A new certified reference material for size and shape analysis of nanorods using electron microscopy

A new certified reference material (CRM) for size and shape analysis of elongated nanoparticles has been developed by the European Commission’s Joint Research Centre. The CRM consists of titanium dioxide nanorods dispersed in 1-butanol, was coded ERM-FD103 and has been certified for different electron microscopy–based operationally defined measurands such as the modal and median values of the particle number-weighted distributions of the minimum and maximum Feret diameter, the maximum inscribed circle diameter, the area-equivalent circular diameter and the aspect ratio. The nanorods have nominal dimensions of 15 nm in width and 55 nm in length. Homogeneity and stability measurements were performed using transmission electron microscopy. The relative standard uncertainty for homogeneity ranged from 0.3 to 1.7%. No significant instability was detected for a shelf life of 18 months and a storage temperature of 18 °C. The certified values have been determined from the results of an interlaboratory comparison in which qualified expert laboratories participated with scanning and transmission electron microscopy. The certified values are traceable to the unit of length in the International System of Units, the metre, and the relative expanded uncertainties (confidence level of approximately 95%) range from 4 to 6%. These properties allow the CRM to be used for quality assurance and calibration of electron microscopy methods for nanoparticle size and shape analysis in ranges relevant for the implementation of EU legislation related to nanomaterials. The presented study discusses the purpose and results of the different steps that were followed to turn an industrially relevant raw titanium dioxide nanorod material into a fit-for-purpose CRM. Graphical abstract Electronic supplementary material The online version of this article (10.1007/s00216-020-02984-z) contains supplementary material, which is available to authorized users.


Aim and scheme of the ILC study
The aim of this interlaboratory comparison (ILC) study is the characterisation and certification of a titanium dioxide (TiO 2 ) nanorod candidate certified reference material (CRM; labelled as ERM-FD103) by SEM and TEM. The basis of this certification approach is the randomisation of the (unknown) laboratory biases. This randomisation is only successful if the measurements within a laboratory are performed under intermediate precision conditions and if the collected results are independent. If one laboratory applies several methods then, these analyses can be treated as if they came from independent laboratories, provided instruments are calibrated differently, and test specimens are prepared separately.
Each contractor will receive three ampoules of the candidate CRM (ERM-FD103), each containing a nominal mass fraction of 1 g/kg TiO 2 nanorods dispersed in about 2 mL of n-butanol.
In addition to the candidate CRM, each contractor will receive one ampoule of a quality control material (QCM) that consists of about 9 mL of a suspension of near-spherical silica (SiO 2 ) nanoparticles with an average particle size in the range of 10 nm to 30 nm and a nominal mass fraction of 10 g/kg.
In total, a minimum of eight independent specimens shall be prepared and analysed: QCM (1 ampoule x 2 replicates) for the following particle size measurand: -Area-equivalent diameter (modal value) 2 specimens Candidate CRM (3 ampoules x 2 replicates) for the following particle size and shape measurands: -Area-equivalent diameter (mode, median) -Minimum Feret diameter (F min ) -Maximum Feret diameter (F max ) -Aspect ratio (as reported by software) and calculated as F min /F max -Maximum inscribed circle 6 specimens

Total 8 specimens
The term 'replicate' shall be interpreted as an SEM or TEM specimen. The latter shall be prepared by transferring a representative portion of the TiO 2 nanorods from the ampoule to an appropriate flat SEM or TEM substrate.
The ILC study is organised as follows:  for each awarded contract, JRC dispatches three ampoules of the candidate CRM and one ampoule of the QCM to the contractor;  the contractor shall analyse the candidate CRM and QCM according to the instructions provided and report within 8 weeks after the date of receipt of samples (based on DHL tracking information);  JRC evaluates and analyses the received datasets using appropriate statistical techniques;  financial settlement of the order(s) by JRC;  results of the ILC study are used in the certification report where they will be presented in an anonymous manner.

Specimen preparation
1) Before opening an ampoule, the ampoule must be gently inverted several times to ensure the homogeneity of the suspension and to re-suspend possibly settled particles. If after homogenisation some of the suspension is still present in the upper part (bulb-like head) of the ampoule, it can be removed by gently flicking the bulb-like head with the forefinger while tilting the ampoule. The ampoule is pre-scored and can be opened by applying moderate pressure with one's thumb to snap off the ampoule's head. Contents of an ampoule must be used the same day as opened. After opening of the ampoule, the ampoule should be either closed with paraffin film or the content should be transferred to a clean glass vial that can be closed with an appropriate cap.
2) The TiO 2 particles, which have external dimensions in the range of 1 nm to 100 nm, should be analysed asreceived (i.e. without filtration, centrifugation or sonication prior to analysis). In case dilution is required, nbutanol anhydrous, purity  99.8 %, must be used.
3) Aliquots shall be taken from the ampoule using clean pipette tips and avoiding to touch the edges of the ampoule. A new (unused) pipette tip shall be used for each aliquot.

4)
In avoiding contamination of the test specimens, it is highly recommended to prepare the SEM or TEM specimens in a low contamination environment (i.e. a clean room or a laminar flow bench).
5) The deposition of TiO 2 nanorod particles onto a suitable SEM or TEM substrate shall be based on the contractor's own established specimen preparation procedure. The applied procedure should ideally generate a uniform distribution of particles across the entire substrate without excessive amounts of agglomerates and touching particles being formed. In particular for particle shape measurements, the used substrate shall be flat over the selected field of view, uniform, and provide good contrast between particles and background.
6) If the minimum required number of particles cannot be observed and measured using one SEM or TEM specimen, then the contractor shall prepare additional specimens from the same ampoule until the minimum required number of particles can be measured.

7)
Opening and preparation of test specimens shall be performed according to the measurement scheme as shown below.

Calibration
1) The contractor shall ensure that the performance of the electron microscope is regularly verified and that it is calibrated to the SI unit of length (metre) at operating conditions similar to those to be used for the analysis of the QCM and the candidate CRM.
2) Calibration materials used must fulfil all requirements of a certified reference material, i.e. they must be homogeneous, stable, and come with a certified value with associated uncertainty and metrological traceability statement.
3) A combined relative standard uncertainty (at a confidence level of 68 %) that is associated to the calibration coefficient of the instrument's length scale shall be estimated and provided together with a detailed description of the applied calibration procedure, in the analysis report.

Image acquisition
1) Images shall be captured using an electron microscope combined with a suitable detector, resulting in the generation of count-or number-weighted particle size/shape distributions.
2) The magnification used for the QCM and candidate CRM shall be large enough to measure the largest external dimension of the nanorods while having sufficient resolution to also accurately measure the rod's smallest external dimension.
3) For particle shape measurements, a higher pixel resolution might be necessary than for particle size measurements. It is, however, up to the user to set the optimal acquisition parameters and pixel resolution and to prove that they were chosen properly. 4) For each specimen prepared from the QCM and the candidate CRM, at least 250 individual (non-touching and not agglomerated) particles shall be imaged and analysed (see Section 5.1) and analysed. If the measurement protocol does not generate sufficient number of individual particles, then the contractor shall measure at least 500 touching (but not overlapping!) particles. The measured particles shall originate from at least five different fields of view (i.e. images). If the required number of particles cannot be reached using five images, then additional images shall be captured until the minimum number of particles is reached. For a given specimen, the different images shall be captured randomly, but widely separated from each other, across the entire surface of the specimen surface.
5) The specimen surface at the field of view must be flat. This is particularly important for particle shape measurements, i.e. where a tilt angle can significantly affect the accuracy of the measurement result obtained on the longest external particle dimension.
Image analysis (according to ISO 13322-1:2014) and statistics 1) For each QCM specimen, at least 250 individual and non-touching particles (or 500 touching but not overlapping particlessee Section 4.4) from at least five different images shall be analysed with respect to the area-equivalent circular diameter. Foreign artefacts (e.g. contamination and dust particles, residues from drying, etc.) as well as particles cut by the measurement frame must be excluded from the data analysis process. In case the contractor decides to analyse touching particles rather than individual non-touching, particles, an appropriate particle separation protocol must be applied and all images showing the detection and separation of particles shall be provided.
2) For each QCM specimen, the obtained raw results must be graphically represented in a histogram (density distribution q 0 according to ISO 9276-1:1998) having a linear abscissa (x-axis). The modal result of this number-weighted particle size distribution shall be reported along with the associated relative (%) expanded measurement uncertainty (at a confidence level of 95 %). The optimal number of size bins for the density distribution shall be defined by the contractor.
3) For each candidate CRM specimen, at least 250 individual (non-touching and not agglomerated) particles, or 500 touching but not overlapping particles, from at least five different images shall be analysed with respect to the following measurands:  Minimum Feret (F min ) diameter according to ISO 9276-6:2008;  Maximum Feret (F max ) diameter according to ISO 9276-6:2008;  Aspect ratio as calculated by the user's software, including an unambiguous description of the parameters used (and/or equation) and as the ratio of F min /F max ;  Area-equivalent diameter according to ISO 9276-6:2008;  Maximum inscribed circle.
Foreign artefacts (e.g. contamination and dust particles, residues from drying, etc.) as well as particles cut by the measurement frame must be excluded from the data analysis process. 4) For each measurand and for each specimen prepared from the candidate CRM, raw measurement results must be graphically represented in a histogram (i.e. as a density distribution q 0 having a linear abscissa) and as a cumulative distribution (Q 0 ). The binarisation process shall be optimised and defined by the contractor. The contractor shall report the modal value of the density distributions q 0 and the median value from the cumulative distributions Q 0 . The median shall be calculated across the following particle size ranges: -Area-equivalent circular diameter: 5 nm to 80 nm -F min : 5 nm to 35 nm -F max : 20 nm to 80 nm -F min /F max : 0.1 tot 0.5 All results shall be reported along with the associated relative (%) expanded measurement uncertainties (at a confidence level of 95 %).
Reporting 1) A detailed and signed analysis report shall be sent by post mail or courier service to the study coordinator within 8 weeks after the date of receipt of samples. The analysis report shall contain as a minimum the information listed in the below mentioned table.

Generic information (according to ISO/IEC 17025)
Name and address of the laboratory, and location where tests were carried out, if different from the address of the laboratory Name and address of the client Unique identification of the analysis report, repeated on each page of the report Page numbering indicated as e.g., "Page 1 of 15" Operator's name Name, function and signature of persons authorising the analysis report Sample information (according to ISO 13322-1) Date of receipt of the candidate CRM and QCM samples Identification of the laboratory samples and, if relevant, identification assigned to the ampoules by the contractor Date when the ampoules were opened and specimens were prepared Identification of the specimens (e.g., ampoule#_replicate#) Complete description of the method used for sub-sampling, if required, and SEM/TEM specimen preparation, with full quantitative details of the nominal mass, volumes and compositions of products, in case dilution was applied Type of the used sample holder/substrate Sample preparation Dilution and dilution medium Sample volume intake Sample preparation/drying Sample grid/stub/sample holder (mesh size, coating, copper/gold/mica) Method and instrument information Make and type of the electron micoscope Date of the last instrument performance check/maintenance Description of the image magnification calibration procedure, including description of the used calibrant(s) Description of the method used (magnifications, CCD camera, nominal camera length, acceleration voltage, tilt angle, spot size, aperture, etc.)

Image analysis and results
Date of performance of the tests For non-touching particles: at least one representative raw micrograph and one annotated micrograph used for particle identification and analysis per specimen For touching particles: all raw micrographs and annotated micrographs For each specimen, and for each measurand (F min , F max , aspect ratio, F min /F max , area-equivalent diameter and maximum inscribed circle), one density distribution and one cumulative distribution based on at least 250 individual particles or 500 touching particles. All particle dimension measurands shall be reported in nanometre (nm). Characteristic values from number-weighted distributions to be reported: Density distribution (linear abscissa): modal value and arithmetic mean Cumulative distribution: median, 25 % and 75 % percentiles Pixel size/resolution (nm) Micrograph size (m) Total area imaged per sample (m 2 ) Counting and analysis procedure and number of counted/analysed particles Binning procedure Description of the estimation of the mode (normal/Gaussian fitting, highest bin, etc.) Estimation of the measurement uncertainty associated to the number-based modal and median particle size and aspect ratio values (see Section 6.2) Description of the image analysis software package used Description regarding adjustment of contrast, brightness, greyscale threshold, etc. Description of the aspect ratio parameter reported by the applied image analysis software Description regarding the usage of image filters (smoothing, NxN, mean, median) Description regarding the image analysis protocol dealing with the separation and identification of touching particles (manually/automatically discard all touching particles, manual or automatic particle separation filters, morphology threshold based separation, etc.) 2) The contractor shall provide an estimation of a realistic relative (%, nm/nm) expanded measurement uncertainty (confidence level of about 95 %) associated to a single measurement result (i.e. the average of the results obtained on  250 (or  500) particles per specimen and for a given measurand), and details on how this uncertainty was derived.
Ideally, the relative expanded measurement uncertainty, U, should be estimated from method validation data. If such data is available, individual relative standard (confidence level of 68 %) uncertainties estimated for method repeatability (u r ), day-to-day variation or intermediate precision (u ip ), instrument calibration (u cal ) and method trueness (u t ) can be combined in order to arrive at a combined standard measurement uncertainty, u meas (see equation below). For a confidence level of 95 %, u meas is multiplied with a coverage factor (k = 2) resulting into a relative expanded measurement uncertainty, U.
In case method validation data is not available, u meas and U, may be estimated following alternative approaches, such as:  in-house quality control charts  results obtained during previous ILC studies  expert judgement Page 8 of 12 2) Measurement procedures used during the ILC study Table S1 Overview of SEM and TEM measurement procedures used by the ILC participants (from: Gerganova T, Roebben G, Kestens V. The certification of size and shape parameters of titanium dioxide nanorods in 1-butanol solution: ERM-FD103. Certification report EUR 29781. Luxembourg: Publications Office of the European Union; 2019)

Laboratory Code
Specimen preparation Instrument Calibration Image acquisition Image analysis and evaluation L1-TEM The as-received material was 5x diluted in 1-butanol. 15 L of the diluted suspension was brought onto a pioloform ® carbon-coated 400 mesh Cu grid. The grid was left in contact with the suspension for 10 min while being covered with an empty petri dish to limit evaporation of 1-butanol. After the incubation period, the grid was blotted dry to remove excess sample and left to air-dry at room temperature.

FEI Tecnai Spirit
Optical diffraction crossgrating with 2160 lines/mm and 463 mm line spacing (Agar Scientific) The TEM instrument was operated at 120 kV and at a spot size of 3. Micrographs were captured using a bottom-mount 4k x 4k Eagle CCD-camera and at a 30,000x magnification. Pixel size = 0.37 nm AnalySIS Solution of iTEM (Olympus Soft Imaging Solutions GmbH) Automatic contrast/brightness correction, manual selection of threshold for particle detection based on mass-thickness contrast, 10x10 filter for reducing background noise, particle detection filters 0-40 nm (min diameter), 10-50 (ECD), 0-0.8 (sphericity), 0.6-1 (convexity) and 0-0.8 (shape factor). Number-based PSDs were iteratively fitted with a Gaussian function.

L2-TEM
For each CRM unit, replicates 1 were prepared by bringing a 10 L drop of the undiluted as-received material onto a carbon-coated 200 mesh Cu grid. Replicates 2 were dipped into the undiluted suspension. Grids were allowed to in an ISO class 5 clean bench.