Our understanding of reverse osmosis is wrong

Researchers from Yale University have now found that the widely accepted solution-diffusion model does not accurately explain water transport in RO membranes.
Tejasri Gururaj
A drop of water
Reverse osmosis as explained by the solution-diffusion model has been shown to be incorrect.

Robert Anderson/Unsplash 

  • Reverse osmosis is an important technique for purifying water that has been in use for a long time.
  • A new study shows that our understanding of how it works may be wrong.
  • This could help develop more efficient clean water systems.

Water is a precious resource that is needed by all living things in order to survive. However, according to the World Health Organization, nearly 2.2 billion people (or 1 in 3) did not have access to safely managed drinking water services as of 2019.

Access to clean drinking water is a crucial global challenge that in some areas relies on desalinating seawater. This is most commonly done using a process known as reverse osmosis (RO). 

The phenomenon of osmosis was observed as early as 1748, but RO technology wasn't commercially viable until nearly two centuries later. Sea-water RO (SWRO) desalination was practically demonstrated in the late 1950s by Srinivasa Sourirajan, Sidney Loeb, and a team of researchers at the University of California, Los Angeles (UCLA). 

Almost every RO system in the world uses membranes to produce potable water from seawater. Now, after more than 270 years since the original observation, scientists think that the basis of our understanding of water transport in reverse osmosis is incorrect. 

Our understanding of reverse osmosis is wrong
A bank of reverse osmosis filters at a public water utility plant.

Scientists from Yale University have found that the solution-diffusion mechanism, which is thought to be the underlying principle of water transport in RO, is wrong, finding instead that water transport in RO is governed by pore flow and pressure changes within the membrane.

Interesting Engineering (IE) spoke to the lead researcher on the study, Professor Menachem Elimelech, the Sterling Professor of Chemical and Environmental Engineering at Yale University, to gain more insights into their work.

"I always thought that the solution-diffusion mechanism for water transport was not intuitive and, to be honest, quite strange with many of its assumptions," said Elimelech, speaking of his motivation behind probing such a well-established phenomenon.

Elimelech collaborated with Professor Ying Li from the University of Wisconsin, Madison and Professor Lianfa Song from Texas Tech University for this research. The team also included postdoctoral researchers, Li Wang, Mohammad Heiranian, Hanqing Fan from Yale, and Jinlong He from the University of Wisconsin, Madison.

Our understanding of reverse osmosis is wrong
Prof. Menachem Elimelech from Yale University led the research

Before we delve into the actual research, it may help to first understand the current explanation of RO, the solution-diffusion mechanism. 

The solution-diffusion mechanism

RO utilizes a permeable membrane to separate impurities and contaminants from water.

According to the solution-diffusion theory, water transport is driven by the processes of solution and diffusion, which are, in turn, driven by water concentration gradients and membrane material characteristics. The water molecules diffuse through the membrane based on their concentration gradient, moving from areas of high concentration to areas of lower concentration.

Seawater is a solution of salts dissolved in water. In this model, the water molecules diffuse across the membrane driven by their concentration gradient, leaving behind the larger salt ions, resulting in clean water.

"I taught the basics of the solution-diffusion mechanism in my class following textbook materials," explains Elimelech, "but I was not comfortable with the core assumptions of the model, especially that the hydrostatic pressure is constant across the membrane and then drops to zero at the membrane exits. I thought this assumption was unphysical."

Our understanding of reverse osmosis is wrong
Reverse osmosis, where water molecules move through the semi-permeable membrane.

"Also, because of this assumption, the model predicts a water concentration gradient across the membrane, which does not make sense. I always asked myself how the concentration of water, which is incompressible, can change with pressure.

Eventually, I decided to unravel the mechanism of water transport. Around that time, Prof. Song from Texas Tech University published a paper on water flow by osmosis, arguing that it cannot be governed by a solution-diffusion mechanism. This motivated me to further explore the nature of water transport in reverse osmosis at the molecular level.

We assembled the team about two years ago. We felt that in addition to experiments and theory, molecular dynamics simulations will be essential to convince the scientific community because we do not have yet experimental techniques to probe water transport within the membrane at the molecular scale. Hence, we decided to team up with UW Madison about one year ago," explained Elimelech, discussing how he came to question the classic solution-diffusion model.

Something didn’t add up

"Our paper, using non-equilibrium molecular dynamics (NEMD) simulations and solvent transport experiments, strongly supports the pore-flow model for water transport in RO membranes while disproving the solution-diffusion model," added Elimelech.

For the solvent transport (or permeation ) experiments, the team used polyamide and cellulose triacetate RO membranes with water and organic solvents. The team made a few key observations which led them to propose an alternative model to the solution-diffusion theory to explain water transport.

Our understanding of reverse osmosis is wrong
Permeation experiments were conducted by the team.

The NEMD simulations revealed that the pressure within the membrane decreased linearly along the direction of water permeation. This suggests the presence of a pressure gradient within the membrane.

Additionally, the NEMD simulations showed no change in water concentration across the membrane , contradicting the main assumption in the solution-diffusion model.

In their permeation experiments, the researchers found that the water flux increased linearly with the applied pressure. This is in complete disagreement with the solution-diffusion model, where the water flux reaches a maximum value called the ceiling flux, after which it doesn't increase with a further increase in the applied pressure.

Lastly, they noted in the NEMD simulations that the water travels in clusters through the membranes via interconnected pores. This observation contrasted the solution-diffusion model's assumption that water molecules permeate as dispersed single molecules.

These findings led Elimelech and his team to propose the solution-friction model to explain water transport in RO membranes. 

The solution-friction model

According to the solution-friction (SF) model, the driving force for water transport is a pressure gradient rather than a concentration gradient. Water and solvents travel as clusters through the membrane pores, experiencing frictional forces along the way.

Our understanding of reverse osmosis is wrong
NEMD simulations showing the trajectory of a water cluster containing five molecules over a duration of 70 nanoseconds

The frictional forces arise from the interactions between the membrane material and the solvent molecules. The model incorporates the concept of friction coefficients, which represent the resistance encountered by the solvent molecules as they move through the membrane pores.

"Pores inside the membrane are hydrated with small clusters of water molecules. For short intervals, some of these pores can be connected together, providing a complete feed-to-permeate passage for water molecules to travel through the membrane. Because of the applied pressure and thermal motion of the polymer matrix, the network of interconnected pores is constantly changing. This means feed-to-permeate passages are constantly being formed and broken up. 

If we tag a cluster of water molecules that enter the membrane from the feed side, we will see that the tagged molecules travel together through the network of interconnected pores. Overall, this is inconsistent with the solution-diffusion model where water molecules are expected to spread out and diffuse individually in the membrane after entering the membrane," explained Elimelech.

The SF model takes into account the pressure gradient within the membrane and the frictional forces acting on the solvent molecules. By considering these factors, the model can predict water and solvent transport in RO membranes more accurately than the traditional solution-diffusion model.

Implications for RO

Considering the longstanding acceptance of the solution-diffusion model, the team could have met with a lot of skepticism from the scientific community. However, according to Elimelech, the team's work has prompted a lively debate among researchers.

"The paper got a lot of attention and stirred a lively debate in the membrane community, more than any topic in the past few decades. In May, when the paper was presented at the North American Membrane Society Annual Meeting, it was the point of discussion by the participants throughout the conference.

Our findings will now prompt researchers to develop experimental techniques to understand the transport of water and salt in RO membranes at the molecular level," he said.

When asked about the implication of their research for the design and optimization of RO systems, Elimelech said, "First, our paper presents a fundamental understanding of water transport in RO and disproves a widely accepted mechanism for water flow in RO.

We also show that water transport in RO can be described by the solution-friction model, which accounts for the friction of water molecules with the membrane. This friction originates from the interaction of water molecules with the membrane material. The chemistry of the membrane will influence interactions and, as a result, friction. Hence, by tuning the pore size and finding materials with lower friction with water, we can design high-performance membranes."

Our understanding of reverse osmosis is wrong
A membrane water filter.

Elimelech believes it is necessary to focus on developing in-situ experimental techniques to probe water and salt transport at the molecular scale within RO membranes. By conducting experiments under pressure, they can gain deeper insights into the mechanisms and dynamics involved and pave the way for more comprehensive research in this field.


The findings of this study challenge one of the most well-established theories in water transport. By providing an alternative and more accurate model for explaining water transport in RO membranes, the researchers may also pave the way toward more efficient and sustainable clean water resources.

Elimelech agreed that their research contributes towards the broader goal of providing clean drinking water access to all.

"A better understanding of water and salt transport in RO membranes at the molecular level will help the development of high-performance membranes, having adequate water permeability and ultra-high salt rejection. This will reduce the cost and energy consumption of desalination, making it a sustainable solution for augmenting water supply in water-scarce regions all over the world," he concluded.

The findings were published in the journal Science Advances.

Study abstract

We performed non-equilibrium molecular dynamics (NEMD) simulations and solvent permeation experiments to unravel the mechanism of water transport in reverse osmosis (RO) membranes. The NEMD simulations reveal that water transport is driven by a pressure gradient within the membranes, not by a water concentration gradient, in marked contrast to the classic solution-diffusion model. We further show that water molecules travel as clusters through a network of pores that are transiently connected. Permeation experiments with water and organic solvents using polyamide and cellulose triacetate RO membranes showed that solvent permeance depends on the membrane pore size, kinetic diameter of solvent molecules, and solvent viscosity. This observation is not consistent with the solution-diffusion model, where permeance depends on the solvent solubility. Motivated by these observations, we demonstrate that the solution-friction model, in which transport is driven by a pressure gradient, can describe water and solvent transport in RO membranes.

Add Interesting Engineering to your Google News feed.
Add Interesting Engineering to your Google News feed.
message circleSHOW COMMENT (1)chevron
Job Board