The grain definition is quite open and depend a bit from the field you are coming. Therefore you will find many answers and it is a very good idea to discuss this first before starting to "fight". As crystallographer I would say that a grain is a volume element having the same orientation. However, what means "the same"? If you discuss this question with XRD people they already include all what is coherently diffracting, i.e. small deviations are allowed described be a so-called mosaic structure. If you discuss this with people from using grain size and shape for materials science they already distinguish between low and high angle boundaries where the low angle is described by local orientation variantions in the scale of 5 degrees in order to describe high-deformed material in a useful way. High-angle boundaries are classified mostly higher than 10 degrees. Although a bit surprising about such a pragmatic defenition, it is a very technical but currently widely accepted condition. If people talk about single-crystalline tourbine blades it is even more funny since the "single grain" is neither chemically homogeneous (it contains millions of well aligned dendrites and interdendritic regions) nor a single phase material since it consists of a comparable fraction of an fcc-phase and an ordered L1_2 phase (Cu3Au).

The grain definition is quite open and depend a bit from the field you are coming. Therefore you will find many answers and it is a very good idea to discuss this first before starting to "fight". As crystallographer I would say that a grain is a volume element having the same orientation. However, what means "the same"? If you discuss this question with XRD people they already include all what is coherently diffracting, i.e. small deviations are allowed described be a so-called mosaic structure. If you discuss this with people from using grain size and shape for materials science they already distinguish between low and high angle boundaries where the low angle is described by local orientation variantions in the scale of 5 degrees in order to describe high-deformed material in a useful way. High-angle boundaries are classified mostly higher than 10 degrees. Although a bit surprising about such a pragmatic defenition, it is a very technical but currently widely accepted condition. If people talk about single-crystalline tourbine blades it is even more funny since the "single grain" is neither chemically homogeneous (it contains millions of well aligned dendrites and interdendritic regions) nor a single phase material since it consists of a comparable fraction of an fcc-phase and an ordered L1_2 phase (Cu3Au).

Thank you for the insight Dr. Gert Nolze. First thing, we don't 'fight/s', we always 'discuss' in science, peace! :). I came from physics background, doing material science now, mostly (rather exclusively) in thin films. My characterization includes both XRD and electron microscopy among others. This question has been always at the back of my mind, never seem to get a convincing answer. So it was natural for me to put forward this question when I joined RG.

Though a 'physicist' and a 'metallurgist' perspective seems right in their own interpretation, finding a scientifically correct agreement between the two is important. Generally the size estimated by XRD is on the lesser side than what is obtained from for eg. SEM. In particular, the phenomenon of columnar growth in thin films is not very clear. The micro-structure in such film is explained through a nucleation and growth process along with the deposition parameters, and so called 'structure zone model' - the competing grain growth, incoming flux, etc. How can one perform a judicious characterization using complementary techniques like XRD and SEM? Where does the size estimated by the XRD stands in the light of the columnar grains probed by a SEM. Can multiple/poly-crystalline phases be contained inside the morphological features of the individual column that we see through SEM? People often say only a TEM can reveal information so as to what is/are the correct crystalline phases in a grain. It only adds up to the confusion. Please shed some light here.

Well, "fight" was certainly used as synonym for nonviolent communication :-), and I don't want argue which definition of grain is the most useful. I onyl wanted point out that before starting a discussion it is always good to define the terms first. In so far your question is really useful, especially since the term "grain" is assumed to be clear. But as I already wrote in my first comment there are very different approaches.

Practically all techniques are from my point of view complementarym, what is the main reason for re-inventions of techniques like Laue diffraction which was assumed to be oldfashion.

Sometimes techniques overlap and give the same or similar results, sometimes they don't for certain reasons. For your example it is caused by the averaging of the XRD signal which will result in a smaller crystal size because of the anisotropic dimensions of your column-shaped grains. You even would get different values for the grain size if you would compare the signanl from the top or in cross section. In SEM you can "see" the difference, but also only for a single cut. Scientifically this is very important, but practically often the question is not correctly formulated, since for thin films the major question is probably how big is the diameter of the column-shaped crystals. The length is usually known since described be the film thickness already.

With respect of the accuracy of a measurement, TEM is certainly a top-technique, but we all know that the extrapolation of very accurately determined results is highly dangerous. In so far, an entire multiscale investigation is highly recommended to prove the existence and relevance of phases determined in nano-scale...except, they cannot be veryfied at lower magnification. Then a result is perhaps correct but the conclusion made are of less reliability and should be discussed in this way...until we can find another technique to prove the assumtion. Powder XRD delivers a "convoluted" signal, SEM is first a technique which works with charged particles what always draws a small question mark behind a result. However, SEM image results in an image which allows to measure the size so that we believe we can use this information as seen in a light mcroscope. This is a big advantage in comparison to XRD, but it also offers problems since small orientation deviations we perhaps interprete as new grain whereas XRD doesn't. Similar problems are known for Sigma3 twins which are electronically irrelevant although high-angle grain boundaries. Therefore, EBSD software allows to subtract them out, but one cannot apply this feature for any material or "problem". What I mean is, we need to define the term grain with respect to the certain problem. It is not a very nice answer, but who told us that science is "nice" :-) ?

Indeed, Science is not nice but interesting :). I agree that XRD gives a global estimate (and sometimes much more meaningful information if we speak macroscopically) while Electron microscopy give local information. But I think it is not right on our part to conclude that the average size given by XRD is the average size of the columnar grains (seen in a cross sectional SEM) unless we prove that the column itself doesn't consist of many smaller crystallites. I also found this topic https://www.researchgate.net/post/What_is_the_difference_between_crystallite_size_grain_size_and_particle_size

I only wanted point out the problem of anisotropic grain size. Rod-like or plate-like crystal individuums cause anisotropic peak broadening, e.g. very nicely visible at clay minerals. Therefore, people prefer an isomorph grain size, beside of the texture problme which is still another effect. Of course, if the grains are not critical in any dimension you cannot see this anymore independently how unisotropic their grain shape is. But if you have very thin plates (lets say a few 10nm) but having a lateral dimension which is much bigger, the reflections will show different widths for diffractions. The same should be oservable for long fibres having a diameter in the scale of a few nm. The fundamental diffraction condition "non-limited translational periodicity" is then no more fulfilled for all reciprocal directions.

Single crystal is a complete periodic array of lattices forming the ingot material

while grain is a small crystal (short range array) inside non-crystalline material or may be several grains forming that ingot or polycrystalline

Grain is a small crytal

Single crystal is a regular repeated unit cell

Grain is a small crystal

Single crystal is a regular repeated unit cells

and polycrystalline if it contains more than one type of single crystals

Dear colleague Singh

Crystallite Size is Different than Particle Size. A particle may be made up of several different crystallites or just one crystallite so in this case (particle size = crystallite size)

Crystallite size often matches grain size, but there are exceptions

Crystallites are coherent diffraction domains in X-ray diffraction.

Particles are chunks/pieces (usually very small, below 1 mm) of solid matter, ensembles of atoms. Particles can be as small as two atoms (the nitrogen particle for example, N2)

Grains are volumes, inside crystalline materials, with a specific orientation.

Particles can be polycrystalline, single crystal or amorphous. A 100 nanometer particle of gold, for instance, can be made of:

a single gold crystal,

many grains with a grain size