Surface roughness in powder-bed fusion is influenced by powder size distribution, partially melted particles, stair-stepping, melt-pool ripples, spatter, scan orientation, and the angle between a surface and the build direction. Down-facing surfaces often receive less favorable thermal support and may interact with loose powder, increasing roughness. Smaller layers reduce stair-step error but increase build time.

In extrusion, surface formation depends on bead width, layer height, nozzle diameter, pressure variation, cooling, and the relation between commanded and actual flow. If volumetric flow is too low, gaps appear between roads; if too high, material accumulates and dimensions increase. At corners, acceleration and pressure dynamics can produce under-extrusion or over-extrusion.

Fluid flow affects all melt-based processes. Surface tension, recoil pressure from evaporation, buoyancy, and Marangoni forces can produce circulation, spatter, and instability. Gas flow in powder-bed systems transports vapor and spatter away from the interaction zone but can also disturb powder or alter oxygen concentration. Inadequate atmosphere control promotes oxidation, which may impair wetting, bonding, ductility, or resin stability.

7. Defects and Their Parameter Signatures

Defects should be diagnosed from evidence rather than from a single parameter value. Important categories include:

Porosity is especially important because it reduces load-bearing area and can act as a crack-initiation site. Its effect depends on volume fraction, shape, location, orientation, and proximity to the surface. Fatigue performance is often more sensitive to defects than static tensile strength because cyclic cracks can initiate at small irregular pores or surface notches.

8. Anisotropy and the Parameter–Structure–Property Chain

AM parts are frequently anisotropic. In metals, anisotropy may result from columnar grains, crystallographic texture, elongated melt pools, layer interfaces, and directional defects. Tensile strength, ductility, fracture toughness, electrical conductivity, and fatigue life can therefore differ in the build, transverse, and scan directions.

In extrusion printing, interlayer strength is often lower than in-road strength because polymer diffusion is incomplete, the interface cools quickly, and voids may form between beads. Build orientation consequently affects both static and fatigue performance. In photopolymerization, anisotropy may arise from layer interfaces, cure gradients, and post-curing conditions.

The parameter–structure–property relationship can be represented as follows:

• Parameters: power, speed, layer thickness, hatch spacing, beam size, exposure, temperature, flow rate, orientation, and scan strategy.
• Process state: thermal field, melt-pool dimensions, flow, cooling rate, degree of cure, and interlayer contact.
• Structure: grain morphology, phase composition, crystallinity, porosity, surface texture, residual stress, and bonding area.
• Properties: strength, stiffness, hardness, ductility, fatigue life, wear, permeability, dimensional accuracy, and corrosion resistance.

No parameter should be optimized in isolation. For example, increasing power may reduce lack of fusion while increasing keyhole risk. Reducing layer thickness can improve resolution and bonding but increases build time. Increasing cooling may improve dimensional stability while reducing polymer diffusion or increasing thermal stress.

9. Process Windows and Qualification

A process window is the region of parameter space that produces acceptable geometry, density, surface quality, and properties. Qualification should combine:

1. Material characterization, including powder morphology, size distribution, moisture, resin viscosity, or filament condition.
2. Equipment calibration, including power measurement, beam focus, temperature sensors, extrusion flow, and exposure uniformity.
3. Designed experiments that vary influential factors systematically.
4. In-process monitoring, such as melt-pool imaging, layerwise optical inspection, acoustic sensing, or chamber monitoring.
5. Destructive and nondestructive evaluation, including density measurement, microscopy, tensile testing, computed tomography, and surface measurement.
6. Statistical control of critical outputs and traceable documentation.

Process monitoring does not automatically guarantee part quality. A sensor signal must be linked to a validated defect or property model. Machine learning can assist anomaly detection and parameter optimization, but training data must represent relevant materials, geometries, orientations, and failure modes. Models should be validated on independent builds and used with human engineering judgment.

10. Applied Case: Distortion in a Laser Powder-Bed Fusion Bracket

Background. An illustrative aerospace bracket is manufactured from a precipitation-strengthened nickel alloy by laser powder-bed fusion. The bracket contains a thin horizontal flange and tall support ribs.

Problem. The flange curls upward during removal from the build plate. Computed tomography also reveals small irregular pores near the underside of the flange. Tensile specimens from the same build meet strength requirements, but fatigue results are inconsistent.

Analysis. The thin flange cools rapidly and is strongly constrained by the substrate and supports. Its repeated thermal cycles generate tensile residual stress. The underside has a poor thermal and geometric condition, while the selected layer thickness and hatch spacing produce incomplete overlap in some regions. The irregular pores are therefore consistent with lack of fusion rather than spherical gas porosity. Surface-connected roughness and pores provide likely fatigue-initiation sites.

Approach. The engineering team changes the build orientation to reduce the unsupported horizontal area, adds supports that improve heat conduction, rotates the scan vectors between layers, and applies a qualified preheat condition. A small design-of-experiments study adjusts speed and hatch spacing while keeping the melt pool in a stable conduction regime. Test coupons are built in orientations representing the bracket. Density, residual distortion, roughness, tensile properties, and fatigue behavior are measured. A stress-relief heat treatment is applied before support removal, and the bracket is machined at fatigue-critical surfaces.

Outcome. Distortion decreases, irregular lack-of-fusion pores are reduced, and fatigue scatter narrows. The final solution is not a single “best” energy density; it is an integrated change to orientation, support conduction, scan strategy, process parameters, post-processing, and inspection.

Learning points. Geometry and parameters cannot be separated. Fatigue qualification must consider defect morphology and surface condition, not only tensile strength. Process changes require requalification because a parameter set suitable for one geometry may fail in another.

11. Best Practices, Limitations, and Responsible Use

Best practice begins with identifying the dominant physical mechanism. If the defect is lack of fusion, changing a heat-treatment schedule alone is unlikely to solve it. If distortion is caused by restraint and thermal gradients, simply increasing source power may worsen the problem. Engineers should control material condition, maintain traceability, use parameter sets validated for the specific machine and alloy, and inspect critical regions using appropriate methods.

Important limitations remain. Material properties vary with temperature and processing history. Effective absorptivity is difficult to measure in situ. Energy-density metrics compress several independent variables into one number and can conceal unstable regimes. Numerical simulations depend on mesh resolution, boundary conditions, calibrated heat-source models, and accurate phase and property data. Laboratory coupons may not represent complex production geometries.

Responsible AM practice includes safe handling of fine powders, control of laser and ultraviolet hazards, ventilation of volatile emissions, prevention of oxygen-related fire or explosion risks, and appropriate management of uncured resin and contaminated waste. Production records should preserve material lot, machine identity, parameter version, build orientation, environmental conditions, monitoring data, and post-processing history. These records support quality assurance, regulatory compliance, recall capability, and ethical communication of part limitations.

Chapter Summary

AM performance is governed by coupled thermal, mechanical, fluid-flow, photochemical, and solidification phenomena. Energy input influences melt-pool formation, but nominal energy density is only a comparative indicator and cannot replace process-specific validation. Heat conduction, melt-pool convection, surface tension, recoil pressure, cooling rate, and repeated thermal cycles determine bonding and microstructure. Insufficient energy commonly produces lack of fusion, whereas excessive localized energy may produce keyhole porosity and instability.

Solidification conditions influence grain morphology, texture, phase constitution, and anisotropy. Thermal contraction, constraint, plasticity, and phase changes generate residual stress and distortion. Surface roughness is shaped by powder, flow, stair-stepping, bead geometry, spatter, and orientation. The final properties depend on the complete chain from parameters to process state, structure, defects, and post-processing. Reliable production therefore requires process windows, monitoring, inspection, qualification, traceability, and recognition of model limitations.

Key Terms

Process window: Range of process conditions that produces acceptable parts.

Melt pool: Localized liquid region created by an energy source.

Line energy: Incident or effective energy per unit scan length.

Volumetric energy density: Approximate energy per unit processed volume.

Absorptivity: Fraction of incident energy absorbed by the material.

Lack-of-fusion porosity: Irregular voids caused by incomplete melting or bonding.

Keyhole porosity: Voids associated with unstable deep vapor cavities.

Gas porosity: Generally rounded pores formed by trapped or dissolved gas.

Marangoni flow: Surface-tension-driven liquid motion caused by temperature or composition gradients.

Thermal gradient: Spatial temperature change per unit distance.

Cooling rate: Rate at which material temperature decreases.

Epitaxial growth: Crystal growth from an underlying grain across a solidification interface.

Residual stress: Stress remaining after external loads or thermal gradients are removed.

Distortion: Permanent geometric deviation caused by stress release or shrinkage.

Anisotropy: Direction-dependent material behavior.

Interlayer bonding: Physical or chemical joining between successive deposited layers.

Cure depth: Depth of photopolymerization produced by a specified exposure.

Process monitoring: Measurement of process signals during manufacturing.

Qualification: Demonstration that a defined process repeatedly meets requirements.

Review Questions

Short Answer Questions

1. Why is volumetric energy density useful but insufficient as a complete AM process descriptor?
2. Distinguish lack-of-fusion porosity from keyhole porosity.
3. How does thermal gradient influence solidification structure?
4. What mechanisms generate residual stress in metal AM?
5. Why can AM components be anisotropic?

Long Answer Questions

1. Explain the thermal history of a point in a powder-bed fusion component and relate it to microstructure.
2. Discuss how power, speed, hatch spacing, and layer thickness jointly affect melt-pool stability and bonding.
3. Explain the origins of distortion and evaluate engineering methods for reducing it.
4. Compare layer formation and bonding in metal powder-bed fusion, material extrusion, and vat photopolymerization.
5. Describe a robust qualification strategy for a safety-critical AM component.

Analytical / Application Questions

1. A process shows irregular interlayer pores and poor fatigue performance, while tensile strength is acceptable. What physical mechanisms are likely, and what parameter and inspection changes would you recommend?
2. Two parameter sets have the same nominal volumetric energy density, but one produces keyhole pores and the other produces lack of fusion. Explain why this is possible.
3. A thin polymer wall warps and has weak interlayer strength. Analyze the competing effects of extrusion temperature, cooling rate, bead geometry, build orientation, and post-processing.

Multiple Choice Questions with Answers

Question 1

Which defect is most directly associated with insufficient melting or overlap between adjacent tracks?

A. Lack-of-fusion porosity

B. Keyhole porosity

C. Gas porosity

D. Polymer crystallization

Correct Answer: A. Lack-of-fusion porosity

Question 2

In the expression E_l = ηP/v, what are the units of E_l?

A. J/m

B. J/m³

C. W/m²

D. K/m

Correct Answer: A. J/m

Question 3

What is a likely consequence of excessive localized energy input in metal powder-bed fusion?

A. Keyhole instability and deep pores

B. Guaranteed elimination of all porosity

C. Lower melt-pool temperature

D. No change in fluid flow

Correct Answer: A. Keyhole instability and deep pores

Question 4

Which mechanism can drive circulation in a molten metal surface?

A. Marangoni flow

B. Photopolymerization only

C. Elastic recovery only

D. Powder sieving

Correct Answer: A. Marangoni flow

Question 5

If P = 200 W, v = 0.8 m/s, and η = 0.5, what is the effective line energy?

A. 125 J/m

B. 250 J/m

C. 400 J/m

D. 800 J/m

Correct Answer: B. 250 J/m

Question 6

Which relationship describes small, free thermal expansion under approximately constant properties?

A. ΔL = αL₀ΔT

B. σ = E/ε

C. Fo = vL/α

D. H = I/t_e

Correct Answer: A. ΔL = αL₀ΔT

Question 7

Why can columnar grains form in metal AM?

A. Directional heat extraction and epitaxial growth

B. Uniform cooling in every direction

C. Complete absence of a thermal gradient

D. Only mechanical polishing

Correct Answer: A. Directional heat extraction and epitaxial growth

Question 8

In vat photopolymerization, what does radiant exposure represent?

A. Irradiance multiplied by exposure time

B. Exposure time divided by irradiance

C. Temperature multiplied by density

D. Resin viscosity divided by layer height

Correct Answer: A. Irradiance multiplied by exposure time

Question 9

Which feature is most likely to increase distortion in a metal AM part?

A. A large thin unsupported flange

B. A compact uniformly supported cube

C. A fully stress-relieved plate

D. A low-temperature inspection coupon

Correct Answer: A. A large thin unsupported flange

Question 10

Why should equal nominal energy-density values not be assumed to produce equal part quality?

A. Power, speed, beam size, timing, and thermal conditions also affect the process

B. Energy density has no units

C. Melt pools do not depend on process parameters

D. All materials have identical absorptivity

Correct Answer: A. Power, speed, beam size, timing, and thermal conditions also affect the process