Supercooling And Liquid Water
Supercooling happens when liquid water cools below 0°C yet does not immediately form ice. The temperature drop does not remove the thermodynamic drive to freeze; it changes the kinetics, meaning the rate at which ice can start growing. In practice, freezing often begins when ice nucleation occurs, then the ice front spreads through the sample. If nucleation is delayed long enough, the water can remain liquid for seconds, minutes, or longer depending on conditions.
Pure water can supercool more than water with dissolved salts, dust, or microscopic bubbles. A common everyday analogy is how a clean, smooth surface can delay freezing compared with a rough surface that traps particles. In a lab, the “cleanliness” of the container and the sample handling steps can shift the observed freezing temperature by several degrees Celsius, which is why two experiments at the same thermometer reading can disagree.
Supercooling is not a magic state that changes water’s chemistry. It is a metastable situation: the liquid is temporarily stuck in a higher-energy arrangement because the pathway to ice formation has a barrier. Once a nucleation event occurs, the system typically releases latent heat and the temperature can jump toward 0°C, which is why a sudden plateau or spike often marks the start of freezing.
What People Get Wrong
Many explanations stop at “water freezes at 0°C,” then treat any liquid below 0°C as a contradiction. Freezing is a process with a rate, not a single instant. The freezing point people quote is usually the equilibrium point for pure water under ideal conditions, while real samples freeze when nucleation happens. That nucleation depends on impurities, dissolved gases, and surfaces that can seed ice-like structures.
Another common misunderstanding is to blame only temperature. Cooling rate matters because it changes how long the sample spends in the supercooled range. If you cool quickly, you may pass through the range before nucleation occurs; if you cool slowly, you give nucleation more time. In a small droplet, the thermal history also differs from a bulk sample because heat leaves the droplet faster, and the droplet’s surface can dominate nucleation behavior.
Supporting technologies often decide whether supercooling is observed. In experiments, thermometers or probes with limited response time can smear the temperature curve, hiding the moment of nucleation. In cryobiology, controlled cooling protocols and container materials influence ice formation, and the presence of cryoprotectants changes viscosity and solute distribution. Even the way a sample is mixed can matter; agitation can introduce microbubbles or detach particles from container walls, which then act as nucleation sites.
One small practical aside: when reading lab reports, check the sampling interval of the temperature sensor. A dataset logged every 1 second can miss a fast nucleation event that lasts a fraction of that time, which then makes “freezing temperature” look artificially low or variable. I’ve seen this in instrument exports from a common data logger setup (firmware version 2.3.x on a bench test), where the plotted curve lagged the actual event.
How To Encourage Or Measure It
Control Surfaces And Particles
To observe supercooling reliably, reduce nucleation sites. Use clean containers with minimal scratches, avoid dust exposure, and minimize handling that can introduce particles or bubbles. Filtration can help in some lab contexts, but filtration itself can add surfaces and shear stress, so it can change nucleation behavior in either direction. If you are comparing results, keep the container type and cleaning method consistent across runs.
In a practical measurement setup, a common approach is to pre-condition the container and sample so they reach thermal equilibrium together. If the probe touches the container wall, it can seed freezing at the contact point, which is why probe placement matters. A small change like moving the probe a few millimeters away from the wall can shift the observed nucleation temperature, and that shift is not “random noise.”
Use Cooling Protocols With Timing
Cooling protocol affects how long water spends in the supercooled range. If your goal is to measure the nucleation temperature, use a controlled cooling rate and record temperature at a high enough sampling frequency to capture the onset of freezing. For many bench setups, logging at 10–100 samples per second is enough to resolve a nucleation event, but the exact requirement depends on sensor response time and thermal lag.
When you see a sudden temperature rise toward 0°C, treat it as evidence of latent heat release rather than a sensor artifact. If the temperature curve shows a smooth decline with no plateau, freezing may not have occurred during the observation window, or nucleation may have been suppressed by the sample’s purity and container conditions.
Interpret Solutes And Cryoprotectants
Dissolved substances change freezing behavior by altering water activity and by changing how ice can form. Salts often reduce supercooling in many contexts because they can promote nucleation or create local environments that freeze more readily. Cryoprotectants such as glycerol or dimethyl sulfoxide (DMSO) can increase viscosity and change solute distribution, which can delay ice formation in some protocols while also introducing new complexities like phase separation.
In cryobiology, the goal is often not “maximum supercooling,” but controlled ice formation and survival outcomes. That means the relevant metric is frequently post-thaw viability or structural preservation, not just the lowest temperature reached before freezing. If a report focuses only on the minimum temperature without describing cooling rate, container type, and solute composition, the result is hard to compare.
Recognize Metastability And Repeat Tests
Supercooled water is metastable, so repeated trials can show variability even with the same nominal temperature. That variability reflects stochastic nucleation: the barrier to forming an initial ice embryo is crossed at random times. To reduce confusion, run multiple replicates and report a distribution of nucleation temperatures rather than a single number.
One practical aside from data handling: when I clean temperature traces in Python (pandas 2.x), I avoid aggressive smoothing because it can erase the sharp rise that marks nucleation. A mild moving average over a very short window can still distort the peak timing, and that timing matters when you compare protocols.
Educational Case Examples
Cold-Room Droplet Observation
An anonymized lab student cools small droplets of distilled water on a polished metal plate in a freezer set to -5°C. The student notices that some droplets freeze quickly while others remain liquid for longer, even though the freezer temperature is constant. After switching to a different plate surface finish and reducing dust exposure during transfer, the student observes fewer “early freezes.” The student then records temperature with a probe placed near the droplet rather than touching the plate, which reduces contact-seeded freezing.
The lesson is not that the freezer temperature changed, but that nucleation sites changed. The student’s results also show that “distilled water” can still contain dissolved gases and trace contaminants, and those factors influence how long supercooling lasts.
Cryopreservation Protocol Comparison
An anonymized research group compares two cooling protocols for a cryopreservation study using a standard cryoprotectant. In both protocols, the sample is cooled from room temperature to subzero temperatures, but one protocol uses a slower cooling rate and a different container geometry. The group sees earlier ice formation in the slower protocol for one container type, while the other container shows delayed nucleation. After reviewing the temperature logging setup, they find that the probe position differed between containers, increasing thermal lag in one case.
The group concludes that observed nucleation timing reflected both real nucleation differences and measurement differences. They adjust probe placement and repeat the comparison, then report nucleation temperature distributions rather than single values.
Checklist For Interpreting Results
| What To Check | Why It Matters | What Good Looks Like | Red Flags |
|---|---|---|---|
| Cooling Rate | Controls time spent in supercooled range | Reported and held constant across trials | Only freezer setpoint given, no ramp details |
| Container And Surface | Surfaces seed ice nucleation | Same material, cleaning method, and probe placement | Probe touching wall or different surface finishes |
| Temperature Logging | Sensor lag can hide nucleation onset | High sampling rate and known sensor response | Sparse sampling and heavy smoothing of traces |
| Replicates And Variability | Nucleation is stochastic | Multiple trials with a distribution of nucleation temps | Single “best” run reported as typical |
Step-by-step checklist for your own interpretation: (1) confirm the sample composition (distilled, tap, saline, cryoprotectant concentration), (2) note the container material and surface condition, (3) verify the cooling ramp and how long the sample sits at subzero temperatures, (4) check temperature sensor placement and sampling rate, and (5) look for evidence of latent heat release such as a sudden temperature rise.
Common Mistakes
One frequent mistake is treating the lowest recorded temperature as the “freezing point.” In supercooling experiments, the lowest temperature often reflects the sensor lag or the time before nucleation, not the equilibrium freezing temperature. A second mistake is comparing results across different container materials without accounting for surface nucleation differences.
Another error is ignoring dissolved gases and handling. Pouring from a height, vigorous mixing, or shaking can introduce microbubbles that act as nucleation sites. If you see faster freezing after a handling change, the cause may be physical nucleation sites rather than a change in water purity.
People also over-trust a single trial. Because nucleation is stochastic, one run can freeze early and another can freeze late under the same nominal conditions. Reporting only one value encourages overconfident conclusions and makes it hard to reproduce results.
Finally, some write-ups omit the probe’s thermal contact details. If the probe is embedded in the sample versus suspended in air near the sample, the measured temperature can differ by several tenths of a degree or more, which can shift the apparent degree of supercooling. That mismatch is subtle, and it rarely shows up unless you compare sensor placement diagrams.
FAQ
Does Water Freeze Exactly At 0°C?
Pure water freezes at 0°C under equilibrium conditions, but real samples freeze when ice nucleation occurs, which can happen below 0°C if nucleation is delayed.
Why Does Distilled Water Still Supercool?
Distilled water can contain dissolved gases, trace impurities, and it still interacts with container surfaces, all of which influence nucleation timing.
What Triggers Ice Formation In Supercooled Water?
Ice formation typically starts when an ice embryo forms on a nucleation site such as a particle, bubble, or surface feature, then ice growth releases latent heat.
How Do Cooling Rate And Time Affect Results?
Cooling rate changes how long the sample stays in the supercooled range, which changes the probability that nucleation occurs during the observation window.
Can Sensors Create False Supercooling?
Yes. Temperature probe lag, poor placement, and heavy smoothing can make the curve appear to stay below 0°C longer than the actual nucleation event.
Author's Insight
Supercooling is best understood as a kinetic delay in a metastable liquid, not as a change in water’s fundamental freezing temperature. Ice nucleation depends strongly on surfaces, impurities, dissolved gases, and thermal history, so two “same temperature” setups can produce different freezing behavior. Measurement details such as sensor response time and probe placement often determine whether the nucleation event is visible in the recorded data. When evaluating claims about how far water can supercool, the most informative details are the sample composition, container material, cooling protocol, and how many replicates were measured.
Key Takeaways
- Water can remain liquid below 0°C when ice nucleation is delayed, creating a metastable supercooled state.
- Freezing timing depends on nucleation sites (particles, bubbles, surfaces) and on how long the sample stays in the subzero range.
- Temperature sensors can mislead if probe placement and response time are not accounted for.
- Use replicates and look for latent-heat signatures like a sudden temperature rise toward 0°C.
- For practical work, control container surfaces and handling steps before comparing cooling protocols.
