^{1}

^{*}

^{1}

^{†}

^{1}

^{†}

^{2}

^{†}

^{3}

^{†}

^{4}

^{†}

These authors contributed equally to this work.

This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution license (http://creativecommons.org/licenses/by/3.0/).

Diffuse reflectance infrared Fourier transform (DRIFT) spectroscopy is a well-known technique for thin film characterization. Since all asbestos species exhibit intense adsorptions peaks in the 4000–400 cm^{−1} region of the infrared spectrum, a quantitative analysis of asbestos in bulk samples by DRIFT is possible. In this work, different quantitative analytical procedures have been used to quantify chrysotile content in bulk materials produced by building requalification: partial least squares (PLS) chemometrics, the Linear Calibration Curve Method (LCM) and the Method of Additions (MoA). Each method has its own pros and cons, but all give affordable results for material characterization: the amount of asbestos (around 10%, weight by weight) can be determined with precision and accuracy (errors less than 0.1).

Italy is a country covering a surface area of 301,340 km^{2}, with about 60,626,442 inhabitants and a population density of 201 inhab./km^{2}. It is thus a very densely populated country, with much of its territory urbanized and requiring careful protection of valuable cultural or natural assets.

This country has been one of the biggest producers of raw asbestos, by mining, and asbestos containing material (ACM), especially for building applications. Asbestos is a common name for classifying a family of silicate minerals, which are sub-divided in amphibole and layer-silicate asbestos on the basis of their structures and chemical compositions [_{3}Si_{2}O_{5}(OH)_{4}].

Italy was the first European country banning asbestos in 1992: the extraction, import, export, marketing and fabrication of asbestos-based products were forbidden [

Legal prescriptions, published over the last 20 years, covered the main items in occupational fields, such as personal and work environment exposure, threshold limits values (sampling, analysis,

From the analytical point of view, the Italian Ministerial Decree issued on 6 September 1994, indicates some analytical techniques and methods for qualitative and quantitative ACM characterization: X-ray diffraction (XRD), Fourier transform infrared spectroscopy (FTIR) and scanning electron microscopy (SEM), but gives a specific description of the analytical procedures only in case of XRD and SEM. XRD can detect the content of asbestos with a detection limit of 10% weight by weight (^{−1} wavenumber region [

DRIFT is a surface localized FTIR spectroscopy, since it can provide both chemical and structural information for all types of solid surfaces. When infrared radiation reaches a sample surface, one or several processes can occur: light can be adsorbed, reflected from the surface, or it can penetrate the sample before being scattered. If scattering centers, which are fibers in the case of ACM, are randomly oriented, the phenomenon is isotropic and generates a diffuse reflectance [

All type of asbestos minerals exhibit strong absorption in the 1200–900 cm^{−1} band, due to Si–O stretching vibration and in 600–900 cm^{−1}, due to vibration of the silicate chain, to metal-oxygen stretching and Si–O bending vibration [^{−1} [^{−1} region [^{−1} and weaker at 3655 cm^{−1} [^{−1}, Si–O stretching vibration at 1069, 1033 and 959 cm^{−1}, and evident bands at about 606, 434 and 300 cm^{−1} can be assigned to Mg–OH bending frequency [

In this work, we have applied and compared three different analytical procedures (the Method of Addition, partial least squares and the Linear Calibration Curve Method) for the quantitative determination of a few micrograms of asbestos in bulk materials.

Quantitative analyses using calibration curves are based upon measuring a property of a sample that changes with the analyte concentration [_{M}_{x}_{M} = A + B C_{x}_{M}

The relative error is, in this case:

where δ

Once the calibration curve has been calculated, the concentration of an unknown sample of analyte is evaluated by measuring the response of the unknown asbestos sample under the same conditions as used for the standard. From the mathematical point of view, MoA is a regression line formula, and for this reason, its intrinsic precision could be limited with respect to other analytical procedures [

The Partial Least Squares (PLS) Method is currently used to solve both descriptive and predictive problems in experimental life sciences, especially in chemistry [

LCM is commonly used in quantitative chemistry, and it has been adopted successfully for the determination of asbestos in bulk materials by means of XRD [_{M}_{x}

The calibration curve can always be approximated to a straight line, at least for very small ranges of concentration, the equation of which is, again: _{M}_{x}

Relative and absolute errors on this quantity, ε_{r} and ε_{a}, respectively, are calculated by the usual formulas of error propagation:

where the error on _{M}_{M}

^{−1} and 3645 cm^{−1} of chrysotile are evidenced well. Other peaks are ascribed to the C–O (2924–2854 cm^{−1}; 2400–2200 cm^{−1} and 1720 cm^{−1}), Si–O–Si (1090–1000 cm^{−1}) and S–O (1200–650 cm^{−1}) chemical groups present in the compound that constitutes the cement matrix in which the asbestos is embedded (see also the ACM description in the Experimental section).

The MoA linear curves are shown in ^{−1} peak, an asbestos concentration of 11.23% ± 0.06% is estimated; in the case of a peak at 3645 cm^{−1}, the concentration is calculated as 11.20% ± 0.05%. This result demonstrates that the DRIFT technique could be simple, fast and also very accurate and precise.

In

The TQ Analyst software elaborates data from the DRIFT spectra and, as a result, produces a calibration curve and a validation curve, as reported in

If we apply the calibration model to the unknown sample, previously analyzed by MoA, we obtain the results reported in

The LCM curves, calculated for both analytical peaks of chrysotile, are shown in _{M}

As in the case of MoA, we can quantify the asbestos content in the ACM sample by using the curve parameters of both peaks at 3688 cm^{−1} and 3645 cm^{−1}, by using the formulas reported in the Theory section, we obtained 11.13% ± 0.02% and 11.30% ± 0.07%, respectively. These numbers show that also, in the case of LCM, the DRIFT spectroscopy is precise and accurate.

As the first conclusion, we can state that from the quantification point of view, the three analytical procedures are equivalent, since in all the experiments that were realized, it was found that the chrysotile content was the same within the experimental errors. For all three methods, the sample preparation and data acquisition procedure are important in order to maximize accuracy and precision. While MoA and LCM require peak height calculation, which means the individuation of peaks and baseline correction, PLS is completely automated,

The need of a critical comparison between different FTIR methodologies, highlighting the pros and cons of each one, arises form a lack of technical prescription in Italian regulation: while X-ray diffraction and optical, as well as electronic, microscopy are well described, FTIR is allowed, but not standardized. We believe that in the near future, there will be a strong need of fast, accurate and precise quantitative methods for monitoring ACM and also asbestos containing wastes or contaminated soils; in Italy there are 10 Superfund,

In this framework, DRIFT spectroscopy represents a rapid-screening method for the quantification and classification of materials. The procedures tested in this work can be successfully applied to different bulk asbestos materials for the qualitative and quantitative determination of all types of asbestos.

Asbestos containing materials (ACM), were supplied by “Ambiente s.r.l.” in the framework of a local research project “Progetto Rifiuti” (INAIL, Regional Direction for Campania) and came from the remediation of industrial building roofs. The ACM was classic cement reinforced by asbestos fibers; this kind of material has been deeply characterized by standard chemical and structural techniques (in our case, X-ray diffraction using a Philips PW3020 X’Pert Diffractometer under the following operating conditions: Bragg–Brentano configuration, Θ–2Θ; tension, 40 kV; current, 40 mA; anode, copper; scanning, step; step size, 0.01°; time per step, 1 s.; data not shown here); and the compositional information agrees with that supplied by the technical sheet of the manufacturer: chrysotile, calcium carbonate, calcium sulfate and alumina. In such materials, the asbestos content was always between 10% and 15%, weight by weight. ACM handling requires special health security procedures: ACMs were received sealed in double polyethylene bags; each bag was opened inside a laminar flow hood to prevent any fiber dispersion into the laboratory, according to the Italian Environment Ministry Decree 6 September 1994, and related acts. Researchers, during laboratory activity, wore protective disposable full-body overalls and the prescribed facial masks. ACM was gently dry-crushed in an agate mortar and then finely ground in a ring mill (FRITSCH, model Pulverisette 9, rotational speed 750/1000 rpm) enclosed in a vial. In all the above operations, the air was monitored by filtration through cellulose filters and fibers counted using phase contrast optical microscopy (PCOM). Fiber concentration, in all analyses, never exceeded the threshold limit (100 fibers/L).

An FTIR spectrometer Nicolet 6700 (Thermo Nicolet Corp., Madison, WI, USA) equipped with a diffuse reflectance accessory (DRIFT, Thermo Nicolet Corp., Madison, WI, USA) was used to obtain the FTIR spectra of the samples. Spectra were collected in the range 4000–400 cm^{−1} by 32 scans and at resolution of 4 cm^{−1}. All spectra were analyzed by the software of the OMNIC operating system (Version 7.0 Thermo Nicolet, Thermo Nicolet Corp., Madison, WI, USA) and normalized against an air background. After every measurement, a new reference air background spectrum was taken.

Mixtures of ACM and known quantities of standard chrysotile (NIST standard SRM 1866b) were prepared by multiple additions in the range of 4%–42% weight by weight. Samples were mixed and homogenized in an agate mortar for a few minutes. The heights of the chrysotile characteristic peaks at 3688 cm^{−1} and 3645 cm^{−1} were registered for each addition, and the data were plotted for linear regression. An unknown asbestos concentration is obtained by the intersection of the best-fit line and the

The software, TQ Analyst, was used for PLS. This statistical analysis requires two elaboration steps: calibration and validation. In the calibration procedure, the software searches for a relation between the dependent variable,

The Linear Calibration Curve Method (LCM) is commonly used in quantitative chemistry, and it has been adopted successfully for the determination of asbestos in bulk materials by means of XRD [^{−1} and 3645 cm^{−1} were registered for each mixture and the data plotted against the chrysotile concentration for linear interpolation. The parameters, and their errors, of the best-fit linear curve can be used for the estimation of unknown samples.

In conclusion, the analytical procedures investigated for the quantitative determination of asbestos in bulk materials based on DRIFT spectroscopy are precise and accurate in the explored range of asbestos concentration, even if DRIFT does not allow for a very low (about 1%

The authors thank INAIL, Regional Direction for Campania, for the economic and technical supports received.

The authors declare no conflict of interest.

Diffuse reflectance infrared Fourier transform (DRIFT) spectrum of the asbestos containing material (ACM) sample.

The Method of Additions (MoA) linear curves of ACM.

TQ Analyst software calibration and validation curves.

Linear Calibration Curve Method (LCM) linear curves for ACM.

Absorption peaks for different types of asbestos.

Type of asbestos | Analytical band (1) (cm^{−1}) |
Analytical band (2) (cm^{−1}) |
Analytical band (3) (cm^{−1}) |
Analytical band (4) (cm^{−1}) |
---|---|---|---|---|

Chrysotile | 3697–3686–3650–3640 | 1078–1020–960 | 654–615–605–550–481–450–440–432 | 400–305 |

Amosite | 3656–3640–3618 | 1128–1082–996–981 | 775–750–703–638–528–498–481–440 | 385 |

Crocidolite | 3636–3620–3610 | 1143–1110–939–897 | 778–775–770–725–694–668–636–630–540–504–495–450 | 320 |

MoA curve parameters.

Analytical band (cm^{−1}) |
_{B} |
_{A} |
^{2} | |
---|---|---|---|---|

3688 | 0.021 | 0.00260 ± 6 × 10^{−5} |
0.0292 + 0.0015 | 0.9928 |

3645 | 0.006 | 4.76 × 10^{−4} ± 2 × 10^{−5} |
0.0056 + 0.0002 | 0.9919 |

Mixtures used in the partial least squares (PLS) estimation.

Index | Spectrum title | Usage | % Chrysotile | % ACM |
---|---|---|---|---|

1 | STANDARD1 | CALIBRATION | 4.0 | 96.00 |

2 | STANDARD2 | VALIDATION | 6.00 | 94.00 |

3 | STANDARD3 | CALIBRATION | 10.00 | 90.00 |

4 | STANDARD4 | CALIBRATION | 11.00 | 89.00 |

5 | STANDARD5 | CALIBRATION | 12.00 | 88.00 |

6 | STANDARD6 | CALIBRATION | 18.00 | 82.00 |

7 | STANDARD7 | VALIDATION | 20.00 | 80.00 |

8 | STANDARD8 | CALIBRATION | 23.00 | 77.00 |

9 | STANDARD9 | CALIBRATION | 26.00 | 74.00 |

10 | STANDARD10 | CALIBRATION | 27.00 | 73.00 |

11 | STANDARD11 | CALIBRATION | 35.00 | 65.00 |

12 | STANDARD12 | CALIBRATION | 36.00 | 64.00 |

13 | STANDARD13 | CALIBRATION | 38.00 | 62.00 |

14 | STANDARD14 | CALIBRATION | 42.00 | 58.00 |

PLS results.

Index | Component | Concentration | Unit | Uncertainty |
---|---|---|---|---|

1 | Chrysotile | 11.06 | % | 9.461 |

2 | ACM | 88.94 | % | 9.461 |

LCM curve parameters.

Analytical band (cm^{−1}) |
_{B} |
_{A} |
^{2} | |
---|---|---|---|---|

3688 | 0.021 | 0.0169 ± 0.0004 | −0.167 ± 0.005 | 0.9945 |

3645 | 0.0069 | 0.0010 ± 0.0002 | −0.0438 ± 0.0002 | 0.9962 |