Homemade FT-NMR

Out of all analytical chemistry instruments, a nuclear magnetic resonance (NMR) spectrometer makes one of the best to try to make from scratch. When pondering instruments to try making, optical spectroscopic methods were ruled out due to the need for well aligned optics, a photo-multiplier tube, CCD/CMOS/PD array, stable light source, or interferometer. These projects seemed to put purchased (expensive) parts, or access to precise machining which are not ideal attributes for a hobby project. Mass spec was ruled out to difficulty of most ionization methods and achieving a high vacuum even though a quadrupole ion trap would be great fun. Similar reasons ruled out SEM, TEM, and STEM. NMR stands out as a cheap option with relatively simple construction requirements. The main challenge comes in the planning of the project which makes for a great hobby project. Described is my jab at making a NMR with the goal of analysing things such as d methanol in ethanol/water (poorly distilled spirits or denatured alcohol), or ethanol in water (spirits) with C or H NMR. The worse you make an NMR the easier it becomes. High frequency RF is very tough as is producing a strong electromagnet. Thankfully, a weak magnet requires lower frequencies making everything simpler.

Design Requirments

Three primary variables were targeted as important to the design considerations. Magnetic field strength/uniformity, signal sampling frequency, and achievable resolution. Field strength and the opererating frequency relationship is described by the gyromagetic ratio and Zeeman splitting. H.E. Gottlieb et al. (1997) was referenced in order to understand the expected signals. The peaks are in ppm referenced to TMS in two deuterated solvents. D.I. Hoult (1977) provides great detail on design parameters of transmit and receive coils.

The gyromagetic frequency will be the base frequency for the NMR. This frequency is dependent on the magnetic field strength that the nuclei are exposed to. A lofty dream would be to have a gradient field strength, so that the detected RF frequency would be the convolution of the element, its chemical enviroment, and its position within the NMR. For a first attempt, we will not build a low resolution MRI. A highly uniform magnetic field can be applied to the sample via a set of Helmholtz or Maxwell coils. The electronic requirements and magnet requirements must be determined in tandom. Per the table above, ethanol and water produce peaks that are seperated by 1.2ppm. A 20mT Helmholtz coil magentic field would have a base frequency of ~850kHz and therefore a 1.02Hz difference. This target field strength allows the coil design to begin. $$B= \frac{I \mu_o n}{R (\frac{5}{4})^\frac{3}{2}}$$