Droplet deposition, powder binding, large-scale deposition, emerging methods, and how to select a process
By completing this chapter, you will be able to explain the following:
Chapter 1 covered material extrusion (MEX), and Chapters 2 and 3 covered vat photopolymerization (VPP) and powder bed fusion (PBF). This chapter takes a cross-cutting look at the remaining major processes: material jetting, binder jetting, directed energy deposition, and sheet lamination, and finishes by addressing the question of "which method to choose, and when." The goal is less about operating specific machines and more about acquiring the axes for comparing processes.
Material jetting (MJ) jets liquid material as tiny droplets using the same principle as an inkjet printer, then cures them in place with ultraviolet (UV) light to build up layers. It is also called PolyJet, after the Stratasys trademark. The print head carries hundreds to thousands of nozzles that deposit droplets of photocurable resin (photopolymer) exactly where needed on each layer, and a UV lamp immediately behind the head solidifies them at once.
Because MJ handles an entire area with a nozzle array at once, it is faster than point-scanning SLA while remaining high-resolution thanks to the small droplet size. On the other hand, its essential constraints are that usable materials are limited to photocurable resins and that mechanical properties (strength, heat resistance) remain moderate.
Whether a droplet forms "cleanly as a single drop" is decided by the balance among viscosity, surface tension, and inertia. The following dimensionless numbers express this. Each term is defined on first use.
Empirically, the window for stable droplet formation is roughly 1 < Z < 10 (up to about 14 depending on the source). If Z is too small (too viscous), the droplet does not detach; if it is too large, it trails a tail and produces satellite droplets. In addition, a sufficient We is needed for ejection.
The 1 < Z < 10 window is an empirical rule derived from many experiments; the actual window shifts with the resin's non-Newtonian behavior (shear-dependent viscosity), the drive waveform, and nozzle geometry. It is useful for early-stage screening, but the honest approach is to leave the final judgment to droplet observation on the real machine (a drop watcher).
MJ's greatest strength is that it can assign different materials to different nozzle groups and switch materials or colors freely within a single build. It can create "digital materials" that blend rigid and soft resins in a gradient to produce intermediate hardnesses from rubbery to rigid, and full-color builds of more than 10 million colors by combining CMYK plus white and clear resins. In medical anatomical models, bone (hard), soft tissue (soft), and blood vessels (clear) can be built into a single piece, which is valuable for surgical planning.
In MJ, a separate support material (often gel-like or wax-like) is jetted at the same time as the build material to support overhangs and hollows. Removal is mainly of two kinds:
The cost of support material and the labor of removal directly affect the effective cost and yield of MJ. Even though the build itself is high-precision, it is important to evaluate it by total cost including post-processing.
Binder jetting (BJ) jets a liquid binder onto a thin powder bed with an inkjet head, gluing the powder particles together to form each layer. Because it uses no laser or heat source to melt the material, the build itself proceeds quickly at room temperature. The formed body straight off the build is called the green part (an unsintered, fragile state).
The green part lacks strength on its own, so it is densified in post-processing. For metals and ceramics, the main routes are:
During sintering, as the relative density (the fraction of theoretical density) increases, the volume decreases and the part shrinks. Assuming isotropic shrinkage, the linear shrinkage can be expressed as the cube root of the density ratio.
Here ρ_green is the green relative density and ρ_sinter is the sintered relative density. To obtain the target dimension, you design the green part larger to anticipate this shrinkage; the scale factor is (ρ_sinter / ρ_green)^(1/3). It is not unusual for linear shrinkage in metal BJ to reach 15-20%, and without compensating for it the part is unusable. This calculation is carried out in Code Example 2 below.
Directed energy deposition (DED) feeds metal powder or wire while melting it with a laser, electron beam, or arc, building up material on a substrate. Because the nozzle and energy source move together, mounting them on a multi-axis robotic arm imposes few limits on build size and can handle large parts. LENS (Laser Engineered Net Shaping) is a representative trademark for the laser-powder variant.
Rather than "creating a precise shape from scratch," DED shows its real strength in "depositing large and fast" and "repairing what is broken." Because finishing presupposes machining, it is closely tied to the hybrid manufacturing discussed in the next section.
Sheet lamination (SL) stacks sheet materials such as paper, metal foil, or plastic film, bonds or welds them, and cuts the contour of each layer. The two representative technologies are:
Hybrid manufacturing is an approach that performs additive processing (AM) and subtractive processing (machining) alternately within a single machine. By depositing a preform with DED, milling the surface to a finish while it is still accessible, and then depositing again, it combines AM's geometric freedom with the surface and dimensional precision of CNC. The ability to perform deposition repair on an existing part plus finishing in one continuous process is another reason industrial adoption is growing.
None of the methods above is "universal." Selection is an evaluation of trade-offs against the application requirements. The main comparison axes are summarized below.
| Process | Accuracy / surface | Speed | Main materials | Strength | Best-suited use |
|---|---|---|---|---|---|
| Material Jetting (MJ) | Very high | Medium | Photocurable resin (multi-material, full-color) | Low to medium | Appearance models, medical anatomical models |
| Binder Jetting (BJ) | Medium | High | Metals, ceramics, sand, gypsum | Medium (after sintering) | Sand molds, metal mass production, full-color figures |
| DED / LENS | Low (needs post-machining) | Very high (deposition) | Metal (powder / wire) | High | Repair, large parts, graded materials |
| Sheet Lamination (LOM / UAM) | Medium | High | Paper, metal foil | Low to medium | Visual models, embedded sensors |
A common mistake in selection is deciding on a method by a single metric (e.g., accuracy) alone. Making a single sand mold does not need MJ's ultra-high precision, and BJ cannot be used for repair. The practical principle is to choose the minimally sufficient method that meets the requirement, judged by total cost (material + post-processing + yield). Code Example 3 quantifies this idea as weighted scoring.
Bioprinting dispenses a "bio-ink" (living cells plus a hydrogel carrier) to construct tissue and organ models. To keep cells viable, low-pressure, low-viscosity conditions that limit shear stress during jetting are required, so the droplet-physics knowledge covered in this chapter applies directly. At present the focus is on tissue chips for drug screening and small pieces of skin or cartilage; transplantable organs are still at the research stage. It is important to avoid overhyping and to assess maturity honestly.
4D printing designs parts so that their shape changes after the build in response to stimuli such as temperature, humidity, or light. The "fourth dimension" is time, that is, the shape changing over time. Shape-memory polymers or materials that swell with moisture are placed with directionality to create structures that self-fold from flat to three-dimensional. Deployable antennas, self-assembling parts, and soft robots are candidate applications, but here too practical use remains limited to a few cases.
Let us confirm the key points of this chapter with runnable Python code. All outputs below are actual results executed with python3 (using NumPy).
For representative jetting fluids, we compute the dimensionless numbers and judge whether they fall inside the printable window (1 < Z < 10 and We > 4).
import numpy as np
# Material Jetting droplet printability: Ohnesorge / Reynolds / Weber / Z number
# Z = 1/Oh ; printable window commonly cited as 1 < Z < 10 (some report up to 14)
# Oh = mu / sqrt(rho * sigma * D)
def dimensionless(rho, mu, sigma, D, v):
Oh = mu / np.sqrt(rho * sigma * D)
Z = 1.0 / Oh
Re = rho * v * D / mu
We = rho * v**2 * D / sigma
return Oh, Z, Re, We
# Representative jetting fluids (SI units)
# rho [kg/m3], mu [Pa.s], sigma [N/m], D nozzle [m], v drop [m/s]
fluids = [
("UV acrylate resin (PolyJet)", 1100, 0.012, 0.030, 30e-6, 8.0),
("Molten wax (support)", 900, 0.020, 0.025, 30e-6, 6.0),
("Water-thin binder (BJ)", 1000, 0.001, 0.072, 40e-6, 9.0),
("Nanoparticle metal ink", 1500, 0.015, 0.035, 20e-6, 7.0),
("Over-viscous resin (fail)", 1150, 0.080, 0.030, 30e-6, 8.0),
]
print(f"{'Fluid':32s}{'Oh':>8s}{'Z=1/Oh':>9s}{'Re':>8s}{'We':>8s} Printable(1<Z<10)")
print("-"*80)
for name, rho, mu, sigma, D, v in fluids:
Oh, Z, Re, We = dimensionless(rho, mu, sigma, D, v)
ok = "YES" if (1.0 < Z < 10.0 and We > 4.0) else "NO"
print(f"{name:32s}{Oh:8.3f}{Z:9.2f}{Re:8.1f}{We:8.1f} {ok}")
print()
print("Interpretation:")
print(" - Z < 1 : too viscous, droplet won't form cleanly")
print(" - Z > 10 : satellite droplets / instability")
print(" - We < 4 : insufficient energy to eject a droplet")Execution result:
Fluid Oh Z=1/Oh Re We Printable(1<Z<10)
--------------------------------------------------------------------------------
UV acrylate resin (PolyJet) 0.381 2.62 22.0 70.4 YES
Molten wax (support) 0.770 1.30 8.1 38.9 YES
Water-thin binder (BJ) 0.019 53.67 360.0 45.0 NO
Nanoparticle metal ink 0.463 2.16 14.0 42.0 YES
Over-viscous resin (fail) 2.487 0.40 3.5 73.6 NO
Interpretation:
- Z < 1 : too viscous, droplet won't form cleanly
- Z > 10 : satellite droplets / instability
- We < 4 : insufficient energy to eject a dropletThe numbers confirm that an overly viscous resin (Z = 0.40) will not detach into droplets, while a low-viscosity binder like water (Z = 53.7) is prone to satellite droplets. The PolyJet resin and the nanoparticle ink fall inside the window.
From relative density we compute linear and volumetric shrinkage, and the green-design scale factor needed to obtain the target dimension.
import numpy as np
# Binder Jetting: green part -> sintered part shrinkage from densification.
# Isotropic linear shrinkage from relative density change:
# L_sinter / L_green = (rho_green / rho_sinter)^(1/3)
# Linear shrinkage (%) = (1 - (rho_g/rho_s)^(1/3)) * 100
def linear_shrinkage(rho_green, rho_sinter):
ratio = (rho_green / rho_sinter) ** (1.0/3.0)
lin = (1.0 - ratio) * 100.0
vol = (1.0 - rho_green / rho_sinter) * 100.0
return lin, vol
# rho values are RELATIVE density (fraction of theoretical)
cases = [
("316L stainless (metal BJ)", 0.55, 0.98),
("Ti-6Al-4V (metal BJ)", 0.50, 0.96),
("Alumina ceramic", 0.45, 0.95),
("Bronze-infiltrated steel", 0.60, 0.90),
]
print(f"{'System':30s}{'rho_green':>10s}{'rho_sint':>9s}{'Lin.shr%':>10s}{'Vol.shr%':>10s}")
print("-"*70)
for name, rg, rs in cases:
lin, vol = linear_shrinkage(rg, rs)
print(f"{name:30s}{rg:10.2f}{rs:9.2f}{lin:10.2f}{vol:10.2f}")
# Compensation: to hit a 50.00 mm target after sintering, scale the green CAD.
target = 50.00 # mm final dimension
rg, rs = 0.55, 0.98
scale = (rs / rg) ** (1.0/3.0) # green must be LARGER by this factor
green_dim = target * scale
print()
print(f"Design compensation (316L, rho_g=0.55 -> rho_s=0.98):")
print(f" required green scale factor = {scale:.4f}")
print(f" to obtain {target:.2f} mm final, model green part at {green_dim:.3f} mm")Execution result:
System rho_green rho_sint Lin.shr% Vol.shr%
----------------------------------------------------------------------
316L stainless (metal BJ) 0.55 0.98 17.51 43.88
Ti-6Al-4V (metal BJ) 0.50 0.96 19.54 47.92
Alumina ceramic 0.45 0.95 22.05 52.63
Bronze-infiltrated steel 0.60 0.90 12.64 33.33
Design compensation (316L, rho_g=0.55 -> rho_s=0.98):
required green scale factor = 1.2123
to obtain 50.00 mm final, model green part at 60.617 mmFor 316L stainless steel, linear shrinkage reaches about 17.5% and volumetric shrinkage about 44%. To obtain a final dimension of 50.00 mm, the green part must be designed at 60.6 mm, showing how essential shrinkage compensation is.
We assign weights to six evaluation axes (accuracy, surface, speed, material range, cost efficiency, strength) and compare processes quantitatively per use case. We also show that changing the weights changes the recommendation (sensitivity).
import numpy as np
# AM process selection by weighted scoring.
# Criteria scored 1-5 (5 = best for that criterion).
criteria = ["accuracy", "surface", "speed", "material_range", "cost_eff", "strength"]
weights = np.array([0.25, 0.15, 0.15, 0.15, 0.15, 0.15]) # sums to 1.0
# rows = processes, cols = criteria (expert-assigned 1-5)
processes = {
"Material Jetting (MJ)": [5, 5, 3, 2, 2, 2],
"Binder Jetting (BJ)": [3, 3, 5, 4, 4, 3],
"DED / LENS": [2, 1, 4, 4, 3, 5],
"Sheet Lamination (LOM)": [2, 2, 4, 2, 5, 2],
"PBF (SLM/SLS)": [4, 3, 2, 4, 2, 5],
"Material Extrusion (FDM)":[2, 2, 3, 3, 5, 3],
}
print(f"weights: {dict(zip(criteria, weights))}")
print()
print(f"{'Process':28s}{'Score':>7s} Ranked criteria contribution")
print("-"*70)
results = []
for name, sc in processes.items():
sc = np.array(sc, dtype=float)
total = float(np.dot(weights, sc))
results.append((name, total))
for name, total in sorted(results, key=lambda x: -x[1]):
bar = "#" * int(round(total*6))
print(f"{name:28s}{total:7.3f} {bar}")
best = max(results, key=lambda x: x[1])
print()
print(f"Recommended (accuracy-weighted use case): {best[0]} (score {best[1]:.3f})")
# Re-run with a "cheap large metal part" weighting to show sensitivity
w2 = np.array([0.05, 0.05, 0.25, 0.15, 0.30, 0.20])
print()
print("Re-weighted for 'low-cost large metal part' (cost & speed heavy):")
r2 = [(n, float(np.dot(w2, np.array(s, dtype=float)))) for n, s in processes.items()]
for name, total in sorted(r2, key=lambda x: -x[1])[:3]:
print(f" {name:28s}{total:7.3f}")Execution result:
weights: {'accuracy': np.float64(0.25), 'surface': np.float64(0.15), 'speed': np.float64(0.15), 'material_range': np.float64(0.15), 'cost_eff': np.float64(0.15), 'strength': np.float64(0.15)}
Process Score Ranked criteria contribution
----------------------------------------------------------------------
Binder Jetting (BJ) 3.600 ######################
PBF (SLM/SLS) 3.400 ####################
Material Jetting (MJ) 3.350 ####################
DED / LENS 3.050 ##################
Material Extrusion (FDM) 2.900 #################
Sheet Lamination (LOM) 2.750 ################
Recommended (accuracy-weighted use case): Binder Jetting (BJ) (score 3.600)
Re-weighted for 'low-cost large metal part' (cost & speed heavy):
Binder Jetting (BJ) 3.950
DED / LENS 3.650
Material Extrusion (FDM) 3.500Under accuracy-weighted weights, binder jetting, PBF, and material jetting are close; but reweighting for a "low-cost, large metal part" by emphasizing cost and speed swaps the ranking. This sensitivity analysis plainly shows that selection changes with the weights (that is, the requirements).
These exercises check your understanding. Think it through yourself before opening the answers.
Exercise 1 (Basics): Matching processes
For each use below, choose the most suitable AM method from MJ / BJ / DED / SL, and add a one-line reason.
(a) Repairing a worn turbine blade (b) A sand mold for an engine block (c) A medical anatomical model with distinct hard and soft regions (d) A structure with sensors embedded inside metal foil
(a) DED: deposition repair onto an existing part is the only practical option.
(b) BJ: forms sand molds fast and large with no sintering.
(c) MJ: can build multiple hardnesses and clear material in a single build.
(d) SL (UAM): its low-temperature solid-state joining allows sensors to be embedded inside.
Exercise 2 (Calculation): Ohnesorge number and Z number
For a resin with density ρ = 1100 kg/m³, viscosity μ = 0.010 Pa·s, surface tension σ = 0.030 N/m, and nozzle diameter D = 30 μm, find the Ohnesorge number and Z number, and judge whether it falls in the printable window (1 < Z < 10).
Oh = μ / √(ρ σ D) = 0.010 / √(1100 × 0.030 × 30e-6) = 0.010 / √(9.9e-4) = 0.010 / 0.03146 ≈ 0.318.
Z = 1/Oh ≈ 3.15. Since 1 < 3.15 < 10, it is inside the printable window. You can check this by passing the same values to the dimensionless function in Code Example 1.
Exercise 3 (Calculation): Dimensional compensation for sintering shrinkage
For a material with green relative density 0.52 and sintered relative density 0.97, you want a final dimension of 40.0 mm. At what size should the green part be designed? Also find the linear shrinkage.
Scale factor = (0.97/0.52)^(1/3) = (1.865)^(1/3) ≈ 1.231. Green dimension = 40.0 × 1.231 ≈ 49.2 mm.
Linear shrinkage = (1 − (0.52/0.97)^(1/3)) × 100 = (1 − 0.812) × 100 ≈ 18.8%.
Exercise 4 (Discussion): Satellite droplets
You warmed a resin to lower its viscosity, and the Z number rose to 12. What print-quality problem would you expect, and how could you address it?
When Z exceeds the upper bound of the window (about 10), the droplet trails a tail and breaks up, making satellite droplets more likely. This leads to landing-position errors and mist contamination. Remedies include: (1) heat less to raise the viscosity slightly and bring Z back within the window, (2) tune the drive waveform for cleaner tail break-off, and (3) optimize the droplet velocity (We). However, the final check requires direct observation with a drop watcher.
Exercise 5 (Application): Designing selection weights
For the requirement "produce high-precision full-color dental models in low-volume, high-mix," how would you set the weights for the six axes in Code Example 3 (accuracy, surface, speed, material range, cost efficiency, strength), and which method do you expect to be chosen?
Weight accuracy, surface, and material range (full color) heavily, and speed, strength, and cost lightly (e.g., accuracy 0.30 / surface 0.25 / material 0.20 / speed 0.10 / cost 0.05 / strength 0.10). Under these weights, material jetting (MJ), which combines full color and high precision, is expected to rank at the top. Rewriting weights in Code Example 3 lets you feel how weight design drives the conclusion.
In this chapter we learned the remaining major AM processes, following material extrusion, vat photopolymerization, and powder bed fusion, along with how to select among them. The key points are:
With this, the overall picture of the major AM processes is complete. In the next chapter, we integrate the knowledge so far and take on simulation and analysis of 3D printing in Python.
In Chapter 4, we took a cross-cutting look at material jetting, binder jetting, directed energy deposition, sheet lamination, and hybrid manufacturing, and surveyed a quantitative approach to process selection along with emerging trends. In the next chapter, Chapter 5, we take on simulation and practical analysis of 3D printing using Python.