Terahertz - Science topic

Dear Lavi Kumar Vaswani,

Do you know how to plot in the CST software? Could you please share the way?

Thank you so much

The solution for this is: The photodiode saturates when incident power is more then 1mW even the thershold is 20mW. So check the manuel of the photodiode and keep the incident power less then the saturation power you can use ND filter to do this.

Chris Bohr, Hello sir i am also facing the same issue at low power approx 1mW quarter waveplate remin stable for more time but eventually the photo diode (A-B) saturates to max or min. voltage. Also if i reduce power the signal become noisy. Have you found solution of the problem please let me know. In some setups I notice people use mirror after crystal or WP to change the direction of beam before photodiode so is this calibration issue is due to reflection?

Have you addressed the issue please let me know solution.

To add an ion gel layer with a refractive index of 1.42 and a thickness of 0.15 µm in CST Microwave Studio, create a new material, define the geometry as a rectangular prism, and assign the material to this layer. Ensure the layer is correctly positioned within the simulation stack and configure the simulation parameters before running the analysis.

Hi!

Thank you for your answers.

I will try to calculate.

Sorry, but I'm not criticizing your work.

Don't take it so personally.

1) The spectral bandwidth of STEs is up to 30 THz, thus, enabling ultra-broad spectral measurements of various materials.

2)TE-based THz spectroscopy system used to measure the THz-frequency photoconductivity of perovskite samples as a function of the doping density and temperature.

3)THz emission spectroscopy from STEs is also an excellent tool for studying the spin–charge conversion and related phenomena

That is a loaded question, when you say Terahertz Spintronics I assume you are talking about using THz pulses to produce spin currents. In this case in a Ferromagnet/Normal Metal heterostructure, a pulsed laser is used to promotes electrons from below the Fermi energy to bands above it, due to the fact that ferromagnets have majority and minority energy bands a transport of spin polarization i.e a spin current will be inject into the adjacent layer, for a more details I recommend the following article "Nature Nanotech 8, 256–260 (2013)". https://doi.org/10.1038/nnano.2013.43

An alternative meaning to THz Spintronics is the energy dispersion/dispersion relation of Antiferromagnets (AFM) (and in some cases some Ferrimagnets), depending on the anisotropies of an AFM you can have spin waves propagating at THz frequencies, experimentally it is possible to probe some points of the Brillouin zone with Brillouin/Raman light scattering with neutron scattering and with antiferromagnetic resonance although the last one need really high magnetic fields. For more details I recommend the following book "Fundamentals of Magnonics" chapter 5.

Büyük tümörlerde terahertz frekans aralığı yeterli olmayabilir ve cihaz bunu algılamayabilir. Bunun için büyük tümör saptanmasında gigahertz kullanılıyor. Terahertz ise genellikle küçük ve fazla etkili olmayan büyük tümörlerde kullanılır. Meme bölgesindeki küçük tümörler genellikle meme bölgesi sağlıklı ise kendiliğinden sindirilir. Istiyorsanız "kadınlarda meme kanseri" yazısını okuyabilirsiniz.

Hi Luke Peters.. Thank you so much for the reply.. Good to hear that you belong to THz Imaging domain..

It is not possible to discuss the details of your optical scheme without drawings. You can contact me by e-mail <aatarasov@laseroptek.com> if you need.

Adnan Zahid thank you very much for mentioning the paper.

Yes.

First of all you need specifically determine detection mechanism, then you be able to find description path. In my view, if the target virus shows distinct absorption and /or reflection versus THz radiation, you can develop a meta-surface according to target frequencies. Backing to your question regarding simulations, I have to say that several possible simulations are available to you. You can use COMSOL, CST, Lumerica and MATLAB to describe your device. Even you can use equivalent circuit model for speedier simulations. In this way you have to describe your materials and structures. For instance see:

Biabanifard, Sadegh, et al. "Tunable ultra-wideband terahertz absorber based on graphene disks and ribbons." Optics Communications 427 (2018): 418-425.

Chris Bohr Spectrally broad spectrum. You can look for some papers on the internet. I would recommend to collect all the pump residue from LNO and inject it into the spectrometer. Be careful not to damage the spectrometer.

In principle yes however there is serious problem: how to achieve good sensitivity? To extract information, one needs to perform down-conversion or envelope demodulation / pulse position detection. For Thz continuous wave one can do down-conversion by some kind of mixer however of Thz pulses we mostly left with option of square (envelope) detector and it requires much stronger signal. Some limited distance can obtained by providing rather high power of transmission.

In short THz communications will be part of 6G. Due to its extremely short range due to poor propagation performance. Such a technology is promising very low latency and very high data-rates, but not intended to provide all kind of services. only expected for some specific use-cases like Indoor connectivity, inter-datacenter communications, UAV/satellite communications, etc.

Hey,

Just a little disclaimer I am an active researcher in THz but i do not tend to do spectroscopy so my information may not be completely up to date.

As far as i am aware this is no commerical software that does this, some THz antennas setups are sold with their own software (such as the BATOP setups)

In the vast majority of cases THz still tends to be lab based research, as such each individual lab tends to have their own fitting codes they have produced themselves. One option would be for you to do this, then you would have a code working exactly like you wished. Depending on the parameters you wish to extract if you shoot me a message/email I can give you some guidance on how to achieve this :)

Alternatively i know there is a github project focusing on this, although i haven't ever tested it myself i've attatched a link here: https://github.com/THzbiophotonics/Fit-TDS

You may consider using some silicon wafer with high brightness light image projected on it e.g. https://core.ac.uk/download/pdf/41979793.pdf

Good question

Hello:

I would suggest a decent background in:

1) Electromagnetism--Dipole models, plasma and material interactions, dispersion, Gaussian beam propagation, etc.

2) Wave/Fourier Optics--We take lots of Fourier transforms for spectroscopy

3) Nonlinear Optics--Some of the best sources are made by optical rectification

4) Lasers--most coherent sources require ultra-short pulsed laser

Thank you Manoj Kumar Banjare sir for the publication. I am reading it now

For THz can used nanowire dipole and monople antennas, dielectric resonators antennas as well.

https://www.youtube.com/watch?v=5OQYVaiqD1k

Hi Masoud,

this open access article seems to give a good overview of the required algorithms for the extraction of the relative permittivity and loss constant:

Usually, most research institutions with TDS systems also have the scripts and are able to post process the raw data after the measurements.

If this is not your case, I found an open source code written in Python:

The code is explained in the following paper:

I hope it helps!

I am not sure whether this would be applicable to your specific setup but a general method is to measure the input impedance :

1. while antenna is radiating in free space ( at such high frequencies chamber may not be required). This would show power loss plus power radiated.

2. Cover the antenna structure with a metallic casing and measure the input resistance. (this would show the power loss)

3. Subtract 2. from 1.

It depends on the thermal part of incident energy, which in turn rising the charge carriers from valence to conduction band

Hello Esraa,

You might be interested by our work in silk foam THz characterization:

We measuredthe refractive index of silk foam. Because of it's high porosity, we found a refractive index of 1.06 very close to that of air. I guess cotton would show similar values very close to 1 (if its porous).

Otherwise, you can also check our review on THz imaging. Specifically, section 2.2 where we show how to calculate refractive index from transmission/reflection measurements. Here:

Also, check the terahertz literature, characterization of cotton may have already been demonstrated.

Best

When you refer to a THz laser source are you referring to something similar to a QCL? In which case i am not aware of anything which can be synthesised without expensive equipment. To my knowledge every THz-QCL requires photolithography and wet etching processes (an example paper on high paper THz-QCL: https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=6746300).

If however you are after Terahertz Time-Domain spectroscopy, there are plenty of inexpensive options for generation materials, such as ZnTe, GaAs, PCAs, InAs etc... This technique has the advantage of being broadband but requires a femetosecond laser source to drive the operation which in itself can be quite expensive.

However i hope that some inexpensive options for QCLs will become available soon, and if anyone has any further information on this please let me know!

THz radiation is made up of femtosecond pulses so you need a femtosecond laser (or femtosecond something) to generate it, as Paul Kinsler describes.

The absorption of the water molecules to the electromagnetic waves is due to the polarization mechanism where there is orientation polarization at relatively low frequencies after which comes the absorption due to ionic polarization at the long infrared and after that comes the electronic polarization at extremely high frequency. There is also the attenuation due to the conductivity of the water because of the salt contained in it. May be the salty water is very similar in response to the human body.

Best wishes

You must have a load resistor connected in the circuit. Also you cannot use a current source since its available power is infinite. You must add a source resistance in parallel with the current source, equal to the load. Then the efficiency can readily be calculated, and is left to an exercise by the student.

As said by Zahed Hossain you should not select THz frequencies for wireless communication that have high water vapor absorption. In addition to wise selection of THz frequencies and high gain antennas you can overcome high pathloss penality by beamforming with ultra-massive MIMO antennas.

Thanks Rajib Biswas and Ahmed Bakr .

How can I access to your special suggestions , give me the link please .

It depends on frequency. Note that at 13 GHz the imaginary part of

conductivity of YBCO is about 10^8 S/m while the real part is two orders of

magnitude smaller. At 1.3 THz the real and the imaginary part of conductivity

will be similar (the imaginary part vary with frequency as f^-(1), while the real part

can be consider as a constant value).

Also when you use Drude model for the real part of conductivity it is necessary

to know plasma frequency for YBCO.

Thank You All,

Indeed infra-red properties of graphene monolayers depend strongly on the doping, and temperature as established by L.A. Falkovsky. With appropriate modelling of Fresnel coefficient as well as the permittivity, it possible to extract the complex refractive index.

With reference to the to the paper suggested by Dr Azeez, I think the real index would be approximately 2.0 in THz range at weak doping.

1) continue with the left overs of 5G (Massive MIMO, mmWave, Terahertz, D2D, UAVs, NOMA, ...etc )

2) Adding Artificial Intelligence, specifically Machine Learning to the Loop

SPP is supported by the plasmonic metamaterials named spoof surface plasmon polaritons like graphene. Surface plasmon polaritons (SPP) are collective oscillations of free electrons trapped at metal-dielectric interfaces. When electrons in graphene are excited by an incoming electromagnetic wave, they start moving back and forth. This global oscillation of electrical charge results in a confined electromagnetic wave on top of the graphene layer. For transmitting, the SPP waves can be created by injecting electrons into the dielectric layer beneath the graphene sheet. In antennas, microstrip SPP transmission line is used as the feeding line to improve the beam scanning range.

For figure12 (b) In this article, it is stated that the each of the higher orders (5th and 7th) have had some arbitrary scaling applied to fit on the same figure as the 3rd order process. I believe this is the root of your confusion.

To my knowledge third-order processes are more efficient for THz generation than the higher order processes

It mostly depends on the density of the carriers. You could use Drude model to simulate the optical response of a film under different density. The carriers' effect would influence the film's optical state (transparent-loss increasing-loss decreasing-reflectant) with the increase of the density.

Have a look at these articles too.

a)

https://core.ac.uk/download/pdf/43097244.pdf.

b)

c)

https://arxiv.org/pdf/1806.00735

NOMA is useful when two (or more) users have similar channel directivity, so that they can be covered by the same beam, but different pathlosses. One of the key properties of massive MIMO is called "favorable propagation" and means that the chance of having two users with similar channel directivity reduces as you add more antennas. If you go up in frequency (and keep the aperture fixed), then the beams will be even more directive.

In other words, more antennas and higher frequency are two properties that reduce the NOMA gain until the point where it disappears. At terahertz frequencies, I doubt that NOMA will be useful.

This is not a bad thing, it is a good thing! It means that using massive MIMO is sufficient - no need for an additional NOMA layer.

The "only" reason to go up in frequency is to have access to more bandwidth. More bandwidth also requires higher sampling rate and higher frequencies require transmitter and receivers that are capable of modulating signals of that kind.

Visible light is one example of terahertz signals. Just as visible light is blocked walls and other objects, terahertz applications will be limited in the same way.

The decomposition temperatures for different precursors used in MOCVD vary but are all so high that you cannot grow at extremely low temperature, so no LT-GaAs.

hey nimisha...you need to have two modules for that... RF module or wave optics module and semiconductor module... I have designed PCA using COMSOL...if you need any help then drop me a mail at Abhishek.Gupta@eli-alps.hu

Ok, Thanks.

Basically, there is 'working' range of THz TDS apparatus given by its construction and can be reached for very low absorbing samples. It is defined by the signal-to-noise ratio > 1 in the spectrum of the collected time-domain signal.

In praxis, due to absorption in the tested object, the useful 'working' range

of the apparatus is reduced, typically on the high-frequency side of the THz spectrum.

I recommend you to read this paper

A negative extinction coefficient essentially means that you have amplification or gain rather than absorption. If gain is unlikely with your sample then you must have some kind of error either in the measurement or calculation.

I hope this helps.

There is a short and rude answer to this question. Zero.

The conductivity is the product of mobility and the concentration of charge carrier at the Fermi level. Mobility is known 15000 cm2⋅V^−1⋅s^−1 (from Wikipedia). There are no charge carriers at the Fermi level in a separate sheet of graphene. It has to be doped to shift the Fermi level. Everything depends on doping.

Conductivity of typical high quality epitaxial graphene is 

about 6x10^6 S/m (APL 96, 082101, 2010) but it would

depend on technology of its deposition. It is nonmagetic material 

so permeability can be assumed as unity. Permittivity up to THz

frequencies is not measurable because its imaginary part related

to conductivity is much larger than the real part.

hope...........is still on the ground

You can use amber.

Hi,

The peak difference you are talking will only give you the Rough estimate to begin your calculation (Time delay between peaks will give you N and Height difference will give you K). The references given above by Márta Unferdorben will show you the precise method to calculate them. I hope this helps. 

Happy New Year :-)

Your question was how can one improve the resolution. The limit if of course diffraction. However, there are some practical ways to improve resolution. First, I'd recommend using focusing mirrors instead of lenses whenever possible. Off-axis paraboloids are conic sections and show no spherical aberrations when focusing an incident plane wave. Use two of them for a confocal imaging system. Even better, if you can measure amplitude and phase, then you don't even need a mirror, but the knowledge of phase and amplitude in the diffraction pattern will allow you to reconstruct the object, much like it is done in holography. The problem here may be signal to noise for time domain systems. I also suggest you set up your system with a visible light source. If it looks good for visible light, it will look good for THz.   

The fundamental problem with many THz imaging systems is that we use fairly small optics to get the job done, often with unsatisfactory results. Remember, THz wavelengths are 1000 times longer than visible wavelenghts, but we don't have 1000 times larger optics of course. Still, go big, like 5 inch mirrors, or even larger. This can become costly if you make such a mirror at visible light quality, but you don't have to. You can save a lot of money by relaxing the tolerances to THz wavelength quality. 

Lastly, coherence can be a problem in the form of stray light, multi-path, etc. Make sure you take that into account when setting up your imaging system. 

If you consider the physical meaning of the phase matching term within the generation process,

I_THz ~ L²Sinc^₂(dK L/2)

for a particular THz and optical frequency we can say that delta k will be constant but non-zero, now by increasing the length of the crystal (L) the generation efficiency grows. It continues to grow until 

L = 2pi/dK

at this point the any newly generated terahertz radiation is pi out phase with any terahertz radiation generated L before it and so destructive interference means they cancel out and the efficiency decreases. 

From the same equation 

I_THz ~ L²Sinc²(dK L/2)

It is also possible to see now that by allowing dK to vary with THz frequency it is possible to have a frequency dependent efficiency. For instance it can occur that if the crystal is a certain thickness, the condition 

L = 2pi/dK

is met for low frequencies but not met for frequencies in a higher range. In its simplest form each frequency component has its own optimal crystal thickness (given by  L = 2pi/dK) and so for a given thickness each frequency component will have a slightly different efficiency.

I hope this helps.  

Hell

I've worked in the field of THz antenna simulation. I work with software COMSOL.  but I not could get an acceptable answer,With this software. Because this software solution is  FEM not FDTD. It is better to use you of the CSTsoftware.

Thanks for your kind information, Mahmut!

Just the Lorentz formula, 

ε = cte + summation_i [ A_i/(  ω²_0,i - ω² - j Γ_iω) ]

where ε is the complex dielectric constant (can be converted to absorption and refractive index). The fitting values are cte, A_i, ω²_0,i Γ_i. Where A is the strength of the i th resonance at frequency ω₀ with a damping of Γi (determines the broadness of the resance).

Database, or academic articles. Just go searching with key words 'Terahertz database', or 'Terahertz semiconductor', etc.

graphene

Dear Dmitriy,

The attached review describes the equations for calculating a frequency dependent dielectric matrix in terahertz regime.

PHYSICAL REVIEW E 75, 036614 2007

Terahertz plasmonic composites

Syrus C. Nemat-Nasser,1,* Alireza V. Amirkhizi,1 Willie J. Padilla,2,† Dimitri N. Basov,2 Sia Nemat-Nasser,1,‡

Derek Bruzewicz,3 and George Whitesides3

1

Center of Excellence for Advanced Materials, Department of Mechanical and Aerospace Engineering, University of California,

San Diego, Mail code: 0416, 9500 Gilman Drive, La Jolla, California 92093-0416, USA 2

Department of Physics, University of California, San Diego, Mail code: 0416, 9500 Gilman Drive, La Jolla,

California 92093-0416, USA 3

Department of Chemistry and Chemical Biology, Harvard University, Cambridge, Massachusetts 02138, USA

Received 27 October 2006; revised manuscript received 20 December 2006; published 27 March 2007

The dielectric response of a polymer matrix composite can be substantially modified and tuned within a broad frequency band by integrating within the material an artificial plasmon medium composed of periodically distributed, very thin, electrically conducting wires. In the microwave regime, such plasmon/polymer

composites have been studied analytically, computationally, and experimentally. This work reports the design, fabrication, and characterization of similar composites for operation at terahertz frequencies. Such composites require significant reduction in the thickness and spacing of the wires. We used numerical modeling to design

artificial effective plasmonic media with turn-on frequencies in the terahertz range. Prototype samples were produced by lithographically embedding very thin gold strips into a PDMS polydimethylsiloxane matrix.

These samples were characterized with a Fourier-transform infrared interferometer using the frequencydependent transmission and Kramers-Kronig relations to determine the electromagnetic properties. We report the characterization results for a sample, demonstrating excellent agreement between theory, computer design,

and experiment. To our knowledge this is the first demonstration of the possibility of creating composites with tuned dielectric response at terahertz frequencies.

Hoping this will be helpful,

Rafik

Its available in TIFR Mumbai

Assalam......

In quasistatic measurements, capacitance is measured directly by integrating charging current. However, CV measurements can also be made by superimposing a small sinusoidally oscillating (AC) signal on the voltage sweep and measuring the corresponding impedance directly as a function of bias voltage. (This requires the use of a high precision impedance meter.) In this case, capacitance measured under conditions of accumulation and depletion can be expected to be the same as observed in quasistatic measurements, i.e., conditions still remain near equilibrium. However, if the frequency of the AC signal is sufficiently high, capacitance measured under a condition of inversion is not the same as in the quasistatic case. The explanation for this is quite simple and is a direct consequence of non-equilibrium behavior of the inversion layer. Physically, any inversion layer must be formed from minority carriers generated in the depletion region and swept to the surface by the electric field. (Of course, minority carriers may be also generated in the bulk and diffuse into the depletion region.) Equilibrium conditions imply that there is sufficient time (by definition) for the inversion layer carrier concentration to respond to any changes in applied field. However, if the material quality of the silicon is good, carrier generation-recombination processes occur very slowly, with

a time constant on the order of milliseconds. Therefore, for an applied AC voltage in the megahertz range, the response of the inversion layer is simply too slow to “follow” the signal and similar to ionized dopant impurity atoms, the inversion layer appears fixed with respect to the AC component of the bias. (Of course, the inversion layer does respond to the primary voltage sweep.) Therefore, for high frequency conditions, the capacitance per unit area measured in inversion is the series combination of oxide capacitance per unit area and capacitance per  unit area of the depletion region. Furthermore, since, the depletion width reaches a

maximum value, the combined capacitance per unit area saturates at Cmin.

for more..

The physical mechanism is that filament induced birefringence in gases provides a phase delay between the two orthogonal components of the THz field, leading to an elliptically polarized THz emission. By sending a sequence of two femtosecond IR laser pulses at 800 nm separated by less than 3 ns, forming two overlapping filaments in air (so called bifilamentation), they can generate 1 order of magnitude more intense THz emission than the transient-Cherenkov THz emission from the plasma filament. The origin is attributed to the emission from a bimodal transmission line created by a pair of neighboring plasma filaments. (http://www.col.org.cn/abstract.aspx?id=COL201311011401-06)

Dear Kamiar,

Both AuGe based alloys and Ti/Au metal layer stacks are widely used as ohmic metal contacts for photoconductive THz antennas made of low temperature grown GaAs. The power emitted from the antennas with AuGe metallization is 50% higher than that of antennas with a Ti/Au metal layer. From a comparison with a photomixer model it can be concluded that the higher output power results from a lower contact resistance of the AuGe contacts leading to an increased current flow. However, Ti/Au contacts have a higher thermal stability which might be advantageous if high system stability is called for.

More detailed information is contained in:

Regards, Leonid A. Skvortsov

The top layer of metal is to shield alternate electrode spacings.  If you look at figure (b) the authors have noted the E field direction in each inter-electrode spacing - the field alternates between a "positive" and "negative" direction.  As explained in the earlier papers on this technique (links below) if both positive and negative directions were allowed to radiate terahertz radiation then the generated pulses would cancel each other in the far field.  The overcoating of gold prevents the excitation by the laser pulse thereby only generating terahertz pulses of the same polarity, producing constructive interference in the far field.  Fig. 2a in the APL link below shows the effect of not having the top layer of gold to prevent the destructive interference.

The SiO2 layer is used to insulate the second layer from the first.

Dear Johneph,

in terahertz area of frequencies index of refraction of substance n (ω) and its coefficient of absorption α (ω) calculate by means of measured amplitudes and phases of the radiation which passed through a sample by means of expressions [1]:

                     n(w) = 1 +[c/wd] Im[lnE_tr(w)/E₀(w)]   ,                                     

where E₀ (ω) and E_tr (ω) – intensity of electric field of radiation before passing of radiation through a sample respectively; d – sample thickness; ω – radiation frequency; c – velocity of light.

Reflection coefficient R (ω) and index of refraction n (ω) are connected with each other by an approximate ratio (near normal falling):

R(w) ~ [(n(w)-1)/n(w)+1]²

[1]. Kemp M. Proc. of SPIE, 6402, 64020D (2006).

See also:

“Determination of complex refractive index of thin metal films from terahertz time-domain spectroscopy”, Началоформы

Da-Xiang Zhou, E. P. J. Parrott, Douglas J. Paul and J. Axel  ZeitlerJ. Appl. Phys. 104, 053110 (2008).

2.      “Measurement of Terahertz Refractive Index of Metal with Terahertz Time-Domain Spectroscopy”, Hiroaki Yasuda and Iwao Hosako  Jpn. J. Appl. Phys. 47, 1632 (2008). 

3.      Yun-Shik Lee. “Principles of Terahertz Science and Technology”, 2009.

https://books.google.ru/books?id=4ZqkuUURxFcC&pg=PA161&lpg=PA161&dq=refractive+index+of+the+metal+in+the+terahertz+frequency+range&source=bl&ots=tbnvcAPsAw&sig=_ezofjaejcch- RIc4PJ2WLFqZyA&hl=ru&sa=X&ei=luCfVNuJDITXyQOIj4LoCw&ved=0CB0Q6AEwADgK#v=onepage&q=refractive%20index%20of%20the%20metal%20in%20the%20terahertz%20frequency%20range&f=false

4. THz Spectral Database - the NIST WebBook http://webbook.nist.gov/chemistry/thz-ir/

I hope that this information will be useful to you.

 Best regards, Leonid

I recommend you to use as a reference M.A. Ordal et al., Optical Properties of metals Al, Co, Cu, Au, Fe, Pb, Ni, Pd, Pt, Ag, Ti and W in the infrared and far-infrared, Applied Optics 22, 1099, (1983). There you can find all the necessary values to fit your dielectric function (of several metals) to experimental data.

What about the natural sources  in THz range ?

See Thz Photonics book by X.-C.Zhang.  Also some of his review articles discuss these issues.

It also depends on what you are looking at?  Highest power/energy of THz radiation?  Wide bandwidth?? Low cost??

Hi Anselm.  

Thanks for your clarification. We have been actually using two more parabolic mirrors to direct the THz beam to the sample and guide the reflected beam to the parabolic mirror infront of the photodetector. So I guess, we are not being able to properly guide and collect the waves from the sample surface. 

Yes, 50 % loss is there, we can get more or less the same bandwidth but the SNR is obviously low. 

Guess, I need to put in more effort. Thanks once again.

Just FYI, you can obtain much higher DR with a CW system based on harmonic multiplication of stabilized microwave sources, up to about 1.4 THz. The ABmillimetre VNA produces about 10 microwatts at 1 THz. Not much help if you've already invested in a time-domain system, I understand.

Also see:

and:

Optical Materials Express, Vol. 3, Issue 2, pp. 260-269 (2013)