As I said in my previous posting, photons are somewhat special, because they are massless vector bosons. They have spin 1 but only 2 polarization degrees of freedom (and not 3 as massive vector would have, because they have 3 eigenvalues of the spin-##z## component, 1,0, and -1). This specialty is closely related to the fact that in relativistic QFT the massless vector bosons must be described as gauge bosons. The natural way to describe the 2 "spin components" of the photon is its helicity, which is the projection of the total angular momentum on the direction of its momentum, and this observable called helicity can take the two values 1 or -1. This corresponds to single-particle Fock states of left- and right-circularly polarized electromagnetic waves. You can of course have any superposition of these polarization states, and among these superpositions are also all linearly polarized photon states in any direction.
Last but not least, you should forget about "wave-particle duality". This is an old-fashioned concept from the time before 1925, where the modern quantum theory has been discovered by Born, Jordan, and Heisenberg (including field quantization, which was particularly worked out by Jordan in one of the famous papers written by Born, Jordan, and Heisenberg). Shortly thereafter the same theory has also been discovered by Schrödinger in the form of wave mechanics (for non-relativistic particles) and by Dirac in its representation free form. The upshot is: The only really adequate description of photons is the quantized electromagnetic field and applying the statistical rules developed by Born as for any other "quantum system". A single photon has observable properties of both particles and waves, which was described as an engimatic "wave-particle duality" by Einstein before the discovery of the modern relativistic QFT.
Whether you observe "particle properties" or "wave properties" depends on what you decide to measure. If you have a single photon, e.g., emitted by an excited atom, and you put a detector like a photo plate or a CCD cam) around this atom, you'll register with some probability a single point somewhere on the photo plate with probabilities given by the specific properties of this photon, described by its quantum state. The same holds true for the polarization. You can measure whether a photon is in a certain linear polarization state by putting a polarization filter between the photon source and the photon detector. If the photon goes through the polarizer (with a probability given by the specific property of the photon encoded in its quantum state) you know it was polarized in the direction determined by the orientation of the polarizer.
You can also observe "wave properties" of single photons. The most famous age-old experiment to demonstrate that light is some kind of wave is Young's double-slit experiment, where you get interference effects leading to the well-known diffraction fringes. If you do such an experiment with single photons, each single photon still leaves only one spot on the detection screen, but using very many photons (all prepared in the same quantum state) you'll get the diffraction pattern as predicted for waves. In the wave picture these patterns are caused by the superposition of partial waves going through either one or the other slits (aka "Huygens's principle" used to explane these diffraction patterns). On the other hand when detecting the single photon with a screen, it always makes only a single spot not the extended diffraction pattern. So you have in some sense a particle property, when detecting the single photon but on the other hand also wave-like behavior in the explanation how the photon goes somehow through the two slits. The consistent picture of modern QFT resolving this contradictory "wave-particle duality" of the old point of view is that observables of a quantum object do not necessarily have well-determined values when prepared in some quantum state, but the quantum state only implies the probabilities for finding one of the possible values of an observable when this specific observable is measured.
