Issue link: https://iconnect007.uberflip.com/i/1062752
DECEMBER 2018 I DESIGN007 MAGAZINE 65 is lower in conductivity, then the skin depth will increase, which is what happens when some types of final-plated finish are applied to the conductor for a PCB. The influence of the plated finish is a rather complicated issue, but I wrote on this subject a few months ago if you want more information. One particular plated finish that is used a lot in the industry is electroless nickel immer- sion gold (ENIG), and the effects of the ENIG are related to an edge effect of the conductor. At the edges of the conductor where the con- ductor meets the substrate, there is naturally higher current density, and the difference in the conductivity of metal at these edges will cause differences in RF performance. In the case of ENIG—assuming very low frequency where the skin depth is thick—the conductiv- ity at the edges of the conductor will be a com- posite conductivity made up of copper-nickel- gold. When the frequency is increased, there will come a point where the composite con- ductivity will be nickel-gold. At much higher frequencies, the conductivity will be mostly related to the gold layer. To give you a sense of the different conduc- tivities for these metals and with the unit being 10 7 S/m, copper is 5.8, nickel is 1.5, and gold is 4.5. These values are for the pure metal. In reality, these metals are usually an alloy for how they are used in PCB processing and are slightly different for conductivity, but these are good reference values. Nickel is approxi- mately ¼ the conductivity of copper, which is a double-edged sword for RF issues. The lower conductivity will cause more insertion loss and also increase the skin depth, which means the RF current is using more of the lossy metal. There is another twist for ENIG that is a potential issue related to magnetic issues. Pure nickel has a very high relative permeability (μ) of about 500, but the nickel used for ENIG is an alloy with a lower μ value than pure nickel—although it is still significant. With an increase in μ, it can be seen in the skin depth formula that the skin depth will decrease. This is an offsetting factor to the lower conductiv- ity of nickel. There are also magnetic losses associated with the metal, which are akin to the losses associated with dielectric. The dielectric losses are related to the dissipation factor (Df), and the magnetic losses are simi- lar—but related to—the magnetic properties of the metal. Nickel does have higher magnetic losses than copper. The following is an example of a real-world issue related to ENIG and skin depth. We had a customer tell us they were experiencing sig- nificantly different RF losses for their circuits when looking at the performance of many cir- cuits of the same design. It was basically cir- cuit-to-circuit insertion loss variation. It turned out that the operating frequency was at 800 MHz (0.8 GHz), which is an interesting fre- quency for skin depth as it relates to ENIG. The skin depth in copper at that frequency is about 2.3 microns (approximately 92 micro- inches), and would be a little thicker for ENIG. The nickel layer of the ENIG can vary from 50–250 microinches depending on many fac- tors. Normal circuit-to-circuit variation for ENIG is not that extreme, but there is a normal nickel thickness variation for ENIG that will be different for many different reasons. What was found was that the nickel thickness variation was in the right thickness range to impact skin depth variation, which was related to the composite conductivity of copper-nickel- gold changing from one value to another with a different nickel thickness. At this frequency, a change in the nickel thickness had a signifi- cant influence on the skin depth and associ- ated insertion loss. If the same nickel thickness variation occurred for an application operating at 24 GHz where the skin depth was about 17 microinches, the composite conductivity would not be impacted because the composite is made up of approximately eight microinches of gold and the rest is all nickel. Again, this is just at the edge of the conductor where the ENIG influences insertion loss. DESIGN007 John Coonrod is technical marketing manager at Rogers Corporation. To read past columns or contact Coonrod, click here.