Materials that live inside us
In 1949, the London eye surgeon Harold Ridley implanted the first artificial lens in a human eye. He had noticed that wartime pilots with splinters of acrylic canopy in their eyes tolerated them without inflammation, and he made his lens of the same polymer, PMMA. In 1962 John Charnley, at Wrightington Hospital in the north of England, began the modern hip replacement: a small stainless-steel ball on a polyethylene socket, both fixed with acrylic cement, after an earlier version with a PTFE socket had worn through within a few years. In Sweden, Per-Ingvar Brånemark found that bone grows into intimate contact with titanium, and in 1965 placed the first titanium dental implants in a patient.
Today more than a million hip replacements are performed each year worldwide, and knee replacements, heart valves, coronary stents, dental implants, intraocular lenses, pacemakers, breast implants and sutures add many millions more. Each is a materials problem of an unusual kind. The environment is warm salt water full of proteins and cells that actively respond to what is put there; the loads are cyclic, millions of times a year; and failure is measured in pain, further surgery and sometimes lives.
This lesson draws on all the families you have studied — metals, ceramics, polymers and composites — to ask what makes a material suitable for the body, and how engineers, surgeons and regulators decide.
What biocompatibility means
It is tempting to think a good implant material is one the body ignores. The field learned otherwise. The consensus definition, framed in the 1980s by David Williams and refined since (Williams), is about performance in context.
Definition— biocompatibility
Biocompatibility is the ability of a material to perform its intended function in a specific application while provoking an appropriate response from the host. It is a property of a material in a use, not of a material alone: titanium is biocompatible as a hip stem in bone, but would be a poor choice for a blood-contacting heart valve leaflet.
Within seconds of implantation, proteins from blood and tissue fluid coat the surface, and the cells that arrive later meet that protein layer, never the bare material. Its composition depends on surface chemistry, charge, wettability and roughness, and it changes over minutes to hours as more abundant proteins are displaced by ones that bind more strongly (the Vroman effect). Then comes the response to the surgical wound itself: acute inflammation over days, as neutrophils and macrophages arrive. Around a material that cannot be digested, macrophages may fuse into foreign-body giant cells, and after a few weeks the implant is usually walled off by a thin fibrous capsule of collagen. A thin, stable capsule is a normal, acceptable response for many devices. A thick or contracting one, persistent inflammation, blood clotting on a stent or valve, infection by bacteria that form a protective biofilm on the surface, or toxic, allergic or carcinogenic effects of released ions and particles are the failures biocompatibility testing looks for.
Materials are often grouped by the response they are designed for:
- bioinert (better, nearly inert): provoke minimal reaction and become encapsulated or, in bone, closely apposed: titanium, alumina, zirconia, polyethylene;
- bioactive: bond chemically to tissue by forming a layer of bone-like apatite on their surface: hydroxyapatite, certain calcium phosphates, and the bioactive glass discovered by Larry Hench in 1969 (Hench et al.);
- bioresorbable: dissolve or degrade, ideally as the tissue heals and takes over: poly(lactic acid) and its relatives, tricalcium phosphate, magnesium alloys.
Biological safety is assessed under the ISO 10993 series of standards, which set out which tests — cytotoxicity, sensitisation, irritation, systemic toxicity, genotoxicity, implantation, blood compatibility — a device needs, according to what it touches and for how long.
Metals, ceramics and polymers in the body
The total hip replacement shows all three families working together, each chosen for one job.
Metals carry load. 316L stainless steel is cheap and tough and is used mainly for temporary fixation — plates, screws, wires — but releases nickel, to which some people are allergic, and is prone to crevice corrosion over decades. Cobalt–chromium–molybdenum alloys are hard and wear-resistant, the classic material for femoral heads and knee components. Titanium and Ti‑6Al‑4V resist corrosion superbly, are about half as stiff as steel, and allow bone to grow into close contact (osseointegration), so they make stems, acetabular shells, dental implants and spinal hardware. Nitinol, a nearly equiatomic nickel–titanium alloy, is superelastic at body temperature: its austenite transforms to martensite under stress and back again on unloading (a martensitic transformation used in reverse), so a nitinol stent can be crushed into a catheter and spring back to shape in an artery. Implant-grade alloys have their own specifications, such as ASTM F136 for Ti‑6Al‑4V ELI and F75 for cast cobalt–chromium–molybdenum, with tighter limits on impurities and inclusions than industrial grades.
Ceramics carry wear and contact. Alumina and zirconia-toughened alumina femoral heads are harder and smoother than metal, giving very low wear against polyethylene or against each other; their risk is rare brittle fracture, minimised by the fracture-mechanics design and proof testing you met in ceramics and glasses. Zirconia makes dental crowns and implants. Hydroxyapatite, , close to the mineral of bone, is plasma-sprayed onto implants to speed bone bonding, and porous calcium phosphates and bioactive glasses fill bone defects.
Polymers carry everything else. Ultra-high-molecular-weight polyethylene (UHMWPE) is the bearing surface of most hip and knee replacements. PMMA is bone cement, mixed in theatre as a liquid monomer and powder that polymerise in the patient within minutes, and the material of hard contact lenses and many intraocular lenses. Silicones make catheters, shunts and breast implants; PEEK makes spinal cages; expanded PTFE and woven polyester make large vascular grafts; polyurethanes are used in catheters and pacemaker leads; hydrogels, including silicone hydrogels, make soft contact lenses; and degradable polyesters make sutures and screws.
That last point became the central story of polyethylene. Wear particles from early liners provoked macrophages to release signals that caused bone around the implant to resorb (osteolysis), loosening it after ten to fifteen years. Cross-linking the polyethylene by irradiation, introduced around 2000, cut wear dramatically; adding vitamin E protects the irradiated material from oxidation. Registries now record long survival for most modern hip designs. Metal-on-metal hips taught the opposite lesson, told in the section on regulation below.
Stress shielding
Bone is a living material that adapts its mass and architecture to the loads it carries: it thickens where it is strained and is resorbed where it is not. A stiff metal stem inside the femur takes over part of the load that bone used to carry, the bone around it is under-strained, and over years it can thin, a phenomenon called stress shielding. How much load does the stem steal? Treat the femoral shaft, cement and stem as perfectly bonded concentric cylinders: under axial load they share one strain, so each carries load in proportion to its axial stiffness ; in bending, in proportion to its bending stiffness .
import numpy as np
# A femoral shaft with a cemented stem, modelled as three perfectly bonded concentric
# cylinders: stem (solid), PMMA cement mantle, cortical bone tube. Dimensions in mm, E in GPa.
D_bone, d_canal, d_stem = 28.0, 16.0, 14.0
E_bone, E_cement = 18.0, 2.5 # cortical bone along the femur; PMMA bone cement
def area(do, di=0.0):
return np.pi / 4 * (do**2 - di**2)
def second_moment(do, di=0.0):
return np.pi / 64 * (do**4 - di**4)
stems = {"CoCrMo alloy": 220.0, "316L stainless": 195.0, "Ti-6Al-4V": 112.0,
"carbon-fibre PEEK (approx.)": 18.0, "no stem (intact bone)": None}
print(f"{'stem':28} {'bone share, axial':>18} {'bone share, bending':>20}")
for name, E_stem in stems.items():
parts_EA = [E_bone * area(D_bone, d_canal)]
parts_EI = [E_bone * second_moment(D_bone, d_canal)]
if E_stem is not None:
parts_EA += [E_cement * area(d_canal, d_stem), E_stem * area(d_stem)]
parts_EI += [E_cement * second_moment(d_canal, d_stem), E_stem * second_moment(d_stem)]
print(f"{name:28} {parts_EA[0] / sum(parts_EA):17.0%} {parts_EI[0] / sum(parts_EI):20.0%}")With a cobalt–chromium stem, the bone carries only 18 % of an axial load and 54 % of a bending moment; since the parts share strain, the bone's strain, and so the stimulus it senses, falls in the same proportion. A titanium-alloy stem, half as stiff, leaves the bone 30 % and 69 %. A stem as compliant as bone, such as carbon-fibre PEEK, would leave it 72 % and 93 %. Bending is shielded less because the bone tube lies outside the stem, where it contributes most to .
The model is crude — real stems taper, load is transferred mostly near the top of the femur, and bonding is imperfect — but its lessons hold in finite-element studies and in X-rays of patients: bone loss around stiff stems is concentrated near the top of the femur, and less stiff designs preserve more bone. Why, then, are stems not made of compliant materials? Because a compliant stem moves more relative to the bone at their interface, and that micromotion can stop bone growing onto the stem and loosen it. The answer engineers have converged on is a titanium-alloy stem, tapered and porous-coated, sometimes with a lattice section made by additive manufacturing, which is the subject of the scaffold section below.
Degradable materials
Many implants are needed only while tissue heals: a suture, a screw holding a fracture, a stent propping open an artery while its wall recovers. A material that degrades after doing its job avoids a second operation to remove it, and it hands load back to the healing tissue gradually, the opposite of stress shielding.
Degradable polyesters — poly(glycolic acid) (PGA), poly(lactic acid) (PLA) and their copolymers (PLGA), and polycaprolactone (PCL) — break down by hydrolysis of their ester bonds (polymers). Water enters the whole bulk faster than the chains break, so they undergo bulk erosion: molar mass falls throughout the part, strength is lost long before mass, and only when chains become short enough to dissolve does the part lose weight and finally disappear. The acidic chain ends catalyse further hydrolysis, so thick parts can degrade faster inside than at the surface. Degradation rates span a wide range: PGA sutures lose most of their strength in weeks, semicrystalline poly(L-lactic acid) screws take years to resorb. Other polymers, such as some polyanhydrides, erode only from the surface, like a bar of soap, which is useful for steady drug release.
Degradable metals — magnesium, zinc and iron alloys — are stronger than polymers. Magnesium corrodes in body fluid, , and magnesium-alloy bone screws have been approved in Europe since 2013, and a magnesium coronary scaffold since 2016. The engineering question is the rate.
import numpy as np
# A biodegradable magnesium-alloy bone screw, idealised as a solid cylinder that corrodes
# uniformly inwards from its surface: Mg + 2 H2O -> Mg(OH)2 + H2
r0, length = 1.75, 30.0 # mm
rho_Mg, M_Mg = 1.74, 24.305 # mg/mm^3, mg/mmol
V_gas = 25.4 # mL of gas per mmol at 37 °C and 1 atm
V0 = np.pi * r0**2 * length
n_H2 = V0 * rho_Mg / M_Mg # one H2 per Mg
print(f"screw volume {V0:.0f} mm^3 -> {n_H2:.1f} mmol of H2 = {n_H2 * V_gas:.0f} mL over its life")
print("\ncorrosion rate time to vanish H2 in the first year per day strength left after 6 months")
for rate in (0.1, 0.25, 0.5, 1.0): # corrosion rates to compare, mm per year
life = r0 / rate
r_1 = max(r0 - rate, 0.0)
gas_1 = np.pi * (r0**2 - r_1**2) * length * rho_Mg / M_Mg * V_gas
r_half = max(r0 - rate / 2, 0.0)
strength = (r_half / r0) ** 3 # bending strength of a round section ~ r^3
print(f"{rate:9.2f} mm/yr {life:12.1f} yr {gas_1:16.0f} mL {gas_1 / 365:9.2f} mL {strength:16.0%}")A 3.5 mm screw 30 mm long releases about half a litre of hydrogen over its life. At 0.25 mm a year it would vanish in about seven years and keep 80 % of its bending strength after six months — about right for a healing fracture — while releasing roughly 0.4 mL of gas a day. At 1 mm a year it loses nearly two-thirds of its strength in six months and releases three times as much gas.
Tissue engineering and scaffolds
Tissue engineering, named in an influential 1993 paper by Robert Langer and Joseph Vacanti (Langer and Vacanti), aims to regenerate tissue rather than replace it: cells, a scaffold to host them, and signals (growth factors or mechanical cues) to guide them. A scaffold for bone must be biocompatible and, usually, resorbable at the rate new bone forms; it must be porous and interconnected, with pores of a few hundred micrometres, large enough for cells to migrate and blood vessels to grow in (for bone, pores above about 300 µm are generally recommended, Karageorgiou and Kaplan); and it must be stiff and strong enough to work in place without shielding the tissue it is meant to stimulate.
Porosity and stiffness are linked. For open-cell foams and lattices whose struts bend, Gibson and Ashby showed that the modulus scales with the square of the relative density, , and the strength as about . So a porous metal can be tuned to the stiffness of bone:
import numpy as np
# Gibson-Ashby scaling for open-cell lattices of a solid with modulus Es and yield strength sy:
# E*/Es ~ (rho*/rho_s)^2, sigma*/sy ~ 0.3 (rho*/rho_s)^1.5 (bending-dominated cells)
Es, sy = 112.0, 880.0 # Ti-6Al-4V: GPa, MPa
def lattice(rel_density):
return Es * rel_density**2, 0.3 * sy * rel_density**1.5
print("relative density porosity E* (GPa) strength* (MPa)")
for rd in (0.6, 0.4, 0.3, 0.2, 0.1, 0.05):
E, s = lattice(rd)
print(f"{rd:12.2f} {1 - rd:11.0%} {E:10.1f} {s:13.0f}")
# Which porosity matches a target bone stiffness?
for tissue, E_target in [("cortical bone", 18.0), ("dense trabecular bone", 1.0)]:
rd = np.sqrt(E_target / Es)
print(f"to match {tissue} ({E_target} GPa): relative density {rd:.2f}, porosity {1 - rd:.0%}, "
f"strength about {lattice(rd)[1]:.0f} MPa")A Ti‑6Al‑4V lattice that is 60 % porous matches cortical bone's stiffness, with a strength of about 67 MPa; one that is 91 % porous matches dense trabecular bone at about 8 MPa. Such lattices are now made routinely by laser or electron-beam powder-bed fusion (see additive manufacturing in Lumenary ME) for acetabular shells, spinal cages and custom implants for bone lost to tumours, where they combine bone-like stiffness with pores for bone to grow into. Their weak point is fatigue: the rough, as-built struts carry stress concentrations, so lattice implants are tested to millions of cycles and often post-processed to smooth them.
For soft tissue, scaffolds are made from hydrogels, electrospun polymer fibres, or decellularised tissue, in which the cells of a donor organ are washed away, leaving the natural protein matrix. Bioprinting deposits cells within hydrogel "bio-inks" layer by layer. Engineered skin and cartilage products have reached patients, and the first engineered blood vessel was approved in the United States in 2024; whole engineered solid organs remain, as of 2026, a research goal, limited above all by the difficulty of building a working blood supply.
Regulation and ethics in outline
A medical device must be shown to be safe and to perform as intended before it is sold, and it is regulated according to risk. In the United States, the FDA divides devices into Classes I, II and III; most moderate-risk devices are cleared by showing that they are substantially equivalent to one already on the market (the "510(k)" route), while high-risk devices such as heart valves need premarket approval with clinical evidence. In the European Union, the Medical Device Regulation, applicable since 2021, classes devices as I, IIa, IIb or III, puts hip, knee and shoulder replacements in the highest class, and requires independent "notified bodies" to assess them. Japan, China, India, Brazil and other countries have their own regulators, increasingly aligned through international forums. Across all of them, makers must run a certified quality-management system (ISO 13485), assess risk systematically, test biological safety under ISO 10993, and keep watching their devices after sale.
That last duty matters because some problems appear only in many patients over many years. Joint registries — national databases recording every implant and every revision, such as those of the United Kingdom and Australia — compare the survival of thousands of designs and have identified underperforming implants early.
The ethics of biomaterials follow from these facts. Patients must give informed consent knowing what is and is not known about a device's long-term behaviour. Animal studies are still required for many implants, under the principle of the "three Rs" — replace, reduce and refine animal use — and laboratory and computational alternatives are growing. Devices should be designed and tested for the people who will receive them, including women, children, older people and patients of all body sizes and ancestries, which has not always happened. Cost and access matter: an implant that works beautifully but is unaffordable in most of the world serves few. And transparency about failures, which investigations such as the Implant Files found lacking in 2018, is part of the engineer's duty, not an afterthought.
Optional Christian reflection · Fearfully and wonderfully made
A reflection from the Christian tradition that inspires Lumenary Academy. It is not part of the mathematics and you are welcome to skip it.
"I will praise thee; for I am fearfully and wonderfully made" (Psalm 139:14). Every material in this lesson is measured against bone that remodels itself, cartilage that lasts a lifetime, and arteries that heal. Biomaterials engineering is, at its best, a humble kind of service: helping a body that is already a marvel to keep walking, seeing and living, and being honest when our materials fall short of what they replace.
Biomaterials in practice and research
Biomaterials is a field of teams: materials scientists and engineers work with surgeons, biologists, regulatory specialists and patients, in companies, hospitals and universities on every continent, and through societies such as the Society For Biomaterials and the European Society for Biomaterials and their counterparts in Asia and elsewhere. Active research in 2026 includes antibacterial and anti-biofilm surfaces, implants with sensors that report loading or infection, patient-specific implants designed from CT scans and printed in titanium or PEEK, degradable magnesium and zinc alloys with controlled corrosion, immune-modulating materials that steer the foreign-body response, and organoids and organ-on-chip systems that test materials with human cells instead of animals. Computational models of bone remodelling and of device performance are increasingly accepted by regulators as part of the evidence for a new device.
Common misconceptions
Misconception
"Biocompatible means inert." No material is entirely inert in the body, and many of the best are not meant to be: bioactive glass bonds to bone, degradable sutures dissolve. Biocompatibility means an appropriate response for a specific use.
Misconception
"A stronger, stiffer implant is always better." A stem far stiffer than bone shields it from load, and the bone thins. Stiffness must be matched to the job and to the tissue, which is why titanium alloys and porous structures are preferred for many bone implants.
Misconception
"Titanium is biocompatible, so any titanium device is safe." Biocompatibility depends on the application, the surface, the wear and corrosion products, and the mechanics. The same alloy can perform well as a stem and badly as a bearing surface, where it wears.
Misconception
"Biodegradable implants simply disappear harmlessly." Their degradation products, rate and timing all matter: acidic polymer fragments, hydrogen gas from magnesium, and loss of strength before the tissue has healed are the design problems.
Exercises
1. Biomaterials
Warm-upHumanWhich statement best captures what biocompatibility means?
2. Biomaterials
CoreHumanTi‑6Al‑4V is the preferred material for many femoral stems. Why are femoral heads almost always made of cobalt–chromium alloy or ceramic instead?
3. Biomaterials
CoreHumanModel a femoral shaft as a cortical bone tube with outer diameter 30 mm, inner diameter 16 mm and GPa, perfectly bonded to a solid Ti‑6Al‑4V stem of diameter 16 mm and GPa that fills the canal (no cement). Under an axial load, what percentage of the load does the bone carry? Give the nearest whole number.
4. Biomaterials
CoreHumanA superelastic nitinol stent is crushed into a catheter and springs back to shape inside an artery. Its alloy is chosen so that one phase is stable at body temperature when unloaded, and the other forms only under stress. Which phase is stable at body temperature in the unloaded stent? Answer with one word.
5. Biomaterials
CoreHumanA poly(lactic-co-glycolic acid) (PLGA) fracture pin is followed after implantation. Which pattern of change is most typical of this material?
6. Biomaterials
Warm-upHumanUnder the EU Medical Device Regulation, in which risk class is a total hip replacement placed, and what does that imply?
7. Biomaterials
CoreComputationalWrite bone_share(E_bone, D_bone, d_canal, E_stem, d_stem, E_cement=2.5, mode="axial"), returning the fraction of the load carried by the bone when a cortical bone tube (outer diameter D_bone, inner d_canal), a cement annulus (from d_canal down to d_stem, modulus E_cement) and a solid stem (diameter d_stem) are perfectly bonded. For mode="axial" the parts share load in proportion to ; for mode="bending", in proportion to , with for a tube. If d_stem == d_canal there is no cement. Raise ValueError for any other mode, or if d_stem > d_canal.
8. Biomaterials
StretchComputationalUsing the Gibson–Ashby scaling for a bending-dominated open-cell lattice, and , write design_lattice(E_target, E_solid, strength_solid) that returns a tuple (porosity, lattice_strength): the porosity (as a fraction, ) giving the target modulus, and the lattice strength it then has. Raise ValueError unless 0 < E_target <= E_solid.
Further reading and resources
- On the mechanisms of biocompatibility · D. F. Williams · 2008A careful discussion of what biocompatibility means, by the scientist who framed the consensus definition.
- Bonding mechanisms at the interface of ceramic prosthetic materials · L. L. Hench, R. J. Splinter, W. C. Allen and T. K. Greenlee · 1971The first report of a glass that bonds to bone.
- Tissue engineering · R. Langer and J. P. Vacanti · 1993
- Porosity of 3D biomaterial scaffolds and osteogenesis · V. Karageorgiou and D. Kaplan · 2005
- Control of biodegradation of biocompatible magnesium alloys · G. Song · 2007
- US Food and Drug Administration — Medical devicesHow devices are classified and cleared or approved in the United States.
- European Commission — Medical devices sectorThe EU Medical Device Regulation and its guidance.
- National Joint Registry (UK)Outcomes of every hip, knee, ankle, elbow and shoulder replacement in England, Wales, Northern Ireland and other UK nations.
- Australian Orthopaedic Association National Joint Replacement Registry
- The Implant Files (ICIJ)A 2018 international investigation of how medical devices are approved and monitored.
- Society For Biomaterials
- European Society for Biomaterials
- Cellular Solids — Structure and Properties · L. J. Gibson and M. F. AshbyThe source of the scaling laws for foams and lattices used in this lesson.