FTIR amide I analysis of lyophilized peptide solids: what the band shape reports and where interpretation breaks down

Most conformational methods for peptides were developed for solutions. Circular dichroism and solution NMR report on the molecule as it exists in a buffer, which is informative for behavior after reconstitution but says little about the solid that was actually stored. Fourier-transform infrared (FTIR) spectroscopy is one of the few techniques that can examine the freeze-dried cake directly, with minimal sample preparation and without dissolving the material. For that reason it has become a standard tool in the characterization of lyophilized protein and peptide formulations.

This post reviews what the amide I band measures, how spectra from lyophilized peptides are typically processed, and which conclusions are reasonably supported by the data. The emphasis is on interpretive limits, since infrared band shapes are easily over-read, particularly for short peptides.

The physical basis of the amide I band

The amide I absorption, centered near 1650 cm⁻¹, arises predominantly from the C=O stretching vibration of the peptide backbone, with smaller contributions from C–N stretching and N–H bending. Because the carbonyl oscillator is sensitive to its hydrogen-bonding environment and to transition dipole coupling between neighboring amide groups, the exact frequency depends on the local backbone geometry. This is the origin of the long-standing correlation between band position and secondary structure.

Commonly cited assignments for hydrated or solid-state proteins place α-helical contributions near 1650 to 1658 cm⁻¹, extended β-sheet structure in a lower-frequency component near 1620 to 1640 cm⁻¹ (with a weaker high-frequency component near 1680 to 1695 cm⁻¹ for antiparallel arrangements), turns in the 1660 to 1685 cm⁻¹ region, and disordered or irregular structure around 1640 to 1650 cm⁻¹. These ranges overlap, and published tabulations differ by several wavenumbers depending on sample state, hydrogen-bond strength, and the reference set used. They are best treated as guides to relative change rather than as fixed fingerprints.

A second absorption, the amide II band near 1550 cm⁻¹, arises mainly from N–H bending coupled to C–N stretching. It is used less often for conformation but is the basis of hydrogen-deuterium exchange experiments in the infrared, where the shift of amide II after exposure to D₂O reports on solvent accessibility.

Why the solid state changes the picture

In aqueous solution, the strong bending absorption of water near 1640 cm⁻¹ overlaps the amide I region and complicates measurement. Samples are typically prepared in D₂O, which moves the interfering band to approximately 1210 cm⁻¹ and shifts amide I slightly, producing what is usually labeled amide I′. A lyophilized solid avoids this problem because residual water is low, which is one practical attraction of the method.

The solid state introduces its own considerations. Removal of water eliminates the hydrogen bonds that the peptide formed with solvent. Research on freeze-dried proteins has characterized this as a mechanism of dehydration-induced structural perturbation: where stabilizers such as sugars or polyols are absent, backbone carbonyls and amides lose their water partners, and the observed spectrum may broaden or shift toward a less native-like profile. When a suitable amorphous excipient is present, it is thought to substitute for water as a hydrogen-bond partner, and spectra of the dried solid have been reported to resemble those of the solution state more closely. The infrared spectrum therefore reports on the combined effect of the peptide’s intrinsic propensity and the matrix around it, which makes comparison across different formulations difficult unless the composition is held constant.

Short peptides add a further complication. Many sequences of fewer than twenty residues have little stable secondary structure in solution, and their amide I envelope is dominated by a broad, featureless contour. Interpreting a shoulder in such a spectrum as evidence of helix or sheet content requires caution, since the same band can arise from a distribution of conformations averaged across the sample.

Data processing: second derivatives and curve fitting

The amide I envelope is broad because the underlying component bands are separated by only a few wavenumbers, which is narrower than their intrinsic width. Two resolution-enhancement approaches are routinely applied: derivatization and Fourier self-deconvolution. Second-derivative spectra are the more common choice, since the minima of the second derivative coincide with the positions of the component bands in the original spectrum and the transformation is straightforward to reproduce.

Several practical points govern whether a second-derivative spectrum is reliable. Smoothing must be sufficient to suppress detector noise, which the derivative amplifies, but not so aggressive that it merges adjacent components; a common approach is a Savitzky-Golay filter with a window chosen to be small relative to the band width. Water vapor, which has sharp rotational lines throughout the 1800 to 1500 cm⁻¹ range, must be removed by purging the spectrometer with dry gas or by subtraction of a vapor reference, because even small residual lines produce spurious derivative features. Baseline correction and area normalization over the amide I range are generally applied before comparing samples.

Some laboratories go further and fit the original spectrum with Gaussian or Lorentzian components placed at the derivative minima, then assign fractional areas to structural classes. This is where caution is most needed. The fit is under-determined, the number of components is a user choice, and the resulting percentages are sensitive to starting parameters. Fractional areas calibrated against crystallographic structures work reasonably for well-folded globular proteins, but a quantitative secondary-structure estimate for a short peptide is not supported by the same calibration. For this reason, comparative use, such as overlaying spectra of two lots or of one lot before and after storage, is better justified than absolute assignment.

Applications that research has characterized

Where FTIR is most defensible is in detecting change. Several uses appear in the literature on peptide and protein solids:

Comparing lots or process conditions. If two preparations of the same sequence and formulation are freeze-dried under different conditions, differences in second-derivative profile can indicate that the solid-state conformational distribution differs. This is a physical comparability signal and does not by itself establish any functional difference.
Tracking stability-related changes. Growth of a low-frequency component near 1620 cm⁻¹ with a weaker band near 1690 cm⁻¹ has been associated with intermolecular β-sheet formation, which is a recognized signature of aggregation. Because this feature can be seen in the solid without reconstitution, it is a useful complement to size-based assays, which require dissolution and may themselves alter the aggregate population.
Assessing excipient effects. The degree to which a stabilizer preserves the spectral profile of the starting material through drying has been used as a screening metric in formulation studies, often correlated with long-term storage outcomes.
Distinguishing amorphous from ordered peptide. Sharpening of bands and splitting of features can accompany crystallization of the solid, and the method is sometimes paired with X-ray powder diffraction for this question.

Sampling modes and reproducibility

The way a solid is presented to the instrument affects the data. Potassium bromide pellet preparation requires grinding and pressing, and the mechanical stress and the pressure can themselves perturb conformation or cause moisture uptake from the hygroscopic matrix. Attenuated total reflectance (ATR), in which the powder is pressed against a crystal and the evanescent wave samples the surface layer, avoids dilution in a salt matrix and is faster, but contact pressure and penetration depth vary with wavenumber and with how uniformly the powder is packed. Because penetration depth scales with wavelength, ATR spectra differ systematically in relative intensity from transmission spectra and are usually corrected before comparison.

Whichever mode is used, replicate sampling from more than one position in the vial is advisable, since moisture and composition can vary through the cake. Reporting the instrument resolution (typically 2 to 4 cm⁻¹), number of co-added scans, purge conditions, and processing parameters is necessary for anyone to reproduce the analysis. A shift of one or two wavenumbers is meaningful only if the instrument’s wavenumber accuracy and the replicate variability have been documented.

Limits worth stating plainly

Several constraints bound what an amide I spectrum can establish. Band overlap means that no single frequency is diagnostic of one structure. Side-chain absorptions from residues such as arginine, asparagine, glutamine, and tyrosine contribute in the 1600 to 1700 cm⁻¹ window, and for short peptides rich in these residues the side-chain signal can be a sizable fraction of the total. Counterions are another consideration: trifluoroacetate has a strong carbonyl absorption near 1670 to 1680 cm⁻¹ that overlaps the turn region and can distort the apparent profile of a peptide supplied as a TFA salt. This is a particular reason to record the salt form, and where feasible to compare against an acetate or hydrochloride preparation, before attributing a feature near 1675 cm⁻¹ to backbone structure.

Finally, a conformational change observed in the solid does not necessarily predict behavior after reconstitution, and the reverse also holds. The technique characterizes the dried matrix; correlation with solution-state properties is an empirical question that has to be examined for each system.

Synthesis

FTIR spectroscopy of the amide I region offers a direct, low-preparation view of peptide conformation in the lyophilized state, and its resolution-enhanced forms are well suited to detecting relative change between lots, formulations, or storage time points. Its quantitative secondary-structure estimates are less trustworthy for short sequences, and its spectra are shaped by counterions, side chains, residual water, and sampling mode as much as by the backbone. Read as one orthogonal line of evidence alongside circular dichroism, X-ray powder diffraction, and size-based purity assays, it contributes to a physical description of the solid. Read alone and in absolute terms, it tends to claim more than the data can bear.