The science of vitrification

How ice damages tissue, how vitrification turns water into glass instead and what research has shown so far.

Every organ has its own time limit

When blood flow stops, tissue starts to run out of oxygen. This is called ischemia. Cooling slows the damage but cannot stop it, and each organ tolerates it for a different time. A heart can wait about 4 to 6 hours, lungs about 4 to 8, a liver about 8 to 12 and a kidney about 24 to 36 hours.¹²

4 to 6 hours

how long a donor heart can wait in cold storage¹²

24 to 36 hours

how long a donor kidney can wait in cold storage¹²

27.9 %

of recovered deceased donor kidneys in the US were not transplanted in 2023¹¹

Why organs cannot simply be frozen

Below 0 °C, water freezes into ice crystals. Inside cells, the crystals break membranes. Between cells, they tear tissue and blood vessels apart.^{1, 2}

A cryobiology lab bench with reagent bottles on the shelf, pipettes, peristaltic pumps and beakers

Cryoprotectants are toxic at high doses

Cryoprotective agents, or CPAs, keep ice from forming. To vitrify tissue, which means cooling it into a glass without any ice, today's mixtures need CPA to make up around 40 to 65 percent of the solution.⁴ At that level the CPA itself damages cells. Researchers call this the biggest barrier in the field.^{3, 4}

40 to 65percent

of the solution in current vitrification mixtures is CPA⁴

The main CPAs still in use today were discovered in 1949 and 1959.^{13, 14}

CPA concentration

Less ice, but not yet enough protectant to vitrify.

Enough CPA to vitrifyVitrifies100 %50 %0 %

CPA concentration

Ice prevented
Cells surviving, today's CPAs
Cells surviving, our goal

With today's CPAs, the concentration that stops ice also damages cells. We are looking for a molecule that cells survive in that range. Illustration, not measured data.^{3, 4}

Vitrification turns water into glass instead of ice

With enough CPA and fast cooling, water never forms crystals. Typical mixtures turn into a solid glass at around −120 to −130 °C.^{5, 6} This point is called the glass transition. Below it, molecules almost stop moving.

Ice

Glass

Illustration, not to scale. Both panels start from the same liquid. In ice the water molecules lock into a crystal lattice. In a glass they stay disordered, as in a liquid, but almost stop moving.

Vitrified organs are stored below the glass transition

To stay a glass, a vitrified sample has to be kept colder than its glass transition.⁵ Organs are not kept as cold as cells, because at the temperature of liquid nitrogen the glass can crack.^{5, 6}

0 °CWater freezes
−120 to −130 °CTypical CPA mixtures turn to glass^{5, 6}
−140 °COur target for storing organs
About −150 °CWhere vitrified organs are kept in research today^{5, 6}
−196 °CLiquid nitrogen, standard storage for cells and small samples^{5, 6}

Rewarming depends on the CPA

As the glass warms, ice can still form. This is called devitrification.⁵ How easily it happens depends on the CPA. More CPA keeps the glass stable, but more CPA also harms cells.^{4, 5} New methods like nanowarming heat the sample from the inside, faster and more evenly.^{6, 7} They do not solve the toxicity. CPA toxicity still limits the latest results.¹⁵ This is the problem we work on.

Clear infusion bags of cryoprotectant solution with graduated volume marks, labelled and lying in a cold bath

What science has shown so far

Results of the field so far. Vitrification is routine for embryos. In animals, whole organs have been vitrified, and some have been transplanted. A human organ has not.

2026
Pig kidneys stored for over a year, rewarmed and transplanted. Follow-up lasted two hours. Injury was attributed to the CPA.
Preprint¹⁵
2025
A pig liver vitrified at liter scale without ice or cracks. Viability was not tested.
Peer reviewed⁷
2023
Rat kidneys stored for up to 100 days. All five recipients survived on the transplanted kidney.
Peer reviewed⁶
2009
A vitrified rabbit kidney was transplanted. One animal survived, with impaired function.
Peer reviewed⁸
Since 1985
Embryos vitrified. Today this is routine in reproductive medicine.
Routine⁹
Not yet shown
A human organ vitrified, rewarmed and transplanted.

A gloved hand connecting tubing to a clear perfusion bag on a lab bench, a volumetric flask in the background

How we design new cryoprotectants

Every result above is limited by how toxic the protectant is. We screen candidate molecules on the computer first. Our cryobiology lab then tests the most promising ones for ice protection, glass formation and toxicity. We are looking for a cryoprotective agent, or CPA, that stops ice at a concentration cells can survive.

Early stage research. We are now building our discovery engine and validation lab.

We predict how a mixture will behave

Our models look at what a mixture is made of and predict how toxic it is and how well it forms glass. They learn from our own lab data. This lets us rank candidates before any lab work starts.

Predicting properties

We predict toxicity and glass-forming capacity straight from the composition. Glass-forming capacity is how readily a mixture turns to glass instead of ice. The temperature where that happens is called the glass transition.

Scoring plates by image

Vision models look at every well of a screening plate and score it as glass, haze or ice.

Prediction accuracy

IllustrationCycles

Learning from every result

Every screening result goes back into the models as training data, so predictions improve with each cycle.

The animations in this section are illustrations, not model output.

We test every prediction in the lab

Results from the lab go back into the models, so each new round of candidates starts from more data.

1Computational design

Machine learning models predict toxicity and glass-forming capacity from composition and then propose new candidates.

10⁵ to 10⁷candidates in the library
10³ to 10⁵after fast screening
500 to 1,000after the membrane toxicity filter
50 to 200confirmed by molecular dynamics

Planned number of candidates at each stage of our discovery process. This is design capacity, not results.

Molecular dynamics, density functional theory and Monte Carlo simulation
Learned property models that predict the glass transition temperature, toxicity and viscosity
Machine learning guided generation with graph neural networks and Bayesian optimization

2High-throughput screening

We are building robotic screens and vision models to score glass formation in hundreds of variants every week.

0wells scored as glass

Glass
Haze
Ice
Empty

Illustration of a screening plate, not measured data.

3Organ-specific optimization

Every organ differs in its vessels, size and tolerance, so each one gets its own formulation.

Transplanted, the kidney keeps the recipient alive.

The recipient has no working kidney.
The donor kidney is vitrified and stored at −140 °C.
It is rewarmed and starts to work again.
Transplanted, the kidney keeps the recipient alive.

Illustration of our goal. Rats have lived on vitrified, rewarmed kidneys.⁶ In pigs, follow-up so far is two hours.¹⁵

Four tall steel cryogenic dewars on wheels in a blue lit storage hall

Our long-term goal is organ banking

Organs that could be stored and reliably rewarmed would allow better donor matching, more time to prepare patients and fewer lost organs.¹⁰ This is a research goal and not a current capability.

Sources (15)