Wideband evaluation of two types of slow-wave microstrip lines

The design, characterization and comparison of two widely used approaches in realizing slow-wave effect on microstrip transmission lines, that is stub loaded and defected ground structure loaded microstrip lines arepresentedinawidebandwidth(10–67GHz)fortheﬁrsttime.Trans-parentsubstrateanddielectricmaterialarechosentoeasethealignmentofelectrodeandgroundplane.Thindielectriclayerareappliedtomake thecomparisonprominent.TheresultsindicatethatdefectedgroundstructureloadedmicrostriplinehasbetterRFperformanceinterms ofcompactnessandinsertionlossthanstubloadedmethodwithinthewholebandespeciallyinthinﬁlmapplications.

✉ Email: wang@imp.tu-darmstadt.de The design, characterization and comparison of two widely used approaches in realizing slow-wave effect on microstrip transmission lines, that is stub loaded and defected ground structure loaded microstrip lines are presented in a wide bandwidth (10-67 GHz) for the first time. Transparent substrate and dielectric material are chosen to ease the alignment of electrode and ground plane. Thin dielectric layer are applied to make the comparison prominent. The results indicate that defected ground structure loaded microstrip line has better RF performance in terms of compactness and insertion loss than stub loaded method within the whole band especially in thin film applications.
Introduction: A transmission line (TL) is the fundamental and key component for RF systems. It is not only used in connecting RF components, but also in wavelength λ based millimetre wave (mmW) devices, such as branchline coupler [1,2], interference based crossover [3], impedance transformer, virtual coupling [4], and delay line phase shifters and so forth. Therefore, TLs with high performance are crucial for advanced RF systems. Microstrip line (MSL) is being widely used in mmW systems, because of its planarity and cost effective fabrication, which makes MSL suitable for monolithic microwave integrated circuits (MMIC) as well as for high-speed digital PCB. Compared with waveguides, the drawbacks of MSLs are lower power handling capability and higher loss. Hence, MSL with high compactness, low loss and high power handling capacity would be of interesting.
Stub loaded high impedance microstrip line: Compared to a conventional homogeneous 50 MSL, high impedance line loaded with open ended stubs (L-MSL), as shown in Figure 1a, performs longer electrical length within the same physical length, which is referred to as slow-wave (SW) effect. This approach has been widely applied in miniaturizing microstrip devices [1,2]. Its equivalent circuit is given in Figure 2a. Z C and θ C represent the characteristic line impedance and electrical length of The equivalent characteristic impedance Z E and electrical length θ E of L-MSL loaded with two stubs can be derived from ABCD-matrix: This gives: The SW factor K is defined as: According to Equation (3), with Z C = 50 , the achievable SW factor depends on the highest possible Z H that can be realized by microstrip TL. Nevertheless, the highest achievable Z H by narrowing the transmission line width W TL is limited to be around 100 , which limits the SW effect. Moreover, such narrow line causes high metallic loss and low power handling capability which largely limits the applicability of the L-MSL.
Defected ground structure loaded microstrip line: Instead of L-MSL, defected ground structure (DGS) loaded microstrip line (DGS-MSL) is also found to be able to provide SW effect, while keeping a low level loss. The 3-dimensional view of a DGS-MSL with rectangular dumbbell shape etched lattice on the ground plane is shown in Figure 1b, and its equivalent circuit is shown in Figure 2b. DGS introduces serial inductance to the TL by disturbing the return current on ground plane. Z L and θ L represent a low impedance TL, and L is the inductance of the single DGS lattice. Considering the SW effect equation could be derived similarly to Equation (3). It is realized by using low impedance TL with Z L < Z C such that W 3 is wider and leads to less loss than L-MSL. This is a big advantage especially in high frequency applications as well as in applications where MSLs are used with thin dielectrics of less than 20 µm.
Fabrication and measurement: By periodically cascading the elements shown in Figure 1, L-MSL and DGS-MSL realized in suspended microstrip line topology are designed and optimized in CST Microwave Studio (CST), both having Z E = 50 . The cross section and dimensions of the suspended microstrip structure are given in Figure 3. To ease the alignment between DGS and electrode, transparent AF32 glass from Schott GmbH with ε r = 5.1 and tanδ = 0.01 is chosen as the top and bottom substrate, on which electrode and ground are plated by using standard photo lithography process described in [5]. Two glass substrates are precisely aligned with a distance of h 2 = 20 µm held by spacers.
Then a fluid dielectric with ε r = 2.4 and tanδ = 0.002 is filled in. Besides, a conventional MSL without any loading method and thru-reflectline (TRL) kits are fabricated together with the proposed demonstrators as reference TL and post de-embedding kits, respectively. A photograph of all the aforementioned devices is shown in Figure 4. Three DGS-MSLs with different lattice dimensions, but same total length, are fabricated. Detailed dimensions are given in Figure 1. In order to perform on-wafer measurements on suspend microstrip TL with fluid dielectric,  Figure 5 in terms of the absolute value of transmission coefficient |S 21 | and reflection coefficient |S 11 |. All five demonstrators show broadband matching where |S 11 | < −15dB. The raw measurement results before deembedding also show good matching, indicating well-fabricated 50 devices, which is not plotted here for simplicity. Based on the deembedded results, unit length loss α, propagation constant β, which represents compactness, and quality factor Q of the demonstrators are calculated and given in Figure 6. As an overall evaluation of SW effect and loss, Q is calculated as: which represents how lossy the TL is. The higher the Q, the less loss generated by the TL. High Q transmission line is required in wavelength λ based passive mmW components, such as coupler, impedance transformer, delay line phase shifters and so forth, to minimize the dimension, loss and heating issues. Within their pass-bands, DGS-MSLs perform almost the same α as MSL, while L-MSL exhibits around twice the loss. Due to expected SW effect, DGS-MSLs and L-MSL all show higher β than MSL, among which DGS-MSL3 with largest etched area indicates the largest SW factor K of approximately 1.9 times that of MSL at 40 GHz. However, DGS-MSL3 performs low-pass feature and has a cut-off frequency at around 40 GHz due to the high lumped serial inductance compared to DGS-MSL1, which limits its pass-band according to   Table. 1. Compared to widely used L-MSL in miniaturizing microstrip TLs, DGS-MSL shows advantages of less loss, wider SW range and higher quality factor. In addition, the electrode width of DGS-MSL is wider than MSL, which increases its power handling capability and decreases the fabrication tolerance especially in thin film and high frequency applications. This gives a higher versatility of DGS-MSL. However, the etched dimension of DGS-MSL should be carefully chosen to provide suitable SW factor and passband: the higher the miniaturization, the lower the cut-off frequency of DGS-MSL. In addition, alignment of DGSs with electrode on thick, untransparent dielectric is challenging, especially when dimensions of TLs and DGS lattices are small which may cause degraded matching. While L-MSL with only one patterned metal layer is easier to be fabricated. L-MSL's disadvantage of high IL might be not critical in thick dielectric situations. The conclusions above are instructional in choosing either of the methods to design compact and high performance mmW microstrip circuits.