Journal of Infrared and Millimeter Waves, Volume. 43, Issue 1, 85(2024)

An improved method for determination of extrinsic resistances for HEMT devices based on 110 GHz S-parameters on-wafer measurement

Zhi-Chun LI1, Yuan-Ting LYU1, Ao ZHANG2, and Jian-Jun GAO1、*
Author Affiliations
  • 1School of Physics and Electronic Science,East China Normal University,Shanghai 200241,China
  • 2School of Transportation and Civil Engineering,Nantong University,Nantong 226019,China
  • show less

    An improved method for determination of extrinsic resistances for 70 nm InP high electron mobility transistor (HEMT) is proposed in this paper. A set of expressions have been derived from the equivalent circuit model under operating bias points (Vgs > Vth, Vds = 0 V). The extrinsic resistances are iterative determined using the discrepancy between simulated and measured S-parameters as an optimization criterion using the semi-analytical method. Good agreement between simulated and measured S-parameters under multi bias over the frequency range up to 110 GHz verifies the effectiveness of this extraction method.

    Keywords

    Introduction

    Accurate extrinsic resistances extraction for modeling InP HEMT devices is a crucial step in the design and production of high-yield,low-cost millimeter wave circuits1. The common used methods mainly include numerical optimization method,cold FET method and cutoff method. In optimization method,the extrinsic resistances extraction result strongly depends on the initial value and suffers from non-uniqueness and non-physical meaning2. In cold FET method,the large gate currents caused by forward-biased run through the gate Schottky junction,which leads to degradation of the gate3-6. To avoid degradation of device characteristics,some authors proposed cutoff method to extract extrinsic resistances. However,this method is valid only at high frequency(>18GHz),and the extracted resistances fluctuate widely over the whole frequency range of interest7-9.

    In order to overcome these limitations,an improved method for determination of extrinsic resistances is proposed. In contrast with previous publications,this extraction method offers the following advantages.

    1)Under operating bias point(Vgs > Vth,Vds = 0 V),the effect of channel between source and drain can be modeled by resistance Rch,while the capacitance will be dominant under cutoff bias condition.

    2)The semi-analytical method which is a combination of optimization method and analytical direct extraction method has been used to determine the extrinsic resistances.

    3)This extraction method is verified with S-parameters on-wafer measurement up to 110 GHz.

    Section Ⅱ gives the equivalent circuit model under operating bias point(Vgs > Vth,Vds = 0 V)as well as the derivation of analytical expressions. Section Ⅲ gives the procedure of intrinsic parameters extraction. The measured and simulated results are presented in Section Ⅳ. The conclusion is given in Section Ⅴ.

    1 Extrinsic resistances extraction

    1.1 Equivalent circuit model

    Fig.1 shows the small-signal equivalent circuit model under operating bias point(Vgs > Vth,Vds = 0 V). CpgdCpg,and Cpd represent the extrinsic capacitances due to the pad connections. LgLd,and Ls represent the extrinsic inductances of the gate,drain,and source feedlines. Rs and Rd represent the extrinsic resistances of the source and drain. Rg is the distributed gate resistance. Cgsa and Cgda represent intrinsic gain-source and gain-drain capacitances respectively. Rch is the channel resistance when the channel between drain and source is open at zero drain bias and gate voltage above threshold voltage.

    The small-signal equivalent circuit model under operating bias point(Vgs > Vth,Vds = 0 V)

    Figure 1.The small-signal equivalent circuit model under operating bias point(Vgs > Vth,Vds = 0 V)

    The intrinsic part of the equivalent circuit model,which exhibits a PI topology,so it is convenient to describe it by a Y matrix as:

    ,

    Convert -parameters to -parameters:

    ,
    ,
    ,
    ,

    The Z-parameters of intrinsic part with extrinsic resistances can be expressed as following:

    ,
    ,
    ,
    ,

    Therefore,we have

    ,
    ,
    ,

    By neglecting the high order term ,the extrinsic resistances can be obtained by analytical expressions from real part of -parameters:

    ,
    ,
    ,

    where

    ,
    ,

    Cgsa and Cgda can be determined at low frequencies:

    ,
    ,

    1.2 Extrinsic parameters extraction

    The pad capacitances can be determined by measuring an open test structure:

    ,
    ,
    ,

    The extrinsic inductances LgLd,and Ls can be determined from the imaginary part of Z-parameters(transformed from measured S-parameters)of the short test structure directly:

    ,
    ,
    ,

    The extraction of extrinsic resistances can be based on the semi-analytical method which is a combination of optimization method and analytical direct extraction method. The extraction procedure is as follows.

    1)De-embedding the pad capacitances and feedline inductances.

    2)Set the initial value of the channel resistance Rch .

    3)Calculate the extrinsic resistances RgRd,and Rs using(13)-(15)which can be expressed as functions of Rch as well as frequency.

    4)Set error criteria as follows:

    ,

    Where represents the calculated S-parameters,and represents the measured S-parameters. represents the discrepancy between simulated and measured S-parameters. Wpq are the weighting factors.

    5)If error criteria are small enough,the iterative process will be end.

    Flow chart of the algorithm

    Figure 2.Flow chart of the algorithm

    2 Intrinsic parameters extraction

    The small-signal circuit model of intrinsic part of InP HEMT devices is illustrated in Fig. 3 after de-embedding the extrinsic parameters,and the intrinsic Y-parameters are described as3

    ,
    ,
    ,
    ,

    The small-signal circuit model of intrinsic part

    Figure 3.The small-signal circuit model of intrinsic part

    with

    From the analytical expressions(27)-(30),the intrinsic element values can be obtained directly.

    3 Results and discussion

    In this paper,70 nm InP HEMT devices have been used with 2×30 μm gate width(number of gate fingers×unit gate width). The S-paremeters on-wafer measurement up to 110 GHz using N5247 network analyzer with DC bias by an Agilent B1500A.

    Fig. 4 shows the plot of extracted extrinsic resistances versus frequency using the presented semi-analytical method and cutoff method9. As can be seen from Fig. 4,the extracted Rd and Rs using semi-analytical method are nearly constant versus frequency,while extracted Rd and Rs using cutoff method fluctuate widely versus frequency. These results verify that the proposed extraction method is better than cutoff method 10.

    Extracted extrinsic resistances using the proposed semi-analytical method and cutoff method

    Figure 4.Extracted extrinsic resistances using the proposed semi-analytical method and cutoff method

    • Table 1. The values of extrinsic resistances

      Table 1. The values of extrinsic resistances

      ElementValue
      2.66
      4.18
      3.80

    Table 1 gives the values of extrinsic resistances extracted using semi-analytical method. The values of extrinsic capacitances and inductances are given in Table 2. The extracted intrinsic parameters in Section III are shown in Table 3.

    • Table 4. Accuracy of S-parameters

      Table 4. Accuracy of S-parameters

      Bias ConditionS11S12S21S22
      5.45%1.55%4.37%2.59%
      5.46%1.46%4.26%2.45%
      4.99%1.85%4.80%2.74%
      5.82%1.97%4.97%4.04%
    • Table 3. The values of intrinsic parameters

      Table 3. The values of intrinsic parameters

      ElementValueElementValue
      46.9782.2
      9.695.20
      15.783.46
    • Table 2. The values of extrinsic capacitance and inductance

      Table 2. The values of extrinsic capacitance and inductance

      CapacitanceValuesInductanceValues
      13.2622.7
      13.1928.5
      0.4163.25

    The intrinsic elements listed in Table 2 are substituted into the equivalent circuit model for simulation. Fig. 5 shows the comparison between simulated and measured S-parameters under multi bias. Good agreement between simulated and measured S-parameters are observed in the frequency range of 1 GHz to 110 GHz,which verifies the validity of this developed method to determine the extrinsic resistances.

    Comparison between simulated (lines) and measured (squares) S-parameters over the frequency range going:from 1 GHz to 110 GHz under multi bias

    Figure 5.Comparison between simulated (lines) and measured (squares) S-parameters over the frequency range going:from 1 GHz to 110 GHz under multi bias

    Table 4 outlines a selection of the accuracy of S-parameters for four different bias points. As can been seen from this table,S11 and S21 maintain an accuracy of around 5%,S12 has an accuracy of less than 5%,and S22 has an accuracy of between 2%~5%.

    4 Conclusion

    An approach for determination of extrinsic resistances for 70 nm InP HEMT devices is proposed in this paper. Extrinsic resistances are described as functions of the intrinsic channel resistance,and optimum values can be obtained using semi-analytical method under operating bias point(Vgs > Vth,Vds = 0 V). Verification of this extraction method is presented by the good agreement between the simulated and measured S-parameters under multi bias over the frequency range up to 110 GHz.

    [6] M Y Jeon, B G Kim, Y J Jeon. A Technique for Extracting Small-Signal Equivalent-Circuit Elements of HEMTs. IEICE Transactions on Electronics, 82, 1968-1976(1999).

    Tools

    Get Citation

    Copy Citation Text

    Zhi-Chun LI, Yuan-Ting LYU, Ao ZHANG, Jian-Jun GAO. An improved method for determination of extrinsic resistances for HEMT devices based on 110 GHz S-parameters on-wafer measurement[J]. Journal of Infrared and Millimeter Waves, 2024, 43(1): 85

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Save article for my favorites
    Paper Information

    Category: Research Articles

    Received: Mar. 31, 2023

    Accepted: --

    Published Online: Dec. 26, 2023

    The Author Email: GAO Jian-Jun (jjgao@ee.ecnu.edu.cn)

    DOI:10.11972/j.issn.1001-9014.2024.01.012

    Topics