Cthugha said:
Direct generation of photons in an entangled basis of left and right circularly polarized photons has been demonstrated for example in Nature 465, 594–597 (2010) by Salter et al. in terms of a cascaded decay. If I remember correctly the references inside also give hints at how to perform Bell measurements on such states.
The abstract from that reference:
"An optical quantum computer, powerful enough to solve problems so far intractable using conventional digital logic, requires a large number of entangled photons1, 2. At present, entangled-light sources are optically driven with lasers3, 4, 5, 6, 7, which are impractical for quantum computing owing to the bulk and complexity of the optics required for large-scale applications. Parametric down-conversion is the most widely used source of entangled light, and has been used to implement non-destructive quantum logic gates8, 9. However, these sources are Poissonian4, 5 and probabilistically emit zero or multiple entangled photon pairs in most cycles, fundamentally limiting the success probability of quantum computational operations. These complications can be overcome by using an electrically driven on-demand source of entangled photon pairs10, but so far such a source has not been produced. Here we report the realization of an electrically driven source of entangled photon pairs, consisting of a quantum dot embedded in a semiconductor light-emitting diode (LED) structure. We show that the device emits entangled photon pairs under d.c. and a.c. injection, the latter achieving an entanglement fidelity of up to 0.82. Entangled light with such high fidelity is sufficient for application in quantum relays11, in core components of quantum computing such as teleportation12, 13, 14, and in entanglement swapping15, 16. The a.c. operation of the entangled-light-emitting diode (ELED) indicates its potential function as an on-demand source without the need for a complicated laser driving system; consequently, the ELED is at present the best source on which to base future scalable quantum information applications"
And a related article:
http://arxiv.org/abs/1103.2969
"A practical source of high fidelity entangled photons is desirable for quantum information applications and exploring quantum physics. Semiconductor quantum dots have recently been shown to conveniently emit entangled light when driven electrically, however the fidelity was not optimal. Here we show that the fidelity is not limited by decoherence, but by coherent interaction with nuclei. Furthermore we predict that on 100\mu s timescales, strongly enhanced fidelities could be achieved. This insight could allow tailoring of quantum logic to operate using quantum dots in the fault tolerant regime."
Of course, even with these articles it appears that the circular polarization vs linear is not really an important distinction (it's not mentioned). From the editor's summary of the 2010 article:
"For optical quantum computation and related information technologies to fulfil their promise, they will require a source of entangled photons that can be delivered efficiently on demand. Existing entangled-light sources are laser driven, and involve bulky and complicated optics. Salter et al. have now developed a compact light-emitting diode with an embedded quantum dot that can be driven electrically to generate entangled photon pairs. Much simpler than its laser-driven counterparts, this ELED (entangled-light-emitting diode) device, based on conventional semiconductor materials, is a promising start point for the development of an entangled light source for quantum information applications."
On demand entanglement! Cool. I presume that one would simply use a wave plate or similar if you specifically needed to have linear polarization for an application.