Issue link: https://iconnect007.uberflip.com/i/623116
January 2016 • The PCB Magazine 69 as decibels per distance unit, typically dB/cm, whereby decibel is the ratio of the logarithms of the signal strengths at the beginning and the end point of the unit length. The former (high frequency) is not so intuitive. First, the trend to higher clock speeds and greater bandwidths to achieve higher data rates is well documented. The problem lies in the fact that the loss tan - gent (Df) of a glass-reinforced dielectric is not a constant but a function of the signal frequency, and, unfortunately, loss increases with frequen- cy. Therefore, it is important to choose a dielec- tric that has a low Df that remains reasonably flat over a range of frequencies. The dielectric resin's Df as a function of fre- quency varies: Df values may be fairly flat, or show an increase in a certain frequency range, or, more typically, decline slightly at higher fre- quencies, reflecting the frequency-dependent activation of chemical bonds. We find such low- loss dielectrics in microwave, radio frequency (RF), and radar boards and modules such as cell phone base station transmitters and receivers or cell phone RF modules. Without low-loss dielec- trics, transmitters would lose most of their sig- nal in the form of heat, and receivers could not maintain a weak incoming signal well above the background noise level. The issue is similar in digital applications where the trend is towards lower voltages. Very low-loss dielectric material such as PTFE (poly-tetrafluoroethylene) can also be found in high-end packages. It is of interest to explore why FR-4 has a rel - atively high loss while PTFE is a low-loss mate- rial. This has to do with the molecular structure of the resin, of the composite or the ceramic. If the resin has predominantly chemical bonds that are characterized by charge separation (i.e., polar bonds or ionic bonds as opposed to co- valent bonds), and if these dipole charges can easily move due to thermal or other activation, then such a resin will have a high dielectric con- stant and a high dissipation factor. On the other hand, a resin high in hydrocarbon domains (al- iphatic and/or aromatic) with no or few polar bonds is likely to have a lower Dk and lower Df. Low-loss resins comprise materials such as PTFE, polyphenylene oxides, functionalized polyethers such as allylated polyphenylene ether, building blocks such as butadiene, sty- rene, and maleic anhydride, and aromatic poly- esters such as LCPs (liquid crystal polymers). The problem is that without functionalization these materials are thermoplastics (not thermo- sets), so that FR-4-type multilayer pressing with prepreg is not possible. The way out of this di- lemma is the functionalization of the pure resin (e.g., with allyl-groups) and/or the construction of prepreg composites that contain some epoxy or similar thermoset; however, such maneuvers will dilute the desirable electrical properties to some extent. There is another ticklish matter: as the resin becomes more hydrocarbon-like in nature, adhesion to copper, or even to treated copper, becomes more problematic. Not only that, the resin itself is more prone to cohesive failure (i.e., the bonds between resin molecules are weaker than the polar bonds and hydrogen bridge bonds found in FR-4). It is of course pos - sible to create multilayers with unadulterated non-polar resins by approaching the melt point in a melt-fusion lamination which fabricators, who are used to FR-4-like processing, don't em- brace. Low moisture absorption of the base ma- terial is desirable from several points of view. Moisture absorption has been linked to failure of solder joints in surface mount ("popcorning" in the pressure cooker test). Absorption of mois- ture also lowers the de-facto Tg, which is detri- mental to dimensional stability and a problem with higher melting, lead-free solder composi- tions. This is not the end of the problem list. Moisture absorption generally leads to a dete- rioration of electrical properties (i.e., Dk and Df increase with increasing moisture content). The chemical approach to combat this problem is not unlike the one to gain better Dk and Df val- ues in the first place: hydrophobic hydrocarbon- like structures are preferred. Cross-linked, dense structures or crystalline moieties with low water affinity and low moisture diffusion rates are use- ful (e.g., LCPs). The need for high temperature dimen- sional stability (high Tg) stems from the advent of lead-free, higher temperature eutec- tic solders. Other drivers are the use of devices that generate a lot of heat such as chips that run at high clock speeds, larger sized chips, or op- toelectronic components with heat generating HIGH-PERFORMANCE LAMINATES karl's tech talk