Hi Maxim,

It looks your waveguides are open-ended. Thus the most of the power is reflected due to the mismatch between the waveguide and air. You can add grating couplers to the input and output for the waveguide excitation and light detection. I have attached two papers on SOI grating couplers. I hope they help. 

Saman

Hello Maxim,

the typical way of doing it is trough grating couplers (vertical coupling), nevertheless I see that the chip is already fabricated and does not include couplers. You should also be able to do it via facet coupling, although the coupling losses would be much higher because of the sub-micron size waveguide. For this, I suggest you to cleave your sample in such a way that both input and output waveguides are at the chip facet. Then you have two alternatives, either to (1) use a lensed fiber close to the facet to couple light into your waveguide, or (2) use free-space coupling, i.e. collimate your light and then focus it into the waveguide(s) with a microscope objective. You might have troubles to split the optical signals (because the wgs seem close) and the coupling loss will be high, but I see no other way of doing it without doing further chip processing to fabricate dedicated coupling structures. I hope it helps!

Victor

Dear Saman and Victor,

Thank you for your helpful answers.

We have attempted to perform free-space coupling and as I wrote, we could not receive any signal from our chip in spite of the fact that we carried out the similar experiment with the chip fabricated by another group. The layout of our experiment is in the attached files. Thank you in advance.

Maxim

Grating are a great solution to couple 1.55 µm light in Si, with a good tolerance to mis-alignement of your optical beam (coming from an optical fiber). It si quite easy to put an input and output grating to get the response of the resonateur, in different optical fibers.

Hello Fabien,

Thank you for your answer. I believe that our next approach will be associated with fabrication of the grating. We will try it.

Best regards

Maxim

Base on your SEM picture, it seems that the waveguide is not cut to the edge, and this will result in a difficulty in the efficient butt coupling.  In addition, if  NA of your MO is not big enough to produce a beam size roughly of ~um, the coupling efficiency will be very low such that no light is coupled into the waveguide.  If the butt coupling is preferred, you can try to put mode conversion couplers in the input/output waveguides. Or just as what  other specialists said , put input/output gratings in the front/rear of the input/output waveguides.  Then your characterization setup should be modified. 

In addition to the answers above: silicon is not like III/V, and does not cleave well. You'd have to polish the facets. Furthermore, with input and output on the same side, it becomes virtually impossible to align well. Your setup is creative, but very annoying to operate.

Hi Maxim, first of all, on your SEM image can be see both input and output is in the same side. I would try to put in oposite sides in order to minimize set-up complexity. Another think is that can be see that the end of the wave guide is not in the edge of the sample, and is needed the end of the waveguide reaches the edge of the sample to obtain a efficient result using horizontal coupling. Despite all, I think vertical coupling using grating coupler will result in the best solution as Saman Jahani have said in his answer.

Alvaro

Dear Chyong,

Thank you for your answer.  Could you please describe what do NA and MO mean in your phrase: "if  NA of your MO is not big enough"?

Best regards Maxim  

NA is the Numerical Aperture of a Microscope Objective (MO). Alternatively you can use normal lenses, if your bandwidth is limited (avoid chromatic aberrations).

I see that your output WG is only ~150µm away from the input WG.  I've never seen anyone couple the input AND output beams through the same objective when the waveguides are a full 150µm apart.  Perhaps it works - I don't know - but as other people said I'm much more familiar with cleaving/polishing right through the access waveguides (so the ridges are flush with the edge of the chip), and also having the input/output on different sides of the chip (thus two different objectives or lensed/cleaved fibers), so each port can be optimized independently.  Your way certainly could be made to work, it just seems very difficult.

To stick with your current setup, I would at least check both the input & output sides independently,  by coupling visible light through the optics, to get the focal point of each beam path somewhat close to the waveguides.  As other people said, the spot size of your input and output beams should closely match your waveguide mode size.  So you'll need to make sure that the lenses on both the I/O sides, when sent through your coupling objective, result in a spot size that overlaps the WG mode well.  Then they both also have to be positioned right on the WG accurately. You may be able to determine this somewhat empirically by sending visible light (or better, IR light but viewing with a nice IR camera on your microscope) through the ports towards each WG and making sure the focal spot is right on the WG.  (The optimal focus for IR vs. visible will be different, but this may get you close.)  If you have an IR camera on your top-down scope, you might even see the waveguide input/output light up as you manage to get the beams coupled in from either end (need a high-sensitivity camera for this though, which is not cheap).  If you do go with the top-down IR camera viewing, it's helpful to angle your chip by a few degrees, so substrate propagation goes straight while WG coupled propagation is at an angle and you can then tell the difference.

Then once you know the optics are aligned by sending light *into* both waveguide ports, then you switch the input & output WG to instead go to your laser source & detector, respectively - hopefully without bumping any optics.  Well-secured Fiber ports can be helpful for this.

I think a much easier way is to polish the device again, so that the Input WG is on one cleaved facet, and the output WG is on a different cleaved facet (eg. the perpendicular facet, or remove the bends entirely and couple in/out from opposite sides of the chip), and use two objectives.  Again, couple light INTO both of the the waveguides first to align the optics, confirming either with top-down IR camera or getting close via visible light (put the focal spot on the waveguide end) and then switching to your IR source/detector.

Another difficulty: you have no way of knowing if the loss of your waveguide, or bends, or coupler, or ring, is simply so high that no light will make it to the output end of the waveguide (your design sounds like it would work, but you never know).  A way to ensure your setup works is to start with a short waveguide with no ring, and make sure you can couple to this first.  A short straight waveguide would be ideal, but is annoying if your ring device is a completely different configuration, requiring you to mangle the setup to test it (again a reason to convert your ring into in/out on opposite sides).

Lastly, another idea to try, if you have the equipment, is to use an IR camera and objective to actually Image (on a TV monitor) your waveguide facet, from the side.  Once it's in focus (illuminate with IR via std. light-pipe microscope illumination), you know that a beam coming Out of this port will be almost in-focus through the objective. Again, you then switch this beam to your detector/source via a flip-mirror or something.  This *could* be done with a visible camera, but again the difference in focus between visible & IR might complicate things a bit.

Good luck!

It would become problem when you do not have any taper waveguide in your structure because the mode profile diameter in the waveguide usually pretty small. I think you will be able to couple the light if you use taper or lens fibers that create the mode profile similar to the mode profile of your waveguide, at least few percents of the light will couple to your structure.

Good luck!

You have to cleave your device.

Its always good to use gratings in such cases. You can either use butt coupling or endfire coupling method but since you don't have tapered waveguide, there will be some issues.

Coupling TE or TM modes for single mode analysis will be another issues for you.