Internal R&D – the burnt stripe effect

Extended material is available  –  AMPERE 2009 presentation on the burnt stripe effect  –  IMPI 2007 presentation on “puddles and droplets”.

The so-called cold rim effect is strong for thin semidry loads with radii less than about 35 mm (at 2450 MHz), which thus get a centre heating that cannot be modified much by the design of the microwave heating system. But a particular amplification of the heating in a typical narrow zone across the diameter in load items such as potato chips may also occur – see the picture below to the left.

If there would be only drying-out by the microwave energy, this would be faster in the whole central region. Since the permittivity then decreases and by that the absorption capability of these dryer parts, a negative feedback would then occur and even out the drying. But as is seen in the photo to the left,  the burnt stripe effect is indeed a runaway phenomenon, where an already dried-out region absorbs additional power so burn marks are created. It is concluded that liquid water transport is necessary for the effect to occur.
What happens is illustrated in the numerical modelling scenario in the image to the right. This shows a circular 24 mm diameter thin potato chip at 100 °C (permittivity ε = 50–j16 ) having an elliptical inclusion with axes 18 and 4 mm, representing a partially depleted region (permittivity ε = 8–j1,6). This region is marked with a black dotted curve. The potato chip is irradiated from the left with a free space TM-polarised 2450 MHz wave with incidence angle θ = 82°. The magnetic source field is thus y-directed and the main current x-directed; the dominating electric field is z-directed.

The following phenomena can be distiguished:

• There is a heating (and thus current) concentration in the high-ε part close to the low-ε area. One can explain this as being due to the current following the shortest path with high ε and thus "avoiding" the low-ε area. The boundary is concave and not convex as is the case with the cold rim effect, so conditions are now reversed.
• There is a significant heating in the centre region of the low-ε area. This is due to direct action by the strong external z-directed electric field. The phenomenon is amplified by the fact that the load is thin. It is to be noted that the heating intensity in the centre is quite strong: almost half of the maximum in the non-depleted area, in spite of the ε″ being ten times lower and the penetration depth being more than three times larger than in the non-dried region. This power density will indeed cause overheating, in consideration of the limited amount of remaining water for evaporative cooling in this semidry region.
• There is a significant heating intensity minimum just inside the low-ε area boundary. This is a result of the z-directed electric field being almost short-circuited there, by the high-ε area nearby. One can also explain the minimum as a so-called magnetic wall effect. The minimum intensity is less than half of that in the centre of the low-ε area. These sharp heating gradients result in the boundary between the outer relatively moist and the inner dry regions to become relatively constant, while the inner region continues to dry out. The resulting narrow burnt regions are clearly seen in the photo to the left.
  
  Can this effect of uneven heating be avoided? Load rotation or changes of the polarisation of the TM field will not help much; the inner overheated region will then instead become symmetrically circular. The use of other microwave polarisations will result in edge overheating as well as a risk of lower microwave system efficiency. There are no easy solutions.