The existence of sharp reflections is interpreted by the theory of an unlimited and perfect crystal (not really correct since then the intensity shouldn't be described by the kinematic theory, but let us assume this). If you reduce a crystal until the diffracting lattice become smaller and smaller you are observing the same as in school using a light and a double slit and compare it with a multiple slit system. The first will generate a strong but broad peak in the center with very weak higher order diffraction phenomena and the second will generate very sharp  and multiple orders. Crystals have a) a lattice which forms perfect reflections, and b) bondings which also display some kind of first order periodicity. Therefore, diffraction experiments are even performed at liquids and glass in order to identify bonding lengths. The problem is only that you don't see these effects as strong peak bit as modulation which need a Fourier analysis to visualize them. Fourier analysis is practically also used for the structure factor equation of crystals. Your broad peak at 25 degrees is actually nothing else as the statistical effect of your SiO4 tetrahedra which are quite perfect in first order but do not show a higher ordering as in crystals. If you would start to increase the order you would see that this peak would become sharper and sharper and in dependence of the specific structure (quartz, cristobalite of tridymite) a very intense peak would appear there. Not exactly at the same position since the SiO4 thetrahedra also varies a bit, and the lattice periodicity as additional parameter becomes more important, since the structure factor is the projection of the atomic arrangement onto the lattice plane normal hkl. Anyway, I am surprized since silica is as far as I know SiO2 and when the particles are very small I would actually expect a broad peak. Maybe your silica is really very small in grain size?

Doesn't have the amorphous SiO2 a broad peak around 23-27°? I guess it could be just normal peak of your holder? Check the article in the link, if it answers you questions :-)

http://www.sciencedirect.com/science/article/pii/002230939500095X

Janez is correct that amorphous SiO2 gives a very broad signal around 25° 2theta. Since the material is amorphous it has no crystal structure and therefore it has not no planes with specific Miller indices. For diffraction to occur you need a regular crystal structure with planes where refraction can take place. Therefore, amorphous material will not give you any sharp reflections but only a very broad background signal. If you do get a sharp reflection it is definitely not from the amorphous silica.

The existence of sharp reflections is interpreted by the theory of an unlimited and perfect crystal (not really correct since then the intensity shouldn't be described by the kinematic theory, but let us assume this). If you reduce a crystal until the diffracting lattice become smaller and smaller you are observing the same as in school using a light and a double slit and compare it with a multiple slit system. The first will generate a strong but broad peak in the center with very weak higher order diffraction phenomena and the second will generate very sharp  and multiple orders. Crystals have a) a lattice which forms perfect reflections, and b) bondings which also display some kind of first order periodicity. Therefore, diffraction experiments are even performed at liquids and glass in order to identify bonding lengths. The problem is only that you don't see these effects as strong peak bit as modulation which need a Fourier analysis to visualize them. Fourier analysis is practically also used for the structure factor equation of crystals. Your broad peak at 25 degrees is actually nothing else as the statistical effect of your SiO4 tetrahedra which are quite perfect in first order but do not show a higher ordering as in crystals. If you would start to increase the order you would see that this peak would become sharper and sharper and in dependence of the specific structure (quartz, cristobalite of tridymite) a very intense peak would appear there. Not exactly at the same position since the SiO4 thetrahedra also varies a bit, and the lattice periodicity as additional parameter becomes more important, since the structure factor is the projection of the atomic arrangement onto the lattice plane normal hkl. Anyway, I am surprized since silica is as far as I know SiO2 and when the particles are very small I would actually expect a broad peak. Maybe your silica is really very small in grain size?

It was explain in great detail by Gert, below is the simulation from SiO2 (Crystobalite and Quartz) with crystallite size of 1.5 nm (instument peak width of 0.1). Which one is close to ur pattern ?

Roughly, this peak corresponds to the Si-O bond length. As it is not crystalline, there is no crystal plan.

usually, this peak is broad and weak. If this bothers you, another technique would be to use monocrystalline silicium and remove the known background after measurement.

Hi Christophe, for what do you want use single-crystalline Si? As monochromator or sample holder?