Practical Quantitative Solid-State NMR Spectroscopy


  • ˜Using a variety of sophisticated one dimensional (1D) and two dimensional (2D) experiments, solid-state nuclear magnetic resonance (SSNMR) spectroscopy has evolved to become an integral and information-rich technique in the field of pharmaceutical science. The Food and Drug Administration (FDA), which requires the control of the solid forms used in drug product, has recognized the need for SSNMR characterization. SSNMR can be used to characterize and quantify solid forms present in drug substances and drug products.  However, quantification of components in mixed powdered solids is almost always more challenging than in liquid solutions, particularly related to particle size and sample homogeneity.
  • ˜Ranitidine HCl (Figure 1) is a type 2 histamine antagonist for treatment of ulcers, indigestion, acid reflux, and heartburn.  In this work, we present the quantification of ranitidine-HCl in three different formulations of over-the-counter (OTC) tablets.  Using the two pure polymorphic forms (form 1 and form 2) of commercially available ranitidine, we developed a quantitative SSNMR (qSSNMR) predictive model for mixtures containing form 1, form 2, and the major excipients (e.g. croscarmellose sodium, microcrystalline cellulose, magnesium stearate, and hypromellose) used for the tablet formulations.
  • The results were compared to simulated models developed from linear combinations of SSNMR spectra of each solid form and the excipients within the formulation.  The SSNMR data clearly showed specificity of each polymorph and enabled reasonable reverse engineering of the commercial tablet formulations.  The model used to evaluate the OTC tablets had excellent linearity (R2>0.99) for a solid-state method.  It was found that the detection limits of the method were around 0.5% for both Ranitidine HCl form 1 and form 2.  These limits are similar to other experimental techniques such as X-ray powder diffraction (XRPD).  The experimental quantitation of ranitidine-HCl in the tablets was on average 95.70±3.17 % in agreement with our predicted model. The use of linear combinations of spectra was shown to: a) help identify the integral regions for the signals of interest; b) accurately predict whether the mixtures used for modeling have been made correctly; c) permit visual estimation of the detection limit for the components of mixtures; and d) guide the choice of methods used for statistical analyses.  This model also demonstrated that the exact target sample composition (e.g., tablet formulation) was not required to adequately develop a quantitative method.


  • ˜The purpose of this work was to develop a quantitative model using chemometrics for assessing the levels of Form 1 and Form 2 Ranitidine HCl in three different OTC tablet formulations.   Linear combinations of the pure components were also used to demonstrate the abilitiy to effectively predict detection limits with only a limited number of spectral acquisitions.

NMR Method

The following parameters were used for data collection:

  • 45 kHz spectral width
  • SPINAL-64 1H decoupling at 86.2 kHz
  • RAMP-CP on 13C channel
  • ~1.5 hours total data acquisition time
  • 30 ms acquisition time
  • 6 ms contact time – optimized for API
  • 5 s pulse delay – optimized for API
  • 1000 scans – optimized for API
  • 2700 acquired points
  • zero filled to 32K points
  • 10 Hz exponential line broadening

The following instrument set up was used for data collection:

  • 9.4 T magnetic field
  • 13C frequency = 100.54 MHz
  • 1H frequency = 399.79 MHz
  • ambient temperature
  • Varian T3 HX narrow bore MAS probe
  • 4 mm zirconia rotors (~30-40 mg sample)
  • 12 kHz MAS
  • glycine was used as an external chemical shift reference


Results and Discussion

  • ˜The specificity using XRPD of the two polymorphic forms of Ranitidine HCl (form 1 and form 2) and the expected excipients in the drug product formulation are shown in Figure 4.  As can be seen, there is excellent specificity between Forms 1 and 2, however the excipient blend used in the formulation has a large amorphous halo which can make quantification extremely difficult.  As is seen below in Figure 5, there are no such issues with the 13C CP/MAS SSNMR spectra of these three components.

  • ˜The specificity using 13C CP/MAS SSNMR of the two polymorphic forms of Ranitidine HCl (form 1 and form 2) and the expected excipients in the drug product formulation are shown in Figure 5.  As can be seen, there are several regions of specificity that were used for developing the predictive model.  The regions used for developing the multiple linear regression (MLR) model are also shown in Figure 5.

  • ˜A series of Ranitidine HCL mixtures were prepared and analyzed.  A linear combination of the pure components of Ranitidine HCl form 1, form 2 and the excipient blend was also performed.  A comparison of the raw data versus the linear combinations are shown in Figure 6 and Figure 7.  The visual detection limit is consistent between the raw data and the linear combinations.



  • ˜A multiple linear regression (MLR) model was developed using the mixtures seen in Figure 6 and Figure 7.  The MLR model was used to predict the % Form 1 and %Form 2, the predicted versus actual % Form 1 and %Form 2 were plotted as shown in Figure 8.  The linearity of the model was excellent for a solid-state method (R2>0.99).

Results and Discussion



  • ˜The three OTC Ranitidine HCl tablet formulations above were predicted using the developed MLR model.  The results of the prediction are summarized in Table 2.

Next Steps

  • ˜Compare spectral deconvolution to simple integration of regions to determine whether a more accurate model with better DL and QL can be obtained.
  • Produce a validation data set to obtain accuracy and reproducibility.  This will complete most of the steps necessary for a validated quantitative SSNMR method according to ICH guidelines.
  • ˜Extend ruggedness and robustness testing to determine whether these are introducing significant error to the method.
  • ˜Test the method on other formulated ranitidine HCl tablets.
  • ˜Extend the quantitative procedure to several other model compounds that may be more difficult due to greater overlap of the resonances of each polymorph.
  • ˜Compare MLR method results with PLS or PCA results.


  • ˜SSNMR is a useful quantitative technique provided that the method is designed carefully.
  • Relative quantification of each component resonance within a SSNMR spectrum for a single sample is not necessary when a calibration curve is based on the signal response of the separate components.
  • ˜SSNMR visual detection limits are less than 1% for ranitidine HCl, but errors accumulated due to mixture inhomogeneity and signal overlap in calibration spectra.  These errors need to be lower to get statistically low detection and quantitative limits.
  • ˜The exact target sample composition (e.g., tablet formulation) is not required to adequately develop a quantitative method.


Relative Quantification – optimize relaxation parameters

R.K. Harris,  Analyst 110, 649-655 (1985)

G.R. Hatfield, Anal. Chem. 59, 172-179 (1987)

P. Gao, Pharm. Res. 13, 1095-1104 (1996)

G.A. Stephenson, et al., Adv. Drug Del. Rev. 48, 67-90, 2001

D.C. Apperley, et al., J. Pharm. Sci. 92, 2496-2503 (2003)

R. Fu, et al., J. Magn. Reson. 168, 8-17 (2004)

*T.J. Offerdahl, et al., J. Pharm. Sci. 94, 2591-2605 (2005)

G. Hou, Chem. Phys. Lett. 421, 356-360 (2006) – 13C labeling required

Absolute Quantification – calibration curve or internal standard

Y. Tozuka, et al., Chem. Phar. Bull. 50, 1128-1130 (2002)

*R. Lefort, et al., Int. J. Pharm. 280, 209-219 (2004)

R.K. Harris, et al., J. Pharm. Biomed. Anal. 38, 858-864 (2005)

B.T. Farrer, et al., J. Pharm Sci. 96, 264-267 (2007)

S. Sanchez, et al., J. Pharm Biomed. Anal. 47, 683-687 (2008)

J. Liu, et al., Drug Dev. Ind. Pharm. 35, 969-975 (2009)

R. Suryanarayanan & T.S. Wiedmann, Pharm. Res. 7, 184-187 (1990)

Y.S. Avadhut et al., J. Magn. Reson. 201, 1-6 (2009)

Absolute Quantification – ERETIC signal and sample restriction in rotor

F. Ziarelli & S. Caldarelli, SSNMR 29, 214-218 (2006)

F. Ziarelli et al., J. Magn. Reson. 188, 260-266 (2007)


  • ˜David Engers, Valeriya Smolenskaya and Yonglai Yang for the solid form selection and formulation.
  • Don Hallenbeck for the MLR model development.