I don't know about a fiber readily available. But you may want to contact Schott AG (www.schott.com). They are IMHO the most knowledgeable company regarding optical glas properties.
Waveguide dispersion in single mode fibres depends on the core-cladding index difference, the refractive index profile, and the operating wavelength or normalised frequency. For typical straightforward refractive index profiles, waveguide dispersion increases as wavelength increases beyond the cut-off wavelength (decreasing as normalised frequency V increases).
Material dispersion is likely to be in the range +10 to +20 ps/nm/km at 1670 nm, depending on the composition. Chromatic dispersion of silica is reduced by germania doping in the core.
Applying a Gaussian approximation for field and index profile (Snyder & Love, Optical Waveguide Theory, 1983, chapter 15):
Waveguide dispersion (group delay derivative with respect to wavelength) is ~ -2 n_core / (vely_light x wavelength) x Delta / V^2
So for V=2 we require relative index difference Delta around 0.01, (NA = 0.2), with core diameter around 5.2 micron.
A commercially available fibre which may come close is Corning's RC 1550 fibre. https://www.corning.com/media/worldwide/csm/documents/RC%201300%20and%20RC%201550%20Specialty%20Fiber.pdf
This is a small core high index design with a reduced (80 micron) cladding diameter, intended for 1550 nm operation with tight bend radii. Loss and dispersion at 1650 - 1700 nm is not specified on the data sheet, but attenuation should be OK with more usual bend radii.
Lots if assumptions and approximations here, but worth investigating whether this fibre or something similar is suitable.
Zero dispersion wavelength fo silica fiber is 1.3 microns. However, since lowest loss is at 1.55 um, telecom people use, for long-distance communications, single-frequency lasers, whose light is not affected by chromatic dispersion.
1550 nm is favoured for long distance telecommunications, but not because laser light is unaffected by chromatic dispersion. To carry information at a useful rate, the laser must be modulated. This broadens the spectrum, and the modulated signal is affected by chromatic dispersion.
For a 10 Gbit/s intensity modulated, non-return to zero signal, there is a significant penalty from around 1000 ps/nm chromatic dispersion, corresponding to roughly 60 km of standard single mode fibre (dispersion zero 1300 nm). The penalty varies quadratically with modulation rate, so propagation over 1000 km is possible at 2.5 Gbaud without dispersion compensation or regeneration of the signal. At higher modulation rates dispersion compensation becomes essential.
Dispersion shifted fibre (DSF) is available with much lower chromatic dispersion and only slightly higher attenuation at 1550 nm. Although it was used for some early fibre deployments, it is now much less widely used. Modern high capacity transmission systems employ multiple high bit rate signals, each propagating at a different laser wavelength (wavelength division multiplexing). In such WDM systems Kerr effect non-linearities such as four wave mixing and cross-phase modulation introduce unacceptable crosstalk penalties in multi-span systems when channel spacings are narrow and fibre chromatic dispersion is close to zero.
Near-zero (or non-zero) dispersion-shifted fibres were developed in the 1990's, offering a compromise between WDM capability and single-channel dispersion compensation requirements. More recent WDM systems can operate at closer channel spacings (50 or 100 GHz), and favour higher chromatic dispersion.
The introduction of coherent modulation schemes brought with it a capability for electronic dispersion compensation at the receiver. The need (and cost) of dispersion compensating fibre modules is eliminated, and silica-based fibres with low attenuation and relatively high chromatic dispersion offers the best performance for new-build systems. Standard single mode fibre with dispersion zero around 1300 nm meets this requirement cost-effectively. Pure silica core fibres with slightly lower attenuation and higher dispersion are manufactured for specialist applications such as long distance undersea cables.
Another reason for the use of 1530 - 1560 nm wavelengths for long-distance communications is the availability of high performance erbium-doped fibre amplifiers, able to compensate the attenuation of multiple WDM signals in a single cost-effective module.
In this case the system architects sometimes employ a balanced structure where
sections of dispersive fibre with different dispersion characteristics are joined to
form the span. The idea here is that no section of fibre has zero dispersion but
that different sections have dispersion of opposite sign so that the total at the end
of the link (span) is zero.
Of course, if we are planning to operate in the 1650 nm band we could install
Dispersion Shifted Fibre (DSF). This has a dispersion of zero at 16500 nm.
However, WDM systems have a severe problem if the fibre dispersion is really zero! This problem is called 4-wave mixing. “ We could use fibre with a dispersion of 4 ps/nm/km to mitigate FWM but in very long amplified links (such as many undersea cables) even this minimal level of dispersion is a limitation.
Arkady is interested in dispersion at 1650-1700 nm (176 to 182 THz), rather than at lower THz frequencies. This is rather close to the edge of the TOPAS COC transmission window shown on page 6 here: http://www.topas.com/sites/default/files/files/optical_e.pdf
Do you have dispersion and attenuation data measured over transmission paths longer than 2 mm?
Arkadiy is not asking for zero dispersion at all wavelengths. He is asking for a fibre in which the sign of dispersion coefficient changes from negative to positive (normal to anomalous dispersion) at one or more wavelengths between 1650 nm and 1700 nm.
As Vincent Lecoeuche has already explained, there is a change of group velocity with wavelength arising from transverse confinement by the waveguide. Together with the material dispersion of the medium, this "waveguide dispersion" determines the net chromatic dispersion of a fibre mode.
Single mode fibres with tailored chromatic dispersion have been in routine commercial production since the 1980's. "Zero dispersion wavelength" is often specified in manufacturer's product data sheets.
The wavelength range 1650-1700 is not commonly specified, but there are specialist fibres which may have dispersion properties close to Arkadiy's requirements. It would certainly be possible to design and manufacture a fibre with zero dispersion wavelength in this range.
Agreed with Boualem. As I said earlier, zero dispersion is only observed in empty space, or could be engineered through the use of metamaterials (in a narrow frequency regime though!)
I am not denying the validity of the Kramers Kronig relations, or claiming that optical waveguides can be made dispersion-free at all wavelengths.
What has been demonstrated without doubt is that the material dispersion of silica changes sign between 1250 nm and 1300 nm, and that with appropriate fibre design the zero dispersion wavelength of a fibre waveguide can be shifted to much longer wavelengths.
This was established in the early days of fibre optic development, with many theoretical papers from 1966 to the 1970's and wide spread commercial fibre production from the 1980's. A W Snyder and J D Love "Optical Waveguides Theory" (1983) present the theory in some detail, but there have been many other papers and texts published.
You speculated in an earlier post that dispersion could be manipulated using metamaterials. Why not accept that this can be achieved more easily by changing the waveguide properties?
I can't vouch for the accuracy of most of the content, but Google found these two articles which present background which you may find useful.
https://spie.org/Documents/Publications/00%20STEP%20Module%2007.pdf - see pages 269 and 278 figure 7-16.
I found the replies by Alan most useful and relevant for the above question. I am experienced with design and manufacturing of single mode fibers of different categories like G652D, G655, G657 etc. But as Alan said the required zero dispersion wavelength range is not found with conventional telecom fibers. So, a special fiber need to be designed and manufactured for this purpose. I have no idea If somebody is already supplying fibers with above zero dispersion wavelength range.
Zero dispersion can be obtained in communication fibers that are used in present optical networks. In conventional single-mode fibers, the zero-dispersion wavelength (ZDW) is at the second optical window, i.e. 1300 nm. In dispersion shifted fibers, the SDW is at 1550 nm.
Currently, in optical communication systems, use of fibers with zero dispersion is avoided because of nonlinearity effects at high bit rates.
The fiber with ZDW in the range of 1650-1700 is not yet available on the market. But it may be manufactured by order.