There are a number of issues that contribute to broadening of the spectrum in NDP analysis. Many of the issues were described on the web page "Sources of Uncertainty." On this page, some additional items and details for consideration are listed related to energy broadening in an NDP spectrum.
First, as a side note, perhaps the best charged particle spectrum energy resolution would be obtained by using a time-of-flight technique. The better the time resolution, the better the energy resolution. Ideally this would include magnetic separation of particles to prevent overlap of the two (or more) neutron reaction particles. The approach would also substantially eliminate radiation background noise.
Below issues are listed that primarily related to solid-state semiconductor detectors (of any composition), such as surface barrier, PIN diodes, passivated implanted planar silicon (PIPs) detectors, etc., and issues related to sample and detector systems.
1. Contributions from the Detector
1.1. Native resolution of detector
1.2. Temperature – increased phonons
1.3. Vibration of detector – increased phonons
1.4. Mass and composition of detector from induced prompt radiation.
1.5. Long term: radiation damage
1.6. Acceptance angle of particles from across the sample
1.6.1. Off-normal positioning of detector to sample face can increasing broaden the acceptance angle of particles into the detector. This issue can be reduced by use of an appropriately sized aperture.
1.7. Contamination on the surface of the detector
1.7.1. Particles on detector face
1.7.2. A film contamination (typically an oil film on the detector face originating from a vacuum pump)
2. Contributions from the signal processing electronics
2.1. Wire length between detector and the preamp
2.2. Optimal voltage (bias) of detector. Requirements are that the bias be high enough for maximum resolution, but no more than needed to collect signal so to minimize collection of noise from deeper in detector than range of charged particles.
2.3. Non-optimal shaping time constant of amplifier
2.4. Pulse pileup
2.4.1. From excessive background
2.4.2. From excessive dead time
2.4.3. From non-optimal setting of lower-level discriminator (LLD)
3. Contributions from the Sample
3.1. Stopping force of the sample composition
3.1.1. Straggling of particle
3.1.3. Variations from the ideal Bragg estimate of stopping force for multi-element samples
22.214.171.124. For example, not correcting for atomic bonding, especially in the case of bonded light elements in sample
3.2.1. Uniformity of composition (in all three dimensions (four dimensions counting time)
3.2.2. Uniformity in density (lateral)
3.2. Contamination on sample surface
3.3.2. Film over sample and its uniformity
126.96.36.199. Oil from vacuum pump
3.3. Diffusion of sample composition during data acquisition
3.4.1. From thermal effects
3.4.2. From electronic induced effects (such as, cycling a lithium ion battery)
4. Sample surface roughness – a more complicated source of signal broadening. Presented here as a separate topic as it is not typically considered. These issues are typically not a complicating factor for NDP for a smooth surface, as for a glass surface or surface deposit. However, the issue is the roughness can possibly result in variations in the path lengths of reaction particles from the average depth of the reaction from the surface. Surface roughness is described as the arithmetic average (Ra) of measurements for “peak” and “valley” heights/depths along a line taken across the sample surface. Root mean square (RMS) roughness is another method of expressing the surface height-valley variations where the results of the height/valley measurements are express as the RMS of the values (see for instance: http://www.harrisonep.com/electropolishing-ra.html ) A few examples of how the roughness changes the NDP spectrum are given here, where the Ra is taken as 500 nm.
4.1. With roughness, for near-surface signal origin the greater potential for signal broadening due to greater probability of large-angle scattering of reaction particles
4.1.1. Not a significant issue if the signal origin is on the surface or very near the surface of the sample