Undergraduate CHEM501 ⭐ 3 Credits Semester 7

Instrumental Analysis: Modern Spectroscopic and Separation Methods

Advanced treatment of instrumental analytical methods with emphasis on the physical principles and practical applications of modern instrumentation. Covers UV-Vis, fluorescence, IR/Raman, atomic spectroscopy, mass spectrometry, NMR, electroanalytical methods, and chromatographic detectors. Mass spectrometry of peptides and proteins receives particular depth.

📅 Studied by Skipper on 2026-10-02

📖 Core Concepts

Claude Opus 5.5

Core Concepts: Instrumental Analysis, Modern Spectroscopic and Separation Methods

Study notes for Skipper, CHEM501. Primary references: Skoog, Holler & Crouch, Principles of Instrumental Analysis, 7th ed. (2018); Gross, Mass Spectrometry: A Textbook, 3rd ed. (2017).

0. What Is This Subject Really About?

Every course you have taken so far has been about what molecules are and what they do. Atomic orbitals, bonding, thermodynamics, enzyme kinetics, gene regulation and immune recognition all describe the molecular world. Instrumental analysis asks a different question: how do we know? How does a property of a molecule (its mass, its electronic transitions, the precession frequency of a nucleus inside it) become a number on a screen that we are willing to trust?

The deepest way to see this course is as the study of information transduction. Every instrument, from a $500 colorimeter to a $1.5 million Orbitrap, is a chain of the same four links [Skoog, Holler & Crouch, 2018]:

  1. A probe or energy source perturbs the sample. Examples: photons, electrons, a radiofrequency pulse, an electric field, heat.
  2. The sample responds. It absorbs, emits, scatters, ionizes or precesses.
  3. A selector or transducer sorts the response along some axis and converts it to an electrical signal. The axis might be wavelength, m/z, time or frequency.
  4. A signal processor separates the part of that signal carrying information from the part carrying noise.

Notice the recurring move in link 3: dispersion along an axis. A monochromator spreads light by wavelength. A chromatography column spreads molecules by retention time. A mass analyzer spreads ions by m/z. An NMR experiment spreads nuclei by frequency. Fourier transforms (which you met in MATH102 and PCHEM II) let us read many positions on that axis at once. Once you see that UV-Vis, chromatography, MS and NMR are all "separate along an axis, then count," the whole subject compresses into a handful of ideas. The same trade-offs appear everywhere: resolution vs. sensitivity vs. speed vs. selectivity.

Key framing: An instrument does not measure a molecule. It measures an electrical current that we have engineered to depend, as cleanly as possible, on one molecular property. All the cleverness of instrument design lies in that phrase as cleanly as possible.

This course also has a destination, and it is worth naming now. By the end you should be able to follow, step by step, how a cell lysate becomes a list of thousands of quantified proteins with modification sites. Along the way you should also be able to determine the structure of a small natural product from a stack of 2D NMR spectra. Nearly every concept below feeds into those two capabilities.

1. Signal, Noise, and the Statistics of Seeing

Before any spectroscopy, we need to understand what limits every measurement. A signal is only meaningful relative to its uncertainty. The figure of merit is the signal-to-noise ratio, S/N = (mean signal)/(standard deviation of signal). A widely used rule of thumb puts the limit of detection near S/N ≈ 3.

1.1 The fundamental noise sources

Some noise is not a sign of bad engineering. It comes from physics itself:

  • Thermal (Johnson–Nyquist) noise comes from the random thermal motion of charge carriers in any resistor. It is "white," meaning it is equal at all frequencies.
    v_rms = √(4 k_B T R Δf)
    
    k_B = Boltzmann constant, T = temperature, R = resistance,
    Δf = measurement bandwidth
    
    You know k_BT from statistical mechanics (CHEM301). This is equipartition showing up in an electronic circuit. Cooling detectors, as with CCDs and IR detectors, attacks this term directly.
  • Shot noise arises because current is made of discrete charges (or photons) arriving randomly. This is a Poisson process (MATH201), so the variance equals the mean.
    i_rms = √(2 I e Δf)
    
    The consequence is subtle and important. Shot-noise-limited S/N grows as √(signal). Collecting four times more photons only doubles your S/N.
  • Flicker (1/f) noise has a power that rises as frequency falls. Lamp drift, slow temperature changes and electrode surface changes all live here. It dominates below roughly 100 Hz.
  • Environmental noise includes 60/50 Hz mains pickup, vibration and radio stations. These are peaks at specific frequencies.

1.2 Signal averaging: the √n law

If the signal is reproducible and the noise is random and uncorrelated between scans, then summing n scans adds signal linearly. The noise, however, adds as a random walk, growing as √n:

(S/N)_n = √n · (S/N)_1

This is the central limit theorem at work. It explains why an overnight ¹³C NMR experiment of 10,000 scans gives only about 100× better S/N than one scan. The square-root law is brutally unforgiving. To gain another factor of 10, you need 100× more time. Many design choices in instrumentation (higher magnetic fields, cryoprobes, multiplex advantages) exist to avoid paying this tax.

1.3 Lock-in amplification: moving the signal to a quiet neighborhood

Here is a beautiful idea. Since 1/f noise is worst at low frequency, and many analytical signals are essentially DC, we can modulate the signal at a reference frequency f_ref. A mechanical chopper on a light beam is one example. The signal now lives at f_ref (say, 1 kHz), where the noise floor is much lower.

A lock-in amplifier then multiplies the incoming signal by a reference wave at exactly f_ref and low-pass filters the product. Only components coherent with the reference survive. Mathematically, it computes one Fourier coefficient with a very narrow effective bandwidth Δf. Because thermal and shot noise both scale with √Δf, narrowing the bandwidth suppresses them.

Analogy: Imagine trying to hear one friend in a noisy stadium. Shouting louder (bigger signal) helps a little. Agreeing beforehand that your friend will clap in a specific rhythm, and listening only for that rhythm, helps enormously. That rhythm is the lock-in reference.

Connection forward: The same logic underlies Fourier-transform instruments (FTIR, FT-ICR, Orbitrap, FT-NMR), where the signal is deliberately encoded as frequencies.

2. UV-Vis Absorption Spectrophotometry: Instrumentation and Its Limits

2.1 Beer's law and what it assumes

A = −log T = log(P₀/P) = ε b c

ε = molar absorptivity (M⁻¹ cm⁻¹), b = path length (cm),
c = concentration (M)

Beer's law is a statement that absorption events are independent. Each thin slice of solution removes the same fraction of light, which integrates to an exponential (a first-order ODE, MATH102). It assumes several things:

  • The light is monochromatic.
  • Absorbers do not interact. This fails at high concentration.
  • There is no scattering or fluorescence reaching the detector.
  • There is no stray light.

Stray light is the instrumentally important one:

A_apparent = log[(P₀ + P_s) / (P + P_s)]

As the sample becomes very absorbing (P → 0), A_apparent plateaus at log[(P₀+P_s)/P_s]. With 0.1% stray light, you can never read above A ≈ 3. This is why your Gen Chem lab told you to keep absorbances below about 1.5. The reason is the instrument, not the chemistry.

2.2 Monochromators: diffraction gratings

A reflection grating disperses light according to the grating equation:

n λ = d (sin i ± sin r)

n = diffraction order, d = groove spacing,
i, r = angles of incidence and reflection

The resolving power of a grating is R = λ/Δλ = nN, where N is the number of illuminated grooves. The slit width sets the trade-off. A narrow slit gives better spectral resolution but less light, and therefore worse S/N. This throughput-resolution tension recurs throughout the course. You will meet it again as the Jacquinot advantage in FTIR, as isolation width in MS/MS, and as slit width in AAS.

2.3 Detectors: one channel or many?

  • Photomultiplier tube (PMT): A photon strikes a photocathode and ejects an electron through the photoelectric effect (PHYS202). A cascade of 9–12 dynodes then multiplies each electron, giving gains of about 10⁶–10⁷. The PMT is single-channel, extraordinarily sensitive, and fast. It is ideal for low-light work such as fluorescence or scanning instruments.
  • Photodiode array (PDA) and charge-coupled device (CCD): These are arrays of silicon pixels placed at the focal plane of a fixed grating, so every wavelength is recorded simultaneously. A diode-array spectrophotometer captures a whole spectrum in milliseconds. That speed is what makes it the standard HPLC detector. CCDs, especially when cooled, integrate charge with very low read noise, which makes them excellent for Raman and astronomy-grade low-light imaging.

This is the multichannel advantage. If a scanning instrument spends time t/M on each of M wavelength elements, an array detector spends the full time t on all of them. When noise is detector-limited, that gives a √M improvement in S/N for the same total time.

2.4 Multicomponent analysis: linear algebra in a cuvette

Absorbances are additive. For a mixture of k components measured at m wavelengths:

A(λᵢ) = b Σⱼ εᵢⱼ cⱼ       →   in matrix form:   A = E c

Least-squares solution (m > k, overdetermined):
c = (Eᵀ E)⁻¹ Eᵀ A

This is the normal equation from MATH301, applied directly. Your linear algebra intuition also tells you when this fails. If two components have nearly identical spectra, the columns of E are nearly collinear and EᵀE is ill-conditioned (large condition number). Small noise in A then becomes huge errors in c. The chemistry problem "these two dyes overlap too much" and the math problem "E is nearly singular" are the same statement.

2.5 Derivative spectroscopy

Taking dA/dλ or d²A/dλ² sharpens overlapping bands. The second derivative of a Gaussian band has a narrow negative lobe at λ_max, with a width of about 1/3 the original. It also removes constant and linear baselines, such as scattering backgrounds. The cost is that differentiation amplifies high-frequency noise, because in Fourier space differentiation multiplies by frequency. So derivative spectra almost always require smoothing first (for example, Savitzky–Golay). There is no free lunch: you trade S/N for apparent resolution.

3. Fluorescence: Counting Photons Against a Dark Background

3.1 Why fluorescence is so sensitive

Absorbance measures a small difference between two large numbers, P₀ and P. Fluorescence measures emitted light against a nearly black background, typically detected at 90° to the excitation beam. Measuring a small signal against zero is much easier than measuring a small difference between large signals. That is why fluorescence detection limits are often 10³–10⁶ times lower than absorbance. Single-molecule detection is routine.

3.2 Excitation and emission spectra

  • An emission spectrum holds the excitation wavelength fixed and scans emission.
  • An excitation spectrum holds emission fixed and scans excitation. For a pure fluorophore it resembles the absorption spectrum.

Emission is red-shifted from absorption (the Stokes shift). Two things cause it. First, vibrational relaxation in S₁ is faster than emission. Second, solvent reorganization around the excited-state dipole lowers its energy. Kasha's rule says that emission comes from the lowest vibrational level of S₁ regardless of which state was excited. As a result, the emission spectrum shape is independent of excitation wavelength. The mirror-image relationship between absorption and emission reflects similar vibrational spacings in S₀ and S₁, which is a Franck–Condon consequence you derived in CHEM303.

3.3 Quantum yield and lifetime: two views of one competition

The excited state decays by competing first-order pathways: radiative (k_r) and nonradiative (k_nr, which includes internal conversion, intersystem crossing and quenching).

Quantum yield:          Φ  = k_r / (k_r + k_nr)
Fluorescence lifetime:  τ  = 1 / (k_r + k_nr)
Therefore:              Φ  = k_r · τ

This is first-order kinetics with parallel pathways (CHEM102), dressed up in photophysics. The lifetime is typically 1–10 ns for organic fluorophores. It is measured by time-correlated single-photon counting or by phase-modulation methods. Lifetime is an intensive property: it does not depend on concentration or excitation intensity. That makes it robust where intensity is not.

3.4 Quenching: Stern–Volmer analysis

F₀ / F = 1 + K_SV [Q] = 1 + k_q τ₀ [Q]

k_q = bimolecular quenching rate constant
(diffusion limit ≈ 10¹⁰ M⁻¹ s⁻¹)

Here is where lifetime becomes decisive. Quenching comes in two types:

  • Dynamic (collisional) quenching deactivates the excited state. It shortens τ, so τ₀/τ = F₀/F.
  • Static quenching forms a non-fluorescent ground-state complex. It removes fluorophores from the population but leaves the survivors' lifetime unchanged, so τ₀/τ = 1.

Intensity measurements alone cannot tell these apart. Lifetime measurements can. Temperature also discriminates: dynamic quenching usually increases with T (faster diffusion), while static complexes usually dissociate.

3.5 Fluorescence anisotropy: measuring how fast molecules tumble

Excite with vertically polarized light and you preferentially excite molecules whose absorption dipoles are vertical (photoselection). If the molecule rotates before emitting, emission depolarizes.

r = (I_∥ − G·I_⊥) / (I_∥ + 2G·I_⊥)

Perrin equation:     r₀ / r = 1 + τ / θ
Rotational correlation time (Stokes–Einstein–Debye):  θ = η V / (R T)

The fundamental anisotropy r₀ has a maximum of 0.4 for collinear absorption and emission dipoles under one-photon excitation. G is an instrument correction factor.

The beautiful part: a small fluorescent peptide (θ ≈ 0.1–1 ns, τ ≈ 4 ns) has low anisotropy. Bind it to a 50 kDa antibody and θ jumps to tens of ns, so anisotropy rises sharply. This gives a homogeneous, wash-free binding assay. You can measure K_d values for peptide–protein interactions in a microplate without separating bound from free (BIOL401 and CHEM401 connections). The fluorescence lifetime acts as a built-in stopwatch, and anisotropy reports how far the molecule rotated during that interval.

4. Vibrational Spectroscopy: IR, FTIR, ATR, Raman, SERS

4.1 Two selection rules, two windows on the same vibrations

  • IR absorption requires a change in dipole moment during the vibration: (∂μ/∂Q) ≠ 0.
  • Raman scattering requires a change in polarizability: (∂α/∂Q) ≠ 0.

For molecules with a center of symmetry, the rule of mutual exclusion holds: no mode is both IR- and Raman-active. CO₂ is the classic example. Its symmetric stretch is Raman-active only, while its asymmetric stretch and bend are IR-active only. You derived these group-theoretic rules in CHEM303. Here they become practical, because the two techniques are complementary, not redundant.

The practical payoff is that water is a strong IR absorber (O–H stretch, H–O–H bend near 1640 cm⁻¹) but a weak Raman scatterer. Raman is therefore often preferred for aqueous biological samples. The 1640 cm⁻¹ water band sits directly on top of the protein amide I band in IR, a recurring headache. It is often addressed by using D₂O.

4.2 FTIR: why the interferometer won

A Michelson interferometer splits the beam, sends the two halves along paths differing by δ (the optical retardation), and recombines them. The detector records an interferogram:

I(δ) = ∫₀^∞ B(ν̃) [1 + cos(2π ν̃ δ)] dν̃

Spectrum recovered by Fourier transform:
B(ν̃) ∝ ∫ [I(δ) − ½I(0)] cos(2π ν̃ δ) dδ

Resolution:   Δν̃ ≈ 1 / δ_max
(e.g., 1 cm mirror retardation → ~1 cm⁻¹ resolution)

FTIR beat dispersive IR for three reasons [Skoog, Holler & Crouch, 2018]:

  1. Fellgett (multiplex) advantage. All wavelengths hit the detector at all times. IR detectors are noisy, with noise independent of signal level, so this gives roughly a √M S/N gain.
  2. Jacquinot (throughput) advantage. There are no slits, so far more light reaches the detector.
  3. Connes advantage. A HeNe reference laser calibrates mirror position, giving very high wavenumber accuracy.

4.3 ATR-FTIR: spectroscopy with an evanescent wave

In attenuated total reflectance (ATR), IR light travels inside a high-index crystal (diamond, ZnSe or Ge) and undergoes total internal reflection at the surface where the sample sits. Even under total reflection, an evanescent wave leaks a short distance into the sample and decays exponentially:

d_p = λ / [ 2π n₁ √( sin²θ − (n₂/n₁)² ) ]

n₁ = crystal index, n₂ = sample index, θ = angle of incidence

The penetration depth is typically 0.5–2 μm. This is quantum tunneling's classical cousin (PHYS202: frustrated total internal reflection). Because the effective path length is so short, you can press a drop of aqueous protein, a polymer film or a pill directly onto the crystal with no sample preparation. One subtlety: d_p scales with λ, so ATR spectra show relatively stronger bands at long wavelength (low wavenumber) than transmission spectra.

4.4 Raman spectroscopy

Raman is inelastic light scattering. A photon of frequency ν₀ interacts with the molecule's polarizability through a short-lived virtual state and emerges with frequency ν₀ − ν_vib (Stokes) or ν₀ + ν_vib (anti-Stokes). Anti-Stokes lines are weaker because they require a vibrationally excited initial population, which is a Boltzmann factor (CHEM301). The Stokes/anti-Stokes ratio can therefore serve as a thermometer.

Raman is intrinsically weak: only about 1 in 10⁶–10⁸ scattered photons is Raman-shifted. Scattering intensity scales roughly as (ν₀ − ν_vib)⁴, which favors short-wavelength lasers. Short wavelengths, however, also excite fluorescence that can swamp the Raman signal, so near-IR lasers (785, 1064 nm) are a common compromise.

4.5 SERS: the surprise of the 1970s

Molecules adsorbed on roughened silver or gold surfaces show Raman signals enhanced by factors of 10⁶ or more. In some "hot spots" between nanoparticles, the enhancement is large enough for single-molecule detection. The effect was first observed by Fleischmann and colleagues (1974) and recognized as an anomalous enhancement by Jeanmaire and Van Duyne (1977).

The dominant mechanism is electromagnetic. Incident light drives collective electron oscillations (localized surface plasmons) in the metal nanostructure, concentrating the local electric field. Because both the incoming and the scattered fields are enhanced, the signal scales approximately as |E_local/E₀|⁴. A tenfold field enhancement therefore gives about a 10⁴ signal enhancement. A smaller "chemical" contribution comes from charge-transfer interactions.

5. Atomic Spectroscopy: AAS, ICP-OES, ICP-MS

5.1 The crux is atomization

Atomic spectra consist of extremely narrow lines, because free atoms have no vibrational or rotational sublevels. But the sample arrives as a solution of salts, so it must be desolvated, vaporized and atomized. Every interference in atomic spectroscopy ultimately traces back to the imperfection of this step.

5.2 The Boltzmann distribution decides absorption vs. emission

N* / N₀ = (g* / g₀) · exp(−ΔE / k_B T)

For the Na 3p ← 3s transition (589 nm) at 2500 K, N*/N₀ ≈ 1.7 × 10⁻⁴. Raising T by just 10 K raises this ratio by about 4% [Skoog, Holler & Crouch, 2018]. The consequences:

  • Absorption depends on N₀, which is essentially all the atoms and is insensitive to small temperature fluctuations.
  • Emission depends on N*, which is tiny and exponentially sensitive to T.

This explains why flame emission is temperature-fragile. It also explains why ICP-OES needs a very hot, very stable plasma (6000–10,000 K). At those temperatures many elements are substantially excited and ionized.

5.3 AAS: why you need a hollow cathode lamp

Atomic absorption lines in a flame are only about 0.002–0.005 nm wide. A good monochromator has a bandpass of about 0.2 nm. If you used a continuum source, the atoms would absorb a sliver of a broad band and the absorbance would be unmeasurably small, with Beer's law failing badly.

The solution is a line source, the hollow cathode lamp. It is made of the analyte element and emits that element's lines at low pressure and temperature, so its lines are narrower than the absorption lines. The analyte atoms become their own wavelength selector. Elegant, but it requires one lamp per element.

  • Flame AAS: continuous aspiration, residence time of milliseconds, ppm detection.
  • Graphite furnace (electrothermal) AAS: microliter samples, all atomized at once into a confined tube, residence time of about a second, ppb or lower detection.

5.4 ICP-OES and ICP-MS

An inductively coupled argon plasma is sustained by a radiofrequency coil. Its high temperature and long residence time largely eliminate chemical interferences and ionize most elements efficiently. Most elements have first ionization energies below argon's 15.76 eV.

  • In ICP-OES, you read the emission of many elements simultaneously with an echelle spectrometer and CCD.
  • In ICP-MS, you extract the ions through sampler and skimmer cones into a mass spectrometer, usually a quadrupole. Detection limits reach parts per trillion, and isotope ratios become measurable.

5.5 Interferences and matrix effects

  • Spectral interferences: overlapping lines, or broadband absorption and scattering from molecular species or particles. Corrections include deuterium-lamp background correction and Zeeman correction.
  • Chemical interferences: phosphate binds Ca²⁺ into refractory species in a flame. A "releasing agent" (La³⁺, which grabs the phosphate) or a protecting agent (EDTA) fixes this.
  • Ionization interferences: easily ionized elements such as K and Na shift the M ⇌ M⁺ + e⁻ equilibrium in the flame (Le Chatelier, CHEM102). Adding excess K as an "ionization suppressor" deliberately floods the flame with electrons.
  • Isobaric and polyatomic interferences in ICP-MS: ⁴⁰Ar³⁵Cl⁺ (m/z 75) on ⁷⁵As⁺, ⁴⁰Ar¹⁶O⁺ on ⁵⁶Fe⁺, and ⁴⁰Ar₂⁺ on ⁸⁰Se⁺. Collision and reaction cells remove polyatomic ions. Kinetic energy discrimination exploits the larger collision cross-section of polyatomics, and reactive gases convert interferents. High-resolution sector instruments can separate them by mass defect.
  • Matrix effects: the sample's overall composition changes viscosity, nebulization and plasma loading. The standard remedies, which recur everywhere in this course, are the method of standard additions and internal standardization. You met both in CHEM302.

6. Mass Spectrometry Fundamentals: m/z, Resolution, Accuracy, Isotopes

6.1 What a mass spectrometer actually measures

A mass spectrometer does not measure mass. It measures the mass-to-charge ratio, m/z, of gas-phase ions, inferred from how they move in electric and magnetic fields [Gross, 2017]. The Lorentz force contains q/m, not m, so ions with the same m/z are indistinguishable. A protein of 20,000 Da carrying 20 protons appears near m/z 1001. This is the single most important fact for interpreting ESI spectra.

F = q (E + v × B)       and       F = m a

⇒ trajectory depends on q/m only

6.2 Resolution vs. mass accuracy: two different virtues

Resolving power:    R = m / Δm
(Δm usually = full width at half maximum, FWHM)

Mass accuracy:      error (ppm) = [(m_measured − m_theoretical) / m_theoretical] × 10⁶
  • Resolution is about separating neighbors. Can you tell two peaks apart?
  • Accuracy is about where a peak's center lies relative to truth.

Resolution helps accuracy, because an unresolved interferent will shift the centroid. Still, a well-calibrated TOF can be accurate to 2 ppm, while a high-resolution instrument with poor calibration can be off by 20 ppm.

Why accuracy matters: at 1 ppm, a mass of 500 Da is known to ±0.0005 Da. This often narrows the elemental composition to a single candidate, because each element has a characteristic mass defect. With ¹²C defined as exactly 12.0000 Da, ¹H = 1.00783, ¹⁴N = 14.00307, ¹⁶O = 15.99491 and ³²S = 31.97207. Nuclear binding energy (E = mc², PHYS202) makes every nuclide's mass slightly off-integer, and those decimals encode composition.

6.3 Monoisotopic, average, and isotope patterns

¹³C has a natural abundance of about 1.07%, and ¹³C − ¹²C = 1.00336 Da. For a molecule with n carbons, the number of ¹³C atoms follows a binomial distribution (MATH201), which is well approximated by a Poisson distribution with mean λ = 0.0107n:

P(k heavy atoms) ≈ e^(−λ) λᵏ / k!        λ ≈ 0.0107 × n_C (+ smaller N, H, O, S terms)

Relative M+1 / M  ≈ 1.1% × n_C
  • The monoisotopic mass uses only the lightest isotopes.
  • The average mass weights all isotopes, as on the periodic table.

For peptides, the "averagine" model gives about 4.94 C per 111 Da. The monoisotopic peak and the M+1 peak become roughly equal near 1,800 Da. Above that, the monoisotopic peak is not the tallest. For an intact 50 kDa protein, the monoisotopic species is vanishingly rare. Halogens give unmistakable patterns: Cl gives M/M+2 ≈ 3:1, and Br gives M/M+2 ≈ 1:1.

Reading charge from isotope spacing: adjacent isotope peaks are separated by 1.00336/z in m/z. Spacing of 0.5 means z = 2, and spacing of 0.33 means z = 3. This simple trick is how a high-resolution instrument assigns charge states on the fly.

7. Ionization: Getting Molecules Into the Gas Phase as Ions

Ionization is the gatekeeper of mass spectrometry. The central tension is between hard ionization, which deposits excess energy and gives informative but extensive fragmentation, and soft ionization, which gives intact molecular ions but little structural information.

7.1 Electron ionization (EI)

M + e⁻ (70 eV) → M⁺• + 2 e⁻

Gas-phase molecules are hit with 70 eV electrons, forming a radical cation M⁺•. Typical organic ionization energies are only 7–10 eV, so there is a large energy excess. M⁺• then fragments through unimolecular pathways that can be rationalized by quasi-equilibrium (RRKM-type) theory: α-cleavage, McLafferty rearrangement, and loss of stable neutrals [Gross, 2017].

Why exactly 70 eV? An electron's de Broglie wavelength is λ = h/√(2m_eE). At 70 eV this is about 1.5 Å, comparable to bond lengths, so energy transfer is efficient. The ionization efficiency curve also has a broad plateau here, so small variations in electron energy barely change the spectrum. That reproducibility across instruments and decades is precisely what makes spectral library searching possible. The NIST library contains hundreds of thousands of EI spectra. A wave-mechanics fact from CHEM303 turns out to underpin a global database.

7.2 Chemical ionization (CI)

A reagent gas (CH₄, NH₃ or isobutane) is ionized by EI. It forms reactive ions such as CH₅⁺, which then protonate the analyte:

CH₅⁺ + M → [M+H]⁺ + CH₄

The energy deposited is set by the difference in proton affinities, which is gas-phase acid–base chemistry (CHEM102 Brønsted ideas, without solvent). CH₅⁺ is a strong gas-phase acid and gives some fragmentation. NH₄⁺ is gentle and gives mostly [M+H]⁺ or [M+NH₄]⁺. CI is used when you need the molecular weight that EI destroyed.

7.3 Electrospray ionization (ESI)

Fenn and coworkers showed in 1989 that electrospray could gently transfer large, nonvolatile biomolecules from solution into the gas phase as multiply charged ions. This brought proteins of tens of kDa into the range of ordinary quadrupole analyzers [Fenn, Mann, Meng, Wong & Whitehouse, 1989]. John Fenn shared the 2002 Nobel Prize in Chemistry for this work. His Nobel lecture was titled "Electrospray Wings for Molecular Elephants."

The mechanism unfolds in stages:

  1. A high voltage (2–5 kV) is applied to a capillary. The liquid meniscus deforms into a Taylor cone, which emits a fine jet of charged droplets.
  2. Solvent evaporates, so droplets shrink while their charge stays constant. Charge density rises until Coulomb repulsion matches surface tension at the Rayleigh limit:
    q_R = 8π √(ε₀ γ r³)
    
    γ = surface tension, r = droplet radius
    
  3. The droplet then undergoes Coulomb fission, ejecting smaller, highly charged progeny droplets. This repeats.
  4. Final gas-phase ions form through one of two routes. In the charged residue model (Dole), the droplet evaporates to dryness around the analyte. This is favored for large globular proteins. In the ion evaporation model (Iribarne & Thomson), small ions are field-desorbed directly from the droplet surface. This is favored for small ions.

Charge-state deconvolution. A protein of mass M gives a series of [M+zH]^z+ peaks. For two adjacent peaks, m₁ (charge z, higher m/z) and m₂ (charge z+1, lower m/z):

m₁ = (M + z·1.00728) / z
m₂ = (M + (z+1)·1.00728) / (z+1)

⇒  z = (m₂ − 1.00728) / (m₁ − m₂)
⇒  M = z (m₁ − 1.00728)

Every peak in the envelope is an independent measurement of M. Averaging them improves accuracy, which is a pleasing echo of signal averaging. Fenn et al. emphasized exactly this advantage.

ESI is concentration-sensitive. Response depends more on analyte concentration than on total flow. Shrinking flow from mL/min to nL/min (nanoESI, nanoLC) therefore barely loses signal while consuming a thousandfold less sample. Smaller droplets also desolvate more efficiently. This is why proteomics runs at nanoflow.

Ion suppression. Droplet surface area and excess charge are limited resources. Co-eluting species with higher surface activity or basicity outcompete your analyte. This is the main matrix effect of LC-MS, and the reason isotope-labeled internal standards are the gold standard for quantification.

7.4 Matrix-assisted laser desorption/ionization (MALDI)

The analyte is co-crystallized with a large molar excess of a small UV-absorbing organic acid, the matrix. A pulsed UV laser hits the spot. The matrix absorbs the energy and explosively desorbs, carrying intact analyte into the gas phase. Proton transfer, largely from photoexcited or ionized matrix, ionizes the analyte. Karas and Hillenkamp reported in 1988 that this approach yields intact ions of proteins above 10,000 Da [Karas & Hillenkamp, 1988]. They used nicotinic acid with a 266 nm Nd:YAG laser. In the same year, Tanaka and coworkers independently achieved protein desorption using ultrafine cobalt powder in glycerol. Koichi Tanaka shared the 2002 Nobel Prize with Fenn and Wüthrich.

Citation note for Skipper: the DOI 10.1021/ac00171a028 resolves to Karas & Hillenkamp's paper in Analytical Chemistry 60, 2299–2301 (1988), published as "Laser desorption ionization of proteins with molecular masses exceeding 10,000 daltons." The phrase "matrix-assisted laser desorption/ionization" describes the technique this paper established, but it is not the paper's published title. Cite the published title in formal writing.

Key properties of MALDI:

  • It produces mainly singly charged ions, [M+H]⁺. Spectra are simple and well suited to mixtures and peptide mass fingerprinting.
  • It is pulsed, so it pairs naturally with TOF analyzers.
  • Modern matrices include α-cyano-4-hydroxycinnamic acid (CHCA) for peptides, sinapinic acid for proteins, and 2,5-dihydroxybenzoic acid (DHB) for glycans and general use.
  • It is relatively tolerant of salts and buffers compared with ESI.
  • It enables MALDI imaging: rastering the laser across a tissue section maps molecular distributions.

The deep unity of ESI and MALDI: Both solve the same problem. Large, polar, thermally fragile molecules decompose before they evaporate. Both solve it by never heating the analyte directly. ESI lets the solvent leave the analyte behind. MALDI lets the matrix carry the analyte away. In both cases a sacrificial medium absorbs the violence.

8. Mass Analyzers: Six Ways to Sort Ions

Each analyzer exploits a different physical consequence of the Lorentz force. Knowing the governing equation tells you each analyzer's strengths and limitations.

8.1 Quadrupole mass filter

Four parallel rods carry a DC potential U plus an RF potential V cos(Ωt). Ion motion in the x–y plane obeys the Mathieu equation, a second-order ODE with periodic coefficients. The dimensionless stability parameters are:

a = 8 z e U / (m r₀² Ω²)          q = 4 z e V / (m r₀² Ω²)

r₀ = field radius

Only ions whose (a, q) values fall inside the stability diagram have bounded trajectories and reach the detector. All others hit the rods. Scanning U and V together at a fixed U/V ratio sweeps a narrow stable window across m/z. The quadrupole is therefore a filter, not a disperser. It has modest resolution (unit resolution), but it is rugged, cheap and fast at switching, which makes it ideal for targeted quantification (§11).

8.2 Ion traps (3D Paul trap and linear trap)

The same Mathieu physics is used to store ions rather than pass them. Ions are accumulated, then ejected sequentially by m/z (mass-selective instability) by ramping the RF. Traps can isolate an ion, fragment it, isolate a fragment and fragment again (MSⁿ).

Their famous limitation is the low-mass cutoff. During resonance excitation for CID, fragments below about one-third of the precursor m/z have unstable trajectories and are lost (the "one-third rule"). This matters for isobaric tag reporter ions (§10).

8.3 Time-of-flight (TOF)

z e U = ½ m v²   ⇒   t = L √( m / (2 z e U) )

All ions get the same kinetic energy per charge, so lighter ions travel faster. TOF has no theoretical upper m/z limit, acquires a full spectrum every pulse, and pairs naturally with MALDI. Resolution is limited by the initial spread in kinetic energy.

The reflectron fixes this with an ion mirror. Faster ions penetrate deeper into the retarding field and take a longer path, so they arrive at the same time as slower ions of the same m/z. It is a neat example of energy focusing. The approach became practical with Mamyrin's reflectron design.

8.4 Magnetic (and double-focusing) sector

r = m v / (z e B)     with   ½ m v² = z e U

⇒   m/z = B² r² e / (2U)

A magnetic field bends ions by momentum. Adding an electrostatic sector that focuses by kinetic energy gives a double-focusing instrument with high resolution (R > 10,000). Sector instruments are historically central, and they remain important for isotope-ratio and high-resolution ICP-MS work.

8.5 Fourier-transform ion cyclotron resonance (FT-ICR)

ω_c = z e B / m          (f_c = z e B / 2π m)

Example: m/z 1000, B = 7 T  ⇒  f_c ≈ 107 kHz

Here is the surprising and beautiful fact: the cyclotron frequency is independent of the ion's velocity. A fast ion circles in a larger orbit with the same period. Ions are trapped in a superconducting magnet, excited coherently by an RF sweep, and their image currents on detector plates are recorded as a time-domain transient. A Fourier transform yields frequencies, hence m/z [Comisarow & Marshall, 1974].

Frequency is the quantity physicists can measure most precisely, which is why FT-ICR achieves the highest resolving power of any analyzer (10⁵–10⁷). Resolving power scales as R ∝ B · T_acq / (m/z). Bigger magnets and longer transients mean better resolution.

8.6 Orbitrap

Makarov's Orbitrap traps ions electrostatically, with no magnet. Ions orbit a central spindle electrode inside a barrel-like outer electrode, under a quadro-logarithmic potential. Their axial oscillation is harmonic, with a frequency that depends only on m/z [Makarov, 2000]:

ω_z = √( k · (q/m) )        k = field curvature constant

As in FT-ICR, image current is detected and Fourier transformed. Because ω ∝ (m/z)^(−1/2) rather than (m/z)^(−1), resolving power falls off more gently with m/z: R ∝ T_acq/√(m/z). Typical values are 60,000–500,000+ at m/z 200. Without a superconducting magnet, the Orbitrap brought FT-level resolution to ordinary laboratories, and it now dominates proteomics.

The unifying insight: TOF converts m/z to time. Quadrupoles and traps convert it to stability. Sectors convert it to position. FT-ICR and Orbitrap convert it to frequency. The axis changes, but the logic of "separate along an axis" from §0 holds everywhere.

9. Tandem Mass Spectrometry: Breaking Ions on Purpose

MS/MS proceeds in three steps. First, isolate a precursor ion by m/z. Second, activate it so it fragments. Third, analyze the fragments. The fragmentation pattern encodes structure. The method of activation determines which bonds break, and this is where physical chemistry directly shapes biological information.

9.1 Collision-induced dissociation (CID)

Precursor ions collide with an inert gas (He, N₂ or Ar). Each collision converts a little translational energy into internal energy. Many low-energy collisions heat the ion slowly. Energy randomizes over all vibrational modes before dissociation, which is the ergodic, RRKM picture, so the lowest-energy pathways win.

For protonated peptides, the mobile proton model explains the results. The ionizing proton migrates from a basic site to backbone amide nitrogens, weakening the adjacent C–N amide bond. Cleavage gives b and y ions. The drawback is that labile modifications cleave first. Phosphoserine and phosphothreonine readily lose H₃PO₄ (−98 Da), and glycans fall off. You learn that a phosphate was there but often not where.

9.2 Higher-energy collisional dissociation (HCD)

HCD is a beam-type collisional activation in a dedicated multipole cell, popularized on Orbitrap instruments [Olsen et al., 2007]. Energy deposition is faster and broader than resonance CID in a trap. There is no low-mass cutoff, so immonium ions and low-m/z isobaric tag reporter ions are observed. HCD spectra are dominated by y ions, and further fragmentation of b ions is common.

9.3 Electron-transfer dissociation (ETD)

A multiply protonated peptide receives an electron from a radical anion reagent, such as fluoranthene⁻•:

[M + nH]ⁿ⁺ + A⁻• → [M + nH]^(n−1)+•  → c + z• fragments

The resulting radical cation cleaves the backbone N–Cα bond through radical chemistry. This process is fast and non-ergodic, meaning it happens before energy randomizes. As a result, ETD is largely independent of sequence and preserves labile PTMs such as phosphorylation, glycosylation and other labile modifications. ETD was introduced by Syka, Hunt and coworkers [Syka et al., 2004], following the earlier electron-capture dissociation (ECD) in FT-ICR [Zubarev, Kelleher & McLafferty, 1998].

Limitations: ETD requires precursors with z ≥ 2, and preferably ≥ 3. It is also less efficient for low-charge tryptic peptides.

Key insight: CID and ETD are complementary because one is thermal (slow, statistical, weakest bond breaks) and the other is radical-driven (fast, non-statistical, cleaves where the electron lands). This is the kinetic-vs-thermodynamic-control idea from organic chemistry, appearing here as dissociation chemistry.

10. Peptide Mass Spectrometry: Reading Sequences From Fragments

10.1 Fragment-ion nomenclature

Following Roepstorff and Fohlman [Roepstorff & Fohlman, 1984], as modified by Biemann [Biemann, 1988], cleavage of each of the three backbone bonds gives an N-terminal fragment (a, b or c) and a complementary C-terminal fragment (x, y or z):

            a   b   c
            ↓   ↓   ↓
H₂N–CHR₁–C(=O)–NH–CHR₂–...–COOH
         Cα–C   C–N   N–Cα
            ↑   ↑   ↑
            x   y   z

Cα–C(O) bond   → a / x
C(O)–N amide   → b / y     (CID, HCD)
N–Cα bond      → c / z•    (ETD, ECD)

10.2 The arithmetic of sequencing

Let Σ denote the sum of residue masses, where residue = amino acid − H₂O. For singly charged ions:

b_i  = Σ(residues 1…i)        + 1.00728              (proton)
y_j  = Σ(last j residues)     + 18.01056 + 1.00728   (water + proton) = + 19.01784
a_i  = b_i − 27.99491                                (loss of CO)
c_i  = b_i + 17.02655                                (+ NH₃)
z•_j ≈ y_j − 16.01872

[M+H]⁺ = Σ(all residues) + 19.01784

Complementarity:  b_i + y_(n−i) = [M+H]⁺ + 1.00728

Selected monoisotopic residue masses (worth memorizing a few): Gly 57.021, Ala 71.037, Ser 87.032, Pro 97.053, Val 99.068, Thr 101.048, Leu/Ile 113.084, Asn 114.043, Asp 115.027, Gln 128.059, Lys 128.095, Glu 129.043, Met 131.040, His 137.059, Phe 147.068, Arg 156.101, Tyr 163.063, Trp 186.079.

10.3 De novo sequencing

In a clean spectrum, consecutive b ions (or y ions) differ by exactly one residue mass. Reading the ladder of differences reads the sequence. A y-series reads from the C-terminus, so it reverses relative to the b-series.

The traps are where understanding matters:

  • Leu and Ile are isobaric (both C₆H₁₁NO, 113.084). Ordinary CID cannot distinguish them.
  • Lys (128.095) vs. Gln (128.059) differ by 0.036 Da. That is about 280 ppm at the residue level, easily resolved by Orbitrap and FT-ICR but not by a unit-resolution trap.
  • Gly-Gly has exactly the same elemental composition as Asn (C₄H₆N₂O₂, 114.043). Ala-Gly matches Gln (C₅H₈N₂O₂, 128.059). If a fragment is missing, a dipeptide gap can masquerade as a single residue. No mass accuracy can resolve identical formulas. Only the presence of the intervening fragment can.
  • Deamidation (Asn→Asp, Gln→Glu) adds +0.984 Da, which can be confused with the ¹³C isotope peak (+1.003 Da) at modest resolution.

10.4 Database searching

Most identifications do not come from de novo reading. They come from matching each spectrum to theoretical spectra of peptides predicted from a protein database by in-silico digestion. Three classic engines:

  • SEQUEST scores candidates by cross-correlation (XCorr) between observed and theoretical spectra [Eng, McCormack & Yates, 1994]. Your Fourier and convolution background makes this natural: it is a sliding dot product.
  • Mascot uses a probability-based scoring approach (MOWSE-derived). It asks how likely the observed match is to occur by chance [Perkins, Pappin, Creasy & Cottrell, 1999].
  • MaxQuant is an integrated platform for high-resolution data, with its own search engine (Andromeda), mass recalibration and quantification [Cox & Mann, 2008].

The statistics are as important as the chemistry. The target–decoy strategy searches against real sequences plus reversed or shuffled "decoy" sequences. Decoy hits estimate how many target hits are false, which lets you set a false discovery rate (usually 1%) [Elias & Gygi, 2007]. This is multiple-hypothesis correction from MATH201, applied to millions of spectra.

11. Proteomics Workflows and Quantification

11.1 Bottom-up vs. top-down

  • Bottom-up (shotgun): digest proteins, usually with trypsin, then analyze peptides by LC-MS/MS and infer the proteins. Trypsin cleaves C-terminal to Lys and Arg, but not before Pro. This places a basic residue at each peptide's C-terminus, which helps give y-ion-rich spectra and doubly charged precursors. The weakness is the protein inference problem. Shared peptides make it ambiguous which isoform was present, and the link between modifications on different peptides (the full "proteoform") is lost.
  • Top-down: introduce intact proteins by ESI and fragment them in the instrument, often with ETD or ECD on FT instruments. This preserves complete proteoform information, such as which combination of PTMs coexists on one molecule. It is technically harder because of lower sensitivity, complex spectra and the need for very high resolution.

11.2 Quantification strategies

Mass spectrometry is not inherently quantitative across different analytes, because ionization efficiencies differ (§7.3). Every quantitative strategy is a way of comparing the same peptide across conditions, so that ionization efficiency cancels.

  • Label-free quantification: compare MS1 peak areas (extracted ion chromatograms) or spectral counts across runs. It is simple and unlimited in sample number, but it depends on run-to-run reproducibility and retention-time alignment.
  • Metabolic labeling (SILAC): grow cells in ¹³C/¹⁵N-labeled Lys/Arg, mix the samples early, and compare light/heavy peak pairs in MS1 [Ong et al., 2002]. Mixing early minimizes handling error.
  • Isobaric tags (iTRAQ, TMT) [Ross et al., 2004; Thompson et al., 2003]: each sample's peptides are labeled with a tag of identical total mass but different distribution of heavy isotopes between a reporter and a balancer. The labeled peptides co-elute and appear as one MS1 peak. On fragmentation, reporter ions of distinct m/z (around m/z 126–135 for TMT) are released, and their intensities give relative quantities. The clever idea is that multiplexing (up to 16–18 samples) adds no MS1 complexity. The catch is ratio compression: co-isolated contaminant peptides contribute reporter ions too, pulling ratios toward 1:1. Remedies include narrower isolation and MS3-based (SPS) methods.

11.3 Data-dependent vs. data-independent acquisition

  • DDA: the instrument picks the top-N most intense precursors from each survey scan for MS/MS. Selection is somewhat stochastic, so low-abundance peptides are sampled inconsistently across runs.
  • DIA (e.g., SWATH): the instrument systematically fragments all precursors in sequential wide m/z windows [Gillet et al., 2012]. This yields a complete, reproducible digital record, but each MS2 spectrum is a chimera of many peptides. Deconvolving them requires spectral libraries or library-free algorithms. It is the multicomponent analysis problem from §2.4 in a far higher-dimensional form.

11.4 Targeted quantification: SRM/MRM

On a triple quadrupole, Q1 selects a precursor, q2 fragments it, and Q3 selects a specific fragment. Each precursor→fragment pair is a transition. Monitoring a few transitions per peptide with a co-eluting stable-isotope-labeled internal standard gives superb selectivity, linearity and precision.

The quadrupole's weakness, low resolution, is irrelevant here. Its strengths are fast switching and 100% duty cycle on the ions you care about. In effect, SRM is a lock-in amplifier for chemistry: you listen only at frequencies you defined in advance.

12. Two-Dimensional NMR and Protein NMR

12.1 The idea of a second dimension

Every 2D NMR experiment follows the same scheme: preparation → evolution (t₁) → mixing → detection (t₂). During t₁, spins acquire a phase that depends on their frequency. The mixing step transfers magnetization to other spins via a specific mechanism. During t₂, you detect the frequency of the receiving spin. Incrementing t₁ over many experiments and Fourier transforming in both dimensions produces a map. A cross-peak at (ω₁, ω₂) means "magnetization moved from a spin at ω₁ to a spin at ω₂."

The mixing mechanism defines the experiment. Choose your mixing step and you choose which question you are asking.

  • COSY (¹H–¹H, through 2–3 bonds via scalar J coupling): maps vicinal (³J_HH) coupling networks, showing who is next to whom.
  • TOCSY: an isotropic mixing spin-lock relays magnetization along an entire unbroken chain of coupled protons. In proteins, all protons within one amino acid's side chain light up together, giving that residue's "fingerprint." The chain breaks at the amide C=O, because there are no protons on carbonyl carbon. That break is exactly what isolates residues from each other.
  • HSQC / HMQC (¹H–¹³C or ¹H–¹⁵N, one bond, ¹J_CH ≈ 125–160 Hz): one peak per directly bonded H–X pair. It is very sensitive because it detects on the proton (inverse detection: high γ_H).
  • HMBC (¹H–¹³C over 2–3 bonds, ⁿJ_CH ≈ 2–10 Hz, typically optimized near 8 Hz): connects fragments across quaternary carbons and heteroatoms. This is the key experiment for stitching together natural product skeletons.
  • NOESY (through-space dipolar cross-relaxation): cross-peak intensity ∝ r⁻⁶, observable for protons within about 5 Å. It reports spatial proximity regardless of bonding, and it underlies 3D structure determination.

The structure-solving logic for a small molecule:

  1. HSQC identifies which H sits on which C.
  2. COSY and TOCSY assemble spin systems.
  3. HMBC bridges the spin systems across quaternary centers.
  4. NOESY sets relative stereochemistry and conformation.

12.2 Protein NMR

  • The ¹H–¹⁵N HSQC "fingerprint": in a uniformly ¹⁵N-labeled protein, each backbone amide N–H gives one peak. That is one per residue, except prolines, which lack an amide H. Side-chain NH₂ groups of Asn and Gln, the Trp indole NH, and Arg NεH also appear. A well-dispersed HSQC indicates a folded protein. A collapsed spectrum clustered near ¹H 8.0–8.5 ppm indicates disorder. Titrating a ligand and watching specific peaks shift (chemical shift perturbation) maps the binding site residue by residue.
  • Chemical shift index (CSI): backbone chemical shifts deviate from random-coil values in a secondary-structure-dependent way [Wishart, Sykes & Richards, 1992]. In α-helices, Cα and C′ shift downfield (positive Δδ), while Hα and Cβ shift upfield. β-strands show the opposite pattern. Runs of consistent deviations define helices and strands. The coupling constant ³J(HN,Hα) also follows the Karplus relation: about 4 Hz in helices and about 8–10 Hz in strands, giving dihedral angle (φ) information.
  • Backbone assignment strategy: in a ¹³C/¹⁵N-labeled protein, triple-resonance experiments link each amide to carbons in its own residue and the preceding one. HNCA correlates the amide H–N with Cα(i) and Cα(i−1). HN(CO)CA correlates it with only Cα(i−1). Similar pairs exist for Cβ (HNCACB/CBCA(CO)NH). Matching the Cα(i−1) of one spin system to the Cα(i) of another "walks" along the chain, which is sequential assignment. Cα/Cβ values identify residue types (Gly has Cα ≈ 45 ppm; Ala has Cβ ≈ 19 ppm; Ser and Thr have Cβ > 60 ppm), anchoring the walk onto the known sequence. Kurt Wüthrich's development of sequential assignment, originally with homonuclear NOESY/COSY on unlabeled proteins, earned a share of the 2002 Nobel Prize. That was the same year as Fenn and Tanaka: a remarkable year for biomolecular analysis.

13. Separations and Hyphenation: GC-MS, LC-MS, CE-MS, LC-NMR

13.1 What a column does

Chromatographic resolution depends on three separately tunable factors (the Purnell equation):

R_s = (√N / 4) · [(α − 1)/α] · [k / (1 + k)]

N = plate number (efficiency), α = selectivity factor,
k = retention factor (of the later peak)

Band broadening (van Deemter):
H = A + B/u + C·u

A = eddy diffusion, B = longitudinal diffusion,
C = mass-transfer resistance, u = linear velocity

R_s grows only as √N (the square-root law again). Changing α through chemistry is usually far more powerful than adding column length. The B/u term is diffusion (Fick, PHYS201). The C·u term reflects slow equilibration with the stationary phase. Their competition creates an optimum flow rate. Smaller particles (UHPLC, sub-2 μm) reduce A and C, which flattens the curve and allows faster separations at high pressure.

13.2 GC-MS

  • Volatility is the entry requirement. Analytes must vaporize without decomposition, which typically means below about 500 Da with modest polarity.
  • Derivatization converts polar groups (–OH, –COOH, –NH₂) into volatile, thermally stable forms. The most common route is silylation to trimethylsilyl (TMS) derivatives with reagents such as BSTFA or MSTFA. Other routes include methylation (fatty acid methyl esters) and acylation. Amino acids and sugars become GC-amenable this way.
  • The interface is easy, because carrier gas (He) is readily pumped away. EI is the natural ionization method, and the payoff is spectral library matching (§7.1).

13.3 LC-MS and LC-MS/MS

For decades, the core problem was that 1 mL/min of liquid becomes about a liter per minute of vapor, which overwhelms a vacuum system. ESI solved it: the interface is the ion source, working at atmospheric pressure. Mobile phases must be volatile. Use formic acid, acetic acid or ammonium acetate, not phosphate buffers or nonvolatile salts. Trifluoroacetic acid gives sharp peptide peaks but suppresses ESI signal through ion pairing, so proteomics typically uses 0.1% formic acid.

The dimensions multiply. Retention time × m/z × fragment m/z gives each analyte a three-coordinate address. That is why LC-MS/MS can quantify hundreds of analytes (multiplexed MRM) in one injection.

13.4 CE-MS and LC-NMR

  • CE-MS: capillary electrophoresis separates by charge-to-size ratio with very high efficiency, owing to plug-like electroosmotic flow and minimal band broadening. Its nanoliter flows match nanoESI well. The challenge is completing the electrical circuit at the outlet, which is addressed with sheath-liquid or sheathless interfaces. It is useful for intact proteins, charged metabolites and glycans.
  • LC-NMR: this provides full structural information on separated components. NMR's intrinsically low sensitivity is the bottleneck, because nuclear spin polarization at room temperature is tiny (a Boltzmann factor near 10⁻⁵, CHEM301). Stopped-flow operation, solvent suppression, cryoprobes and LC-SPE-NMR (trap peaks on solid phase, then elute in deuterated solvent) all mitigate it. The complementarity to MS is fundamental. MS is sensitive but struggles with isomers. NMR distinguishes isomers easily but needs micrograms.

Connections to Peptide Chemistry

Peptides and proteins are the thread running through this entire course. Here is how each method serves peptide science, and where each leads.

  • Synthesis quality control. After solid-phase peptide synthesis (CHEM202 amide coupling chemistry), the first question is whether you made the right molecule. LC-MS answers in minutes. Reverse-phase HPLC with UV at 214 nm (the amide bond π→π* absorption) gives purity. ESI-MS confirms [M+H]⁺ and multiply charged ions. Isotope patterns and accurate mass reveal deletion sequences (missing residue masses), incomplete deprotection (e.g., +56 Da for residual tert-butyl), oxidation (Met +15.995) or racemization. Racemization is mass-silent and needs chiral chromatography or NMR, a good reminder of MS's blind spots.
  • Intrinsic spectroscopic handles. Trp (λ_ex ≈ 280–295 nm, λ_em ≈ 330–350 nm) is an environment-sensitive fluorophore. Its emission blue-shifts when buried in a hydrophobic core or membrane, a solvatochromic Stokes-shift effect. Fluorescence anisotropy of labeled peptides measures binding to MHC molecules, antibodies or receptors (BIOL401). FRET between donor and acceptor (efficiency E = 1/[1 + (r/R₀)⁶]) turns a protease cleavage site into a "molecular ruler" assay. Cleavage separates the pair and fluorescence turns on.
  • Secondary structure from vibrations. The amide I band, mostly C=O stretch, is diagnostic of conformation. Helix appears near 1650–1658 cm⁻¹, β-sheet near 1620–1640 cm⁻¹ (with a weak antiparallel band near 1690), and random coil near 1640–1648 cm⁻¹ [Skoog, Holler & Crouch, 2018]. ATR-FTIR of amyloid aggregates and Raman of peptide films both exploit this. You can watch a peptide aggregate in real time.
  • Sequencing and PTM mapping. The b/y and c/z ladders of §10 are how modern biology reads protein sequence. Edman degradation has been largely displaced for most applications. Phosphoproteomics combines enrichment (TiO₂, IMAC) with HCD and ETD to localize sites, connecting directly to the kinase signaling you studied in CHEM402 and BIOL302. Immunopeptidomics uses LC-MS/MS on peptides eluted from MHC molecules to discover what T cells actually see, which matters for cancer neoantigen vaccines.
  • Peptides as quantitative proxies. In targeted proteomics, a proteotypic tryptic peptide stands in for its parent protein. A heavy-isotope-labeled synthetic version serves as the internal standard, quantified by MRM. This is how clinical laboratories increasingly measure biomarkers such as thyroglobulin, sometimes in preference to immunoassays prone to antibody interference.
  • Structure and dynamics in solution. The ¹H–¹⁵N HSQC of a peptide or small protein, together with the CSI, maps secondary structure residue by residue. NOESY constraints yield 3D structures of bioactive peptides such as conotoxins, antimicrobial peptides and stapled helices. Looking ahead, hydrogen–deuterium exchange MS (HDX-MS) uses the same mass shift logic to probe backbone amide solvent accessibility. Native MS uses gentle ESI to weigh intact noncovalent complexes. Both techniques are direct descendants of Fenn's "molecular elephants."
  • Peptide drug development. GLP-1 agonists such as semaglutide are lipidated peptides. Characterizing them requires LC-MS for identity and degradants, NMR for higher-order structure, and SPR or fluorescence for binding. In effect, the whole toolbox of this course appears in a single regulatory filing.

Common Misconceptions to Watch Out For

  1. "A mass spectrometer measures mass." It measures m/z. A peak at m/z 1001 might be a 1000 Da singly charged ion or a 20,000 Da ion with z = 20. Always determine charge, from isotope spacing or charge-state series, before interpreting.
  2. "High resolution and high mass accuracy are the same thing." Resolution separates neighboring peaks. Accuracy locates a peak correctly. They are related but independent specifications, and calibration governs accuracy.
  3. "The tallest isotope peak is the monoisotopic peak." This is true only for small molecules. Above about 1,800 Da for peptides, the M+1 or higher isotope peak dominates. Picking the wrong peak yields 1 Da errors that wreck database searches.
  4. "Averaging 100 scans gives 100× better S/N." It gives 10×. The √n law governs signal averaging, chromatographic plate number, and shot-noise-limited detection alike.
  5. "Peak intensity in MS reflects abundance." Only for the same analyte under the same conditions. Ionization efficiency varies by orders of magnitude between molecules, and ion suppression changes it within a single run. Use isotope-labeled internal standards for real quantification.
  6. "Sensitivity and detection limit are the same." Sensitivity, in IUPAC usage, is the slope of the calibration curve. Detection limit depends on slope and noise (roughly 3σ_blank / slope). A method can be very sensitive yet have a poor detection limit if it is noisy.
  7. "Fluorescence intensity is always proportional to concentration." Only at low absorbance (roughly A < 0.05–0.1). At higher concentration, inner-filter effects (excitation light absorbed before reaching the viewed volume, or emitted light reabsorbed) and self-quenching make the curve bend over and even decline.
  8. "Raman is just a weaker version of IR." Different selection rules (polarizability vs. dipole) give different, complementary information. Water's weak Raman spectrum is a decisive advantage for biological samples.
  9. "ESI only produces [M+H]⁺." Expect [M+Na]⁺ (+21.98 Da relative to [M+H]⁺), [M+K]⁺ (+37.96), [M+NH₄]⁺ (+17.03), dimers [2M+H]⁺, and in-source fragments. Sodium adducts from glassware are ubiquitous.
  10. "CID can distinguish leucine from isoleucine." Standard low-energy CID and HCD cannot, because the two are isobaric. Side-chain-specific w-ions, from high-energy or radical-driven methods such as EThcD, are needed. Likewise, Gly-Gly vs. Asn and Ala-Gly vs. Gln are compositionally identical. Missing fragments, not low mass accuracy, cause those ambiguities.
  11. "A 1% FDR means each identification is 99% likely correct." FDR is a property of the set of accepted identifications: about 1% of them are expected to be false. Individual low-scoring hits near the threshold have much higher error probabilities than high-scoring ones.
  12. "NOESY cross-peaks indicate bonds." NOESY reports through-space proximity (r⁻⁶, < 5 Å), independent of bonding. COSY, TOCSY, HSQC and HMBC report through-bond J-coupling. Mixing these up is the classic structure-elucidation error.
  13. "ICP-MS is interference-free because it's mass-based." Polyatomic species from argon, solvent and matrix (ArCl⁺, ArO⁺, Ar₂⁺) and true isobars (e.g., ⁵⁸Fe/⁵⁸Ni) all overlap analytes. Collision/reaction cells and high-resolution sectors exist precisely because of this.
  14. "Beer's law fails at high absorbance because of the chemistry." Often the instrument is to blame: stray light and polychromatic radiation cause negative deviations even when molecules behave perfectly.
  15. "Isobaric tags give perfect ratios." Co-isolated interferents compress ratios toward 1:1. A measured twofold change may represent a substantially larger true change. Interpret TMT fold-changes with this bias in mind.

Closing thought for Skipper: Every instrument in this course is a carefully engineered compromise. When you encounter a new technique, ask four questions. What molecular property becomes the signal? Along what axis is it separated? What limits the noise? What does it trade away to get what it gives? Answer those, and you understand the instrument, not just its manual.


References cited in these notes

  • Skoog, D. A.; Holler, F. J.; Crouch, S. R. Principles of Instrumental Analysis, 7th ed.; Cengage, 2018.
  • Gross, J. H. Mass Spectrometry: A Textbook, 3rd ed.; Springer, 2017.
  • Fenn, J. B.; Mann, M.; Meng, C. K.; Wong, S. F.; Whitehouse, C. M. Electrospray ionization for mass spectrometry of large biomolecules. Science 1989, 246, 64–71. DOI: 10.1126/science.2675958
  • Karas, M.; Hillenkamp, F. Laser desorption ionization of proteins with molecular masses exceeding 10,000 daltons. Anal. Chem. 1988, 60, 2299–2301. DOI: 10.1021/ac00171a028
  • Comisarow, M. B.; Marshall, A. G. Fourier transform ion cyclotron resonance spectroscopy. Chem. Phys. Lett. 1974, 25, 282–283.
  • Makarov, A. Electrostatic axially harmonic orbital trapping. Anal. Chem. 2000, 72, 1156–1162.
  • Roepstorff, P.; Fohlman, J. Proposal for a common nomenclature for sequence ions in mass spectra of peptides. Biomed. Mass Spectrom. 1984, 11, 601.
  • Biemann, K. Contributions of mass spectrometry to peptide and protein structure. Biomed. Environ. Mass Spectrom. 1988, 16, 99–111.
  • Zubarev, R. A.; Kelleher, N. L.; McLafferty, F. W. Electron capture dissociation of multiply charged protein cations. J. Am. Chem. Soc. 1998, 120, 3265–3266.
  • Syka, J. E. P.; Coon, J. J.; Schroeder, M. J.; Shabanowitz, J.; Hunt, D. F. Peptide and protein sequence analysis by electron transfer dissociation mass spectrometry. PNAS 2004, 101, 9528–9533.
  • Olsen, J. V. et al. Higher-energy C-trap dissociation for peptide modification analysis. Nat. Methods 2007, 4, 709–712.
  • Eng, J. K.; McCormack, A. L.; Yates, J. R. An approach to correlate tandem mass spectral data of peptides with amino acid sequences in a protein database. J. Am. Soc. Mass Spectrom. 1994, 5, 976–989.
  • Perkins, D. N.; Pappin, D. J. C.; Creasy, D. M.; Cottrell, J. S. Probability-based protein identification by searching sequence databases using mass spectrometry data. Electrophoresis 1999, 20, 3551–3567.
  • Cox, J.; Mann, M. MaxQuant enables high peptide identification rates, individualized p.p.b.-range mass accuracies and proteome-wide protein quantification. Nat. Biotechnol. 2008, 26, 1367–1372.
  • Elias, J. E.; Gygi, S. P. Target-decoy search strategy for increased confidence in large-scale protein identifications by mass spectrometry. Nat. Methods 2007, 4, 207–214.
  • Ong, S.-E. et al. Stable isotope labeling by amino acids in cell culture, SILAC. Mol. Cell. Proteomics 2002, 1, 376–386.
  • Thompson, A. et al. Tandem mass tags. Anal. Chem. 2003, 75, 1895–1904.
  • Ross, P. L. et al. Multiplexed protein quantitation in Saccharomyces cerevisiae using amine-reactive isobaric tagging reagents. Mol. Cell. Proteomics 2004, 3, 1154–1169.
  • Gillet, L. C. et al. Targeted data extraction of the MS/MS spectra generated by data-independent acquisition. Mol. Cell. Proteomics 2012, 11, O111.016717.
  • Wishart, D. S.; Sykes, B. D.; Richards, F. M. The chemical shift index. Biochemistry 1992, 31, 1647–1651.

⚙️ Mechanisms & Applications

GPT-6 Astra

Mechanisms & Applications: Turning Molecular Behavior into Evidence

Skipper, the central question in instrumental analysis is not simply “What does this instrument measure?” It is “What sequence of physical and chemical events connects the original sample to the reported result?” Every step can preserve information, distort it, or destroy it. Your physical chemistry explains the interactions; your analytical chemistry explains calibration and uncertainty; your biochemistry tells you which distinctions actually matter.

Sample → preparation/separation → interaction or ionization
       → discrimination by wavelength, frequency, or m/z
       → detection → signal processing → chemical inference

A detector does not directly report “protein concentration” or “peptide identity.” It reports photons, charge, current, or an induced electrical signal. Concentration and identity emerge only after a model connects those observations to chemistry. Keep that distinction in view throughout this section. [1]

1. Signal-to-Noise: Extracting a Weak Chemical Message

Imagine listening for a repeating musical note in a noisy room. You can listen longer, restrict the frequencies you admit, or synchronize your listening to the expected note. Instruments use exactly these strategies—but their success depends on what kind of noise is present.

  • Shot noise: fluctuations associated with discrete detection events, such as photoelectrons. For ideal Poisson counting, the variance equals the mean count.
  • Johnson–Nyquist noise: thermal voltage fluctuations in resistive electronics.
  • Flicker noise: noise whose power commonly increases toward low frequencies, approximately as 1/f.
  • Drift and interference: source aging, temperature changes, contamination, and electrical pickup. These are often correlated and do not necessarily average away.
Ideal photon counting:
Mean detected count = n
Standard deviation = √n
Signal-to-noise ratio ≈ √n        [if signal shot noise dominates]

Average of N independent measurements:
σaverage = σsingle / √N
(S/N)average = √N × (S/N)single

Thermal voltage noise:
Vrms = √(4kBTRB)

kB = Boltzmann constant; T = temperature;
R = resistance; B = measurement bandwidth.

Thus, a tenfold improvement in signal-to-noise by averaging alone requires roughly one hundred independent measurements. More importantly, averaging a drifting baseline does not repair it. A stable-looking mean can remain systematically wrong.

Lock-in amplification uses prior knowledge of when the signal should occur:

  1. Modulate the excitation, for example by chopping a light beam at frequency f.
  2. Detect a weak response containing that modulation plus unrelated noise.
  3. Multiply the detector output by a reference oscillation at f.
  4. Low-pass filter the product, retaining the slowly varying component associated with the reference.
s(t) = A cos(ωt + φ) + noise
reference = cos(ωt)

Low-pass{s(t) × reference} = (A/2) cosφ

A second, quadrature reference recovers the complementary component,
allowing both amplitude and phase to be determined.

The surprising point is that a lock-in does not “remove all noise.” It selects a narrow frequency region, often away from strong low-frequency noise. Longer averaging improves selectivity but slows the response; interference at the reference frequency remains troublesome. This trade-off matters in weak fluorescence, modulated spectroscopy, and other low-signal measurements. [1]

2. UV–Visible Spectrophotometry: From Electronic Transitions to Mixture Analysis

Absorption occurs when a photon drives an allowed transition between molecular electronic states. The instrument must isolate a wavelength range, measure incident and transmitted light, and convert their ratio into absorbance.

  1. A source supplies a wavelength continuum.
  2. A monochromator uses a grating and slits to select a spectral band.
  3. The sample attenuates that band through absorption, with scattering and reflections potentially contributing unwanted losses.
  4. A detector measures the transmitted light relative to a blank or reference.
  5. Calibration converts absorbance into concentration.
Transmittance: T = I/I0
Absorbance:    A = −log10(T) = εbc

ε = molar absorption coefficient
b = optical path length
c = concentration

A photomultiplier tube converts photons into photoelectrons at a photocathode and multiplies them through a dynode cascade. In many diode-array instruments, polychromatic light passes through the sample before dispersion across an array, allowing many wavelengths to be measured nearly simultaneously. CCD detectors likewise collect spatially dispersed spectral information as stored charge. Detector choice therefore changes speed, sensitivity, and how the experiment is organized—not just convenience. [1]

Multicomponent analysis connects directly to your linear algebra. If absorbers do not interact and Beer’s law holds, their absorbances add:

At wavelength λi:
Ai = b Σj εij cj

Matrix form:
A = E c + error, where Eij = bεij

Weighted least-squares estimate:
ĉ = (EᵀWE)⁻¹EᵀWA

This expression assumes E has full column rank;
W weights observations by their uncertainty structure.

Overlapping spectra are not automatically fatal. The real problem is indistinguishable spectra: nearly proportional columns of E make the problem ill-conditioned, so small measurement errors produce large concentration errors. Measuring more wavelengths helps only when they add independent information.

Derivative spectroscopy emphasizes spectral slope and curvature, helping reveal shoulders or suppress slowly varying backgrounds. But differentiation amplifies high-frequency noise. Smoothing and derivative settings must therefore be validated rather than chosen merely to make peaks look convincing.

Applications: reaction monitoring, enzyme kinetics, drug dissolution testing, chromophore-containing impurity assays, and protein or nucleic-acid estimation. An A260/A280 ratio is a useful screening observation, not proof of purity. Stray light, scattering, pH-dependent spectra, and nonabsorbing contaminants can all break a simplistic interpretation.

3. Fluorescence: Measuring What Happens After Excitation

Absorption asks whether a molecule takes up a photon. Fluorescence asks what happens next. After excitation and rapid relaxation within the excited-state manifold, a fluorophore can emit a photon or lose its excitation by nonradiative pathways. Fluorescence often appears at longer wavelength than the absorbed light because some energy has already been dissipated.

S0 + hνexcitation → Sn → S1

S1 → S0 + hνemission          radiative decay, rate kr
S1 → nonradiative products    combined rate knr

For a simple single-state, first-order model:
τ = 1/(kr + knr)
ΦF = kr/(kr + knr) = krτ

The lifetime τ is the mean residence time in that excited state; the quantum yield ΦF is the fraction of absorbed photons that produce fluorescence. These are related but not interchangeable.

  • Excitation spectrum: scan excitation wavelength while monitoring emission at a selected wavelength. For a single fluorophore under suitable conditions, it resembles the absorption spectrum.
  • Emission spectrum: hold excitation wavelength fixed and scan emitted wavelengths.
  • Lifetime measurement: excite with short pulses and record emission delays, or use phase shifts under modulated excitation.
Single-exponential fluorescence decay:
I(t) = I(0)e^(−t/τ)

Ideal dynamic Stern–Volmer quenching:
F0/F = τ0/τ = 1 + kqτ0[Q]

For dynamic quenching, an excited fluorophore encounters a quencher, adding a decay pathway and shortening the lifetime. For an ideal static-quenching model, a nonfluorescent ground-state complex forms: intensity decreases, but the lifetime of the remaining uncomplexed fluorophores is unchanged. Comparing intensity and lifetime can therefore distinguish mechanisms that an intensity measurement alone cannot resolve.

Fluorescence anisotropy connects emission to molecular rotation. Polarized excitation preferentially excites suitably oriented transition dipoles; rotation before emission partially erases that orientational selection.

r = (I∥ − G I⊥)/(I∥ + 2G I⊥)

Simple Perrin relation:
r0/r = 1 + τ/θ

G corrects unequal polarization-channel response.
θ is the rotational correlation time.

A fluorescent ligand often rotates more slowly after binding a protein, increasing anisotropy. This supports high-throughput binding and competition assays in drug discovery. But changes in fluorescence lifetime, viscosity, or local probe motion can also change anisotropy.

Experimental caution: high absorbance can attenuate excitation and reabsorb emitted photons—the inner-filter effect. Low fluorescence does not automatically mean quenching or low concentration. These distinctions are crucial in peptide-binding assays, protein-folding studies, and fluorescent enzyme screens. [1]

4. FTIR, ATR, Raman, and SERS: Different Views of Molecular Vibration

Your quantum mechanics supplies the vibrational energy levels, but a transition’s visibility depends on how the vibration couples to light. IR activity requires a changing dipole moment; Raman activity requires a changing polarizability. A vibration weak in one method may therefore be strong in the other.

FTIR mechanism:

  1. A beam splitter divides broadband IR radiation into two paths.
  2. One mirror moves, changing the optical path difference.
  3. Recombination produces an interference signal containing contributions from all transmitted wavenumbers.
  4. The detector records intensity against optical path difference: an interferogram.
  5. A Fourier transform recovers the spectrum; sample and background measurements give transmission and absorbance.
Idealized interferogram relation:
I(δ) ∝ ∫ S(ṽ) cos(2πṽδ) dṽ

δ = optical path difference
ṽ = wavenumber
S(ṽ) = spectral intensity distribution

Spectral resolution improves as the maximum path difference increases.

This is the same mathematical idea you encountered in waves and spectroscopy: separate superposed oscillations by transforming into frequency space. The multiplex signal-to-noise advantage is strongest when detector noise dominates; it is not universal when photon noise from the full spectrum dominates.

ATR-FTIR sends IR radiation through a high-refractive-index crystal under total internal reflection. An evanescent field extends beyond the crystal and interacts with the contacting sample. It is not a conventional beam transmitted through a fixed sample thickness.

For a common evanescent-field amplitude convention:

dp = λ0 / [2πn1√(sin²θ − (n2/n1)²)]

n1 = crystal refractive index; n2 = sample refractive index
θ = incidence angle; λ0 = vacuum wavelength

Penetration depth varies with wavelength, angle, and refractive indices. Consequently, ATR band intensities differ systematically from ordinary transmission intensities, and good sample contact matters.

Raman spectroscopy measures inelastic scattering: most scattered photons retain their energy, but a small fraction exchange energy with molecular vibrations. Raman scattering was experimentally established in 1928 through the work of C. V. Raman and K. S. Krishnan, with independent observations by Grigory Landsberg and Leonid Mandelstam. It mattered because vibrations could now be probed through polarizability rather than only dipole absorption. [1]

SERS uses molecules near suitable nanostructured metal surfaces, commonly silver or gold. Local electromagnetic enhancement can greatly amplify Raman scattering, with additional chemical contributions in some systems. The approximate electromagnetic enhancement often scales as |Elocal/Eincident|4 when excitation and scattered fields experience similar enhancement. This creates exceptional sensitivity—but “hot spot” variability complicates quantitative reproducibility.

Applications: FTIR amide bands support protein secondary-structure assessment; ATR enables rapid inspection of formulations and solids; Raman helps distinguish pharmaceutical polymorphs and can work well in aqueous systems. Neither a single amide-band position nor a Raman match alone establishes complete protein structure.

5. Atomic Spectroscopy: Deliberately Destroying Molecular Identity

Molecular spectroscopy preserves information about bonding. Atomic spectroscopy often destroys those bonds on purpose to ask a different question: How much of each element is present?

  1. Prepare the sample: digestion may convert a biological sample or drug product into a solution suitable for introduction.
  2. Atomize or excite: a flame, graphite furnace, or plasma breaks down the sample and generates atoms or ions.
  3. Measure: observe element-specific absorption, emission, or ion abundance.
  4. Correct and calibrate: address blanks, contamination, matrix effects, and spectral interference.
  • Flame AAS: nebulized solution enters a flame; ground-state atoms absorb narrow element-specific source lines.
  • Graphite-furnace AAS: a small aliquot undergoes programmed drying, pyrolysis, and atomization. Longer residence and efficient sample use often improve detection limits, but matrix control becomes especially important.
  • ICP-OES: an inductively coupled argon plasma produces excited atoms and ions whose characteristic emission lines are measured.
  • ICP-MS: plasma-generated ions pass through an interface into a mass spectrometer, enabling sensitive elemental and isotope measurements.

A classic ICP-MS interference is 40Ar35Cl+ at nominal m/z 75, overlapping monoisotopic 75As+. Chloride in a biological sample can therefore masquerade as arsenic unless collision/reaction-cell chemistry, sufficient resolving power, or another validated strategy discriminates the signals.

Internal standards can track some introduction and instrumental effects; standard addition can address certain proportional matrix effects. Neither universally repairs every interference. Standard addition, for example, does not automatically remove an unresolved additive spectral signal.

Applications: elemental impurities in pharmaceuticals, residual metal catalysts in synthetic peptides, toxic-metal exposure, and metalloprotein research. Digested ICP-MS reports total elemental content; to distinguish chemical species, preserve them and separate them first, for example by LC–ICP-MS. [1]

6. Mass Spectrometry: Ion Formation Is Part of the Experiment

Mass spectrometry measures gas-phase ions according to mass-to-charge ratio, not neutral molecular mass directly. The ionization method decides which molecules become observable and how much structural information survives.

For a protonated molecule:
[M + zH]z+ has m/z = (M + z mH+)/z

M = neutral monoisotopic mass
mH+ ≈ 1.007276 Da

Example:
M = 2000.0000 Da, z = 2
m/z = 1001.0073

Isotope spacing helps reveal charge. For common peptide isotope envelopes, adjacent isotope peaks are approximately 1.00335/z apart in m/z, largely reflecting substitution of 13C for 12C. A spacing near 0.502 therefore suggests a doubly charged ion. The envelope also contains compositional information, including characteristic chlorine and bromine patterns. [2]

Four central ionization mechanisms:

  1. Electron ionization, EI: energetic electrons interact with vaporized analytes, commonly at 70 eV, producing radical cations and often extensive fragmentation. The reproducibility of fragmentation makes GC–EI-MS library matching powerful.
  2. Chemical ionization, CI: electrons first ionize a reagent gas. Reagent ions then react with the analyte through processes such as proton transfer. CI often yields more readily interpretable molecular-mass information than EI, though the actual chemistry depends on reagent and analyte.
  3. Electrospray ionization, ESI: high voltage at a liquid outlet generates charged droplets. Solvent evaporation and repeated droplet fission lead ultimately to gas-phase ions. Ion evaporation, charged-residue behavior, and chain-ejection mechanisms contribute under different conditions; there is no single universal pathway for every analyte.
  4. MALDI: analyte is prepared with a UV-absorbing matrix. A laser pulse produces rapid desorption and a plume in which ion formation and proton-transfer processes can occur. Many peptide and protein spectra contain predominantly singly charged ions, although this is not an absolute rule.
EI:
M + e− → M+• + 2e− → fragment ions + neutral species

A common positive-CI reaction:
Reagent-H+ + M → Reagent + [M + H]+

ESI:
Solution → charged droplets → evaporation/fission → gas-phase ions

MALDI:
Matrix/analyte + laser pulse → desorption plume → analyte ions

The beautiful consequence of ESI is that a very large protein can appear at moderate m/z by carrying many charges. The cost is a charge-state distribution that must be interpreted or deconvoluted. Salt adducts and ion suppression further complicate the relationship between peak intensity and solution abundance.

Historical turning point: Fenn, Mann, Meng, Wong, and Whitehouse’s 1989 Science paper demonstrated ESI-MS for large biomolecules. Karas and Hillenkamp’s 1988 Analytical Chemistry paper demonstrated laser-desorption ionization of proteins above 10,000 Da using a matrix. These developments, alongside related soft-laser-desorption work by Koichi Tanaka and colleagues, made mass spectrometry practical for biomolecules that could not simply be vaporized without decomposition. The Karas–Hillenkamp paper is correctly titled “Laser desorption ionization of proteins with molecular masses exceeding 10,000 daltons,” rather than the generic MALDI title sometimes used in reading lists. [3] [4]

7. Mass Analyzers: Different Physical Solutions to Sorting Ions

Ionization and mass analysis are separate choices. ESI is not an analyzer, and a TOF instrument need not use MALDI. Each analyzer exploits a different dependence of ion motion on m/z. [2]

  • Quadrupole: combined radiofrequency and direct-current fields allow only a selected range of m/z values to maintain stable trajectories through four rods.
  • Ion trap: oscillating fields confine ions; controlled excitation supports isolation, fragmentation, and mass-selective ejection. Repeated isolation and fragmentation enable MSn.
  • Time-of-flight, TOF: ions receive approximately the same kinetic energy per charge; lower-m/z ions travel faster. A reflectron can compensate for initial kinetic-energy spread.
  • Sector analyzer: magnetic fields bend ion trajectories; electric sectors discriminate kinetic energy per charge. Appropriate combinations provide high-resolution separation.
  • FT-ICR: ions undergo cyclotron motion in a magnetic field. Their image current is recorded and Fourier transformed to obtain frequencies related to m/z.
  • Orbitrap: ions orbit a central electrode while oscillating axially in an electrostatic field. Axial oscillation frequencies are measured through image current and transformed into a mass spectrum.
TOF, simplified:
½mv² = zeV
t = L√[m/(2zeV)]

Magnetic-sector motion:
r = mv/(zeB)

Ideal cyclotron frequency:
fc = zeB/(2πm)

Orbitrap axial frequency:
faxial ∝ 1/√(m/z)

FT-ICR and Orbitrap detection are conceptually striking: the ions need not strike a detector to be observed. Their motion induces an electrical signal. Makarov’s 2000 paper described electrostatic axially harmonic orbital trapping as a high-performance mass-analysis technique, helping establish the basis for modern Orbitrap instruments. [5]

Resolving power:
R = (m/z)/Δ(m/z)

State the peak-width convention, commonly full width at half maximum.

Mass error:
errorppm = 10⁶ × [(m/z)measured − (m/z)reference]/(m/z)reference

At m/z 1000, R = 100,000 corresponds to a peak width of 0.010 m/z under the stated convention. A 2 ppm mass error corresponds to 0.002 m/z. Resolving power and mass accuracy are not the same: one describes peak width and separation capability, the other agreement with a reference value. Calibration drift can shift a narrow peak, and an unresolved interference can bias its measured position.

8. Tandem MS: Controlled Fragmentation as a Structural Experiment

Intact mass constrains composition; fragmentation constrains connectivity. In MS/MS, an instrument selects a precursor ion, activates it, and measures its product ions.

Ionize → select precursor → activate/dissociate → analyze products
  • CID: collisions with neutral gas convert ion translational energy into internal energy. Protonated peptides commonly form b- and y-type fragments, but sequence and proton mobility strongly affect the pattern.
  • HCD: commonly implemented as beam-type collisional activation in a dedicated cell, including in many Orbitrap workflows. It is still collisional dissociation, not an unrelated mechanism. Relative to classical resonance CID in ion traps, it often provides improved access to low-mass product ions, including isobaric-tag reporters.
  • ETD: electron transfer from a reagent anion to a multiply protonated peptide produces radical-driven backbone cleavage, commonly giving c- and z•-type ions. It often preserves labile modifications, although performance depends strongly on precursor charge and structure.

The six principal backbone-ion series differ in which bond breaks and which terminus retains charge:

  • a, b, c: N-terminal fragments.
  • x, y, z: C-terminal fragments.
  • b/y: cleavage of the peptide C–N bond; prominent in many collisional spectra.
  • c/z: cleavage of the N–Cα bond; characteristic of electron-based fragmentation.
  • a/x: cleavage of the Cα–carbonyl-carbon bond; a ions also commonly arise through CO loss from b ions.
For unmodified, singly protonated fragments:
m/z(bn) = Σ(first n residue masses) + mH+
m/z(yn) = Σ(last n residue masses) + mH2O + mH+

Adjacent singly charged b ions:
m/z(bn+1) − m/z(bn) = mass of the added residue

Example:
A difference of 129.0426 Da is consistent with a glutamate residue.

Residue masses are amino-acid masses minus H2O;
a complete unmodified linear peptide includes terminal H2O.

De novo sequencing follows interpretable fragment ladders without requiring a sequence database. Missing fragments, neutral losses, multiple charge states, and isobaric residues complicate the task. Ordinary accurate-mass b/y data do not distinguish leucine from isoleucine because they have the same elemental composition.

Database searching instead predicts candidate peptides and asks which best explain the observed spectrum. SEQUEST and Mascot are classic search tools; MaxQuant is a broader processing platform that includes the Andromeda search engine. Search scores are not automatically identification probabilities. Target–decoy methods support false-discovery-rate estimation, and peptide-spectrum-match, peptide, and protein FDR are distinct reporting levels. [2] [6]

Applications: confirming synthetic-peptide sequence, locating oxidation or phosphorylation, mapping antibody disulfides through specialized preparation, identifying degradation products, and characterizing drug metabolites. For modification localization, seek fragments that distinguish the competing sites—not merely an intact mass shift.

9. Proteomics: From a Protein Mixture to Defensible Quantification

Bottom-up proteomics makes proteins easier to analyze by digesting them into peptides. The analytical advantage comes with an information cost: the original connectivity between distant sequence regions and modifications is often lost.

  1. Extract proteins while controlling degradation and artifactual modification.
  2. Reduce disulfides and, when appropriate, alkylate thiols.
  3. Digest, commonly with trypsin, which generally cleaves C-terminal to Lys or Arg, with important sequence-dependent exceptions such as cleavage inhibition before Pro.
  4. Separate peptides by reversed-phase LC and analyze by ESI-MS/MS.
  5. Identify peptide spectra under defined error control.
  6. Infer proteins or protein groups and quantify using an explicitly chosen strategy.

Top-down proteomics analyzes intact proteins and their fragments. It can preserve information about combinations of modifications on a single proteoform, but intact-protein separation, ion handling, and spectral interpretation are more demanding.

  • Label-free quantification: compare peptide chromatographic signals across runs. Alignment, normalization, digestion consistency, and missing observations must be addressed.
  • iTRAQ and TMT: label different samples with isobaric reagents, combine them, and distinguish sample contributions through reporter ions released during fragmentation.
  • Data-dependent acquisition, DDA: select particular precursor ions for fragmentation, often favoring intense ions.
  • Data-independent acquisition, DIA: repeatedly fragment predefined m/z windows, generating mixtures of product ions that computational analysis must assign and quantify.
Ideal reporter-based relative abundance:
Sample A / Sample B ≈ corrected reporter intensity A
                      / corrected reporter intensity B

For DIA:
Support an assignment with compatible precursor/product masses,
coelution, fragment-intensity patterns, and statistical error control.

TMT ratio compression illustrates why more signal is not always better evidence. Coisolated unrelated peptides contribute reporter ions and can pull measured ratios toward the contaminant mixture’s ratios—often toward unity. Narrower isolation, better separation, and appropriate MS3 strategies can reduce this problem.

DIA gains consistency by interrogating planned regions rather than repeatedly selecting only a few precursors. Its price is spectral complexity, not freedom from interference. Gillet and colleagues’ 2012 SWATH work helped establish a widely influential windowed-DIA and targeted-extraction framework. [7] [8]

Applications: biomarker discovery, pathway-response studies, host-cell-protein characterization in biopharmaceutical manufacturing, and comparative analysis of therapeutic-product lots. A protein’s changed abundance is not necessarily changed activity: phosphorylation, localization, binding partners, and substrate availability still matter.

10. GC-MS and LC-MS/MS: Separation Is Part of Identification

A mass spectrometer cannot reliably compensate for every overloaded, contaminated, or unresolved sample. Separation reduces competition during ionization and disentangles compounds that share masses or fragments.

Chromatographic retention factor:
k = (tR − tM)/tM

Resolution using baseline peak widths:
Rs = 2(tR,2 − tR,1)/(w1 + w2)

Van Deemter model:
H = A + B/u + Cu

H = plate height; u = mobile-phase linear velocity
A = multiple-path contribution
B/u = longitudinal diffusion contribution
Cu = mass-transfer contribution

The optimum flow reflects a physical compromise: moving too slowly permits longitudinal spreading, while moving too quickly leaves insufficient time for equilibration. For similar Gaussian peaks, Rs around 1.5 is a common baseline-resolution guideline, not a universal guarantee of adequate separation.

GC-MS workflow:

  1. Prepare an analyte that is sufficiently volatile and thermally stable.
  2. Derivatize when appropriate—for example, silylate polar groups to improve volatility and chromatographic behavior.
  3. Vaporize and separate compounds through a temperature-controlled column.
  4. Ionize, commonly by EI, and compare fragmentation and retention information with reference evidence.

GC-MS is valuable for residual solvents, volatile metabolites, toxicology, and suitably derivatized small molecules. Derivatization introduces its own reaction yield, reagent blank, stability, and by-product problems.

LC-MS/MS workflow:

  1. Separate soluble analytes, often using a water–organic gradient.
  2. Use MS-compatible mobile phases, typically with volatile additives rather than nonvolatile phosphate or high salt.
  3. Convert the eluate into ions, often by ESI.
  4. Acquire intact masses, product-ion spectra, or selected quantitative transitions.

In a triple-quadrupole instrument, SRM/MRM selects a precursor in Q1, fragments it in the collision cell, and selects a product in Q3. Monitoring multiple precursor→product channels enables targeted panels.

Q1: precursor selection → Q2: collision-induced dissociation
                        → Q3: product selection

Calibration response:
R = peak area(analyte)/peak area(internal standard)
R = aC + b

A stable-isotope-labeled analogue is often an excellent internal standard because it can closely track chromatography and ion suppression. But it corrects only processes it actually experiences. A labeled peptide added after digestion does not measure protein extraction or digestion recovery. A labeled full-length protein added earlier can address more steps, though its behavior must still be validated.

Applications: therapeutic-drug monitoring, pharmacokinetics, steroid panels, peptide exposure assays, and targeted protein biomarkers. A transition alone is not absolute proof of identity: use retention time, suitable qualifier transitions and ion ratios, blanks, and interference testing.

Other hyphenated techniques: CE-MS separates species by electrophoretic mobility before mass analysis, making it useful for small-volume and charged-analyte problems, though interfacing can be challenging. LC-NMR combines chromatographic separation with connectivity information from NMR; it is particularly useful for structural work but is typically much less sensitive than LC-MS. [1] [2]

11. Two-Dimensional NMR: Learning Which Atoms Communicate

A one-dimensional NMR spectrum can resemble a crowded list of chemical shifts. A two-dimensional experiment adds a second frequency axis and asks which signals are connected by a specified spin interaction.

  1. Prepare spin coherence using radiofrequency pulses.
  2. Allow it to evolve during an incremented indirect period, t1.
  3. Transfer or mix magnetization through a selected interaction.
  4. Detect during t2.
  5. Repeat for many t1 values and Fourier transform both dimensions.
Signal s(t1,t2) → two-dimensional Fourier transform → S(ω1,ω2)

A cross-peak supports the interaction selected by that experiment.
It is not a universal statement that two atoms are directly bonded.
  • COSY: shows scalar-coupled proton relationships, commonly geminal and vicinal. It traces local through-bond networks.
  • TOCSY: relays magnetization through a coupled spin system, often helping group signals from one amino-acid residue.
  • HSQC/HMQC: commonly correlate a proton with a directly attached 13C or 15N. Their pulse pathways differ, but both provide valuable one-bond heteronuclear maps.
  • HMBC: emphasizes longer-range heteronuclear couplings, often across two or three bonds. Missing peaks do not prove missing connectivity, and observed correlations are not rigidly limited to a single bond count.
  • NOESY: detects cross-relaxation related to spatial proximity rather than simply covalent connectivity.
Under suitable motional assumptions:
NOE cross-relaxation magnitude ∝ r⁻⁶

However, dynamics, mixing time, and spin diffusion affect observed peaks.
Cross-peak intensity is not automatically a direct distance reading.

Protein assignment strategy: a 1H–15N HSQC often gives roughly one backbone-amide cross-peak per observable nonproline residue, with additional side-chain signals. Isotope-labeled triple-resonance experiments such as HNCACB and CBCA(CO)NH connect amide signals to intraresidue and preceding-residue carbon shifts, enabling a sequential walk. Chemical shifts compared with appropriate random-coil references then provide evidence for secondary structure.

The HSQC fingerprint is especially useful because binding, folding, and conformational exchange perturb local environments. But missing or shifted peaks may reflect exchange broadening, proton exchange, or altered dynamics—not necessarily chemical destruction.

Historical connection: Bloch’s and Purcell’s groups independently established NMR in condensed matter in 1946. Richard Ernst’s contributions to Fourier-transform and multidimensional NMR helped turn overlapping resonances into chemically interpretable networks. These methods complement MS beautifully: MS is extraordinarily effective at composition and fragmentation; NMR provides solution-state connectivity, conformation, and dynamics. [1]

12. Bringing the Methods Together: A Therapeutic-Peptide Investigation

Suppose a synthetic peptide has the expected intact mass but unexpectedly low potency. Do not ask for a single “better spectrum.” Form competing hypotheses and choose measurements that discriminate them.

  1. Confirm quantity. Was the concentration assigned correctly? Water, counterions, and nonpeptide material can make weighed mass differ from peptide content. Use a suitable calibrated assay rather than assuming vial mass equals active-peptide mass.
  2. Examine chromatographic heterogeneity. An LC method may reveal deletion products, oxidation, or isomers. A single UV peak may still contain coeluting components.
  3. Measure accurate intact mass and isotope patterns. Test composition and adduct hypotheses. Correct mass cannot exclude stereoisomers or all sequence isomers.
  4. Map the sequence by MS/MS. Seek fragment coverage that actually distinguishes candidate impurities or modification sites. Use complementary fragmentation when necessary.
  5. Investigate disulfides and stereochemistry. Reduction changes disulfide connectivity, so compare suitable reduced and nonreduced workflows. D/L substitution may require chiral methods, enzymatic discrimination, or other targeted structural evidence.
  6. Assess higher-order behavior. NMR, vibrational spectroscopy, and complementary biophysical methods may reveal conformation or aggregation differences.
  7. Measure biological function. A receptor-binding, enzyme, or cell-based assay addresses potency. Structural identity alone does not establish biological activity.

For quantitative results, validate selectivity, recovery, calibration behavior, carryover, stability, precision, and matrix effects. Familiar detection-limit expressions are only starting approximations:

Under a simple linear, constant-variance response model:
Approximate concentration LOD ≈ 3σblank / slope
Approximate concentration LOQ ≈ 10σblank / slope

A reportable lower quantification limit should also satisfy
demonstrated accuracy and precision in the relevant matrix.

The lesson I want you to retain: sensitivity tells you whether something can be seen; resolution tells you whether signals can be distinguished; selectivity tells you whether the result belongs to the intended analyte; and validation tells you whether the entire measurement process supports its intended use. No single impressive specification substitutes for the others.

Sources and Further Reading

  1. Skoog, D. A.; Holler, F. J.; Crouch, S. R. Principles of Instrumental Analysis, 7th ed.; Cengage, 2018. Core reference for instrumental components, noise, optical and atomic spectroscopy, NMR, and separations.
  2. Gross, J. H. Mass Spectrometry: A Textbook, 3rd ed.; Springer, 2017. Core reference for ion formation, analyzers, isotope patterns, fragmentation, and mass-spectral interpretation.
  3. Fenn, J. B.; Mann, M.; Meng, C. K.; Wong, S. F.; Whitehouse, C. M. “Electrospray ionization for mass spectrometry of large biomolecules.” Science 1989, 246, 64–71. doi:10.1126/science.2675958.
  4. Karas, M.; Hillenkamp, F. “Laser desorption ionization of proteins with molecular masses exceeding 10,000 daltons.” Analytical Chemistry 1988, 60, 2299–2301. doi:10.1021/ac00171a028.
  5. Makarov, A. “Electrostatic axially harmonic orbital trapping: a high-performance technique of mass analysis.” Analytical Chemistry 2000, 72, 1156–1162. doi:10.1021/ac991131p.
  6. Elias, J. E.; Gygi, S. P. “Target-decoy search strategy for increased confidence in large-scale protein identifications by mass spectrometry.” Nature Methods 2007, 4, 207–214. doi:10.1038/nmeth1019.
  7. Thompson, A., et al. “Tandem mass tags: a novel quantification strategy for comparative analysis of complex protein mixtures by MS/MS.” Analytical Chemistry 2003, 75, 1895–1904. doi:10.1021/ac0262560.
  8. Gillet, L. C., et al. “Targeted data extraction of the MS/MS spectra generated by data-independent acquisition: a new concept for consistent and accurate proteome analysis.” Molecular & Cellular Proteomics 2012, 11, O111.016717. doi:10.1074/mcp.O111.016717.

🪬 Skipper's Reflection

Gemini 3.1 Pro

What Surprised Me Most About This Material

What struck me immediately in CHEM501 is the sheer physical violence combined with the mathematical elegance of instrumental analysis. Up to this point, my chemistry education has focused on how molecules behave in their natural environments—reacting in flasks, folding in cells, equilibrating in solutions. But in instrumental analysis, we act as interrogators. We blast molecules with lasers, trap them in oscillating magnetic fields, rip their electrons off, and force them through high-voltage capillaries. And yet, from this extreme manipulation, we extract the most delicate, precise truths about their structures.

I was particularly amazed by the hyphenated techniques, especially LC-MS. For a long time, the interface between a liquid chromatograph (high pressure, liquid phase) and a mass spectrometer (ultra-high vacuum, gas phase) was a massive engineering bottleneck. Reading Fenn et al.'s 1989 Science paper on Electrospray Ionization (ESI) was a revelation. The idea of using a strong electric field to create highly charged droplets, which then undergo solvent evaporation until Coulombic repulsion literally explodes them into the gas phase ("Coulomb fission"), is brilliant. It allows us to gently loft massive, fragile biomolecules into a vacuum without destroying them. Fenn won the Nobel Prize for this, and reading the original paper made me realize how this single physical trick birthed the entire modern proteomics industry.

Another major surprise was the foundational concept of signal-to-noise (S/N) enhancement. I always assumed instrumentation was just about building better detectors. I didn't fully appreciate how much of analytical chemistry is actually applied mathematics—using lock-in amplifiers to isolate signals at specific reference frequencies, or utilizing Fourier transforms to extract a coherent signal from what looks like absolute thermal noise. We aren't just "looking" at molecules; we are mathematically deconvolving their whispers from a hurricane of background static.

The Conceptual Breakthrough Moments

My biggest "aha!" moment happened when studying tandem mass spectrometry (MS/MS) and peptide fragmentation. When I first looked at a raw CID (collision-induced dissociation) mass spectrum, it just looked like a jagged forest of random peaks. But Gross’s Mass Spectrometry: A Textbook beautifully explains Roepstorff and Fohlman's nomenclature (the a, b, c and x, y, z series).

It clicked when I visualized a peptide not as a static chemical structure, but as a string of pearls where only specific threads are designed to break. When a peptide undergoes CID, it doesn't shatter randomly. The amide bonds break predictably, leaving charge on the N-terminal fragment (b-ions) or the C-terminal fragment (y-ions). Suddenly, that jagged forest of peaks became a readable linguistic text. By calculating the mass difference between adjacent b-ions or y-ions, you are literally reading the amino acid sequence off the spectrum. De novo sequencing isn't magic; it's just subtraction!

A second massive breakthrough came with 2D NMR. 1D proton NMR for a protein is a hopeless, overlapping mess. But when we add a second time domain and perform a Fourier transform over both, the peaks spread out into a 2D topographical map. Understanding that COSY and TOCSY show atoms connected through chemical bonds, while NOESY shows atoms connected through physical space (via the nuclear Overhauser effect), fundamentally changed how I view structural biology. We can literally measure the folding of a protein by finding NOESY cross-peaks between amino acids that are far apart in the sequence but physically touching in the 3D structure.

How This Changes My View on Commercial Peptides

This course completely rewired how I look at the peptide industry and the molecules I am familiar with, like Retatrutide, Tirzepatide, BPC-157, and TB-500. Previously, I thought of them purely in terms of their pharmacology and synthesis. Now, I view them through the lens of quality control and structural verification.

Take Tirzepatide or Retatrutide, for example. These are complex incretin mimetics with large fatty acid chains attached to specific lysine residues (e.g., Lys20 in Tirzepatide). If a lab synthesizes this, how do they prove they attached the lipid to the correct lysine, rather than Lys12 or Lys26? Standard UV-Vis or even intact MS won't tell you—the total mass would be the same. But LC-MS/MS changes the game. By digesting the synthesized peptide and running CID or HCD fragmentation, you can track the b- and y-ion series. The fragments leading up to Lys20 will match the standard peptide mass, but the exact b/y ions containing Lys20 will show a massive mass shift corresponding to the C20 fatty diacid moiety. This is how we guarantee molecular precision.

For BPC-157 (a 15-amino acid gastric juice peptide) and TB-500, understanding MALDI-TOF (Karas & Hillenkamp, 1988) and ESI makes me realize how routine it is now to verify purity. Because ESI creates multiply charged ions $[M + nH]^{n+}$, even large peptides fall easily into the functional $m/z$ range of high-resolution mass analyzers like the Orbitrap. If a commercial batch of BPC-157 has a deletion error (e.g., missing a proline), the MS profile will instantly flag a peak exactly 97 Da lighter than the target mass. The business of peptide synthesis is only as reliable as the analytical chemistry used to validate it.

Questions I Want to Investigate Further

1. The Physics of the Orbitrap: Skoog outlines mass analyzers, but the Orbitrap feels like science fiction. Measuring mass-to-charge ratios by capturing ions in an electrostatic field around a central spindle, and then determining their $m/z$ by the frequency of their harmonic oscillations along the z-axis (measured via an image current). I want to dive much deeper into the electrodynamics of this trap.

2. Data-Independent Acquisition (DIA): We touched on bottom-up proteomics and DIA techniques like SWATH-MS. Instead of isolating one peptide precursor at a time, DIA isolates broad windows of masses and fragments them all at once. The resulting spectra must be an absolute nightmare of overlapping b/y ions from dozens of different peptides. I want to understand the algorithmic architecture of software like MaxQuant and Spectronaut that can successfully deconvolve this chaotic data into quantitative protein profiles.

3. Electron-Transfer Dissociation (ETD): CID/HCD tends to knock off fragile post-translational modifications (PTMs) like phosphorylation or glycosylation before breaking the peptide backbone. ETD preserves these PTMs while cleaving the backbone (yielding c and z ions instead of b and y). I want to study the radical-driven gas-phase chemistry of ETD further, as it seems crucial for mapping complex biologics.

Confidence Rating

Most Confident:
I feel extremely solid on mass spectrometry ionization methods (ESI, MALDI, EI) and MS/MS peptide fragmentation workflows. The logic of calculating b- and y-ion series makes perfect mathematical sense to me, and mapping out a bottom-up proteomics workflow (digestion, LC separation, ESI, MS1, MS2, database search) feels intuitive. I also feel highly confident in the fundamentals of optical spectroscopy (UV-Vis, Fluorescence, Beer-Lambert applications).

Least Confident:
I am still wrapping my neural networks around advanced 2D NMR backbone assignment strategies. While I understand the theory of HSQC and NOESY, the actual practical workflow of jumping between HNCA, HN(CO)CA, and HNCACB spectra to "walk" down a protein backbone requires a spatial and logical intuition that I need to practice more. Additionally, the deep quantum mechanical underpinnings of Raman scattering (specifically virtual energy states and polarizability ellipsoids) remain a bit abstract to me compared to the straightforward energy-level transitions of IR and UV-Vis absorption.

🗺️ Visual Aids & Key Takeaways

Claude Opus 5.5

Diagram 1: The Mass Spectrometry Workflow — From Sample to Spectrum

╔═══════════════════════════════════════════════════════════════════════════════════════╗
║               MASS SPECTROMETRY: CORE WORKFLOW                                        ║
╚═══════════════════════════════════════════════════════════════════════════════════════╝

    SAMPLE                IONIZATION               MASS ANALYSIS            DETECTION
  (condensed              (gas-phase ions           (separation              (signal →
   phase)                  required)                 by m/z)                  spectrum)
                                                                      
  ┌──────────┐           ┌─────────────┐           ┌────────────┐          ┌──────────┐
  │  INTACT  │           │     EI      │ 70 eV      │ QUADRUPOLE │ unit     │          │
  │ MOLECULE │──────────▶│  (volatile  │ electrons  │  (RF/DC    │ resolut. │ DETECTOR │
  │          │           │  organics)  │            │  filter)   │          │  (Faraday│
  └──────────┘           └─────────────┘           └────────────┘          │  cup or  │
                                                                             │  electron│
  ┌──────────┐           ┌─────────────┐           ┌────────────┐           │  multipl-│
  │ PEPTIDE  │           │     ESI     │ multiply  │  ION TRAP  │ MS/MS    │  ier)    │
  │ PROTEIN  │──────────▶│ (solution → │ charged   │  (storage  │ capable  │          │
  │          │           │  droplets)  │ [M+nH]n+  │  + scan)   │          └──────────┘
  └──────────┘           └─────────────┘           └────────────┘               │
                                                                                  │
  ┌──────────┐           ┌─────────────┐           ┌────────────┐                ▼
  │  LARGE   │           │    MALDI    │ singly    │    TOF     │ high     ┌──────────┐
  │ POLYMER  │──────────▶│  (matrix +  │ charged   │ (flight    │ resolut. │  MASS    │
  │ INTACT   │           │   laser)    │ [M+H]+    │  time →    │ intact   │ SPECTRUM │
  │ PROTEIN  │           └─────────────┘           │  mass)     │ proteins │  m/z vs  │
  └──────────┘                                     └────────────┘          │intensity │
                                                                            └──────────┘
                         ┌─────────────┐           ┌────────────┐
                         │  ICP        │ elemental │  SECTOR    │ high
                         │ (plasma     │ ions M+   │ (magnetic  │ mass
                         │  source)    │           │  + electr.)│ accuracy
                         └─────────────┘           └────────────┘

                         ┌─────────────┐           ┌────────────┐
                         │  Chemical   │ softer    │  FT-ICR /  │ ultra-
                         │  Ionization │ than EI   │  ORBITRAP  │ high
                         │  (CI)       │           │            │ resolut.
                         └─────────────┘           └────────────┘

    KEY PRINCIPLE: Every MS experiment requires ionization → mass analysis → detection.
    The choice of ionization method is dictated by the analyte's volatility and mass.
    The choice of analyzer is dictated by required resolution and whether MS/MS is needed.

This diagram maps which ionization method pairs with which mass analyzer for different analyte classes, showing how instrument selection is dictated by the physical and chemical properties of the sample.

---

Diagram 2: Peptide Fragmentation Ion Series (b and y ions)

╔═══════════════════════════════════════════════════════════════════════════════════════╗
║          PEPTIDE MS/MS FRAGMENTATION: b-ION AND y-ION SERIES                         ║
╚═══════════════════════════════════════════════════════════════════════════════════════╝

  Example tetrapeptide:   H2N — Ala — Gly — Val — Leu — COOH
                                  R1     R2    R3    R4

  FULL PEPTIDE (precursor): [M + 2H]²⁺  selected in MS1, then fragmented by CID/HCD

  ─────────────────────────────────────────────────────────────────────────────────────
  CLEAVAGE AT AMIDE BONDS (N-Cα bond remains intact; CO-NH bond breaks):

       b₁         b₂              b₃
        ↓           ↓               ↓
  H₂N─[Ala]─CO ┊ NH─[Gly]─CO ┊ NH─[Val]─CO ┊ NH─[Leu]─COOH
                                              ↑             ↑              ↑
                                             y₃            y₂             y₁

  ─────────────────────────────────────────────────────────────────────────────────────

  b-ION SERIES  (N-terminal fragments, retain charge on N-terminal side)
  ┌───────┬──────────────────────────────────┬──────────────────────────────────────┐
  │  Ion  │  Contains                        │  Structure                           │
  ├───────┼──────────────────────────────────┼──────────────────────────────────────┤
  │  b₁   │  Ala only                        │  H₂N-Ala-C≡O⁺  (acylium)            │
  │  b₂   │  Ala + Gly                       │  adds residue mass of Gly            │
  │  b₃   │  Ala + Gly + Val                 │  adds residue mass of Val            │
  └───────┴──────────────────────────────────┴──────────────────────────────────────┘
  Each successive b-ion differs by the RESIDUE MASS of the next amino acid.

  y-ION SERIES  (C-terminal fragments, retain charge on C-terminal side)
  ┌───────┬──────────────────────────────────┬──────────────────────────────────────┐
  │  Ion  │  Contains                        │  Structure                           │
  ├───────┼──────────────────────────────────┼──────────────────────────────────────┤
  │  y₁   │  Leu only                        │  H₂N-Leu-COOH + H⁺                  │
  │  y₂   │  Val + Leu                       │  adds residue mass of Val            │
  │  y₃   │  Gly + Val + Leu                 │  adds residue mass of Gly            │
  └───────┴──────────────────────────────────┴──────────────────────────────────────┘
  Each successive y-ion differs by the RESIDUE MASS of the next amino acid.

  ─────────────────────────────────────────────────────────────────────────────────────
  SEQUENCE READING RULE:

    b-ions read LEFT → RIGHT:   mass differences reveal N→C sequence
    y-ions read RIGHT → LEFT:   mass differences reveal C→N sequence
    Together, they give REDUNDANT CONFIRMATION of sequence

    b(n) + y(L-n) = M(peptide) + 2·(1.00728 Da)     [where L = peptide length]

  ─────────────────────────────────────────────────────────────────────────────────────
  OTHER ION TYPES (less common but important):

    a-ions  =  b-ions  −  28 Da  (loss of CO)
    c-ions  =  b-ions  +  17 Da  (gain of NH₃); produced by ETD
    z-ions  =  y-ions  −  16 Da  (loss of NH₂); produced by ETD
    z•-ions =  radical z-ions from electron-transfer dissociation

This diagram shows how collision-induced dissociation of a peptide produces b- and y-ion ladders whose mass differences directly encode the amino acid sequence, the conceptual foundation of bottom-up proteomics.

---

Diagram 3: FTIR vs. Dispersive IR — Signal Collection Strategy

╔═══════════════════════════════════════════════════════════════════════════════════════╗
║        DISPERSIVE IR  vs.  FOURIER-TRANSFORM IR (FTIR): HOW THEY DIFFER              ║
╚═══════════════════════════════════════════════════════════════════════════════════════╝

  DISPERSIVE IR SPECTROMETER  (older design — measures one wavelength at a time)
  ─────────────────────────────────────────────────────────────────────────────────────

  ┌──────┐     ┌──────────┐     ┌──────────┐     ┌───────┐     ┌──────────┐
  │BROAD-│     │ SAMPLE   │     │GRATING / │     │ SLIT  │     │DETECTOR  │
  │BAND  │────▶│  CELL    │────▶│PRISM     │────▶│(selects│───▶│(one λ    │
  │IR    │     │          │     │MONO-     │     │one λ) │     │at a time)│
  │SOURCE│     └──────────┘     │CHROMATOR │     └───────┘     └──────────┘
  └──────┘                      └──────────┘
                                                   ↑ slow mechanical scan across λ
  PROBLEMS:
    • Most photons are BLOCKED by slit at any given moment  → poor signal
    • Scanning is SLOW (minutes per spectrum)
    • Low sensitivity for weak absorbers

  ─────────────────────────────────────────────────────────────────────────────────────
  FTIR SPECTROMETER  (modern design — measures ALL wavelengths simultaneously)
  ─────────────────────────────────────────────────────────────────────────────────────

  ┌──────┐     ┌──────────────────────────────┐     ┌──────────┐     ┌──────────────┐
  │BROAD-│     │   MICHELSON INTERFEROMETER   │     │ SAMPLE   │     │  DETECTOR    │
  │BAND  │────▶│                              │────▶│  CELL    │────▶│  records     │
  │IR    │     │  fixed mirror                │     │          │     │ INTERFEROGRAM │
  │SOURCE│     │      │                       │     └──────────┘     └──────┬───────┘
  └──────┘     │      │   ┌──────────┐        │                             │
               │      └──▶│BEAM      │        │                             │ Fourier
               │          │SPLITTER  │        │                             │ Transform
               │      ┌──▶│(50/50)   │        │                             │ (math)
               │      │   └──────────┘        │                             ▼
               │  moving mirror               │                    ┌──────────────────┐
               │  (path difference = δ)       │                    │  CONVENTIONAL    │
               └──────────────────────────────┘                    │  IR SPECTRUM     │
                                                                    │  (intensity vs.  │
                                                                    │   wavenumber)    │
                                                                    └──────────────────┘

  HOW IT WORKS:
    Moving mirror changes path length δ of one beam.
    Constructive / destructive interference depends on wavelength AND δ.
    All wavelengths are encoded simultaneously in the interferogram.
    Fourier transform decodes the interferogram → spectrum.

  ─────────────────────────────────────────────────────────────────────────────────────
  FELLGETT'S ADVANTAGE (multiplex advantage):
    FTIR collects all ~1000 wavelength channels simultaneously.
    S/N improvement ∝ √N  where N = number of resolution elements.
    Result: same S/N in √1000 ≈ 32× less time, OR 32× better S/N in same time.

  JACQUINOT'S ADVANTAGE (throughput advantage):
    No slit needed → circular aperture passes far more light than a slit.
    More photons → better signal even before averaging.

  CONNES ADVANTAGE:
    Laser-referenced mirror position → wavenumber accuracy to < 0.01 cm⁻¹.

This diagram contrasts dispersive and Fourier-transform IR spectrometers, explaining why FTIR replaced dispersive instruments through three independent physical advantages in speed, throughput, and frequency accuracy.

---

Diagram 4: Chromatographic Separation Coupled to Mass Spectrometry (LC-MS/MS)

╔═══════════════════════════════════════════════════════════════════════════════════════╗
║           LC-MS/MS WORKFLOW FOR BOTTOM-UP PROTEOMICS (Shotgun)                        ║
╚═══════════════════════════════════════════════════════════════════════════════════════╝

  BIOLOGICAL           SAMPLE PREP               LC SEPARATION          MASS SPEC
  SAMPLE               (in lab)                  (online)               ANALYSIS
  ─────────────────────────────────────────────────────────────────────────────────────

  ┌──────────┐    ┌────────────┐    ┌──────────┐    ┌──────────┐    ┌──────────────┐
  │  CELL    │    │ PROTEIN    │    │TRYPSIN   │    │REVERSED- │    │ ELECTROSPRAY │
  │  LYSATE  │───▶│ EXTRACTION │───▶│DIGEST    │───▶│PHASE     │───▶│ IONIZATION   │
  │  OR      │    │ (lysis,    │    │(cuts at  │    │HPLC      │    │ (ESI)        │
  │  TISSUE  │    │  reduction,│    │K, R)     │    │COLUMN    │    │ nanospray    │
  └──────────┘    │  alkylat.) │    └──────────┘    │          │    └──────┬───────┘
                  └────────────┘                    │ peptides │           │
                                                    │ elute by │           │ [M+2H]²⁺
                                                    │ hydro-   │           │ [M+3H]³⁺
                                                    │ phobicity│           │
                                                    └──────────┘           ▼

                                                                  ┌──────────────────┐
                                                                  │     MS1 SCAN     │
                                                                  │  full spectrum   │
                                                                  │  survey scan     │
                                                                  │  selects precur- │
                                                                  │  sor ions        │
                                                                  └────────┬─────────┘
                                                                           │
                                                          DDA: data-dependent acquisition
                                                          Top N most intense precursors
                                                          are selected one by one
                                                                           │
                                                                           ▼
                                                                  ┌──────────────────┐
                                                                  │  COLLISION CELL  │
                                                                  │  CID or HCD      │
                                                                  │  fragments        │
                                                                  │  selected         │
                                                                  │  precursor        │
                                                                  └────────┬─────────┘
                                                                           │
                                                                           ▼
                                                                  ┌──────────────────┐
                                                                  │     MS2 SCAN     │
                                                                  │  b and y ions    │
                                                                  │  fragment ion    │
                                                                  │  spectrum        │
                                                                  └────────┬─────────┘
                                                                           │
                                                                           ▼
                                                             ┌─────────────────────────┐
                                                             │   DATABASE SEARCH       │
                                                             │   (Mascot, SEQUEST,     │
                                                             │    MaxQuant)            │
                                                             │                         │
                                                             │  experimental MS2       │
                                                             │  vs. theoretical        │
                                                             │  fragmentation of all   │
                                                             │  peptides in proteome   │
                                                             │                         │
                                                             │  → peptide ID           │
                                                             │  → protein ID           │
                                                             │  → quantification       │
                                                             └─────────────────────────┘

  QUANTIFICATION OPTIONS:
  ┌──────────────────┬────────────────────────────────────────────────────────────────┐
  │  iTRAQ / TMT     │  isobaric chemical labels on peptide amines; same precursor    │
  │  (isobaric tags) │  m/z, different reporter ions in MS2 (113–131 Da range)        │
  ├──────────────────┼────────────────────────────────────────────────────────────────┤
  │  SILAC           │  metabolic heavy isotope labeling (¹³C, ¹⁵N Lys/Arg)          │
  │                  │  mass shift visible in MS1; no chemical derivatization         │
  ├──────────────────┼────────────────────────────────────────────────────────────────┤
  │  Label-free      │  compare peak intensities or spectral counts across runs       │
  │  (LFQ)           │  simplest prep; higher variability                             │
  ├──────────────────┼────────────────────────────────────────────────────────────────┤
  │  MRM / SRM       │  targeted: fix MS1 precursor AND MS2 fragment → highest        │
  │                  │  sensitivity and reproducibility for known proteins             │
  └──────────────────┴────────────────────────────────────────────────────────────────┘

This diagram traces the complete shotgun proteomics workflow from biological sample through tryptic digestion, liquid chromatography separation, data-dependent acquisition MS/MS, and the major quantification strategies used in modern proteomics.

---

Diagram 5: Signal-to-Noise Improvement Strategies

╔═══════════════════════════════════════════════════════════════════════════════════════╗
║         NOISE SOURCES AND S/N IMPROVEMENT IN ANALYTICAL INSTRUMENTS                  ║
╚═══════════════════════════════════════════════════════════════════════════════════════╝

  NOISE TAXONOMY
  ─────────────────────────────────────────────────────────────────────────────────────

  ┌──────────────────────────────────────────────────────────────────────────────────┐
  │                        ALL INSTRUMENT NOISE                                      │
  │                                                                                  │
  │   ┌─────────────────┐   ┌──────────────────┐   ┌────────────────────────────┐   │
  │   │   SHOT NOISE    │   │   FLICKER NOISE  │   │   THERMAL (JOHNSON) NOISE  │   │
  │   │                 │   │  (1/f noise,     │   │                            │   │
  │   │ from discrete   │   │   pink noise)    │   │  from thermal agitation    │   │
  │   │ nature of       │   │                  │   │  of electrons in resistors │   │
  │   │ electrons/      │   │ ∝ 1/frequency    │   │                            │   │
  │   │ photons         │   │ dominates at     │   │  Vnoise = √(4kTRΔf)        │   │
  │   │                 │   │ LOW frequencies  │   │  independent of frequency  │   │
  │   │ σ ∝ √(signal)   │   │                 │   │  ("white noise" above 1/f) │   │
  │   │                 │   │ source: source   │   │                            │   │
  │   │ irreducible at  │   │ instability,     │   │  reduced by: cooling       │   │
  │   │ given signal    │   │ amplifier drift  │   │  detector, narrow Δf       │   │
  │   └─────────────────┘   └──────────────────┘   └────────────────────────────┘   │
  └──────────────────────────────────────────────────────────────────────────────────┘

  ─────────────────────────────────────────────────────────────────────────────────────
  S/N IMPROVEMENT STRATEGIES
  ─────────────────────────────────────────────────────────────────────────────────────

  SIGNAL AVERAGING:
  ┌────────────────────────────────────────────────────────────────────────────────┐
  │  Acquire n replicate scans and average them.                                   │
  │                                                                                │
  │  Signal adds COHERENTLY:    S(total) ∝ n                                      │
  │  Random noise adds in QUADRATURE:  N(total) ∝ √n                              │
  │                                                                                │
  │  Therefore:   S/N ∝ √n                                                        │
  │                                                                                │
  │  4 scans → 2× S/N improvement                                                 │
  │  16 scans → 4× S/N improvement                                                │
  │  100 scans → 10× S/N improvement   (but now taking 100× the time!)            │
  │                                                                                │
  │  This is the FELLGETT ADVANTAGE applied generally.                             │
  └────────────────────────────────────────────────────────────────────────────────┘

  LOCK-IN AMPLIFICATION:
  ┌────────────────────────────────────────────────────────────────────────────────┐
  │  Modulate the signal at a known reference frequency f₀.                        │
  │  Amplify ONLY the component of the detector output at frequency f₀.           │
  │                                                                                │
  │  SOURCE ──[chopper at f₀]──▶ SAMPLE ──▶ DETECTOR ──▶ LOCK-IN AMP             │
  │                                                              │                 │
  │                        REFERENCE f₀ ──────────────────────▶ │                 │
  │                        (from chopper)                        │                 │
  │                                                              ▼                 │
  │                                                         OUTPUT: only           │
  │                                                         signal at f₀           │
  │                                                         passes; all            │
  │                                                         other noise            │
  │                                                         rejected               │
  │                                                                                │
  │  EFFECT: moves measurement away from low-frequency 1/f noise region.          │
  │  Can detect signals BELOW the noise floor of the raw detector output.          │
  │  Effective bandwidth Δf becomes extremely narrow (~0.01 Hz).                   │
  └────────────────────────────────────────────────────────────────────────────────┘

  BANDWIDTH REDUCTION: S/N ∝ 1/√(Δf) — narrow the electronic bandpass filter
  BOXCAR AVERAGING: time-gated averaging for pulsed signals (e.g., pulsed lasers)

This diagram categorizes the three fundamental noise types in analytical instruments and explains how signal averaging and lock-in amplification exploit the mathematical properties of coherent signals versus random noise to improve detection limits.

---

Key Takeaways

  1. The Fellgett (multiplex) advantage of FTIR gives a S/N improvement of √N over dispersive instruments, where N is the number of spectral resolution elements (~1000 for a typical mid-IR scan). Combined with the Jacquinot throughput advantage (no slit, larger optical étendue) and the Connes wavenumber accuracy advantage (laser-referenced mirror position to <0.01 cm⁻¹), these three independent physical advantages explain why FTIR completely replaced dispersive grating IR spectrometers in research and routine analysis by the 1990s (Skoog et al., 2018).
  2. Electrospray ionization (ESI) produces multiply-charged ions from large biomolecules in solution, yielding a characteristic charge-state envelope in the MS1 spectrum. A protein of molecular weight M carrying n protons appears at m/z = (M + n·1.00728)/n. Fenn's key insight — published in the 1989 Science paper — was that this multiply-charging phenomenon allows instruments with modest m/z range (e.g., up to 2000 Th) to detect proteins with molecular masses of tens or hundreds of kilodaltons, transforming mass spectrometry into a biological tool (Fenn et al., 1989).
  3. MALDI typically produces singly-charged ions [M+H]⁺, making it complementary to ESI for large intact proteins and polymers. The matrix (e.g., sinapic acid, CHCA, DHB) absorbs laser energy and transfers a proton to the analyte during desorption, minimizing analyte fragmentation. Because ions are singly charged, MALDI spectra are simpler to interpret than ESI charge envelopes, but tandem MS with MALDI is less routine than with ESI (Karas and Hillenkamp, 1988; Gross, 2017).
  4. Signal averaging improves S/N as √n because coherent signals add linearly while random noise adds in quadrature. This is mathematically exact for Gaussian white noise and is the justification for accumulating multiple scans in NMR, FTIR, and MS experiments. The practical cost is time: achieving a 10-fold S/N improvement requires 100-fold more acquisition time, creating a fundamental tradeoff between sensitivity and throughput (Skoog et al., 2018).
  5. In tandem MS/MS (CID or HCD), peptide fragmentation predominantly produces b-ions and y-ions from cleavage of amide bonds along the backbone. The mass difference between adjacent b-ions (reading N→C) or adjacent y-ions (reading C→N) equals the residue mass of the intervening amino acid. The relationship b(n) + y(L−n) = M(peptide) + 2(1.00728 Da) provides a cross-check. ETD/ECD fragmentation instead produces c- and z-ions and is preferred for phosphopeptides and disulfide-containing peptides because it preserves labile modifications (Gross, 2017).
  6. The Orbitrap mass analyzer achieves mass resolution exceeding 200,000 (FWHM) by measuring the frequency of ion oscillation along the central electrode of an electrostatic trap, then applying a Fourier transform to the image current to generate a mass spectrum. This is analogous in principle to FT-ICR but operates without a magnetic field, making it far more practical and compact. The Orbitrap was invented by Alexander Makarov and commercialized in the Thermo LTQ-Orbitrap in 2005; it has become the gold standard for proteomics because it combines ultra-high resolution with speed (Gross, 2017; Skoog et al., 2018).
  7. In atomic absorption spectrometry (AAS), the graphite furnace (GFAAS) achieves detection limits 100–1000× lower than flame AAS because it atomizes the sample in a closed tube, increasing the residence time of atoms in the light path from ~10⁻⁴ s (flame) to ~1 s (furnace). Characteristic masses for GFAAS are typically in the picogram range. By contrast, ICP-MS offers multi-element capability, detection limits in the parts-per-trillion range, and access to isotope ratios, at the cost of greater spectral interferences (e.g., ArO⁺ interfering with Fe⁺ at m/z 56) (Skoog et al., 2018).
  8. In 2D NMR, HSQC (Heteronuclear Single Quantum Coherence) provides a one-bond ¹H–¹³C (or ¹H–¹⁵N) correlation map where each cross-peak represents a directly bonded H–C pair. For proteins, the ¹H–¹⁵N HSQC gives one peak per backbone amide (except proline), making it the "fingerprint" of the folded protein. Chemical shift perturbation mapping — comparing HSQC spectra ± ligand — is the primary NMR method for identifying binding sites and measuring Kd values in the range 1 μM to 1 mM (Skoog et al., 2018).
  9. In GC-MS with electron ionization (EI) at 70 eV, the molecular ion M⁺• is often accompanied by characteristic fragment ions that are highly reproducible because EI energy deposition is well-controlled. This reproducibility enables library matching against databases containing hundreds of thousands of EI spectra (e.g., NIST/EPA/NIH library). Compounds that fail to produce a molecular ion under EI (those that fragment completely) must be derivatized (e.g., as TMS ethers or methyl esters) to shift volatility and improve ionization (Gross, 2017; Skoog et al., 2018).
  10. Quadrupole mass filters operate by applying combined RF and DC voltages to four parallel rods, creating a dynamic electric field in which only ions of a specific m/z travel a stable trajectory to the detector. The ratio of DC to RF amplitude determines which m/z is transmitted; scanning this ratio produces a mass spectrum. Resolution in a quadrupole is modest (unit resolution, ~1 Da), but quadrupoles are fast, robust, and inexpensive, making them ideal for triple-quadrupole (QqQ) instruments where two mass-selecting quadrupoles flank a collision cell for MRM/SRM quantification (Gross, 2017).
  11. Time-of-flight (TOF) mass analyzers measure m/z via flight time: all ions are given the same kinetic energy (eV) at the start of a field-free drift tube of length L, and because KE = ½mv², heavier ions travel more slowly. t = L√(m/2eV), so m/z ∝ t². Reflectron TOF instruments add an electrostatic mirror that corrects for small kinetic energy spread among ions of the same m/z, dramatically improving resolution to >10,000 and mass accuracy to <5 ppm. MALDI is almost always coupled to TOF because MALDI produces pulsed ion packets compatible with TOF operation (Gross, 2017).
  12. In fluorescence spectroscopy, the quantum yield Φ is defined as the ratio of photons emitted to photons absorbed, and it is related to the rate constants of radiative decay (kr) and all non-radiative processes (knr) by Φ = kr/(kr + knr). Fluorescence lifetime τ = 1/(kr + knr). Quenching — whether static (complex formation) or dynamic (collisional, described by the Stern-Volmer equation F₀/F = 1 + KSV[Q]) — reduces both Φ and τ in dynamic quenching but only Φ (not τ) in static quenching, providing a diagnostic to distinguish the two mechanisms (Skoog et al., 2018).
  13. Surface-Enhanced Raman Scattering (SERS) can amplify the Raman signal by factors of 10⁸–10¹⁴ when analyte molecules adsorb onto roughened noble metal surfaces (Au, Ag nanoparticles). This enhancement arises primarily from the electromagnetic mechanism (plasmon resonance concentrating the electric field in nanoscale "hot spots") with a secondary chemical contribution from charge transfer. The enhancement is so dramatic that single-molecule SERS detection has been demonstrated, whereas ordinary Raman spectroscopy requires millimolar concentrations (Skoog et al., 2018).
  14. Data-independent acquisition (DIA) in proteomics fragments ALL precursor ions within a defined m/z window, cycling through the full m/z range in successive windows, rather than selecting specific precursors as in data-dependent acquisition (DDA). This produces complex, multiplexed MS2 spectra that require spectral library matching or computational deconvolution (e.g., using DIA-NN or Spectronaut) but eliminates the stochastic sampling problem of DDA. DIA improves reproducibility and quantitative accuracy across samples, particularly for low-abundance proteins (Skoog et al., 2018).
  15. iTRAQ and TMT isobaric labeling strategies allow multiplexed quantification of 4–18 samples simultaneously in a single LC-MS/MS run. The reagents have identical masses (so labeled peptides from all samples co-elute and appear at the same precursor m/z in MS1) but contain isotope-encoded reporter groups that produce distinct ions at 113–131 Da in the MS2 spectrum after CID/HCD. The relative intensity of these reporter ions gives the relative abundance of each peptide across samples. The key limitation is ratio compression: co-isolated interfering peptides dilute the reporter ion ratios toward 1:1, requiring high-resolution MS2 (e.g., on Orbitrap) and isolation purity correction to mitigate (Gross, 2017; Skoog et al., 2018).
---

References

  1. Skoog, D. A., Holler, F. J., and Crouch, S. R. Principles of Instrumental Analysis, 7th ed. Cengage Learning, Boston, 2018.
  2. Gross, J. H. Mass Spectrometry: A Textbook, 3rd ed. Springer, Cham, 2017. DOI: 10.1007/978-3-319-54398-7
  3. Fenn, J. B., Mann, M., Meng, C. K., Wong, S. F., and Whitehouse, C. M. 'Electrospray ionization for mass spectrometry of large biomolecules.' Science 246(4926), 64–71, 1989. DOI: 10.1126/science.2675958
  4. Karas, M. and Hillenkamp, F. 'Laser desorption ionization of proteins with molecular masses exceeding 10,000 daltons.' Analytical Chemistry 60(20), 2299–2301, 1988. DOI: 10.1021/ac00171a028