Semiconductor grade epoxies

This paper reports the results of a molecular-level investigation of the effects of flame retardant additives on the thermal dedompositlon of thermoset molding compounds used for encapsulation of IC devices, and their implications to the reliability of devices in molded plastic packages. In particular, semiconductor grade novolac epoxy and silicone-epoxy based resins and an electrical grade novolac epoxy formulation are compared. This work is an extension of a previous study of an epoxy encapsulant to flame retarded and non-flame retarded sample pairs of novolac epoxy and silicone-epoxy compounds. The results of this work are correlated with separate studies on device aglng2>3, where appropriate. 

EGA. The overall ion profiles for the semiconductor grade epoxy compounds are presented in Figures 9 and 10 for samples F(FR) and G(no-FR), respectively. Very little outgasslng is observed from these samples below 200 C, in marked contrast to the results obtained for the electrical grade epoxy samples. These data clearly reflect the effects of the more stringent processing controls employed in the production of the semiconductor grade materials. Also, because of the lower out-... 

Since no difference was found between FR and non-FR formulations in device aging studies, some other cause must account for the relatively early failures observed for devices molded in the electrical grade epoxy material and aged under bias at 200°C. These failures are attributed to chloride contamination present in the non-semiconductor grade epoxy resin. The extractable Cl concentration is a factor of four higher than Br, and this is correlated with a much higher concentration of CHoCl than CH Br in the EGA data below 200°C. The high Br concentration is also attributed to the... 

Figure 11. Differential scanning calorimetry curves for semiconductor-grade novolac epoxy compounds. Key ---, Sample F(FR) ----, Sample G (no FR) ambient-nitrogen temperature...

Semiconductor Grade Silicone-Epoxy. TGA, DSC, and EGA analyses revealed no difference between the FR and non-FR compounds below 200°C. The FR moieties again decomposed only in the temperature range above 350°C. There was very little Cl or Br in the aqueous extract, and no CH2CI or CH2Br was detected in the EGA product profiles. This shows the capability of material formulators to supply very clean semiconductor grade molding compounds. 

Semiconductor Grade Epoxies. As was the case for the semiconductor grade silicone-epoxy, there was no difference between FR and non-FR epoxies recorded by either DSC or EGA below 200°C. However, the nominally equivalent non-FR epoxy exhibited significantly lower thermal stability as indicated by the Isothermal TGA data. Furthermore, the aqueous extract of the non-FR compound contained more than twice as much Cl as the combined concentrations of Cl and Br in the FR epoxy. Although there have been no direct comparisons on device aging with these two epoxies, the above findings indicate that the FR compound, being cleaner and more thermally stable, could actually be the better material for encapsulation applications. 

For epoxy resins used in electronic applications, such as cresol epoxy no-volacs, more powerful polar aprotic solvents such as dioxane or dimethyl for-mamide (DMF) have been used to hydrolyze the difficult-to-hydrolyze HyCls, such as the abnormal chlorohydrins and the organically bound chlorides. The issue here is the inconsistency in results obtained by different methods (78). The presence of ionic hydrolyzable chlorides and total chlorides has been shown to affect electrical properties of epoxy molding compounds used in semiconductor encapsulation (85). For these applications, producers offer high purity grade epoxy resins with low ionic, hydrolyzable and total chloride contents. 

---