3M TM VCP TM Package Stripline Two Port Characterization David A. Hanson Division Scientist 3M Microelectronic Packaging 3M 21
2 Background To support the requirements of 2.5GBs and 1GBs data communication standards, it has become increasingly important to understand the high frequency performance of transmission lines within a package. To this end a characterization study was undertaken to evaluate the two port transmission and reflection performance of stripline transmission lines embedded in a typical 3M TM, 7 Layer VCP TM package. The effort centered on extracting S11 (reflection) and S21 (transmission) parameters over a frequency range of 45MHz to 2GHz. All the analysis is on done on single transmission lines, since the appropriate test structures and equipment to perform differential measurements are currently not available. Test Vehicle Design An electrical test vehicle was designed to characterize the standard 3M TM 7 layer VCP cross section as shown below. In this cross section, the signal layers are formed on metal layers M2 and M6, and generally form a stripline transmission line. M1 and M7 merged pad and plane layers. M3, M4 and M5 are all solid power planes. Exposed Outer Metal Surface Finish: 12um CU 5um NI.15um AU S: 35um Soldermask In a typical flip chip application, where flatness is critical to successful assembly, careful balancing of metal density needs to be performed across the centerline. In other words, the amount of metal on M1 is very carefully balanced with the amount of metal on M7. In a similar fashion, M2 is matched with M6, and M3 is matched with M5. This creates a more complicated reference plane for the transmission line. Plots of the outer layers are include here, which clearly illustrates this point. As can be seen from the layer images (figures 1-4), the test vehicle consists of 4 sub vehicles, each 4mm square, with a 1521 BGA footprint. The purpose of the different sub vehicles is to analyze varying proximity effects to the holes in reference planes M1 and M7. Of special interest in this study are the two structures on the left. The lower left structure with a solid ground plane on both outer surfaces will be used as a reference for a best case scenario. It also will give a realistic representation of the capability available if an inverted cross section is used. The upper left sub vehicle is a close representation of typical SCM design to date. Three styles of transmission lines were studied: 1. Lines under a solid plane using the lower left sub-vehicle 2. Lines under a holie plane, centered under the holes using the upper left sub-vehicle 3. Lines under a holie plane, offset from the holes again using the upper left sub-vehicle. For each style of transmission line, both signal layers were measured, namely M2 and M6. All lines were 4mm long.
3 Figure 1. Test Vehicle Top Surface Metal (M1)
Figure 2. Test Vehicle Bottom Surface Metal (M7) 4
Figure 3. Test Vehicle Top Signal (M2) 5 Figure 4. Test Vehicle Bottom Signal (M6)
6 Measurement Results All measurements were performed with an HP851 using a full two port calibration. Connection to the test substrate was accomplished with Cascade Microtech air coplanar microwave probes. The frequency was swept from 45MHz to 2GHz, and all two port parameters were extracted. Data was collected on 5 substrates from 5 different manufacturing lots which were built over a period of several months. Figures 5-7 show a summary of the results of the 5 measurements superimposed on one another. From the data in figures 5 and 6, it can be seen that the reflection loss, S11, is generally much less than 15dB over a very wide frequency range for both the solid plane structure and the holie ground plane with the striplines offset from the holes. This is an interesting result, in that it suggests that as long as the lines are away from the holes in the ground plane, the performance characteristics of the line are maintained. Also of interest, is that there is not a significant degradation of performance between the signal layers. This indicates that the effect of the buried via connecting from the top signal layer to the bottom signal layer is not of great impact in the structure. Not suprisingly, the S11 results for the holie ground plane with the striplines centered under the holes is not quite as good, as shown in figure 7. It is, however, still very near 15dB over the swept frequency range. This is a very encouraging result, since rarely does a production design ever approach this worst case over the length of the line. No significant differences are seen among the transmission parameter, S21. In all cases it looks like the line loss is maintained at about.25db/mm at 2.5GHz, increasing to about.5db/mm at 1GHz. Again no great difference is seen between the signal layers. Ground structure has only a minor effect on this parameter. Conclusions The measurements suggest that a design goal of less than -15dB return loss (S11) on 3M TM VCP TM Package transmission lines is very achievable. A couple of options for meeting this goal are available. First critical nets should be designed such that they stay covered by ground planes both above and below the line. This is somewhat restrictive, and may lead to difficulties in some cases. A second approach would be lower the target impedance of a full stripline without holes above or below it. This would have the effect of lowering the return loss of the worst case line covered by holes, at the expense of the more typical case of a line covered by solid metal.
7 Figure 5. Solid Ground Plane M2 Stripline Solid Ground Plane -1-2 -3-4 -5-6 -7 M2 Stripline Solid Ground Plane -1-2 -3-4 -5
M6 Stripline Solid Ground Plane 8-1 -2-3 -4-5 -6-7 M6 Stripline Solid Ground Plane -1-2 -3-4 -5
9 Figure 6. Holie Ground Plane, Solid Cover M2 Stripline Holie Ground Plane, Solid Cover -1-2 -3-4 -5-6 -7 M2 Stripline Holie Ground Plane, Solid Cover -1-2 -3-4 -5
M6 Stripline Holie Ground Plane, Solid Cover 1-1 -2-3 -4-5 -6-7 M6 Stripline Holie Ground Plane, Solid Cover -1-2 -3-4 -5
11 Figure 7. Holie Ground Plane, Holie Cover M2 Stripline Holie Ground Plane, Holie Cover -1-2 -3-4 -5-6 -7 M2 Stripline Holie Ground Plane, Holie Cover -1-2 -3-4 -5
M6 Stripline Holie Ground Plane, Holie Cover 12-1 -2-3 -4-5 -6-7 M6 Stripline Holie Ground Plane, Holie Cover -1-2 -3-4 -5 8-621-2465-3