Chapter 4 Instrumental Analysis
Instrumental settings determine how much chemical information can be captured from a sample and how reliably those signals can be quantified. In GC-MS and LC-MS workflows, full-scan acquisition is often preferred for broad profiling because each scan collects a mass spectrum across the selected m/z range. If the scan range is narrowed while scan time is held constant, each m/z value receives more sampling time and sensitivity improves. If the scan range is widened, sensitivity per m/z decreases. Increasing scan time can also improve sensitivity, but it may compromise chromatographic resolution.
Full-scan acquisition occurs synchronously with chromatographic separation. To preserve peak shape, each chromatographic peak should usually be sampled at least 10 times. For example, if two peaks are separated by 10 s and the half peak width is 5 s, the instrument should acquire roughly 10 spectra within that time window, corresponding to a cycle time near 1 s. If a high-resolution column produces peaks with half widths near 1 s, the cycle time must be closer to 0.2 s. Shorter dwell or cycle times generally reduce sensitivity, so there is always a trade-off between separation and signal quality. In UPLC workflows, where separation may finish within 20 min, it is especially important to check whether the mass spectrometer can keep pace without undersampling peaks. A recent study(Cai and Yan 2021) suggested that, under optimized conditions, even 6 points can still recover peak shapes similar to those obtained with 20 points.
4.1 Ionization mode selection
Ionization mode is one of the most important instrumental decisions in metabolomics because it determines which compounds can be detected efficiently and how severe matrix effects may become. No single source works best for the whole metabolome.
ESI+ is commonly used for compounds that protonate easily, such as amines, many lipids, peptides, and some xenobiotics. It is often sensitive but can be strongly affected by salt and matrix effects.
ESI- is often preferred for acidic compounds such as organic acids, phosphorylated metabolites, bile acids, and many phenolic compounds. It may provide cleaner spectra for some metabolite classes that ionize poorly in positive mode.
APCI is often more suitable for less polar and more thermally stable compounds. It may show reduced ion suppression compared with ESI for some neutral or weakly polar analytes.
APPI is useful for relatively non-polar compounds and certain lipids, steroids, and aromatic compounds that are difficult to ionize efficiently by ESI.
As a practical rule:
use ESI for most broad LC-MS metabolomics workflows
use both ESI+ and ESI- when broader coverage is needed
consider APCI or APPI when the target analytes are less polar, when ESI suffers strong suppression, or when lipid and xenobiotic coverage is unsatisfactory
The pretreatment solvent, chromatographic condition, and analyte class should all be considered together with the ionization mode rather than chosen independently.
4.2 Column and gradient selection
For GC, higher temperature could release compounds with higher boiling point. For LC, gradient and functional groups of stationary phase would be more important than temperature. Polarity of samples and column should match. More polar solvent could release polar compounds. Normal-phase column will not retain non-polar compounds while reversed-phase will elute polar column in the very beginning. To cover a wide polarity range or logP value compounds, normal phase column should match with non-polar to polar gradient to get a better separation of polar compounds while reverse phase column should match with polar to non-polar gradient to elute compounds. If you use an inappropriate order of gradient, your compounds would not be separated well. If you have no idea about column and gradient selection, check literature’s condition. Meanwhile, the pretreatment methods should fit the column and gradient selection. You will get limited information by injection of non-polar extracts on a normal phase column and nothing will be retained on column. This study showed improved chromatography conditions will improve the annotation results(Anderson et al. 2021). You can also install polar and non-polar columns and run separation on one column while condition on another one, which could extend the chemical coverage(Flasch et al. 2022).
Meta-analysis of chromatographic methods in EBI metabolights and NIH Workbench could be a guide for lab without experience on metabolomics chromatographic methods(Harrieder et al. 2022).
This work introduce Sequential Quantification using Isotope Dilution (SQUID), a method combining serial sample injections into a continuous isocratic mobile phase, enabling rapid analysis of target molecules with high accuracy, as demonstrated by detecting microbial polyamines in human urine samples with an LLOQ of 106 nM and analysis times as short as 57 s, thus proposing SQUID as a high-throughput LC–MS tool for quantifying target biomarkers in large cohorts(Groves et al. 2023).
4.3 Mass resolution
For metabolomics, high resolution mass spectrum should be used to make identification of compounds easier. The Mass Resolving Power is very important for annotation and high resolution mass spectrum should be calibrated in real time. The region between 400–800 m/z was influenced the most by resolution(Najdekr et al. 2016). Orbitrap Fusion’s performance was evaluated here(Barbier Saint Hilaire et al. 2018), as well as the comparison with Fourier transform ion cyclotron resonance (FT-ICR)(Ghaste et al. 2016; Huang et al. 2021). Mass Difference Maps could recalibrate HRMS data (Smirnov et al. 2019).
4.4 Instrument comparison
Different mass analyzers provide different trade-offs in resolution, scan speed, sensitivity, dynamic range, and suitability for quantitative work.
| Instrument type | Main strengths | Main limitations | Typical role in metabolomics |
|---|---|---|---|
| Quadrupole | Simple, robust, good filtering | Low resolving power in full scan | Front-end filtering, targeted acquisition |
| Triple quadrupole (QQQ) | Excellent sensitivity and quantitative performance in MRM/SRM | Limited untargeted discovery capability | Targeted quantification and validation |
| Q-TOF | Good balance of high resolution, accurate mass, and MS/MS speed | Usually lower quantitative precision than QQQ for targeted assays | Untargeted profiling and MS/MS acquisition |
| Orbitrap | High mass accuracy and high resolving power | Scan speed and duty cycle may become limiting at very high resolution | Untargeted profiling, formula assignment, annotation |
| FT-ICR | Extremely high resolving power and mass accuracy | Expensive, slower, less common in routine workflows | Advanced structural analysis and specialized applications |
In general, QQQ is preferred when the primary goal is robust quantification of known metabolites, while Q-TOF and Orbitrap platforms are commonly preferred for untargeted metabolomics because they support broad profiling and annotation. The choice is therefore linked to study objective rather than instrument prestige alone.
4.5 Scan speed and dwell time
Practical scan-speed planning is essential because the instrument must sample chromatographic peaks often enough to preserve peak shape and enable reliable integration. A useful rule is to collect at least 8–12 points across a chromatographic peak, although optimized workflows may sometimes work with fewer(Cai and Yan 2021).
For example:
if a peak is 12 s wide, a cycle time of about 1 s still gives about 12 points
if a peak is 3 s wide, the cycle time should be closer to 0.25–0.35 s
if UPLC produces sub-second peak widths, very high-resolution full scans or too many MS/MS events may undersample the peak
For full-scan HRMS, slower scans usually increase ion statistics and sensitivity, but they reduce the number of points across a peak. For data-dependent MS/MS, the total cycle time includes both the survey scan and all triggered fragmentation scans. Therefore, too many MS/MS events per cycle can damage quantification and peak detection in MS1.
For targeted methods such as MRM, dwell time should be long enough to maintain sensitivity but short enough to monitor all transitions with enough peak definition. If too many transitions are monitored at once, each transition gets too little time and quantitative precision will decrease. Scheduled MRM can reduce this burden by monitoring each transition only around its expected retention time window.
Therefore, scan speed should always be tuned together with chromatographic peak width, m/z range, resolving power, and acquisition mode. Fast chromatography without compatible mass spectrometer timing will reduce data quality rather than improve throughput.
4.6 Quantitative Metabolomics
Quantification is a fundamental challenge in mass spectrometry-based metabolomics. The level of quantification achievable depends on the analytical strategy (targeted vs. untargeted) and available resources (standards, labeled compounds). Understanding the quantification hierarchy helps select appropriate methods for specific research questions(Alseekh et al. 2021).
4.6.1 Quantification levels
Quantification in metabolomics can be categorized into several levels with decreasing accuracy:
Absolute quantification uses calibration curves built with authentic chemical standards spiked into a matrix-matched blank. Stable isotope-labeled internal standards (SIL-IS) are the gold standard for correcting matrix effects and ion suppression. This approach provides concentrations in physical units (e.g., nmol/L) and is essential for clinical validation and cross-laboratory comparison.
Semi-quantification uses calibration curves from structurally related surrogate standards or class-representative compounds when authentic standards are not available. The accuracy is lower, but it still provides approximate concentration values. Recent work has explored ionization efficiency prediction to enable semi-quantification without standards(Liigand et al. 2020; Kruve 2020).
Relative quantification compares peak areas or intensities across samples without converting to concentration units. This is the default mode in untargeted metabolomics. Proper normalization and batch correction (see Chapter 9) are critical to ensure that relative differences reflect biology rather than technical variation.
4.6.2 Calibration strategies
For targeted quantification, several calibration strategies are commonly used:
External calibration: Standard solutions of known concentration are analyzed separately from samples. Simple to implement but does not correct for matrix effects.
Standard addition: Known amounts of standard are added to the sample matrix itself. Each sample effectively gets its own calibration curve, making this method robust against sample-specific matrix effects but labor-intensive.
Isotope dilution: Stable isotope-labeled analogues of target compounds are spiked into each sample at known concentrations. The ratio of unlabeled (endogenous) to labeled (spiked) signal is used for quantification. This is the most accurate method because the labeled standard experiences the same matrix effects and losses as the analyte(Roberts et al. 2012).
Single-point calibration is a simplified version of external or isotope dilution calibration with only one standard concentration level. It can be sufficient when the expected concentration range is narrow and the response is linear.
4.6.3 Untargeted quantification challenges
In untargeted metabolomics, absolute quantification of all detected features is impractical because authentic standards are not available for most unknowns. Several strategies have been developed to bridge this gap(Cajka and Fiehn 2016a; Guo and Huan 2024):
Response factor estimation: Ionization efficiency depends on physicochemical properties (e.g., polarity, molecular size). Machine learning models trained on known compounds can predict ionization efficiency for unknowns, enabling rough concentration estimates.
Class-level quantification: Using one or a few representative standards per compound class (e.g., one phospholipid standard for all phospholipids) provides order-of-magnitude estimates.
QC-based normalization followed by relative quantification: While not providing absolute concentrations, this approach combined with proper experimental design can reliably identify biologically significant changes(Broadhurst, Goodacre, Stacey N. Reinke, et al. 2018a).
4.6.4 Practical recommendations
For newcomers to quantitative metabolomics(Ribbenstedt et al. 2018; Alseekh et al. 2021):
If your goal is biomarker discovery, relative quantification with proper QC and normalization is usually sufficient as a first step. Absolute quantification can follow for validated candidates.
Always include at least a few internal standards across different compound classes, even in untargeted studies. These serve as anchors for quality control and semi-quantification.
Document the dynamic range of your instrument for key metabolite classes. Signals outside the linear range require dilution series or alternative quantification approaches.
Be aware that different ionization modes (ESI+, ESI-, APCI) show very different response factors for the same compound. Quantitative comparisons across ionization modes are not straightforward.
4.7 Matrix effects
Matrix effects could decrease the sensitivity of untargeted analysis. Such matrix effects could be checked by low resolution mass spectrometry(Yu et al. 2017) and found for high resolution mass spectrometry(Calbiani et al. 2006). Ion suppression should also be considered as a critical issue comparing heterogeneous metabolic profiles(Ghosson et al. 2021). This work discussed the matrix effects after Trimethylsilyl derivatization(Tarakhovskaya et al. 2023). The study(Dagan et al. 2023) investigated how the complexity of matrices affects nontargeted detection using LC-MS/MS analysis, finding that detection limits for trace compounds were significantly influenced by matrix complexity, with higher concentrations required for detection within the “top 1000” list compared to the first 10,000 peaks, suggesting a negative power law functional relationship between peak location and concentration; the research also demonstrated a correlation between power law coefficient and dilution factor, while showcasing the distribution of matrix peaks across various matrices, providing insights into the capabilities and limitations of LC-MS in analyzing nontargets in complex matrices.