Strong rare-earth permanent magnets have become relatively cheap to produce, and play an important role in the development of modern technology. One example is the integration of magnetophoresis and microfluidics, where permanent magnets are miniaturized and integrated onto lab-on-chip devices with the help of micro-electro-mechanical systems techniques.
A structure of two adjoined giant magnetic anisotropy rare-earth magnets with opposite directions of magnetization produces a very strong and inhomogeneous magnetic stray field. The field is several times stronger than the induction of the rare-earth material itself, and the product of the magnetic stray field (B) and the gradient of the field reaches theoretical and simulated values of 10^8 – 10^10 mT^2/m. Building upon this basis, a new design adds two thin, soft magnetic masks on top of the magnets, forming a small air gap directly above the junction line between them, in order to adjust the shape and strength of the stray field.
Simulations with 50 micrometer thick vanadium permendur masks show that, when the gap size decreases towards 50 micrometer, the tangential component of B increases with a factor of 20%, and narrows in width comparable to the gap size. In a distance of 10 micrometer from the masks the product of B and the gradient of B now exceeds 10^11 mT^2/m. The normal and tangential gradient of B are oppositely directed and on the same order, nevertheless, the tangential field is several times stronger than the normal field, thus the main contributor to the product of B and the gradient of B.
Singularities in the demagnetization field above the corners of the masks are responsible for most of the increase and distortion of the magnetic stray field. However, above a critical distance of 40 micrometer they are undetectable. The stray field is now automatically reduced since the masks increase the absolute distance to the source of the field. Thus depending on mask thickness, all distributions of the product of B and the gradient of B are correspondingly decreased. Magnetic separation of large bulk quantities is thus performed better in a device without masks, while separation of small quantities in confined regions, beneath the critical distance, benefits significantly from the new design.
Experimental results indicate that, the simulations predict close to realistic results above the critical distance. The experiments are, however, not performed close enough to observe the singularities, and the full extent of the simulations can thus not be verified. Nevertheless, as a consequence by the fact that, keeping corners perfectly square and junctions between materials ideal in a real device, simulations where the product of B and the gradient of B exceeds 10^11 mT^2/m are not realistic. A more realistic value is 10^10 mT^2/m, still several times larger than that in simulations of a structure without masks.
The new design is thus superior for separation purposes in microfluidic environments, if the separation distance is less than 40 micrometer. Developing new and better magnet and mask material compounds, as well as perfecting techniques ensuring ideal magnet-mask and magnet-magnet junctions will increase its potency even further.