TSPEM Parameter Extraction Method and Its Applications in the Modeling of Planar Schottky Diode in THz Band

Round 1

Reviewer 1 Report

This manuscript reports on the methodology for the extraction of Schottky diode parameters at high frequencies. It makes an impression to be a  scientifically sound contribution. However, the main claims about the accuracy of the proposed method are not supported quantitatively. Below I would like to address some aspects in detail:

1) In Figure 8, the results are compared with the "traditional formula, " which is neither cited nor presented. Does it include the circuit as presented in Figure 5 and Table 1 or refers to a different set?

2) The experimental validation of model-based simulations appears to be supportive, the performance of the fabricated frequency doubler is not at the state-of-the-art level for this kind of device. Can the model allow better prediction regarding the optimal performance (for example the maximum efficiency, etc.)?

3) The experiment implies measurement of the total emitted power, which should include outcoupling loss. Which value for the efficiency assumed in the model?

 4) In Figure 9 the modeled curve is smooth whereas the experimental one shows oscillations that typically originate from the impedance mismatch. Why the impedance mismatch does not appear in the simulated curve if the model is claimed to be between the most accurate models?

5) It would be of interest if the simulations of the performance of frequency doubler would be extended to the predictions using other, previously addressed, models.

6) The summary section presents the statement that the proposed technique is "more accurate", however, the quantitative analysis is not performed. In their current form, such claims look too vague.  

Author Response

Dear editor：

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

-Some references have been previously published in this journal about modeling of Schottky diodes and Zero Bias diodes for high frequencies. Maybe you could check them.

-Authors have direct access to the technology as they can fabricate several versions of diodes for calibration (FIG 6: without junction but with ommic contact, open , thru, etc.). Maybe you could provide some technology information about GaAs substrate growing, MBE, RIE, anode diameter ranges, etc...

-Line 6. To eliminate--> maybe you mean to de-embeed?

-line 23,24 solid source --> maybe you mean solid-state source?

-Line 24 Maybe you should explain that you refer to Thz sources based on frequency multiplication of lower frequency tones

-Figure 1: I would suppose that the schottky junction should have one port at each side (to connect a nonlinear circuit of the junction in between for fully circuit simulations). To put a single port in the junction point, and to ground it, it is not clear at all. Could you explain ports in fig. 1a and b?

-Line 66 additional parasitic parameters --> maybe you mean effects?

-Line 97: "It is also used for reducing the impact error caused by the discontinuity after the test pad  ¿?¿?  is directly equivalent to a capacitor and the transmission structure is equivalent to an ideal transmission line". Is something missing in this phrase?

-Line 141 Cross-sectional

-Figure 5: You use Rj and Cj to model the intrinsic junction, nonlinearly dependent on applied voltage, even in a zero bias device. Could you provide specific plots about this dependence?

-In Fig. 5 You mention Lpad related to anode and cathode mesas, but you neglect. Could you please indicate Lpad on the figure and/or on the circuit?

-You refer to Fig. 6 like” simulation models”. It means Fig 6 contains drawings for HFSS simulation of the special structures for de-embedding?

-Line 159: You mention extraction at lower frequency ranges. DC measures and higher frequency measures are also used. Maybe you could represent the steps in a flow diagram indicating frequency ranges?

-Line 192: You say: “…can be obtained through parameter fitting through engineering.”  Referring to equation 14. Parameter fitting sounds like optimization, but equation 14 is analytical. Could you please clarify?

-Line 227 you mention fabrication for several anode diameters. Have you consider including physical dependence on the equivalent circuit and comparison with classical model dependent on anode area or diameter?

-Figure 8 shows better agreement for proposed model than for C-V but to be fair the frequency of operation of semiconductor analyzer should be indicated (maybe is in the range of MHz)

-In Figure 8 maximum frequency is 40 GHz, but later the model is used above 80 GHz and measured above 160 GHz. Could you justify this?

- Were the diodes optimized to better suit doubling function (removing of dielectric layers, optimization of anode area…)? Is this optimization also valid for mixing or detection? Maybe you could include here some details of doubler design (14). Are Zero bias diodes the most suitable for a doubler? Could additional DC bias improve the frequency conversion?

Author Response

Dear editor：

Please see the attachment

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

-Authors have addressed many points raised but some doubts remain:

-According to your answer:

 Point 4:I would suppose that the schottky junction should have one port at each side (to connect a nonlinear circuit of the junction in between for fully circuit simulations). To put a single port in the junction point, and to ground it, it is not clear at all. Could you explain ports in fig. 1a and b?

Response 4:

This is a method that combines the intrinsic parameters of the diode and the electromagnetic coupling effect to establish a high-frequency model of the diode[1].Since the wave port can only be set on the ideal conductor surface in the HFSS software, only the setting method in Figure 1 can be used. If you want to have one port at each side, you can set two lumped ports. Both two setting methods are feasible[2].

I understand that maybe in Fig. 1, b, port 1 should be also grounded?

-I cannot find in the paper the expressions and parameter values (ideality factor, built in potential, etc.) of the equations you wrote in answer 7, and I think it could be interesting as the intrinsic part is the core of the diode:

The trend of Rj and Cj with voltage changes is as follows: ?(?)=?0[???(?????)−1]=??? ??=????√2????(???−?)

In fact when you bias the diodes in the multiplier you are using this dependence.

-In this phrase maybe “and” could be omitted: “the TSPEM method is developed to get a more accurate and intrinsic junction capacitance value of the SBD in the Terahertz band”

-Line 303: “…with the range…”, maybe within the range?

-Line 307: trough means Channel, maybe you mean “pass through”?

-Line 308-309: “…resistance during assembly” . Conductor losses and its variation with frequency use to be underestimated by models. Maybe you could consider and mention this possibility as an additional cause.

 

 

Author Response

Please see the attachment

Author Response File: Author Response.pdf