ITLA Emcore TTX1994

Construction analysis

Draft n.1

Part 3

  • The two Etalons in Vernier configuration
  • Back cavity mirror

Note:

the current draft is intended to report on the current status of analysis, and is by no way complete.

It requires feedback to address specific further investigation prior to any dismounting.

Because of the many single elements and the overall complexity of the assembly, any compositional analysis must be delayed to the after-dismounting phase.

Also formatting is not definitive, and omits some references to existing and available documents.

The two Etalons in Vernier configuration.

The most peculiar and intriguing part of the device is the tunable external cavity, where manufacturer’s proprietary solutions are employed.

The general consideration for tuning is that all possible optical frequencies belong to the spontaneous emission spectrum of the active region of the gain chip.

Roughly, the spontaneous spectrum can be described as a rather wide peak, centred at the frequency  for which , (Eg is the bandgap of the active layer in the gain chip) and a linewidth of

It amounts to some 0.01.

Among all possible frequencies in that range, only the resonating ones will be enabled to achieve stimulated emission.

The simple resonance condition is ( that is: the length L of the optical cavity must be equal to an integer number of half a wavelength). The point is that changes when light crosses materials with different refractive indexes.

It is then possible to build up a varying effective cavity length L by inserting elements for which their refractive index can be adjusted by external controls. Temperature is one of the possible controlling parameters in dielectrics, as well as charge injection is for semiconducting materials.

It is a simple matter of calculation to see that for wavelengths in the micrometer range and cavities in the range of some millimetres, m easily achieves the value of 1000 and more. The peak separation due to resonance, that is in the order of then amounts to some 0.001 or less.

This means that tens of peaks (modes) can be simultaneously activated for a given active material and a given cavity length.

This calls for further wavelength selection, by suppressing the unwanted peaks.

The solution proposed by INTEL allows for tunable wavelength selection by means of three elements:

  • two thermally controlled Etalon filters and
  • one thermally controlled (?)* high reflecting back mirror.

The two Etalon filters select a wavelength, and the back mirror adjust the phase (fine adjustment of the cavity length) to achieve exact resonance for the selected wavelength.

* the thermal control of the back mirror is here just inferred. No explicit data are available from INTEL documents.

The general appearance of the external cavity looks coherent with the INTEL documents, at least at a first insight.

The assembly displays the following features:

  • The two etalons are identical and mounted mirror-symmetric
  • They are separately soldered on a patterned sub-board, parallel to the bottom of the case
  • No mechanical element separates the two devices
  • They are carefully slightly rotated with respect to the optical axis
  • The electrical connections correspond to the layout reported on the INTEL documents
  • The back mirror looks as a passive element, mounted on an electrically connected substrate.

The rotation is expected: it inhibits parasitic resonances to setup between the etalons. It also introduces an increase of the effective thickness travelled by the transmitted wavefronts. This can contribute to make the two etalons slightly different, as required by the Vernier configuration. In any case, current remains the dominant control of the respective transmission functions.

Point to be clarified: the back mirror should allow phase control within . According to the INTEL block diagram, the mirror should be a Lithium Niobate element, coated on its front facet with a high reflection layer. In this case, only the mechanical displacement of the surface could produce the phase shift. Thermal expansion could be invoked, but it should be carefully evaluated to check if the required temperature variation is practically achievable.

On the other hand, piezoelectric control of the mirror looks not allowed, because of the lack of any electric contact.

This point, if required, will be clarified when each element will be dismounted.

The most surprising result, to be coupled with the absence of any mechanical element between the two etalons, comes from the perspective view of the Etalons at the SEM. Their structure comes out to be different from that reported in the INTEL papers.

Let us first recall the INTEL official structure.

All previous elements correspond to the observed features, but the Etalon pair looks:

  • Hosted on a dedicated submount-spacer
  • Soldered with the chip connections aligned along a vertical line.

In addition, the structure of the Etalon is declared to be made of:

A disk of Si suspended in the middle of a thin Si3N4 membrane

  • Two concentric circular electric paths, internal to the suspended Si disk
  • The internal path is wider (heating element)
  • The external path is thinner (temperature sensing element).

The optical observation was able to show

  • the absence of the spacer
  • the 90° rotation of the solder line

The SEM analysis shows the most important differences.

Here:

  • The Etalon is made of a Si thick slice, where a passing hole has been etched
  • The optical layer is made of a separate Si slice, glued onto the previous one
  • The patterned metal paths are larger than in the official drawing, and run on the support slice

The latter means that the whole element is heated, which makes comprehensible the removal of the metal spacer, that would have introduced a thermal path between the two etalons.

The reasons of such an important change are not clear.

It is obvious that the solution to heat the whole etalon is not an improvement by itself: the INTEL paper explicitly celebrates the thermal insulating role of the thin Si3N4 membrane, that would allow fast setting of the required temperatures for wavelength tuning.

It may be not a case that in table 4.1 of the Datasheet the power and frequency locking time ranges from 25 (warm start) to 60 sec (cold start).

At the moment, it can only be argued that renouncing to the suspended membrane and disk did cause some kind of problems (mechanical during shock tests?)

Finally, the observed thickness of the optically active Si layer (some 150m), is consistent with the spectrum reported in Intel Technology Journal, Volume 8, Issue 2, 2004, fig.14.

A separate document will be issued on details of the external cavity and its relevant parameters and controls.

Back cavity mirror

The back cavity mirror, that Datasheet claims to be in Lithium Niobate, is about 500 m thick, and should be made of Lithium Niobate, according to the Datasheet.

The first finding is that it is a block of Germanium.

It then results coated with a multilayered structure that, at a first EDX inspection, seems compatible to alternating SiO2 and Ta2O5.

The latter is known (see f.i. ) to be “a high refractive index, low absorption material useful for coatings in the near-UV to IR spectra regions”.

It looks like a bulk passive element, mounted on a Thermo Electric Cooler (TEC).

The separate document dealing with the Etalon theory will also consider the suitability of such a solution to act as a final phase element of the external cavity.