In the synchronous thermal analysis experiment (TG-DSC/DTA), the sources of errors and data correction need to be comprehensively controlled from three dimensions: equipment, operation, and data processing. The following are the key points:
1、 Common sources of error
Equipment factors
Temperature control error: Aging of heating elements or lagging temperature control algorithms can cause temperature fluctuations (requiring temperature control accuracy<± 0.1 ℃), especially during rapid heating, the thermal inertia of the furnace body may cause temperature lag. For example, the thermal decomposition temperature difference of polyimide in air and nitrogen can reach over 50 ℃, and the temperature axis needs to be calibrated in sections using standard substances such as indium and tin.
Sensor drift: When TG sensors are affected by fluctuations in carrier gas flow rate or vibration interference, microgram level mass signals are easily masked by noise; The mismatch in heat capacity between the reference end and the sample end of the DSC sensor (such as impurity deposition) can cause distortion of the heat flow signal and requires regular cleaning and maintenance.
The influence of crucible and atmosphere: Improper selection of crucible materials (such as alumina, platinum) may catalyze reactions; Insufficient gas purity (such as impurities in oxygen) or unstable flow rate (recommended 20-50mL/min) can introduce background interference, requiring the use of high-purity gas and optimization of flow field design.
Operational factors
Sample preparation error: Insufficient sample size (<5mg) will reduce the signal-to-noise ratio, while excessive sample size (>20mg) will hinder heat and mass transfer; Heterogeneous samples (such as blends) need to be ground to a particle size of<100 μ m to avoid stepwise decomposition.
Sample loading deviation: The difference in the placement position of the crucible twice or the vibration of the heating furnace in and out can cause the center of gravity of the balance to shift. It is recommended to use the "in-situ sampling" method (adding samples directly after peeling the empty crucible).
Cross contamination: Residual volatile samples may contaminate subsequent experiments, and the furnace body needs to be cleaned and the reference crucible replaced after testing.
Data processing error
Baseline drift: The difference in heat capacity between the reference end and the sample end needs to be deducted. The baseline drift is significant in the low temperature range (<100 ℃) and can be corrected by polynomial fitting.
Noise interference: TG signals are easily affected by vibration and need to be smoothed with a low-pass filter, but excessive filtering may weaken the true step signal.
Quality normalization error: Initial quality input errors can lead to deviation in percentage weight calculation, and accurate recording of sample quality is required.
2、 Data correction method
Baseline correction
Record the baseline using an empty crucible or an inert reference material (such as α - Al ₂ O3) to eliminate the effects of furnace expansion or gas buoyancy. For example, in the thermal decomposition experiment of calcium carbonate, baseline correction can reduce the weight loss step error of TG curve from ± 5% to within ± 1%.
Calibration of standard materials
Temperature calibration: Using metal melting point standard substances (such as indium 156.6 ℃, tin 231.9 ℃) to calibrate the temperature axis, segmented calibration can reduce nonlinear errors. For example, after calibration, the temperature error of a TG-DSC instrument in the high temperature range (>500 ℃) decreased from ± 3 ℃ to ± 0.5 ℃.
Thermal effect calibration: Using standard substances such as sapphire to calibrate DSC enthalpy values, the enthalpy change error of uncalibrated instruments can reach 10% -20%.
Noise filtering and smoothing
Using a low-pass filter (cut-off frequency adjusted according to sample characteristics) for TG signals and Savitzky Golay smoothing algorithm for DSC signals can reduce noise while preserving peak features. For example, in a polymer glass transition experiment, smoothing treatment increased the glass transition temperature resolution of the DSC curve from ± 2 ℃ to ± 0.5 ℃.
Overlapping peak separation
Use peak fitting (such as Gaussian function) to analyze the enthalpy change and weight loss ratio at each stage of complex reactions (such as multi-stage decomposition). For example, in the thermal decomposition experiment of a compound containing crystal water, peak fitting can accurately distinguish between the two stages of crystal water release (100-200 ℃) and main chain decomposition (300-400 ℃).
3、 Optimization suggestions
Regular maintenance: Verify instrument performance with standard substances every 3 months, clean sensors and furnace bodies.
Standardized operation: fixed sampling techniques (such as compaction degree), recording parameters such as gas flow rate and heating rate for easy replication of experiments.
Cross validation: Verify the thermal analysis results by combining XRD, FTIR, and other techniques, such as confirming through XRD that the phase transition temperature detected by TG-DSC is consistent with the changes in crystal structure.
By systematically controlling the sources of errors and applying correction methods, the accuracy and repeatability of synchronous thermal analysis experiments can be significantly improved, providing reliable data support for research on material thermal stability, reaction kinetics, and other related fields.