Ep 51 — Common Troubleshooting and Real-World Pitfalls

Series: Encyclopedia of Infrared Spectroscopy: From Principles to Practice
Chapter: Part 5 · Instruments and Tools — Troubleshooting and Maintenance (Ep 46–55 Second Half)
Target Audience: Laboratory testing technicians, instrument managers, QC/QA supervisors, new graduate students in labs
Prerequisites: Ep 11 (Dispersive vs FTIR), Ep 12 (Michelson Interferometer), Ep 16 (Sample Preparation), Ep 17 (Instrument Operation), Ep 18 (Spectrum Processing), Ep 50 (Daily Maintenance and ASTM Performance Verification)
Reading Time: Approximately 40 minutes

Introduction: That "Flat" Spectrum

One Monday morning, Xiao Zhang at the QC lab opened the Nicolet iS20 and pressed "Collect Sample." The interferogram signal on the screen was normal, but the resulting spectrum—a line almost hugging 100% T. She thought the sample wasn't placed correctly, tried again, but it was still flat. Changed personnel, changed sample, changed pellet—still the same. A whole morning passed, and the entire QC team became anxious: Is the instrument broken? Is an accessory loose? Should we report for repair immediately?

The engineer arrived at the scene in the afternoon. He blocked the sample compartment beam path with a piece of white paper—the interferogram barely changed. Then he opened the instrument cover and used a multimeter to measure the two ends of the source—open circuit. It turned out that the instrument had been left on over the weekend, and the Globar source had reached the end of its life. Replaced the source, realigned, collected a background, and the spectrum returned to normal after 10 minutes.

"In FTIR troubleshooting, first look at the interferogram, then the spectrum is a golden rule. The interferogram is the 'ECG' of the instrument, telling you whether the problem is in the optical path, electronics, or data acquisition."
— Adapted from Thermo Fisher Application Engineer Training Manual [1]

This is a typical case of a "seemingly complex, yet simple" fault. This episode will systematically explain the fault tree methodology for common FTIR faults, compile anonymized real-world pitfall cases, and provide preventive maintenance strategies, enabling you to calmly and methodically locate faults when issues arise, rather than blindly calling for repair or randomly changing parameters.

💡 Positioning of this episode: Ep 50 discusses 'how to prevent instrument failure', while this episode covers 'what to do when it breaks, and the pitfalls that seem like failures but aren't'. Together, they form an essential toolbox for instrument managers.

1. General Approach to Fault Troubleshooting: Fault Tree Method

FTIR is an opto-mechatronic system, and faults can occur at any link in the chain: source → interferometer (beamsplitter/moving mirror) → sample compartment/accessory → detector → signal amplification and acquisition → software [1][2]. Blindly replacing parts or adjusting parameters is both costly and may introduce new problems.

The core of the fault tree method: Start from the phenomenon, eliminate in the order of 'optical, mechanical, electronic, software', and verify each hypothesis with minimal cost [1][2][3].

Abnormal spectrum
   |
   ▼
1. Phenomenon classification (flat / poor SNR / baseline tilt / water peaks / abnormal peaks)
   |
   ▼
2. First check interferogram: signal amplitude? symmetry? interference fringes present?
   |
   ├─ Interferogram abnormal → optical/electronic issue (source, beamsplitter, detector)
   |
   ├─ Interferogram normal but spectrum abnormal → sample/preparation/software issue
   |
   ▼
3. Troubleshoot step by step along optical path:
   Source → beamsplitter → sample compartment → detector → acquisition parameters
   |
   ▼
4. Change only one variable at a time, record changes
   |
   ▼
5. Re-measure + performance verification (polystyrene film 1601 cm⁻¹ verification)

Figure 1: FTIR fault tree troubleshooting flowchart (compiled from [1][2][3])

Below, we expand on the "five typical phenomena" respectively.

2. Phenomenon 1: Flat Spectrum (No Signal)

Typical manifestation: The spectrum is nearly a straight line between 95%–100% T, with no sample peaks visible; sometimes it is completely flat at 100% T [1].

2.1 Fault Tree

Flat spectrum
   |
   ├─ Interferogram also near zero?
   |   |
   |   ├─ Yes → source/beamsplitter/detector fault (optical path blocked)
   |   |   |
   |   |   ├─ Source resistance infinite? → source damaged (most common)
   |   |   ├─ Source lit but signal weak? → beamsplitter fogging/tilted due to moisture
   |   |   ├─ Detector no signal? → detector not powered / preamplifier failure
   |   |   └─ Optical path blocked? → sample compartment cover not closed / obstruction / aperture closed
   |   |
   |   └─ No → software setting issue (background and sample order wrong, subtraction error)

Figure 2: Fault tree for flat spectrum

2.2 Troubleshooting Steps (From Low to High Cost)

Step 1 — Check the interferogram [1][2]

• Is the center burst amplitude normal (> 1 V)? Optical path is clear;
• Interferogram amplitude very small or no burst? Optical path issue.

Step 2 — Physical inspection (without disassembly) [1]

• Is the sample compartment cover fully closed?
• Is the aperture set to minimum (some software 'low throughput' mode)?
• Is an accessory blocking the beam (e.g., ATR not pressed down, DRIFTS mirror misaligned)?
• Detector preamplifier indicator light (MCT usually has an LED).

Step 3 — Source test (after power off and cooling) [1][4]

• Use a multimeter to measure resistance across the source terminals:
  • Globar: normal 1–5 Ω, open circuit indicates damage;
  • Nichrome wire (Nernst not common, some old instruments) similar.
• Observe if the source is lit (red glow visible through instrument vents, but do not look directly).
• Globar lifetime is typically 1000–2000 hours; frequent switching shortens life [4].

Step 4 — Beamsplitter check [1][4]

• KBr/Ge beamsplitter: after moisture absorption, the surface becomes foggy and white → signal drops significantly or disappears;
• Remove the beamsplitter and observe under dry nitrogen; if visible fogging, replacement is needed;
• Beamsplitter tilt (due to transportation, impact) can also cause loss of interferogram, requiring realignment (using the instrument's built-in Align function).

Step 5 — Detector [1]

• DTGS detector has no external power supply, relies on the pyroelectric effect; very weak signal is usually a preamplifier or wiring issue;
• MCT detector requires liquid nitrogen cooling; liquid nitrogen depletion is a common 'false fault'—add liquid nitrogen and wait 15 minutes to stabilize before measurement;
• Loose MCT wiring or Dewar vacuum failure can also result in no signal.

Step 6 — Software settings [1][2]

• Was the background collected before the sample? Is the background file outdated?
• Was the 'subtraction' mode accidentally selected, resulting in zero?
• Are the resolution/number of scans settings appropriate?

💡 Engineer's experience: 90% of 'flat' cases ultimately boil down to ① source failure, ② beamsplitter moisture absorption, ③ detector liquid nitrogen depletion or not powered. Check these three first to resolve most issues [1][4].

3. Phenomenon 2: Poor Signal-to-Noise Ratio

Typical manifestation: Spectrum shows peaks but with many spikes; small peaks are buried in noise; SNR test (e.g., peak-to-peak noise at 2200–2000 cm⁻¹) is far below instrument specifications [1][2].

3.1 Physical Nature of SNR

SNR is proportional to the square root of the number of scans $N$ [1][2]:

$$\text{SNR} \propto \sqrt{N \cdot T_{\text{single}}}$$

The instrument specification "$2200–2000 \text{ cm}^{-1}$ SNR > 50000:1 (1 min scan, 4 cm⁻¹)" is typically measured under maximum throughput, optimal alignment, and fresh background [1][5]. Any degradation in these factors reduces SNR.

3.2 Fault Tree

Poor SNR
   |
   ├─ Insufficient throughput?
   |   ├─ Source aging (reduced brightness) → replace source
   |   ├─ Beamsplitter slight moisture → dry / replace beamsplitter
   |   ├─ Optical path misaligned → realign
   |   ├─ Aperture too small → increase aperture
   |   └─ Mirror contamination → clean (professional operation)
   |
   ├─ Improper acquisition parameters?
   |   ├─ Too few scans → increase to 32/64
   |   ├─ Resolution too high (1 cm⁻¹) → reduce to 4 cm⁻¹
   |   └─ Incorrect detector gain setting
   |
   ├─ Detector issues?
   |   ├─ DTGS aging → replace
   |   ├─ MCT insufficient liquid nitrogen or overheating → replenish liquid nitrogen
   |   └─ MCT saturation after warming → cool for 15 min
   |
   └─ Sample issues?
       ├─ Too little sample (pellet too thin) → add sample
       ├─ ATR poor contact → increase pressure

Figure 3: Fault tree for poor SNR

3.3 Troubleshooting Steps

Step 1 — Check throughput [1][2]

• In software, display the single-beam spectrum of the background; the maximum signal level (e.g., at 2000 cm⁻¹) should be within typical range (e.g., > 10 V for DTGS).
• Compare with historical data to see if there is a significant drop.

Step 2 — Check alignment [1]

• Run the instrument's alignment routine (e.g., 'Align' in OMNIC).
• If alignment fails or signal does not improve, suspect beamsplitter or mirror issues.

Step 3 — Check acquisition parameters [1][2]

• Increase number of scans (increase by factor of 4 to double SNR).
• Reduce resolution (4 cm⁻¹ is standard for most samples).
• Ensure detector gain is set appropriately (e.g., 'Auto gain' for DTGS; MCT often uses fixed gain).

Step 4 — Check detector status [1]

• For DTGS: verify that the detector is locked in and contacts are clean.
• For MCT: check if liquid nitrogen level is sufficient; refill if needed, wait 15 min.
• If MCT signal is saturated (e.g., overload indicator), allow it to cool.

Step 5 — Check sample preparation [1][2]

• For transmission: ensure pellet is not too thick or thin; sample amount should be 0.5–2% for KBr pellets.
• For ATR: ensure good contact between sample and crystal; increase pressure if necessary.
• For DRIFTS: ensure sample surface is flat and filled level.

💡 Tips: Many 'poor SNR' problems are resolved by collecting a fresh background and increasing number of scans [2]. Avoid blaming the instrument too early.

4. Phenomenon 3: Baseline Tilt

Typical manifestation: Baseline gradually rises or falls from high to low wavenumber, or appears curved, significantly affecting spectral interpretation and quantification [1][2].

4.1 Physical Causes

Baseline tilt is typically caused by light scattering (due to sample particle size, surface roughness) or instrument instability (e.g., source drift, detector nonlinearity) [1][2]. For ATR, poor contact or improper pressure can also cause baseline tilt.

4.2 Fault Tree

Baseline tilt
   |
   ├─ Sample related?
   |   ├─ Particle size too large (scattering) → grind finer, use smaller aperture
   |   ├─ Sample uneven thickness → reprepare
   |   ├─ ATR poor contact → increase pressure or clean crystal
   |   └─ DRIFTS surface uneven → refill sample cup
   |
   ├─ Instrument related?
   |   ├─ Source drift (on after short warm-up) → warm up for 30 min
   |   ├─ Beamsplitter moisture (tilt worsens over time) → dry instrument
   |   ├─ Detector nonlinearity (MCT saturated) → reduce throughput or gain
   |   ├─ Purge insufficient (CO2/H2O fluctuation) → improve purge
   |   └─ Mirror contamination → clean

Figure 4: Fault tree for baseline tilt

4.3 Troubleshooting Steps

Step 1 — Check sample preparation [1][2]

• Observe whether sample scattering is severe (visible turbidity in pellet).
• For ATR, re-clean the crystal and increase pressure; ensure sample covers the crystal area.
• For DRIFTS, ensure the sample cup is filled flush with the edge.

Step 2 — Collect a new background [1]

• If baseline tilt disappears after a fresh background, the issue is likely instrument drift or purge fluctuation.

Step 3 — Check instrument warm-up and purge [1][2]

• Ensure the source has been on for at least 30 minutes.
• Check CO2/H2O levels in the spectrum; if high, improve purge (increase flow rate, seal leaks).

Step 4 — Check detector linearity [1]

• For MCT, reduce aperture or use a neutral density filter to lower throughput; if nonlinear, replace detector.
• For DTGS, nonlinearity is rare but possible if damaged.

Step 5 — Perform baseline correction in software (temporary fix) [2]

• Use automatic baseline correction (e.g., polynomial fit) to remove tilt, but note that this does not address the root cause.

💡 Note: Baseline tilt due to scattering can be minimized by using a smaller aperture or sample dilution [2]. For quantitative analysis, it is best to eliminate physical causes rather than rely on software correction.

5. Phenomenon 4: Water and CO2 Peaks

Typical manifestation: Sharp peaks near 2300–2400 cm⁻¹ (CO2) and 1600, 3700 cm⁻¹ (H2O) appear in the spectrum, obscuring sample peaks [1][2].

5.1 Sources

• Insufficient purge: Ambient air contributes to the spectrum.
• Wet sample: Moisture in the sample or KBr.
• Background expired: If the background was collected under different purge conditions.
• Leaks in purge line.

5.2 Troubleshooting Steps

Step 1 — Check purge [1][2]

• Ensure the instrument is purged with dry air or nitrogen for at least 10-15 minutes.
• Check for leaks at connections; use a flow meter to verify gas flow.
• If using a desiccant dryer, ensure it is regenerated.

Step 2 — Collect a new background [1]

• After purge has stabilized, collect a fresh background and re-measure the sample.

Step 3 — Check sample handling [1][2]

• For solid samples, dry KBr (at 120°C) and sample thoroughly.
• For liquid samples, use a sealed cell or dry solvent.
• For ATR, ensure the crystal is clean and dry.

Step 4 — Software subtraction (temporary) [2]

• Use CO2/H2O compensation algorithms in software, but this is not recommended for quantitative work.

💡 Best practice: Always monitor the water and CO2 lines in the single-beam background spectrum. If they exceed 10% transmittance, purge is insufficient [1].

6. Phenomenon 5: Unexpected Peaks or Spectral Artifacts

Typical manifestation: Peaks appear that do not belong to the sample; typical artifacts include interference fringes (cosine oscillations), ghost peaks (due to detector nonlinearity), or spikes (cosmic rays) [1][2].

6.1 Common Artifacts and Causes

Table 1: Common spectral artifacts and solutions (compiled from [1][2])

6.2 Troubleshooting Steps

Step 1 — Identify the artifact [1][2]

• Use spectral libraries or knowledge to determine if the peaks are real or artifacts.
• Interference fringes are evenly spaced; can be removed by Fourier filtering.

Step 2 — Check sample thickness [1]

• For transmission modes, ensure the pellet thickness is appropriate (0.5–1 mm).
• If using a thin film, try increasing thickness or using ATR.

Step 3 — Check detector linearity [1]

• If ghost peaks appear, reduce the light level (aperture or neutral density filter) and see if they disappear.

Step 4 — Check for cosmic rays [1][2]

• Spikes are random; collect multiple scans and average, or use spike detection software.

Step 5 — Re-collect background [1]

• If negative peaks or accessory bands appear, ensure the background is collected under identical conditions (with the same accessory configuration, without sample, with proper purge).

💡 Pro tip: Always use the same accessory and sample holder for background and sample measurements. If changing accessories, re-collect the background.

7. Real-World Pitfall Case Stories (Anonymized)

Compiled from lab anecdotes and engineering experiences [3][4], these cases illustrate 'non-faults' that mimic failures and common mistakes.

Case 1: 'Liquid nitrogen leak' alarm, but it was just the humidity

Situation: MCT detector kept reporting 'low Dewar vacuum' or 'overheat'. Technicians replaced the detector, but problem persisted.

Root cause: The instrument was in a high-humidity environment (>60% RH). Moisture condensed on the Dewar's cold surface, causing the alarm. After installing a dehumidifier, the problem disappeared.

Lesson: MCT detectors require low humidity (<40% RH) in the surrounding environment. Always consider environmental factors before blaming the detector.

Case 2: ATR spectrum had strong water peaks even after drying

Situation: A diamond ATR spectrum showed strong water bands (~3300 cm⁻¹ and 1640 cm⁻¹) despite sample being thoroughly dried.

Root cause: The technician cleaned the ATR crystal with water-soaked wipes and didn't dry it completely. Water remained in the crystal housing and was absorbed by the sample during measurement.

Lesson: Use only anhydrous solvents (e.g., isopropanol) for cleaning ATR crystals, and dry thoroughly with nitrogen gas.

Case 3: 'No signal' after installing a new beamsplitter

Situation: After replacing the KBr beamsplitter, the interferogram was flat. Technicians suspected the new beamsplitter was defective.

Root cause: The beamsplitter was installed with the orientation reversed. The anti-reflection coating faced the wrong direction, causing near-complete reflection. After flipping, signal returned.

Lesson: Always check the orientation of the beamsplitter with the instrument manual; some beamsplitters are directional.

Case 4: Background collected with sample still in path

Situation: After collecting background and then sample, the spectrum showed 100% absorption at all wavelengths. The software automatically subtracted the background, but since the background already contained the sample, the result was zero.

Root cause: The user forgot to remove the sample from the beam path when collecting background.

Lesson: Always ensure the sample compartment is empty when collecting background.

Case 5: 'Noisy' spectrum due to vibration

Situation: A newly installed instrument produced very noisy spectra. All hardware tests passed.

Root cause: The lab was located above a parking garage. When cars entered or exited, vibrations affected the interferometer. After installing vibration isolation pads under the instrument, the noise disappeared.

Lesson: Environmental vibration is a common source of noise in FTIR. Use vibration isolation tables or pads for critical measurements.

8. Preventive Maintenance Strategies

To minimize faults and pitfalls, implement the following strategies:

8.1 Daily/Weekly Checklist [5]

• [ ] Check source status: power-on time, appearance (no dark spots).
• [ ] Check desiccant: indicator color (blue/green for dry, pink for saturated) and replace if needed.
• [ ] Purge: monitor CO2/H2O levels; keep flow rate at 15–20 L/min for nitrogen.
• [ ] Performance verification: run a polystyrene film daily and check 1601 cm⁻¹ peak position tolerance (±0.05 cm⁻¹).
• [ ] Cleaning: wipe external surfaces; clean sample compartment windows with ethanol.

8.2 Monthly Tasks

• [ ] Align the instrument using the automatic alignment routine.
• [ ] Inspect beamsplitter for fogging or damage.
• [ ] Check purge lines for leaks (soap test).
• [ ] Record throughput at a standard wavenumber (e.g., 2000 cm⁻¹) for historical trend analysis.

8.3 Quarterly Tasks

• [ ] Run ASTM performance tests: wavenumber accuracy, linearity, repeatability.
• [ ] Clean or replace light source if lifetime is approaching (check cumulative operating hours).
• [ ] Check detector: for DTGS, compare signal level with baseline record; for MCT, verify Dewar hold time (typically 8–12 hours).

8.4 Annual Tasks

• [ ] Factory calibration or professional maintenance.
• [ ] Replace consumables: fans, filters, desiccant cartridges.
• [ ] Update software/firmware if applicable.

Table 2: Preventive maintenance schedule (adapted from [5])

Summary

Table 3: Quick reference for common FTIR problems (compiled from [1][2][5])

References

[1] Thermo Fisher Scientific. FTIR Troubleshooting Guide, 2018.

[2] Griffiths, P.R. and de Haseth, J.A. Fourier Transform Infrared Spectrometry, 2nd ed., Wiley, 2007.

[3] Anonymous lab notes from FTIR user groups.

[4] PerkinElmer. FTIR Instrument Maintenance and Troubleshooting, 2015.

[5] ASTM E168-16. Standard Practices for General Techniques of Infrared Quantitative Analysis.

Preview of Ep 52: We will delve into quantitative analysis methods — from Beer-Lambert law to PLS regression, including practical tips for method validation.

└─ Absorbance too strong (black sample) → dilute or change accessory

> Figure 3: Fault tree for poor SNR (compiled from [1][2][5])

### 3.3 Key Troubleshooting Tips

**① Measure SNR** (standardized) [5]
- ASTM E1421 specifies: after background collection, measure the peak-to-peak noise of the 100% T line in the 2200–2000 cm⁻¹ region (where no sample absorbs);
- $\text{SNR} = 100 / \text{peak-to-peak \%T noise}$;
- Compare with the instrument's factory specifications; a difference > 5× indicates a problem.

**② Binary search localization** [1]
- First measure SNR with an **empty sample compartment + fresh background**: if still poor → instrument problem; if good → sample/preparation problem.
- For instrument problems, first realign using the instrument's built-in Align tool, then re-measure SNR; if no improvement, check the source, beamsplitter, and detector in sequence.

**③ Detector temperature** [1]
- MCT heats up after liquid nitrogen is depleted, causing noise to surge by >10×;
- Check if the Dewar still has liquid nitrogen (shake to listen for sound or weigh it);
- After refilling with liquid nitrogen, wait at least 15 minutes for the detector to stabilize, otherwise 1/f noise will appear.

> 🔗 **Further reading**: Strong absorption of water at 3400 and 1640 cm⁻¹ significantly reduces the SNR in those regions. For detailed water molecular vibrations, see [ftir.fun water functional group page](https://ftir.fun/ir/group/water).

---

## Four, Phenomenon 3: Severely Tilted Baseline

**Typical manifestation**: The overall spectrum is not horizontal, high on the left and low on the right or vice versa; sometimes curved in a "smile/frown" shape [1][6].

### 4.1 Physical Origins

The three physical origins of baseline tilt/curvature [1][6][7]:

| Origin | Shape | Typical Scenario |
|--------|-------|------------------|
| **Scattering** | Baseline elevated at high wavenumbers (A ∝ ν⁴) | Coarse KBr pellet particles, uneven powder |
| **Interference fringes** | Sinusoidal ripples | Parallel flat films, parallel liquid cell windows |
| **Sample too thick/strong absorption** | Overall downward pressure, zero transmission | Pellet too thick, liquid path length too long |

> Table 1: Physical origins of baseline anomalies (compiled from [1][6][7])

### 4.2 Fault Tree

Severely tilted baseline
│
├─ Monotonic tilt (high at high wavenumber)?
│ ├─ Scattering dominates → coarse sample particles / uneven pellet
│ ├─ KBr moisture absorption (3400/1640 rise) → dry KBr
│ └─ Large refractive index mismatch between sample particles and KBr
│
├─ Sinusoidal ripples?
│ ├─ Thin film sample with parallel surfaces → slightly wrinkle or use wedged windows
│ ├─ Liquid cell windows parallel → switch to wedge cell
│ └─ Interference fringe period = 1/(2·n·d·cosθ)
│
├─ Overall downward pressure (near 0% T)?
│ ├─ Sample too thick → thin it
│ ├─ Sample absorbance too strong → dilute
│ └─ Totally absorbing region (black sample) → switch to ATR
│
└─ Curved (frown/smile)?
├─ Mie scattering (particles close to wavelength) → grind particles finer
└─ Refractive index change near strong absorption band → real phenomenon, can be post-processed

> Figure 4: Fault tree for severely tilted baseline

### 4.3 Approach to Handling

**Important principle**: Baseline issues should **first be resolved at the sample preparation end, rather than by post-processing** [1][6].

- Severe scattering → re-grind sample + KBr for >2 minutes, particles <2 μm;
- Interference fringes → use a wedged liquid cell or scratch a line on the film to break parallelism;
- Sample too thick → re-make the pellet, controlling 1–2 mg sample + 200 mg KBr;
- True Mie scattering → use specialized Mie correction algorithms (e.g., `mixture` class in pybaselines) for post-processing [6].

Post-processing methods (polynomial, Rubber Band, ALS) have been detailed in Ep 18, so they are not repeated here.

> ⚠️ **Warning**: If the baseline is "severely distorted" (e.g., 1 absorbance unit higher at 4000 cm⁻¹ than at 1000 cm⁻¹), it is most likely a preparation or optical path issue. **Do not just use baseline correction to "flatten" it**—this will mask real faults and introduce large errors in quantitative analysis [1][6].

---

## Five, Phenomenon 4: Water Peak Interference

**Typical manifestation**: Strong water peaks at 3400 cm⁻¹ (broad) and 1640 cm⁻¹ (weak), sometimes masking sample peaks; inconsistent water peak intensity between measurements [1][8].

### 5.1 Sources of Water Peaks

Water molecules are nonlinear asymmetric molecules with three IR-active vibrational modes [8][9]:

- $\nu_1$ symmetric O–H stretch ≈ 3657 cm⁻¹ (gas phase)
- $\nu_3$ antisymmetric O–H stretch ≈ 3756 cm⁻¹ (gas phase)
- $\nu_2$ H–O–H bend ≈ 1595 cm⁻¹ (gas phase)

In liquid water, due to hydrogen bonding, the O–H stretch forms a broad peak at **3400 cm⁻¹**, and the bend at **1640 cm⁻¹** [8][9]. These are the water peaks seen in spectra.

> 🔗 **Further reading**: The three vibrational modes of water and spectral features are detailed at [ftir.fun water functional group page](https://ftir.fun/ir/group/water).

### 5.2 Fault Tree

Water peak interference
│
├─ Insufficient purge?
│ ├─ Instrument not purged with dry gas → connect N₂ / dry air purge for 15 min
│ ├─ Leak in purge gas line → check connections
│ └─ Sample compartment lid opened frequently → reduce openings, let sit before collection
│
├─ KBr not dried?
│ ├─ KBr powder exposed to air → bake at 130 °C for 24 h, store in desiccator
│ ├─ KBr used for pellet contains water → re-bake
│ └─ Pellet die damp → dry the die
│
├─ Ambient humidity too high?
│ ├─ Lab humidity >60% → run dehumidifier to below 40%
│ ├─ Rainy season effect → enhance purge
│ └─ Air conditioner condensation → maintain AC
│
└─ Sample itself contains water?
├─ Sample hygroscopic → dry the sample
├─ Liquid sample contains water → use D₂O or dry solvent
└─ Protein aqueous solution → use ATR with water background subtraction

> Figure 5: Fault tree for water peak interference (compiled from [1][8][9])

### 5.3 Practical Countermeasures

**① Dry KBr** (most basic) [1][8]
- For newly purchased IR-grade KBr powder, immediately bake at 130 °C for 24 h after opening;
- Store the dried KBr in a 130 °C oven or desiccator (indicated by silica gel color change);
- When taking out, do it quickly to avoid prolonged exposure.

**② Continuous purge** [1]
- Purge the optical compartment with dry nitrogen or dry air (dew point < −40 °C) at 5–10 L/min;
- Purge for 15 minutes before collecting background;
- When not measuring for long periods, keep a low flow purge (1–2 L/min).

**③ Atmospheric compensation** [1][2]
- Modern FTIR software (OMNIC, OPUS) has an "Ambient Compensation" function;
- Principle: Real-time monitoring of atmospheric changes between background and sample, automatically subtracting residual CO₂ and water vapor;
- **Note**: Atmospheric compensation is not a substitute for dry purging; it is only used as an auxiliary when residual water vapor fluctuations occur.

**④ Water background subtraction** (e.g., protein aqueous solutions) [8]
- When measuring aqueous solutions with ATR, first collect a pure water background, then the sample;
- Residual water peaks can be further refined using software "water vapor subtraction" functions.

> ⚠️ **Important**: If the background is collected and then the sample is collected 30 minutes later, the concentrations of water vapor and CO₂ will have changed, and residual peaks will inevitably appear. **Rule: The interval between background and sample collection should not exceed 5 minutes** [1].

---

## Six, Phenomenon 5: Abnormal Peaks Appear

**Typical manifestation**: Peaks in the spectrum that should not be present in the sample; unstable positions in repeat measurements, or recurring across multiple different samples [1][10].

### 6.1 Common Sources of Abnormal Peaks

| Abnormal peak position (cm⁻¹) | Source | Identification |
|-------------------------------|--------|----------------|
| 2350, 667                     | CO₂    | Doublet characteristic, disappears after background subtraction |
| 3400, 1640                    | Water vapor / Water | Broad peak or fine lines |

| 1260, 1100, 1020 | Silicone oil / Siloxane | Degradation of detector window or seal |
| 1730 | Phthalate (plasticizer) | Volatilization from plastic cap, PVC hose |
| 1700–1750 | Previous sample residue | Incomplete solvent cleaning, residue on ATR crystal |
| 1380, 870 | Carbonate (CO₃²⁻) | Impurity in KBr or sample contamination |
| 1110, 800, 480 | SiO₂ (quartz) | Agate wear during grinding, mineral sample |

> Table 2: Common sources of abnormal peaks (compiled from [1][10])

### 6.2 Fault Tree

Abnormal peaks appear
│
├─ Appear in multiple measurements of different samples?
│ ├─ Same wavenumber → Instrument contamination (optical path, window, detector)
│ ├─ ATR crystal surface residue → Thoroughly clean with appropriate solvent
│ ├─ Pellet die contamination → Disassemble and clean die
│ └─ Mortar residue → Validate with blank using different samples
│
├─ Appear only in a single measurement?
│ ├─ Impurities in sample itself
│ ├─ Contamination during sample preparation (gloves, tissue, solvent)
│ └─ Previous sample residue → Strengthen cleaning SOP
│
└─ Time-dependent (increasing with measurements)?
├─ Window degradation (KBr in high humidity) → Replace window
├─ Seal aging releasing plasticizer → Replace seal
└─ Aging of plastic parts near light source → Contact engineer
```

Figure 6: Fault tree for abnormal peaks

6.3 Troubleshooting Tips

① Blank verification [1]

• Use pure KBr pellet or empty ATR crystal to collect spectrum, check if abnormal peaks still exist;
• If still present → Instrument/sample preparation tool contamination; If disappeared → Sample issue.

② Solvent ladder flushing (ATR) [1]

• Rinse crystal sequentially with water → ethanol → isopropanol → hexane, verify with spectrum each step;
• Isopropanol effective for organic residues, hexane for oily residues;
• Finally, use a cotton swab with solvent to wipe in one direction to avoid recontamination.

③ Check sample preparation tools [1][10]

• Agate mortar: Clean with water + ethanol + ultrasonic, dry;
• Pellet die: Disassemble all parts, ethanol ultrasonic for 10 min, dry;
• Spatula: Use disposable plastic spatula or dedicated one per material.

7. Classic "Stumbling" Case Collection (Anonymous)

Below are several anonymized real cases, each labeled with "Stumbling reason" and "Avoidance tips".

Case 1: Misidentifying CO₂ peak (2350 cm⁻¹) as sample characteristic

Scenario: A graduate student, while measuring a new porous material, found an "obvious double peak" at 2350 cm⁻¹ and excitedly thought it was a feature of CO₂ adsorption in the material, writing it into a paper draft.

Truth: The advisor asked him to collect background with empty sample compartment and immediately collect sample spectrum; the double peak at 2350 cm⁻¹ still existed; after atmospheric compensation, the double peak completely disappeared [1].

Stumbling reasons:

• Waited 40 minutes after background collection before collecting sample;
• Fume hood just started in the lab, CO₂ concentration fluctuating;
• Unaware that 2350 cm⁻¹ is the characteristic antisymmetric stretching peak of CO₂.

Avoidance tips:

• Remember that 2350 + 667 cm⁻¹ is the "fingerprint" of CO₂;
• Background and sample interval < 5 minutes;
• Turn off fume hood 5 minutes before measurement to stabilize air;
• Enable atmospheric compensation.

🔗 Extension: For detailed analysis of CO₂ vibrational modes and symmetry, see ftir.fun water molecule functional group page (same column has CO₂ analysis link).

Case 2: ATR crystal scratches causing poor repeatability

Scenario: In a QC lab, ATR spectra of three batches showed poor repeatability, with main peak intensity varying by 20%, but the samples themselves were homogeneous within batch.

Truth: Diamond crystal surface had scratches; each time the sample contacted a slightly different position, causing variation in coupling efficiency [1][11].

Stumbling reasons:

• Operator used steel spoon to scoop powder and pressed directly onto crystal;
• During cleaning, cotton swab rubbed back and forth, particles scratched crystal;
• Did not regularly inspect crystal surface with magnifying glass.

Avoidance tips:

• Use PTFE or wooden spoon for sampling, not metal;
• When cleaning, use cotton swab with solvent to wipe in one direction, not back and forth;
• Check crystal surface monthly with 10× magnifying glass; if severe scratches, polish or replace;
• ATR repeatability verification: measure same sample 5 times continuously, RSD of main peak should be < 2% [11].

Case 3: Uneven pellet thickness causing quantitative deviation

Scenario: During validation of a quantitative analysis method, RSD of 1740 cm⁻¹ C=O peak area from 6 parallel pellets was 8%, far exceeding the 2% validation criterion.

Truth: The operator did not spread the powder evenly before pressing, resulting in thickness variation of up to 30% across the pellet [1][6].

Stumbling reasons:

• Powder piled to one side in pellet die;
• Pressurized too quickly, powder pushed to one side;
• Did not use internal standard method to eliminate thickness variation.

Avoidance tips:

• Tap gently after loading powder to level it;
• Press in two steps: first 2 tons pre-press for 1 min, then 8 tons hold for 2 min;
• Release pressure slowly to avoid pellet cracking;
• For quantitative analysis, always normalize with internal standard (KSCN 2050 cm⁻¹) to eliminate thickness error [1].

🔗 Extension: For internal standard method and Beer-Lambert law, see Ep 20. For assignment of 1740 cm⁻¹ C=O peak, see ftir.fun carbonyl functional group page.

Case 4: Ignoring atmospheric compensation leads to false peaks

Scenario: A lab reported a new compound with an "N=C=O isocyanate characteristic peak" near 2300 cm⁻¹, but a reviewer pointed out it might be CO₂ interference.

Truth: Upon retesting with atmospheric compensation and enhanced purge, the "characteristic peak" disappeared; the compound contained no isocyanate group [1][10].

Stumbling reasons:

• Experimenters had insufficient understanding of CO₂ interference;
• Did not perform blank verification first;
• Paper did not mention atmospheric compensation processing.

Avoidance tips:

• Any peak at 2300–2370 cm⁻¹ should first be suspected as CO₂;
• Use triple verification: atmospheric compensation + purge + blank control;
• Method section of paper must describe spectrum acquisition conditions and processing.

Case 5: "Pseudo-fault" from MCT detector liquid nitrogen depletion

Scenario: A user had good spectra in the morning, but in the afternoon SNR suddenly dropped 20-fold; they suspected instrument damage and prepared to call for service.

Truth: MCT detector was filled with liquid nitrogen in the morning, but it depleted by afternoon; after detector warmed up, noise increased sharply [1].

Stumbling reasons:

• Did not regularly check liquid nitrogen level;
• Unaware of the effect of MCT warming on SNR.

Avoidance tips:

• Check liquid nitrogen before starting measurements each day;
• A full MCT Dewar can last 8–12 hours; for long measurements, refill midway;
• After warming, when refilling liquid nitrogen, wait 15 minutes for detector to stabilize;
• When not in use for extended periods, some users let MCT warm naturally (to save liquid nitrogen), but must fully cool before reuse.

Case 6: "Ghost fault" from beam splitter moisture absorption

Scenario: A lab's spectrum SNR gradually decreased over 2 months, and finally one day it went completely flat. During that time, they contacted engineers multiple times for alignment and cleaning, but the problem recurred.

Truth: The KBr/Ge beam splitter absorbed moisture during the rainy season, gradually fogging, and finally failed completely. After replacing the beam splitter and enhancing dry purge, it recovered [1][4].

Stumbling reasons:

• Did not enhance purge during rainy season;
• Did not visually inspect the beam splitter regularly;
• Misjudged "gradual degradation" as "instrument aging".

Avoidance tips:

• Maintain 24-hour low-flow purge during rainy season;
• Remove beam splitter monthly for visual inspection in dry environment;
• KBr/Ge beam splitter recommended replacement every 3–5 years (shorter in high humidity);
• In high humidity, consider CsI beam splitter (far-IR) or moisture-resistant coated models.

💡 Engineer's motto: 90% of "sudden flat line" and "gradual degradation" correspond to specific, locatable faults. Don't use "aging" as a catch-all [1].

8. Preventive Maintenance Strategies

Troubleshooting is "post-event remedy", while preventive maintenance is "pre-event prevention". Below is a preventive maintenance checklist based on ASTM E1421 and manufacturer recommendations [1][5][12].

8.1 Daily Maintenance (Every Day / Each Use)

| Purge before background collection | Each measurement | 15 minutes or more |
| Sample compartment cleaning | After each measurement | Swab with ethanol |

Table 3: FTIR daily maintenance checklist (compiled from [1][5])

8.2 Weekly Maintenance

• Check light source cumulative usage hours; prepare spare if approaching 1500 hours;
• Inspect ATR crystal surface with magnifying glass;
• Disassemble and clean pellet dies with ultrasonic;
• Clean mortar with water + ethanol and dry;
• Dust instrument exterior.

8.3 Monthly Maintenance

• Visually inspect beamsplitter for fogging;
• Performance verification (ASTM E1421): polystyrene film 1601 cm⁻¹ peak position, SNR test, 100% T line flatness [5][12];
• Check software logs for abnormal alarms;
• Replace drying tube/filter.

8.4 Annual Maintenance

• Engineer on-site: optical alignment, detector inspection, electrical check;
• Comprehensive instrument performance verification report (OQ/PQ documents);
• Evaluate key component lifetime (light source, beamsplitter, detector);
• Software upgrade and backup.

8.5 The "Save Money" Philosophy of Preventive Maintenance

"Saving money on FTIR maintenance will eventually cost you more – this is a lesson learned from countless labs."
– Interview with an instrument manager [1]

Negative examples [1][4]:

• Skipping purge to save money → beamsplitter fails every 2 years (¥20,000+ each);
• Using ordinary KBr instead of IR-grade → impurity contamination, spectral distortion, quantitative errors;
• No performance verification → data untraceable, GMP audit failure;
• Running light source until failure → 1–2 weeks downtime affecting production.

Positive examples [1]:

• Continuous low-flow purge + desiccant → beamsplitter lifetime 5+ years;
• Regular performance verification → data traceable, audit worry-free;
• Proactive light source replacement (e.g., every 1500 hours) → avoid sudden downtime;
• Spare parts inventory (light source, O-rings, desiccant) → recovery within 1 hour on failure.

9. Troubleshooting Quick Reference Table

Below is a quick reference table of "Phenomenon → Most Likely Causes → Troubleshooting Steps", recommended to be printed and placed near the instrument.

Table 4: FTIR Troubleshooting Quick Reference Table (compiled from [1][2][5][12])

Summary of This Section

Questions for Reflection

1. You open the instrument and collect a spectrum, finding a completely flat 100% T line with no peaks. Please write the first 5 troubleshooting steps in fault tree order, and explain what hypothesis each step tests.

2. For the same polystyrene film sample, the 1601 cm⁻¹ peak positions measured 5 times are 1601.2, 1600.9, 1601.1, 1600.8, 1601.0 cm⁻¹ respectively. Is this a wavenumber accuracy problem or a wavenumber repeatability problem? How should it be handled? Please provide verification steps referring to ASTM E1421.

3. When measuring an aqueous protein solution, you find that the water peak near 1640 cm⁻¹ masks the protein amide I band. Please design a complete experimental protocol (including sample preparation, background collection, subtraction strategy) to maximize retention of protein information. Hint: refer to ftir.fun water molecule functional group page.

4. A colleague reports finding an "isocyanate characteristic peak" at 2350 cm⁻¹ in a sample spectrum. Explain why this conclusion is likely wrong, and design a 3-step verification process to prove or disprove it.

5. You inherit an old FTIR and find that the SNR test result is only 5000:1, while the instrument specification is 50000:1. List all factors that could cause a 10× decrease in SNR, and sort them by "troubleshooting cost" from low to high.

References

[1] Griffiths P R, de Haseth J A. Fourier Transform Infrared Spectrometry. 2nd ed. Wiley, 2007. Chapter 3, 7, 11. ISBN: 978-0-471-19404-0.

[2] Smith B C. Fundamentals of Fourier Transform Infrared Spectroscopy. 2nd ed. CRC Press, 2011. Chapter 5–6. ISBN: 978-1420069297.

[3] Thermo Fisher Scientific. "FTIR Troubleshooting Guide." Application Note AN-001.
https://www.thermofisher.com/…

[4] Bruker Optics. "Maintenance and Care of FTIR Spectrometers." Technical Note TN-014.
https://www.bruker.com/en/pro…

[5] ASTM International. "ASTM E1421-99(2021): Standard Practice for Describing and Measuring Performance of Fourier Transform Mid-Infrared (FT-MIR) Spectrometers."
https://www.astm.org/e1421-99…

[6] Liland K H, Rukke E H, Olsen E F, et al. "Customized Baseline Correction." Applied Spectroscopy, 2011, 65(8): 907–915. DOI:10.1366/10-06124.

[7] Eilers P H C, Boelens H F M. "Baseline Correction with Asymmetric Least Squares Smoothing." Leiden University Medical Centre Report, 2005.

[8] Bertie J E, Lan Z. "Infrared Intensities of Liquids XX: The Intensity of the OH Stretching Band of Liquid Water Revisited." Applied Spectroscopy, 1996, 50(8): 1047–1058. DOI:10.1366/0003702963905385.

[9] ftir.fun. "Water Molecule Functional Group Page."
https://ftir.fun/ir/group/wat…

[10] Socrates G. Infrared and Raman Characteristic Group Frequencies. 3rd ed. Wiley, 2001. ISBN: 978-0470093078.

[11] Harrick N J. Internal Reflection Spectroscopy. Harrick Scientific, 1987. 3rd printing.

[12] USP <854>. "Mid-Infrared Spectroscopy." United States Pharmacopeia.
https://www.usp.org/

[13] Shimadzu. "FTIR Maintenance and Troubleshooting." Application News No. A528.
https://www.shimadzu.com/an/s…

Next Episode Preview: Ep 52 — Commercial Software Review: OMNIC, OPUS, Spectrum, etc.
The instrument failure is solved, but the "pitfalls" at the software level are just beginning. In the next episode, we will review the four major commercial FTIR software: Thermo OMNIC, Bruker OPUS, PerkinElmer Spectrum, and Shimadzu LabSolutions IR, covering the feature matrix, macro programming capabilities, accessory auto-recognition, and practical tips for data export (CSV / JCAMP-DX / SPC), as well as command-line tools and methods for data format conversion. Whether you are a purchase decision-maker or a daily user, you can find a suitable choice for your workflow.

This article is licensed under CC BY-NC-SA 4.0. The accompanying images are from the public domain or from web resources with indicated sources, and the copyrights belong to the original owners.