High Resolution N.M.R.-Instrumentation : Recent Advances and Prospects BY R. ERNST AND H. PRIMAS Laboratory of Physical Chemistry, Swiss Federal Institute of Technology, Zurich, Switzerland Received 18th June, 1962 The recent advances of n.m.r.-instrumentation will be reviewed and some new types of equip- ment will be described. New stabilizers for n.m.r.-spectrometers, double resonance devices and high-resolution spectrometers with high magnetic field strength sre discussed. By increasing the signal-to-noise ratio essentially over the present usage exciting new applications of n.m.r. -spectroscopy will become feasible. As an example the proton resonance spectroscopy with just a few micrograms of complex organic substances will be sketched and its importance to biology will be stressed.1. INTRODUCTION In spite of the rapid progress in the field of high resolution n.m.r.-spectroscopy we are of the opinion that there is still plenty of room for essential improvements of the instrumentation and even for completely new developments. It is to be expected that in the next few years we will have available significantly improved n.m.r.-spectrometers and that the chemists will therefore be able to extend the range of n.m.r.-spectroscopy to problems outside the range of present possibilities. Some exciting new applications seem to become feasible. We would expect that the future development in n.m.r.-instrumentation will just continue the trend of the last few years with special concentration on the follow- ing topics : (i) enhancement of the sensitivity of n.m.r.-spectrometers by quite a large factor ; (ii) increasing the resolution, (iii) improving the stability of the instruments, (iv) developments of auxiliary methods and the hereto necessary devices which allow to get more information than from the usual n.m.r.-spectra or to get some part of the information in a more convenient and understandable form, (v) developments of spectrometers of greater versatility, (vi) developments of really foolproof instruments.Of course, these points are not all independent of each other and several are even contradictory. A technical review of the state of the art at the end of 1959 was given at another place 1 and we will restrict ourselves here to the more recent improvements that are partially already realized and partially still in the state of further development of great promise.2. HIGH-FIELD N.M.R.-SPECTROMETERS AND THEIR PROBLEMS It is well known that it is often possible to simplify the appearance of n.m.r.- spectra by increasing the field strength of the magnetic field. There exist two 4344 HIGH RESOLUTION N.M.R. INSTRUMENTATION classes of phenomena, those which are independent of the magnetic field strength (spin-spin-coupling, internal rotation, line width) and those proportional to the magnetic field strength (chemical shift). Only with very high fields can the first class of phenomena be easily separated from the second. The usual magnets used for high-resolution n.m.r.-spectroscopy are electromagnets with iron pole tips. The very high homogeneity of such magnets is due to the fact that iron has at moderate field strengths a high permeability and therefore the pole faces are almost exactly equipotential surfaces.The criterion of “ high ” and “ low ” field strength for a magnet with ferromagnetic material is its saturation field strength, i.e., for pure iron 21-62 kgauss, whilst the highest known saturation fields are 2 to 3 kgauss higher with alloys of the type of Permendur (50 % Fe+ 50 % Co). With iron electromagnets of conventional construction a rather abrupt change of the field configuration occurs at about 12 kgauss (fig. 1) which spoils B &GI FIG. 1 .-Inhomogeneity against field strength. (a) for a typical electromagnet of conventional construction ; (b) for an electromagnet with correctly designed pole tips which avoid any local saturation.(r = radial distance from the centre). B~(v) = B(l +r2b2+ . . .) (2.1) the homogeneity of the higher fields. With certain tricks, such as cycling of the magnet, it is possible to extend the useful range of an iron electromagnet for high- resolution n.m.r. to 15 kgauss or even more. Such procedures can be extremely useful but we have to pay for it : we lose the versatility of the instrument because there is usually a sharp optimum of the field strength for maximal resolution. Further, any cycling procedure reflects the irreversible behaviour of the ferromagnetic material and is therefore inherently not stable. A theoretical and experimental study 2 has elucidated this strange behaviour of electromagnets of conventional construc- tion over 12 kgauss. We could show that these phenomena are entirely due to local saturation effects that must appear with the usual cylindrical or conical pole tips.According to a somewhat simplified theory there is just one pole form that avoids any local saturation at any field strength. It was found that this theoretical curve has to be realized with an extreme accuracy to get a completely flat magnetic field with the same pole tips for any field strength from 0 to 22 kgauss. Further, with carefully selected pure iron any cycling effect can be avoided.R. ERNST AND H. PRIMAS 45 Such spectrometers are nowadays a reality and the extension to field strengths of up to 30 kgauss is in the stage of detailed investigation. At least it should be possible to work with a reduced resolution (enough e.g., for 13C-n.m.r.) at fields up to 30 kgauss even with pole tips of iron.Permendur would be preferable, of course, but there are still severe technological difficulties. It is not easy to speculate on the future of magnetic fields essentially over 30 kgauss for n.m.r.-work. It seems not completely impossible to use magnets of con- ventional construction for certain special n.m.r.-investigations at field strength up to 100 kgauss. But it is to be expected that such investigations will, in consequence of the great expense necessary, be restricted to highly specialized laboratories such as the M.I.T. National Magnet Laboratory and that these magnets will not come into general use. Progress in the development of magnets with superconducting material is so ex- citing that it is tempting to speculate on a possible application for n.m.r.-spectros- copy.It seems possible that a resolution of 1 : 106 can be achieved at a field strength of 60kgauss and with newer materials perhaps even over 100kgauss. But now- adays it is still unknown whether it will be possible to get the necessary time stability for such a resolution and whether it may eventually be feasible to shim such a field to even higher homogeneities. 3. STABILITY OF N.M.R.-SPECTROMETERS Besides all the problems of homogeneity of the polarizing magnetic field the time stability of this field is the most difficult problem in the construction of a high- resolution n.m.r.-spectrometer. The approved principle of pre-stabilization of the magnet current and fine stabilization with a flux-stabilizer seems still to be the best solution.A flux-stabilizer needs a d.c.-amplifier of an extremely good zero-point stability. The use of a galvanometer amplifier for this purpose is now superseded by the use of special mechanical chopper amplifiers which are much more robust, very easy to handle and which give an essentially better zero-point stability than any galvanometer.3 Even with the best flux-stabilizer there are still appreciable residual field fluctu- ations that restrict the measuring accuracy and the measuring time. With a flux- stabilizer alone it is not possible to get automatically a reliable calibration of the spectra which, of course, would be very convenient for precise routine n.lz1.r.- measurements.It is evident that the phenomena of nuclear magnetic resonance itself is predestinated for a field-stabilizer of excellent long-time stability and suitable for an absolute and automatic calibration of the spectra. This idea is almost as old as the invention of the n.m.r. phenomena itself. A high resolution n.m.r.- spectrometer stabilized with a second n.m.r.-probe was for the first time described in 1957 by Baker and Burd.4 The construction of a n.m.r.-stabilizer with a second probe was brought to perfection in the commercial routine instrument A-605, a specific proton resonance spectrometer with precalibrated paper and a calibration accuracy which should fulfil most requirements of routine measurements of the organic chemist. There is a second approach to the construction of a n.m.r.-stabilizer which is technically more complicated but which allows a still greater improvement in the long-term stability of the magnetic field.Evidently the stability of a spectrometer with two spacially separated probes-one for measuring purposes and one for the stabilizer-can be only as good as the field between the probes is constant, i.e., this type of stabilizer is inherently not of absolute stability but susceptible to external disturbances. Such troubles can be avoided if only one single probe is used.46 HIGH RESOLUTION N.M.R. INSTRUMENTATION Nowadays it is customary to use an internal standard (e.g., a few % TNS for proton resonance spectra) and this suggests the use of just this internal reference for the signal for the n.m.r.-stabilizer.This idea was first realized in 19606 and proved to give excellent results. With a high-resolution 25 Mc proton resonance spectro- meter we could achieve a long-term stability of better than 3 parts in 1Olor.in.s. (or better than 1, 5 parts in lO9p.t.p.) over several days. With such excellent long-term stability it is, of course, easy to get n.m.r.-spectra automatically calibrated with extreme accuracy. Further, there is no longer any limitation of the measuring time. There is no difficulty in extending the measuring time over days and to getting 1 I I I I I I 4 5 4 3 2 9 p.p.m. from t.m.s. FIG. 2.-N.m.r.-spectrum of 0.085 mg hydroxyprogesterone. IH-spectrum at 25 Mc ; RC-time constant = 50 sec ; sweep rate = 1 p.p.m./h. in this manner an increased sensitivity by a factor 10 to 50.On account of practical considerations one often does not wish to extend the measuring time over, say, 14 h (taking spectra overnight) but even this time gives an enhancement of the signal- to-noise ratio of a factor of 10 compared with the usual measuring time of some minutes. Fig. 2 shows the very first application of this method : the proton reson- ance spectrum of 0.085 mg of a steroid. In spite of the low frequency of 25 Mc the signal-to-noise ratio still allows a useful discussion of the spectrum. 4. EQUIPMENT FOR DOUBLE RESONANCE EXPERIMENTS BETWEEN LIKE NUCLEI Double resonance experiments and very similar experiments with multiple quantum transitions allow us to get more information from a spin system than the usual n.m.r.-measurements and-what is often even more important-under certain conditions we are able to get this information in a much simpler form. Such decoupling experiments between unlike nuclei or between like nuclei with a big dif- ference in chemical shift have been well known for some years and are rather easily realizable.But the extension of double resonance methods to nuclei with only small differences in chemical shift imposes some difficulties on the apparatus side. Using a n.m.r.-stabilizer with an internal standard this problem can be easily over- come and we could show that it is possible to decouple proton resonance lines which are just a few c/sec apart. Fig. 3 shows a principal scheme of such an apparatus.R . ERNST AND H.PRIMAS 47 The exact theory of double-resonance experiments with a two-spin system with strong coupling shows7 that one has to be very careful in the interpretation of a double resonance spectrum with strong coupling. The large direct and indirect Bloch-Siegert shift makes a straightforward evaluation of the chemical shift values complicated. The complications by the direct Bloch-Siegert effect can be eliminated if one works under symmetrical irradiation conditions.8 For a simple double resonance experiment this means that one has to use two double-resonance frequencies which are symmetrical with respect to the measuring frequency. This can be achieved simply by amplitude modulation of the measuring frequency. p1 record e r FIG. 3.-Schematic diagram for double-resonance equipment with an n.m.r.-stabilizer.w = measuring frequency ; w+ Aw = frequency of the n.m.r.-stabilizer ; w + Sw = double resonance frequency. N.m.r. experiments with multiple quantum transitions give similar information to a double resonance experiment. The ideal universal n.m.r.-spectrometer of the future should have the facilities for using high-frequency fields with high field strength together with very stable leakage and modulation systems. 5. SENSITIVITY OF N.M.R.-SPECTROMETERS The sensitivity of an n.m.r.-spectrometer is defined as the ratio of the signal-to- noise voltages at the output of the instrument. If the noise sources consist entirely of the Johnson- and the Shot-noise of the input circuit the optimal linear filter to achieve a maximal sensitivity is determined by the van Vleck-Middleton-criterion2 The maximal sensitivity Y is then given by a relation of the following form :48 HIGH RESOLUTION N.M.R.XNSTRUMENTATION Apart from the factor 09 the frequency dependence of Y is determined by the series- resistance r of the receiving coil, the noise-factor F of the preamplifier and the con- nection receiving coil-preamplifier. Y is, therefore, a complicated function of the frequency which can be approximated by &YO. It is, therefore, possible to increase the sensitivity considerably by increasing the frequency of the high-frequency field (e.g. from 20 to 100 Mc by a factor of 11). Further the sensitivity is proportional to a geometric factor g determined by the dimensions of the receiver coil and the probe.It can be shown that it is possible to increase the sensitivity of most present n.m.r.-spectrometers by a factor of 5-10 by the proper optimization of the receiver coil and of the connection between receiver coil and preamplifier and by the choice of an appropriate preamplifier tube. There can be given iterative methods for the optimization of the factors dJ'F It is seen from (5.1) that the sensitivity can be increased by reducing the sweep rate a. But a smaller sweep rate affords simultaneously a higher time stability of the frequency and of the magnetic field. This implies the necessity of a n.m.r.- stabilizator. An improvement of the sensitivity by a factor of about 10 in relation to the usual measuring time of some minutes is possible. It is to be remembered that the usual measuring conditions are far from stationary conditions.Therefore, the field strength of the high-frequency field has to be lowered for smaller sweep rates to prevent saturation effects. In this way the sensitivity will be decreased by some amount especially if the relaxation time TI is very long. By the combination of some of these improvements it is therefore easy to exceed the sensitivity of all present n.m.r.-spectrometers by at least a factor of 100. This is, of course, of greatest importance for the biologists and the organic chemists as it means a reduction in the analyzable amount of a substance to less than a hundredth of the present minimum amount. 6. DESCRIPTION OF A NEW UNIVERSAL N.M.R.-SPECTROMETER It is impossible to combine in one single instrument the wishes of a pure chemist who likes a simple and foolproof instrument, suitable for routine investigations and those of a physical chemist who prefers a universal spectrometer of maximum efficiency and versatility.So it is unavoidable that in the future we will have two rather different types of n.m.r.-spectrometers. We restrict ourselves here to a short description of a new universal n.m.r.-spectrometer, developed during the last few years in our laboratory. The first criterion in the design of our new system was a reasonable compromise between the maximum possible field strength suitable for high-resolution n.m.r. and the expenses a research laboratory of physical chemistry can afford. According to a criterion of Klerklo there is a logarithmic dependence between the quotient Y/v (Y = volume of the copper and iron of the magnet, v = volume of the field space) and the maximum possible field strength obtainable with such a magnet This means that the weight of a magnet is increasing exponentially with the maximal possible field strength.We decided that a weight of 5 tons (everything included) would make a fine research magnet at a not yet astronomically high price. This magnet was designed to give a resolution of a few parts in lo9 for the usual n.m.r.-probes but we intend to use also much larger probes with only slightly re- duced resolution. These conditions have given the volume of the field space to about 600 to 800 cm3. With a gap of 30 mm we can achieve a field of 20 kgauss with an (cf.fig. 4).R. ERNST A N D H . PRIMAS 49 input power of only 2.7 kW. But the magnet can dissipate 20 kW with a rise in temperature of only a few degrees so that we can get a field strength over 30 kgauss with a gap of 25 mm and a somewhat reduced homogeneity, but still suitable e.g., for 13C investigations. Thanks to the very carefully designed pole tips this magnet shows the same high degree of homogeneity in the whole range of 0 to over 22 kgauss without the necessity of changing the pole tips and without cycling. It is clear that today a high-voltage magnet is an anachronism. Our electromagnet is a transistor-regulated high-current magnet with directly water cooled hollow conductors. Such a magnet shows a much better thermal behaviour which is of utmost importance for a stable operation.log (V/4 FIG. 4.-Maximum field strength B achievable with a magnet of volume Y and a field space u (according to Klerk 10). a = high performance magnet ; b = low performance magnet. Because the use of an n.m.r.-stabilizer is not possible nor convenient for all types of experiments? we used a flux-stabilizer with a drift stability of about 1 part in 1010 per second and additional devices to allow all kinds of sweep and modulation methods. For the most precise measurements a n.m.r.-stabilizer with internal standard will be added. Facilities for n.m.r.-experiments with any nuclei are a matter of course. It is clear that such a universal instrument cannot be a very compact unit nor an extremely simple instrument. We preferred to use a system of standardized units built around the central part of the system, the high performance magnet.7. EXAMPLES OF NEW APPLICATIONS OF N . M . R . In regard to the applications of the n.m.r.-spectroscopy in chemistry the most significant progress is the gain of sensitivity. In the following we will present as examples some new experiments which, as a result of insufficient sensitivity, were not previously realizable and which are now under investigation in our laboratory.50 HIGH RESOLUTION N.M.R. INSTRUMENTATION Preliminary results with our new n.m.r.-spectrometer constructed especially for The necessary number of moles M to get a signal-to-noise ratio V with a sweep high sensitivity suggest that it will be possible to get the following sensitivities : 8 rate a in rad sec-2 and a longitudinal relaxation time TI is given by c is dependent on the magnetic field strength B and on the nuclear P, kgauss c, mole sec) for 1H for 1 X 7 3 6 ~ 10-9 1 8 ~ 10-7 1 4 12 x 10-9 5~ 10-7 21 7~ 10-9 3~ 10-7 28 5~ 10-9 2~ 10-7 ( 7 4 species : 13C-RESONANCE WITH NATURAL ABUNDANCE : It is well known that n.m.r.-spectroscopy with W-resonance can supply valuable information about the structure of organic molecules.The chemical shifts of the W-lines are much more sensitive to small chemical changes in the environment as those of the 1H-lines. A universal method would permit work with the natural abundance of 13C of 1.1 %. Together with the relatively small magnetic moment of 13C this means that the necessary sensitivity of the n.m.r.-spectrometer has to be very high.ExampZe.-13C-resonance with natural abundance. Molecular weight, 100; relaxation time, TI = I sec; sweep rate, 1 rad sec-2; To get a signal-to-noise ratio of 4, 11 mg of this substance would be necessary. A further application of W-spectroscopy would be as a tracer method for marked molecules in organic chemistry and in biochemistry. The radioactive tracer isotope 14C could be replaced by the stable isotope 13C. A great advantage of 13C is that for an analysis it is not necessary to disintegrate the molecule. magnetic field, 21 kgauss. BIOCHEMICAL PROBLEMS WITH LARGE MOLECULES : With a n.m.r.-spectrometer of extreme sensitivity it is now possible to examine solutions of very low concentrations or of solutions of macromolecules.This is important for biological solutions and, e.g., for the analysis of polypeptides and other complicated molecules. ExampZe.--1H-resonance. Molecular weight, 4000 ; relaxation time, TI = 1 sec ; sweep rate, 1 rad sec-2 ; magnetic field, 21 kgauss. To get a signal-to-noise ratio of 4 0.12 mg of this substance would be necessary. This is equivalent to a solution of 0-4 % in our usual spherical probes of diameter 4 mm. With a smaller sweep rate the necessary amount could be further diminished. PHOTOMAGNETISM: Another useful application of the very high sensitivity of our spectrometer is the detection of paramagnetic triplet states with n.m.r. by observation of the Line broadening on illumination of the probe under examination with suitable light. There are, of course, a great many further applications in any chemical problem where only very small amounts of substance are available.t 1 Primas, Arndt, and Ernst, Z. Instr., 1959, 67, 293 ; 1960, 68, 8 ; 1960, 68, 21 ; 1960, 68, 55. 2 to be published in NucZear Instruments and Methods. 3 A high resolution n.m.r.-spectrometer with a chopper-type flux-stabilizer was constructed in 1961 in our laboratory and exceeds the performance of any galvanometer-type stabillzer by far (unpublished, some preliminary details can be found in ref. (1)). 4 Baker arid Burd, Rev. Sci. Instr., 1957, 28, 313. 5 Varian Associates, Palo Alto, California. 6A short communication was given at the 5th European Congr. MoZ. Spectr. in June 1961. 7 to be published in Mol. Physics. 8 Ernst, Thesis, 1962 (Swiss Federal Institute of Technology). 9 Goldman, Injormatiun Theory, p. 233 (Constable, 1953). 10 de KIerk, Ned. Naiuurk., 1960, 26, 65. de Klerk and Gorter, Appl. Sci. Res. B, 8,265.