Solid-State Nuclear Magnetic Resonance Spectroscopy for Analyzing Polymorphic Drug Forms and Formulations

February 1, 2006
Pharmaceutical Technology, Pharmaceutical Technology-02-01-2006, Volume 2006 Supplement, Issue 1

This article discusses the advantages and disadvantages of using solid-state NMR spectroscopy for the analysis of pharmaceutical solids.

Scientists use several common techniques to analyze pharmaceutical solids. These methods, which include single-crystal X-ray diffraction, X-ray powder diffraction, differential scanning calorimetry, thermal gravimetric analysis, infrared, and Raman spectroscopy, have several advantages for the analysis of polymorphic drug forms and formulated products, but, in some cases, they can have some disadvantages (see sidebar, "Pros and cons of techniques for pharmaceutical solids analysis").

Solid-state nuclear magnetic resonance (SSNMR) spectroscopy has several advantages over these techniques. First, it is nondestructive and noninvasive. If little sample is available for analysis, scientists can acquire an SSNMR spectrum and return the sample for any further testing. SSNMR spectroscopy can be used on bulk drugs and on all types of drug formulations (e.g., tablets, capsules, suspensions, creams, and transdermal formulations) with little to no sample preparation. SSNMR spectroscopy is uniquely suited for testing drug formulations because the NMR resonances for most pharmaceutical excipients occur in a narrow range of the NMR spectrum. Thus, it is easy to distinguish the excipient from the active pharmaceutical ingredient (API) NMR resonances. Spectral subtraction can even be used to eliminate excipient peaks from the spectrum.

In addition, the technique can be used to study inclusion compounds such as those with cyclodextrins. Furthermore, the NMR spectrum can be used to examine host–guest interactions by monitoring changes in the NMR peak position and dynamics associated with a given nuclear site.

Moreover, SSNMR spectroscopy is quantitative and selective. Because NMR spectroscopy is inherently a quantitative technique (i.e., signal intensity is relative to the number of nuclear sites at that specific resonance frequency), SSNMR spectroscopy also can provide quantitative information if done properly, as will be discussed later in this article. This technique can quantify mixtures of polymorphic crystalline forms, or of crystalline and amorphous materials. Analyses can all be performed without the need for pure forms or a standard curve.

Pros and cons of techniques for pharmaceutical solids analysis.

The method also is selective; each NMR resonance seen in the NMR spectrum is caused by a specific nuclear site. Most often, 13C NMR spectroscopy is used, but many other nuclei (e.g.,19F, 15N, 23Na, 31P, 17O, or 2H) can be used, depending on what will be studied. NMR spectroscopy is unique because it can monitor a specific nuclear site. For example, it can be used to look for changes in the local chemical environment that may be caused by drug–excipient interactions or changes during the formulation process, along with the formation of different polymorphic or amorphous forms. In addition to the technique's selectivity, isotopic labeling of low natural abundance nuclei such as 13C, 15N, and 17O can be used to increase the sensitivity of a specific nuclear site. An example will be presented later in this article.

SSNMR spectroscopy also can be used to examine the structure of the material and its associated dynamics. For crystalline materials, one can determine the number of crystallographically inequivalent sites in the unit cell. It also can detect conformational changes and hydrogen bonding to some extent. Other, more advanced, two-dimensional techniques such as rotational echo double resonance can be used to measure distances (1). For amorphous materials, SSNMR spectroscopy can be used to examine the degree of disorder because various processing techniques (e.g., lyophilization, spray drying, melt-quench, cryomilling) can vary the overall degree of disorder in the sample.

Sometimes the system's dynamics also can be used to investigate other aspects of the drug being studied. For crystalline polymorphic forms, some forms may be less stable than other forms, which could be caused by the overall molecular mobility. Correlations between the overall mobility or stability of polymorphic forms can be made by measuring the spin-lattice relaxation times (T1 and T). If a new form was discovered, the relaxation time could be measured to yield an early approximation of its overall stability. Sometimes, the same polymorphic form from different lots can exhibit different properties (e.g., stability, dissolution rate) which could be caused by the degree of crystallinity in the sample. The presence of defect sites or less-crystalline domains in the sample may not be observed in X-ray powder diffraction. This is not the case with SSNMR spectroscopy. These sites provide another avenue for the relaxation process of the spin states in the sample. Thus, a sample that has a lower degree of crystallinity would exhibit a faster spin-lattice relaxation time (T1), which may vary by less than a second to several seconds. This technique is not limited to analyzing only the bulk drug. It also can be applied to formulations. Changes in the overall dynamics of the drug during formulation could screen out potential drug formulations because of the increased mobility observed, possibly yielding a final candidate faster.

Despite its advantages over other common techniques, SSNMR spectroscopy also suffers from some disadvantages. It often requires a lot of expertise in the technique to run it properly. And, the equipment can be expensive, though most narrow-bore solution NMR spectrometers can be adapted to perform SSNMR work. Another obstacle is that the experiments are not routine (i.e., the experimental parameters used must be determined for a specific compound before running the NMR experiment). Automation is difficult because the sample often fails to spin when using an automatic program and requires human intervention. Compared with the techniques listed in the sidebar, "Pros and cons of techniques for pharmaceutical solids analysis," SSNMR spectroscopy is the least sensitive and requires a large amount of sample to generate an adequate spectrum when studying low natural abundance nuclei such as 13C.

Analysis times for SSNMR experiments also can be problematic because they can range from a couple of minutes to a few hours, to a couple of days or more, depending on the sample being analyzed and what type of NMR experiment is used. Peak assignment in the SSNMR spectrum can be challenging because multiple peaks could be present for a single nuclear site or they may be overlapping peaks. Another problem is that the chemical shifts in the solid state can often differ as much as 10 ppm from the solution NMR values. A current review article shows many applications of this technique to the analysis of pharmaceutical solids (2).

General theory

Originally discovered in 1945, NMR involves transitions between nuclear spin states that become nondegenerate when placed in a magnetic field. The transitions resonate when an electromagnetic field at a specific frequency (i.e., based on the nucleus and magnetic field) is used.

In general, nuclei have a spin quantum number (I) which may be an integer, half-integer, or zero (not NMR active). A nucleus with integral or half-integer spin has an associated magnetic moment. When placed in a magnetic field, a torque is produced to align the magnetic moment with the magnetic field. Because this effect cannot occur, the torque produces a rotation about the axis of the magnetic field at a specific frequency of:

in which γ is the electromagnetic ratio for the given nucleus and B0 is the magnetic field strength. If a second field, B1, that is static and small is applied to the system, the nuclei nutate or wobble slightly in their orbits. But if the B1 field oscillates at the same frequency as the nuclei, then the field will appear to be static to the nuclei and transitions between the spin states will occur. NMR spectroscopy is so powerful because each nucleus in the molecule is shielded by the surrounding electrons, so each nucleus sees an "effective field" that will change its resonance frequency. The frequency changes observed for the peaks in a molecule are reported in ppm to account for the differences in all of the magnetic fields that are used:

(shift in hertz [Hz]) ÷ (spectrometer operating frequency [Hz]) × 106 .

Two problems for performing NMR spectroscopy in the solid state are dipolar coupling and chemical shift anisotropy. Dipolar coupling arises because of the sample's rigid nature. The equation for dipolar coupling is:

v = va ± 0.5R(3cos2 θ–1)

in which v is the resonance frequency,

for which γ is the magnetogyric ratio for the specific nucleus (A or X), r is the internuclear distance, μ0 is the permeability constant, and θ is the angle between the static magnetic field and the spin pairs.

One must be concerned with the γ and R in these equations. If a nucleus has both a high magnetogyric ratio and short internuclear distance, the dipolar coupling appears very large. For this reason, proton NMR spectroscopy is not performed in the solid state. For similar reasons, natural abundance also must be considered. With 13C at ~1% natural abundance, the probability of having two 13C nuclei next to each other in the molecule is very small, and thus only the heteronuclear dipolar coupling must be addressed. Normally, this dipolar coupling is ~50 kHz and is the strongest for CH and CH2; not as strong for CH3; and much weaker for quaternary carbons. To solve this problem and reduce or eliminate the dipolar coupling, a decoupling pulse is used. The decoupling pulse in SSNMR spectroscopy is similar to that used in solution NMR spectroscopy, but as much higher power (i.e., typically a few hundreds of watts). During the acquisition of the free-induction decay (FID), the radiofrequency (RF) amplifier for the nucleus to be decoupled is turned on. This causes the spins to be flipped rapidly between spin states, thus stopping the dipolar-coupling spin-exchange process and thereby narrowing the resonances (see Figure 1).

Chemical shift anisotropy (CSA) is another main problem. The NMR spectrum of a solid sample can be very broad because of the individual crystallites' multiple orientations in the magnetic field. The equation for chemical shift anisotropy is:

in which σ is the observed chemical shift, σiso is the isotropic chemical shift, σaniso is the anisotropic chemical shift, and θ is the angle between the shielding vectors and the magnetic field. To get high resolution NMR data in the solid state, only the isotropic chemical shift must be obtained. For this to occur, the anisotropic portion must be zero and can be accomplished by setting the angle to 54.7°. By rapidly spinning the sample about this axis (i.e., magic-angle spinning), the anisotropic portion cancels out thereby giving high-resolution spectra (see Figure 1). Spinning speeds can be approximately a few kHz to 35 kHz or more (i.e., 180,000–2,100,000 RPM or more). The 54.7° angle is also the angle of the body diagonal of a cube. By spinning about this angle, each crystallite spends an equal amount of time in the x, y, and z dimensions, thereby averaging out the crystallites' multiple orientations in the sample. If the spinning rate is not fast enough, artifacts, called spinning sidebands, will appear at integral factors of the spinning speed (see Figure 1, only the peaks at ~17 and ~133 ppm are real). Based on the spinning rate compared with the width of the static NMR powder pattern, a certain percentage of the main peak intensity will be lost in the spinning sidebands. This point is important if quantitation is to be performed.

Figure 1: Comparison of the 13C NMR spectra of hexamethyl-benzene acquired using several techniques. DD is dipolar decoupling, MAS is magic-angle spinning, and CP is cross polarization.

The spinning sidebands can be eliminated in two ways. First, the sample can be spun faster than the width of the static powder pattern, which could be as high as 50 kHz or more. Second, a special pulse sequence can be used to cancel out the sidebands. More information about this process will be presented later in this article.

Common and advanced techniques

With 13C SSNMR spectroscopy, the recycle delay between transients can range from a few seconds to an hour or more. For this reason, cross polarization (CP) is often used.

With 1H- 13C CPMAS NMR spectroscopy, relaxation times are not dictated by the 13C relaxation rate, but rather by the 1H relaxation, which is much shorter in most cases. A typical CP experiment consists of four main steps: generate the magnetization on the abundant nuclei (1H), transfer this magnetization to the dilute nuclei (13C), acquire the dilute spin (13C) FID, and decouple the abundant spin (1H). The magnetization from the 1H is placed on the x-axis by applying a π/2 or a 90° pulse. This procedure is followed by another pulse, which is sometimes termed the contact pulse, contact time, or the Hartman–Hahn matching condition.

The process of transferring the magnetization from the abundant to dilute spin is accomplished through the use of the heteronuclear dipolar interactions by a process of spin exchange. This will only work if the nuclei are at the same resonance frequency. Because the main magnetic field cannot be changed, the B1 field must be changed. When the B1 field is applied at the resonance frequency of the specific nuclei, the nuclei will precess as if they were the only applied field. For 1H- 13C CP, if γHBIH = γCBIC, the Hartman–Hahn condition is achieved.

Often, the duration of the contact pulse must be optimized because two main factors affect the signal intensity: the cross polarization time constant (TCH) and the proton spin-lattice relaxation in the rotating frame (T1ρH). Initially, the magnetization builds up because of the TCH, and then the T1ρH will cause the magnetization to decrease. Thus, a contact time period must be chosen that allows for the maximum magnetization transfer or maximum signal intensity. Not all samples will exhibit the same CP profile. Some samples may have a very fast TCH while others are slow. The same scenario applies to the T1ρH. Thus, quantitation by CPMAS NMR spectroscopy must be performed very carefully. Another benefit of CP is that the sensitivity is increased by a factor based on the magnetogyric ratios of the two nuclei. The factor is 4 for 1H- 13C and 10 for 1H- 15N.

A couple of advanced CP techniques such as variable amplitude cross polarization (VACP) and ramped CP also can be used. These techniques are needed because of the rapid spinning rates that are used in modern SSNMR systems. Because CP transfers magnetization through dipolar couplings, the dipolar couplings are partially averaged at fast spinning rates (i.e., the equation for dipolar coupling has the same angular dependence as the chemical shift anisotropy [3cos2 θ – 1]). This will lower the overall CP efficiency, hence yielding lower sensitivity. In both VACP and ramped CP, the intensity of the B1H field is varied, thereby transferring the magnetization that would have been lost in the spinning sidebands. Figure 2 offers a visual representation of the standard CP pulse sequence along with the VACP and ramped CP pulse sequences.

Figure 2. Schematic representation of cross polarization (CP) pulse sequences. X) X channel (i.e., 13C, 15N), a) standard CP, b) ramped CP, c) variable amplitude CP. 1 is 1H 90° pulse, 2 is the contact pulse, 3 is the 1H decoupling pulse.

Another pulse sequence addition is that of total sideband suppression (TOSS). With complex NMR spectra, the presence of spinning sidebands can cause problems for determining which peaks are real and which are artifacts without conducting multiple spinning speed studies. The sequence consists of four 180° pulses in which the sideband phases are modulated. The times between the pulses—based on the rotor period (τR)—are set so that the centerband is refocused and the sidebands cancel. A visual representation of the pulse sequence is shown in Figure 3 with an example spectrum.

This pulse sequence has a few disadvantages. The sideband intensity is not refocused into the centerband. If the spectrum was acquired at a low spinning rate, the centerband intensity can be greatly reduced. Another problem is that if the four 180° pulses are not perfect, then additional artifacts will be present in the spectrum.

Table I: 13C NMR chemical shifts and associated assignments.

Component identification

With SSNMR spectra, there are some general regions that can be used to aid in the assignment of a spectrum (see Table I). For the drug itself, the spectra can vary, but they mostly fall in the same regions because of the common functional groups. Most often, the API consists of aromatics, aliphatics, carboxylic acids, and heteroatoms. Excipients usually will not have any aromatics, aliphatics, or carboxylic acids, thus allowing for the easy identification of these elements in the NMR spectrum (see Figures 4 and 5).

Figure 3. Schematic representation of a variable amplitude cross polarization (VACP) pulse sequence with total sideband suppression (TOSS) and corresponding changes observed in the 13C CPMAS NMR spectra of neotame monohydrate.

But, this is not the case for XRPD. The diffraction lines can occur at any location, which can make component identification difficult in a formulated product (see Figure 5). To show this effect further, several blends of ranitidine HCl (Alfa Aesar) and lactose monohydrate (USP) were prepared and analyzed by XRPD and SSNMR spectroscopy.

Figure 4. 13C CPMAS spectra comparing a commercial ibuprofen tablet and ibuprofen.

When using a standard scan rate of 1°/min, a detection limit of ~5% was achieved (see Figure 6). Many diffraction lines overlapped, thus further limiting the detection of small amounts of ranitidine HCl. This was not the case with the SSNMR experiments (see Figure 7). Using this method for the analysis (8-h experiment time), the peaks were clearly resolved and a detection limit of ~0.5% was calculated based on the signal–noise ratio. It is interesting to note that the lactose peaks in the blends prepared were not as intense as they should have been, based on the percentage of lactose present. This was because the NMR experiments were conducted with optimized parameters for the ranitidine HCl (recycle delay and contact time). With excipients, the relaxation time usually is much longer. For example, for lactose at 9.4 Tesla, the relaxation time was ~200 s, whereas the relaxation time for ranitidine HCl was 10 s. Because of this, the lactose peaks became saturated over the course of the NMR experiment and did not add any signal to the peaks.

Figure 5. 13C CPMAS spectra and X-ray powder diffraction profiles of some common excipients.

In addition, the blend had only two components. If more excipients were added, the detection limit by SSNMR spectroscopy would remain essentially the same because the excipient peaks fall in the same region of the NMR spectrum. If any inorganic excipients are used, they will not even be seen in the NMR spectrum. In the case of XRPD, this cannot be known.

Figure 6. Determination of the detection limit of ranitidine HCl in a lactose blend by XRPD. Arrows indicate resolved diffraction lines.

Another interesting use for SSNMR spectroscopy is to determine pH (3). In a solution, only a standard pH meter is required. In the case of solids, no current analytical method can determine pH either in a formulation or during stability testing. This is important if the compound either undergoes a hydrolysis reaction or is sensitive to changes in the pH. It is possible to perform the measurement with SSNMR spectroscopy. This can be done by the use of a probe molecule, which can be any weak acid or base that has a pKa(s) in the range being examined. This probe molecule also must be 13C enriched ensuring that only 1 to 2 mg is added to 1 g of the total sample so that it will not have any other effect on the product. For example, fumaric acid is readily available, 13C enriched, and has two pKa values (3.03, 4.54) for a wider range of pH values (4).

The relative concentration of the species present with respect to pH shows that, depending on the detection limit of the salts produced, a 1.5–6.5 pH range may be possible (see Figure 8). To determine the pH relating back to a standard pH scale, a standard curve must be determined. This is done by dissolving some standard fumaric acid and adjusting the pH of the resulting solution. Because the solid must be isolated at this specific pH, the sample must be flash frozen to "lock" the molecular state. Then, the frozen solution is lyophilized to get the solid for analysis by SSNMR spectroscopy.

Figure 7. Determination of the detection limit of ranitidine HCl in a lactose blend by 13C CPMAS nuclear magnetic resonance spectroscopy.

Based on the data in Figure 8, fumaric acid can be used to monitor the pH from ~2 to 5 (5). At low pH, only a single carbonyl peak is observed. When the pH is increased, however, other peaks are seen because of salt formation. By plotting the relative values of these peaks at each pH, a standard curve can be prepared and then related back to the sample to yield the pH. The lower and higher pH values were hard to determine because the optimized NMR values for generating the standard curve were those of the stock fumaric acid. This is also why the peaks' relative intensities in the NMR spectrum do not match the theoretical calculations. This is necessary because all NMR experiments for preparing the standard curve and the sample must be measured using identical parameters (except for the number of transients).

Figure 8. Concentration profile and 13C CPMAS NMR spectra at designated pH for fumaric acid.

With the use of selective isotopic labeling, the analyst can monitor very low levels of any impurity, either polymorphic, degradation product or generation of amorphous material. Because the natural abundance of 13C is 1%, isotopic labeling can increase the sensitivity of a single nuclear site by as much as two orders of magnitude. Care must be taken if the entire molecule will be 13C labeled because this will introduce 13C– 13C dipolar interactions, thereby decreasing the overall spectral resolution (6).

Figure 9. 13C CPMAS spectra of 13C labeled aspirin, a) 13C CPMAS spectrum of the sample, b) 13C CPMAS spectrum of aspirin (natural abundance), c) result of spectral subtraction showing only the 13C labeled carbon.

Researchers have examined drug degradation in the solid state by using aspirin as a model (7). Aspirin was chosen because it has been well characterized and it is easily synthesized with 13C isotopic enrichment on the carbonyl carbon. It also is representative of drug degradation by hydrolysis. Some initial work was performed that showed little to no degradation when crystalline aspirin was placed in the stability chamber and subsequently analyzed by HPLC. The drug was pure crystalline, however, and may not be representative of a manufactured tablet. Studies were conducted to examine highly energetic states in the presence of a bulk, less reactive state. Highly energetic states may be formed during the formulation process, and the formulation's overall stability may be governed by these small amounts of highly energetic states. Samples were prepared by either grinding or freeze-drying 13C labeled aspirin with hydroxy-propylmethylcellulose (HPMC) (6 mg 13C labeled aspirin/100 mg HPMC) and then adding it to 294 mg of crystalline aspirin (1.5% impurity level for 13C labeled aspirin). These samples were then placed in a stability chamber at 70° C, 50% RH for seven days and analyzed by SSNMR spectroscopy. Spectral subtraction was used to isolate the NMR peak easily because of the 13C labeled site from the bulk aspirin (see Figure 9). For the formulation blends shown previously, the peaks caused by HPMC did not interfere with any of the aspirin peaks. For the ground samples, no degradation products were observed. The NMR peak of the labeled material decreased in intensity. A decrease in the cross-polarization efficiency caused this effect because the sample was "wet."

Figure 10. 13C CPMAS spectra of a) 13C labeled aspirin after freeze drying with HPMC, b) blend with natural abundance aspirin, c) acquired after stability testing.

With the freeze-dried sample (see Figure 10), the initial sample was amorphous aspirin, which would be expected to be highly reactive. After seven days, however, no peaks from the labeled component were seen. After seven days, however, no peaks from the labeled component were seen. After the sample dried overnight in a 50° C oven and was analyzed by SSNMR spectroscopy, the degradation product peak was observed, which was that of amorphous salicylic acid (see Figure 11). This finding was not expected because it was anticipated that amorphous aspirin would degrade and recrystallize to a crystalline form of salicylic acid. This low level of impurity would have been seen easily when performing an HPLC assay, however it would not have been able to determine whether the impurity was amorphous. It is likely that SSNMR spectroscopy would be the only technique to possibly answer this question, especially at only a ~1% level.

Figure 11. 13C CPMAS spectrum of a) aquired after stability testing, b) after drying the poststability sample, c) obtained after spectral subtraction of 13C CPMAS NMR spectrum of natural abundance aspirin (*) amorphous aspirin, (**) amorphous salicylic acid.

Polymorphism

Polymorphism, or the ability of a substance to adopt two or more configurations or arrangements in the crystal lattice, is a concern whenever a solid dosage form is used. Through either the manufacturing of the API or the formulation process, polymorphs can be generated. Most often, XRPD is the technique of choice for the identification of polymorphic forms because it is a readily available technique. As shown in previous examples, if a polymorphic change occurs during the formulation process, the detection limit may not be that great using XRPD. SSNMR is capable of detecting different polymorphs easily. NMR spectroscopy, unlike XRPD, is sensitive to changes in the local chemical environment, thus slight changes in bond lengths, bond angles, interactions with neighboring molecules or, in the case of pseudopolymorphs, different salts, hydration levels, or solvates can have an effect on the NMR spectrum. The changes seen in the chemical shift can be affected by several factors. The gamma gauche effect can result in a significant change in the chemical shift. Isotactic polypropylene results in a 10-ppm shift from syndiotactic polypropylene. In addition, an eclipsed conformation is ~5 ppm different than a staggered conformation, which is most significant for aliphatic carbons. Conjugation in the molecule is also significant with molecules containing double bonds or carbonyl groups attached to aromatics. The overall amount of conjugation is often determined by the relative orientation of the groups and can cause a 15 ppm shift or more. Some pseudopolymorphs of nedrocromil salts are shown in Figures 12 and 13 (8). As Figure 12 shows with the two sodium salts, a change occurred in the orientation of the carbonyl group attached to the aromatic carbon, resulting in approximately an 8-ppm shift in the chemical shift of the quaternary carbon (No. 8) that is easily resolved. This same quaternary carbon shows similar changes for the pseudopolymorphs analyzed allowing for good resolution at determining the form present. Another possible change is a difference in the arrangement of the molecules in the crystal lattice. This type of change usually will result in a change in chemical shift of 1 to 2 ppm. Because it is difficult to find a system that has a change in the arrangement in the crystal lattice without a corresponding change in the molecular conformation, it is difficult to determine what the actual chemical shift change will be.

Figure 12. Crystal struture and 13C CPMAS NMR spectra of pseudopolymorphs of nedocromil Na with tentative spectral assignments.

Possibly the most important application of SSNMR spectroscopy is quantitation. NMR spectroscopy is inherently quantitative because the signal's intensity is directly proportional to the number of nuclei at that given resonance frequency. This is not the case when cross polarization is used. Because most 13C SSNMR spectra are acquired using CP, this must be considered before performing any type of quantitation work.

Figure 13. 13C CPMAS NMR spectra of some pseudopolymorphic forms of nedocromil salts.

Three main items must be accounted for: the spin-lattice relaxation time for the proton (T1H), the cross polarization constant (TCH), and the spin-lattice relaxation time in the rotating frame for the proton (T1ρH). First, for accurate quantitation, the relaxation time should be equal to or greater than five times the longest T1 of the components present in the sample. This factor will ensure that all of the spin states have returned to equilibrium. If the relaxation delay is too short, then the quantitation values will be off (because of partial saturation of the peaks with the longest T1 value).

Figure 14. 13C CPMAS NMR spectra of anhydrous neotame a) Form G, b) Form A, c) amorphous. Asterisk indicates quaternary aromatic peak used in quantitation.

Another problem is a combination of TCH and T1ρH. If magnetization is not allowed to be transferred fully or if the decrease in magnetization varies for the components in the mixture, the results will be incorrect. For example, neotame, a newly developed sweetener, has multiple polymorphic forms. Form A and G and amorphous material have been quantitated with relatively good accuracy by SSNMR spectroscopy (see Figure 14) (9). In the case of the crystalline polymorphs A and G, the TCH times were nearly identical, however the T1ρH times vary significantly. To account for the decrease in magnetization over time, a multiple contact time approach was used. NMR experiments were conducted at multiple contact times and were fit (see Figure 15). By using this fit and extrapolating back to a contact time of zero, researchers accounted for the decrease in intensity over time by only using an equilibrium magnetization intensity. The results from this work and associated errors are presented in Table II.

Figure 15. Plot of integrated intensities versus contact time obtained when analyzing a 50% (wt–wt) mixture of anhydrous neotame Form A and Form G.

With the amorphous material and either crystalline polymorph, there were many errors with the individual quantitation results for each mixture. Nonetheless, when comparing a plot of the measured with actual amorphous content, the data fit well, leading to the discovery of the presence of amorphous material (<20%) in the crystalline samples. For the NMR spectrum of an amorphous material, peaks are roughly ten times as broad and based on the positions of these amorphous peaks relative to the crystalline peaks, the amorphous peaks sometimes will appear to blend in with the baseline. Additional experiments were conducted that proved the presence of amorphous material in the crystalline sample.

Table II: 13C CPMAS NMR quantitation results of anhydrous neotame polymorphs A and G.

Because amorphous peaks generally are very broad, the detection limit for an amorphous material is ~10–20%, based on the relative location of the amorphous and crystalline peaks in the spectrum. Nonetheless, one can increase the detection limit and even perform quantitation on samples with small amounts of amorphous material. First, amorphous materials generally have T1H times that are much faster than the crystalline forms. Thus, the NMR peaks from the crystalline material can be attenuated by using a relaxation time (or contact time) that is optimized for the amorphous material, thereby saturating or reducing the relative intensity of the crystalline peaks. As for the quantitation of low amounts of amorphous material, a standard addition method can be used if amorphous material can be made and verified to be amorphous by SSNMR spectroscopy. If amorphous material stability is an issue, the NMR experiments can be run at a low temperature to limit any degradation or recrystallization during the course of the NMR experiment. For the actual quantitation experiments, a single sample must be prepared with a known amount of amorphous material added to the sample. After the NMR experiments are completed, the percentage of amorphous material in the original sample can be calculated easily. If a very small amount of amorphous material is expected in the API, multiple quantitation experiments should be performed to generate some statistics to go along with the determined amorphous content.

Conclusions

Solid-state nuclear magnetic resonance (SSNMR) spectroscopy is an excellent technique for performing many different types of analyses on pharmaceutical solids. It has some disadvantages, but it also has many advantages over other techniques. Though SSNMR spectroscopy is often considered a "last chance" technique, it should be considered seriously as the technique of choice if there are doubts about other current analytical techniques' ability to achieve the necessary results.

Acknowledgment

The author would like to thank Dr. Jeff Botts for his assistance in the preparation of some figures in this article.

Thomas J. Offerdahl, PhD, is a senior scientist and manager of the solid-state NMR facility at Aptuit Inc., 10245 Hickman Mills Drive, Kansas City, MO 64137, tel. 816.767.4798, thomas.offerdahl@aptuit.com

References

1. F.G. Vogt et al., "Determination of Molecular Geometry in Solid-State NMR: Rotational-Echo Double Resonance of Three Spin Systems," J. Phys. Chem. B 107 (5), 1272–1283 (2003).

2. P.A. Tishmack, D.E. Bugay, and S.R. Byrn, "Solid-State Nuclear Magnetic Resonance Spectroscopy—Pharmaceutical Applications," J. Pharm. Sci. 92 (3), 441– 474 (2003).

3. B. Henry, P. Tekely, and J.-J. Delpuech (2002). "pH and pK Determinations by High-Resolution Solid-State 13C NMR: Acid-Base and Tautomeric Equilibria of Lyophillized L-Histidine," J. Am. Chem. Soc. 124 (9), 2025–2034.

4. S. Budavari et al., Eds., The Merck Index, (Merck Research Laboratories, Whitehouse Station, NJ, 1996).

5. J. W. Lubach et al., unpublished results.

6. M.T. Zell et al., "Two-Dimensional High-Speed CP/MAS NMR Spectroscopy of Polymorphs 1. Uniformly 13C-Labeled Aspartame," J. Am. Chem. Soc. 121 (6), 1372-1378 (1999).

7. B. Chen et al., "Monitoring the Reactivity of Amorphous Aspirin as a Minor Component with 13C Isotopic Labeling and Solid-State 13C CPMAS NMR Spectroscopy," paper presented at the American Association of Pharmaceutical Scientists National Meeting, Salt Lake City, UT, Oct. 26–30 2003.

8. L.R. Chen et al., "Nuclear Magnetic Resonance and Infrared Spectroscopic Analysis of Nedocromil Hydrates," Pharm. Res. 17 (5), 619–624 (2000).

9. T.J. Offerdahl et al., "Quantitation of Crystalline and Amorphous Forms of Anhydrous Neotame Using 13C CPMAS NMR Spectroscopy," J. Pharm. Sci. 94 (12), 2591–2605 (2005).