Errors and variations in the observed particle size distribution results can be introduced in various stages of analysis, right from sampling to data interpretation. Most instrument manufacturers list sources of error particular to the instrument and suggest means and techniques to avoid them. The following discussion lists briefly in a general manner the various stages at which errors and other variations can be introduced in the analysis process.

1. Sampling and specimen preparation related

a. Errors introduced due to use of non-representative samples, i.e.. errors due to incorrect sampling procedures. This holds true for both dry powders and powders dispersed in suspensions. Proper sampling procedures can ensure that these errors can be minimized, if not eliminated completely.

b. Analysis of powders finer or coarser than the detection limits of the instrument being used. Even when analyzing powders with dimensions close to the upper and lower detection limits, it is a good practice to verify the validity of these limits using suitable primary or secondary standards.

c. Errors introduced upon analysis of non-spherical powders. Due to the assumptions of particle sphericity, any deviations from this shape will cause bias and errors to be introduced in the particle size and size distribution results. Unless analyzed by appropriate algorithms designed as part of the instrument software, or designed for use on the obtained scattering patterns, it may be expected that the magnitude or error will be magnified as the deviation from spherical shapes is increased.

d. Errors associated with optical properties of the material. In most instances it is extremely helpful to know the optical properties and some physical properties such as density of the material being tested. Most instruments require the user to provide this information for calculation of the particle size and size distribution and any errors in these values are reflected in the calculated results.

e. Reliable procedures for creating and maintaining stable dispersions of the powders should be developed. The particles should remain dispersed even in the sample cell, as these instruments lack the ability to distinguish between primary particles and agglomerates. These procedures should not cause fragmentation of friable particles or lead to formation of stable bubbles that can interfere with the measurement and bias the calculated result. Closely related to this is the issue of sample stability, where certain powders may change size over a period of time due to dissolution or precipitation mechanisms. In such cases, samples should be prepared immediately prior to analysis and discarded after analysis.

2. Instrument and procedure related

a. Errors introduced due to non-aligned or misaligned optics. Instrument manufacturers specify procedures and checks to ensure proper alignment.

b. Errors due to lack of background signals or errors in procedures for obtaining background signals. In this instance, the detector elements use an incorrect background signal to ratio the diffracted signal. Thus, if the dispersion medium has been changed then an incorrect signal will be used leading to errors in the calculated size and size distribution.

c. Eight leakage due to stray or extraneous light in the instrument will cause additional signals at the detector elements that will be analyzed as diffracted signals from the particles and be included in the size distribution results.

d. The use of incorrect optical models will have a significant effect on the calculated size distributions. These effects may be manifest not only in the range of the size distribution, but also in the shape of the distribution. Determining whether the optical model used is appropriate can be quite complex. Some manufacturers are working on software routines that will enable determination of a theoretical scattering pattern from the calculated size distribution and based on the chosen optical model. A comparison of the calculated scattering pattern with the observed scattering pattern can provide a check of the model used. In numerous instances, microscopy-based techniques may have to be used to compare the calculated size distributions. In these cases however, appropriate precautions need to be observed. 

e. Software related bias may cause significant errors in the calculated results. Most errors would arise due to the design of the deconvolution and inversion algorithms that may be based on assumptions not reflected or applicable to the particle system under study. An example of such an error would be the use of a model-dependent inversion procedure (i.e., inversion procedure that assumes a particular shape for the powder distribution) on a multi-modal powder system.