Mixed Polymers Form Unique One-Piece Medical Implant

By DR. MARK YEOMAN, R&D DIRECTOR, CONTINUUM BLUE LTD.

Mixing polymers for a medical implant while injecting them into a mold involves setting a wide range of parameters. When they ran into manufacturing problems, engineers developed a COMSOL Multiphysics model that uncovered the cause and have helped bring a similar novel polymeric implant based on this process to clinical trials.

Many soft tissue medical implants are only required for a number of months or years before the body’s own tissue regenerates and heals. At this point it is sometimes desirable to remove them; however, this is not always possible. Such soft tissue medical implants have traditionally been made of fabrics or biotextiles, but with the advancement of specialized polymers there is a huge interest toward the use of hyper-elastic polymeric materials to produce implants that duplicate or augment the natural response of body tissues. Engineers at Continuum Blue Ltd. (UK) are taking that process a big step forward with the development of soft tissue implants made of biodegradable hyper-elastic elastomers that, when they have served their purpose, dissolve naturally within the body to eliminate the need for revision surgery to remove them.

Figure 1. A fabric-based LARS implant shown in a shoulder joint. Note the two screws in the bone that anchor the fixation holes at one end of the implant. Continuum Blue is now working on a multi-elastomeric version of this implant that will biodegrade once the joint has been healed.

Founded in 2004, Continuum Blue specializes in the research, development and analysis of medical devices and implants. It has worked with a number of international companies including Medtronic, Abbott Spine (now Zimmer), Synthes, NuVasive, Scient’X, British Technology Group, Ranier Technology, Aesculap and Blackstone Medical to name a few. The company focuses primarily on the orthopaedic and cardiovascular markets. Novel new ligament implants act as would a rubber band to hold together and support bones while providing flexibility. However, they require anisotropic material properties to address various requirements. For instance, in a rotator cuff LARS (Ligament Replacement and Augmentation System, Figure 1), the implant must be flexible enough to give the patient ease of arm movement without restriction, and at the same time providing sufficient stability for functional use of the arm and shoulder region such a where the patient wants to support or hold an object steady.

Figure 2. Modeling domain showing half the geometry for molding a dual-elastomer LARS. The polymer solution enters at the bottom and pushes air out through the top opening.

Mixing Polymers

A single elastomeric material cannot meet the desirable requirements of such a LARS. Our client developed a novel method of injecting two slow-curing polymers into a mold to create a one-piece implant with the desired anisotropic hyper-elastic properties. With a model, we were able to determine how to best manufacture the 3D implantable device in a single production process. This we have successfully done, and a product based on this process is now undergoing clinical trials. The next step is to add a third polymer that will add the desired biodegradable properties and contribute further to additional anisotropic hyper-elastic material properties.

Such polymer-based LARSs have many benefits. They’re relatively inexpensive and quicker to produce than textile-based LARSs, which require multiple manufacturing steps. As there is human intervention in each of these manufacturing steps, each also requires extensive quality assurance (QA). In contrast, the polymeric implant requires just two steps: the molding and then the cleaning of the final product, thereby slashing manufacturing and QA costs.

Our first prototypes involved two elastomers that flow into the mold in the right concentration, at the right sequence, at the right speed, and under the right conditions. Each end of the molded product consists of 100% of one of the elastomers, and the regions in between have a continuous mix of the two. Our engineers must control the point of injection as well as the temperatures of the injected polymers, the overall mold, and the faces of the mold. Other aspects that must be controlled are when the polymer enters the mold and their combined volumes.

Determining the proper parameters for all of these variables is no easy task. In fact, when developing the first implant of this type, we encountered a manufacturing problem but could not identify the cause in the molding process. We then turned to simulation software to give us more insight into the process, and in the end it was only COMSOL Multiphysics that was up to the task. In particular, its capabilities to handle full 3D dual-polymer injection and control of the injection profile were not available in any of the specialized injection-molding software alternatives we evaluated.

“We then turned to simulation software to give us more insight into the process, and in the end it was only COMSOL Multiphysics that was up to the task.”

Figure 3. Simulation of the polymer as it fills the mold cavity after 20s, 40s, 100s and 3000s. The red isosurface indicates where the boundary between the polymer mixture and air occurs, the green isosurface indicates where the volume fraction between the two polymers is at 50%, while the blue isosurface is where the second polymer has a volume fraction of 100%. Note that some of the mold cavity walls have been removed for illustrative purposes.

As for the model itself, it consists of three domains: a solid region for the mold cavity walls; a liquid region for the injected polymers; and a gas region for the air in the cavity. The co-injected polymers are mixed prior to entering the mold, which was simulated by including a boundary condition that describes the volume fractions of the two polymer solutions as a function of the injection rate. This description was very easy to define in COMSOL Multiphysics but almost impossible in other software.

To handle the complex interactions of the polymers with each other and the surrounding environment, we coupled three transient physics interfaces to each other. First, a two-phase flow phase field interface simulates the liquid flow front as it evacuates the air from the mold cavity. Second, a phase-field flow interface simulates the two dual injected polymer solutions in the liquid phase of the two-phase flow and their interactions with each other. Third, the convection and conduction application mode models the thermal changes. When the injection process comes to a stop, the filled cavity cools down. During that process, the two polymers’ densities and viscosities change, and continued fluid-fluid interaction motion occurs until high viscosities are reached and all flow stops (Figure 3).

This model itself consists of roughly 50,000 elements and almost 300,000 degrees of freedom. It solves in 18 hours on a 2.8-GHz Intel Core 2 quad processor with 8GB of memory running on Windows Vista.

Temperatures and Flows Correspond Nicely

The validation of the model was performed on a different complex three-dimensional body and cavity that cannot be fully disclosed in this article. However, we validated our model in three ways. First was a qualitative evaluation where we compared a video of the liquid-air flow front during the filling process with an animated visualization of the same process in COMSOL. We found that the model is very good at showing the fluid-air interface and the observed flow around various 3D features (such as walls and curved surfaces, including flow baffles) in the mold cavity at different injection rates.

Figure 4. (left) Comparison of temperature readings inside the mold cavity during the filling process compared with model results. The solid lines represent measured data, and the dashed lines indicate the model data. The erratic sections in the measured data (represented by the grey regions in graph) are due to opening and closing solenoids to control the flow of the injected polymers. (right) Thermal plot of the mold cavity during filling.

Next, we placed three thermocouples inside the model at three locations to measure the temperature of the polymers as they flowed into the cavity and cooled. The thermocouples were located at the top, bottom and side of the mold cavity. The readings from these thermocouples were compared with the simulation. Figure 4 shows that there is a very good fit to the physical data. Concerning the discontinuities in the physical injection profile, they are due to the switching of the solenoids that control the polymer flow into the mold. In contrast, the model uses an ideal smooth flow function. For our purposes, these curves show excellent results from the model.

Figure 5. A cross-section from a validation sample compared to COMSOL results (blue line) shows that the outline of the second polymer matches very closely.

Third, once the mold had cooled and the polymer cured, sectional cuts were made in the device to look specifically at the boundary between the two polymer regions and compare them with the model. Once again, the model solution gives a very good estimation of the actual makeup of the implant (Figure 5, and note that this model shows a partial view of a 3D device different from the actual rotator cuff LARS).

“COMSOL Multiphysics has been very valuable to finding a solution to this project, and its capabilities will certainly be beneficial and a key component to our future.”

With the model, it’s now possible for us to investigate many mold variations in a reasonable time and cost. It is estimated that a single mold redesign and trial run with the model takes roughly 1.5 days and costs approximately GBP 850. In contrast, to do a physical sample run, where a specialized silicone mold costs approximately GBP 3000, the estimated total costs are almost GBP 9000, and the time to get the redesigned silicone mold is between 3 to 4 weeks.

Without COMSOL, finding the critical parameters that control polymer location would have taken much longer and would have cost in the order of 20 times more to do so. In addition to cost, due to the opaque nature of the molded implant samples, it is very difficult to easily define the polymer boundaries and graduated regions of the physical samples (as seen in Figure 5). In order to do so, microbubbles or dyes had to be added to the polymer solutions, changing the polymers’ flow characteristics. Another more expensive option was to use mechanicalindent analysis to map physical material differences at points throughout the mold. In contrast the model gives far better insight into what is happening during the mold filling process, where the client can easily visualize the end product and the mixed regions between the two polymers. Having the COMSOL model and the resulting visualizations provide clear cost- and time-efficient benefits when convincing customers of the best mold process to provide a viable polymeric LARS solution.

Adding Biodegradability

With the validated model in hand, our group has many projects in mind. First, we want to increase the number of injected polymers to allow for a biodegradable product. We also want to allow for multiple injection points. Most ambitious of all is to add parameterization to the model so that the software can help us quickly find the process parameters for new products. It will be necessary to change the size and shape of a LARS implant to account for different anatomy, joint complications and disease. We thus hope to create an optimized model whereby we need only describe the end product’s geometry, and the software will suggest the best mold design along with the process parameters that will result in the desired physical properties. The model will hopefully cater to any mold and material design change we might encounter. In summary, COMSOL Multiphysics has been very valuable to finding a solution to this project, and its capabilities will certainly be beneficial and a key component to our future.

ACKNOWLEDGEMENTS

The author would like to thank Robert Snell at Ranier Technology for his assistance and help with this work.

About the Author

The author, Mark Yeoman (right), receives an award from Svante Littmarck, President & CEO of COMSOL, Inc., during the Milan COMSOL Conference 2009.

Dr. Mark Yeoman, R&D Director at Continuum Blue, is a medical-device development specialist with over 12 years of experience and expertise in this area. He has a PhD in computational modeling and applied mathematics, where his postgraduate studies focused on the design and optimization of cardiovascular implants using computational techniques and genetic algorithms for Medtronic. Dr. Yeoman has also worked for Disa Vascular, which specializes in stents. He also lectured at the graduate and undergraduate levels for a number of years on Dynamics and Computational Methods & Techniques.