Prevention-through-design approach to mitigate workers’ exposure in the graphene production processes

The growing development of new and advanced nanomaterials calls for a responsible approach to evaluate and prevent health and safety risks for workers, who could be exposed in their whole life cycle. Since many uncertainties still remain about health effects and as long as occupational exposure limits will not be available, Prevention-through-Design (PtD) approach may be proposed as a framework aimed at preventing risks, taking into account health and safety aspects starting from the design stages of innovation production processes. PtD principles could be applied to NMs, including strategies to mitigate emissions and minimize risks related to the manufacturing and use. In the present study, this approach has been successfully applied to different case studies of graphene-based NMs production in research and development laboratories, with promising applications in the transition towards the industrial scale. The methodology includes the integration of ISO control banding tool and OECD multi-metric and tiered approach to assess exposure by inhalation, improving the reliability of risk analysis framework. The findings support the complementary use of qualitative and quantitative data to identify tailored control measures and prevent risks in parallel with the development of NMs production processes, by giving also the opportunity to evaluate their effectiveness.


Introduction
Since the discovery and characterization of graphene in 2005 [1], its applications are increasing worldwide in different industrial sector.Graphene has impact on the plastic market, providing extraproperties to polymer composites, by increasing their mechanical properties and the electrical/thermal conductivity.Furthermore, it has extensive use in the energy, opto-electronics, sensors and construction fields [2].Graphene was also included in the working description of advanced materials proposed by the Organisation for Economic Cooperation and Development (OECD), as "material rationally designed to have new or enhanced properties, and/or targeted or enhanced structural features" [3].
The growing development of such advanced nanomaterials (NMs) call for a responsible approach to evaluate and prevent health and safety (H&S) risks for workers, who could be exposed in their whole life cycle.Some studies in literature have shown that the exposure to these materials can cause adverse health effects [4].Additionally, the unusual aerodynamic behavior of 2-D graphene structures makes concerns for exposure by inhalation [5].
At present, occupational exposure limits (OELs) specific for NMs are not enforced by law [6].Evidences from subchronic inhalation studies in rats led Lee et al. 2019 [7] to propose an OEL of 18 g/m 3 for graphene.In terms of particle number concentration (PNC), a nano reference value 8 h timeweighted average (NRV8h TWA) of 40,000 part/cm 3 for particles having density lower than 6 g/cm 3 was proposed [8], but no NRVs are still recommended for non-spheroidal NMs, such as graphene based materials [9].Furthermore, for peaks lasting few seconds the NRV 15 minutes time-weighted average (NRV15min TWA) was calculated as two times the NRV8h TWA by van Broekhuizen et al. 2012 [10].
Limited data are also available on airborne graphene-based material concentrations in occupational settings [11].Spinazzé et al. 2016 [12] studied the exposure in the production of graphene family NMs by a multi-metric direct reading approach, highlighting that workers with high potential for occupational exposure are those directly involved in material sampling for quality control.Lee et al. 2016 [13] showed airborne graphene-like structures during the weighing and sonication operation of graphene nanoplatelets (GNPs), despite particle concentration values were similar to the background ones.GNPlike materials were also found in the workplace air during the weighing of powder, the addition of liquid and mixing procedures [14].
Since many uncertainties still remain about the health effects and as long as OELs will not be officially recognized, Prevention-through-Design (PtD) approach has been proposed as a framework aimed at preventing risks, taking into account H&S aspects starting from the design stages of innovation production processes [15].PtD is also part of the Safety and Sustainability by-Design principles [16] which describe all the safety measures for the prevention of health or environmental damages that are applied during the design phase of a structure, process, material or product, taking into consideration the entire production process line (synthesis, harvesting, purification, drying, packaging and end of life).PtD principles could be applied to NMs, including graphene, providing the design of strategies to mitigate the emissions and minimize the risks related to the manufacturing processes.The strategy is based on the early and iterative implementation of the traditional hierarchy of control measures: 1. elimination, replacement or modification of the risk factor; 2. use of engineering processes to minimize or eliminate exposure; 3. implementation of administrative controls that limit the amount or duration of exposure; 4. use of collective protective equipment and/or personal protective equipment (PPE) [17].In particular, according to the European Commission Directive 2019/1832 that modified the annexes I, II e III of the Directive 89/656/CEE on PPE, face masks, gloves and suits use is mandatory to protect workers in the activities in which NMs exposure risk has been recognized.
In this framework, Inail-Dimeila developed a methodological approach to support the enterprises in the design of OSH risk prevention when they start a NMs manufacturing process [18].In the present study, this approach has been tested and applied to different case studies of graphene based NMs production in Research and Development (R&D) laboratories, with promising applications at the industrial scale production.

Methodological framework
The characterization of exposure scenarios and the identification of NMs emissions that could be inhaled (ingested or absorbed through the skin) by workers are key steps in implementing the PtD approach in parallel with the process design and development.
In the analysis of occupational exposure to NMs by inhalation, the concentration contribution by mass alone (generally used for the definition of OELs) may not to be quantitatively representative but must be integrated with other parameters, such as particle number concentration (PNC), lung deposition surface area (LDSA), size distribution (SD) or average diameter (Davg) of the particulate matter [19].
The methodology applied in this study for characterization of different graphene exposure scenarios [20], includes the integration of ISO Control Banding (CB) tool [21] and OECD multi-metric and tiered approach [22] also integrated in the CEN guidelines [23].This provides for a tier 1 of information gathering by using technical sheets to collect data and parameters provided by the manufacturer on the process/material to be evaluated.A preliminary visit in the workplace supports the identification of structures, equipment, ventilation/air exchange systems, production phases, working times and protective devices.Furthermore, a trial sample of the produced materials can be obtained for setting up experimental tests and laboratory simulations.The use of qualitative risk analysis tools based on control banding techniques (e.g. according to ISO/TS 12901-2:2014 [21]) is recommended at this stage.
When the release of NM cannot be excluded from the first level analysis, tier 2 should be performed, which consists in carrying out PNC measurements during work processes with portable and easy-to-use real-time devices, and time-integrated sampling to collect airborne NMs for the following off-line characterization by electron microscopy.It will be necessary to move to a further level of analysis if the resulting PNC measured during the production is greater than the background plus three times the standard deviation of the background itself, and/or when the off-line analysis provides evidence of the presence of airborne nano-objects.
In tier 3, all available techniques, including sampling in the workers' personal breathing zone (PBZ), are integrated to provide a detailed analysis for NM quantification in the work environment.The realtime instruments used include condensation particle counters (CPC) or diffusion chargers (DC) to measure PNC, Davg and LDSA (usually alveolar fraction) at high frequency (1 Hz = 1 measurement per second), optical, aerodynamic or electric mobility based particle meters (i.e.Optical particle Sizer, OPS or Fast Mobility Particle Sizer, FMPS) to obtain NMs SD.These devices are integrated with off-line chemical, gravimetric and morphological analysis, in order to characterize the airborne nano-objects and confirm if the exposure is properly mitigated or if it is necessary to implement further risk management measures.
The results of measurements and data investigations can be also used to improve the qualitative risk analysis carried out with control banding tools.The integration of the exposure scenario characterization with the bio-monitoring of involved workers represent an added value for an effective implementation of mitigation measures.

Workers' exposure scenarios in different graphene production processes
The synthesis of graphene relies on the bottom-up approach, in which the chemical vapor deposition (CVD) is the most representative and industrially-relevant technique, and the top-down approach, when graphite crystals are exfoliated to achieve ultra-thin flakes.In the present study three different graphenebased materials production processes have been compared by the point of view of occupational exposure: 1. CVD of graphene on silicon carbide substrates [24].2. Thermal exfoliation and spray casting deposition of graphene nanoplatelets (GNPs) [25,26].
As summarized in Table 1, far-field (FF) and near-field (NF) average background and standard deviation values of real-time PNC have been measured in order to calculate the significant levels (PNCSL, i.e. the average background plus three times the related standard deviation).The ratio ( > 1) between the PNC level during the production phase and the PNCSL may correspond to the extent of NMs emission.Evidences obtained by off-line morphological and elemental characterization (SEM-EDS and/or TEM) of airborne NMs sampled in the workplace during each phase are used to complete the assessment.Based on such results and data analysis integration, control bands (CB) have been identified for each production phase in a scale from 1=very low to 5=very high according to Boccuni et al. 2020 [21] and primary risk management measures are recommended to mitigate workers' exposure.

CVD of graphene
In the process 1 the phases related to sample preparation (1a), CVD growth inside the reactor including the periodical reactor opening (1b) and the cleaning of the reactor components in the furnace (1d) showed no significant variations in the PNC compared to the background.Moreover, it is important to note that during the phase 1b a decrease in PNC levels happened that might be related to the pure nitrogen flux originating from the CVD reactor during the vacuum chamber opening phase, which tends to lower the particulate concentration in the area of the laboratory where air is sampled.Otherwise the process phase 1c including the graphite spraying of some reactor components, shows PNC values two time greater than the significant value.According to the optical counter (CPC, size range 10-1000 nm) measurements, the Fast Mobility Particle Sizer (FMPS, size range 5.6-560 nm) signal gives the same features with a delay probably due to the different distances of the two instruments from the source of the emitted particles: CPC was near the operator PBZ (personal), while FMPS was at about 1.5 m distance (near-field).The lower peak intensity measured by FMPS is probably due to the different dilution rates depending on the locations of the two instruments and on the different size ranges.(Figure 1).In any case such values were not associated with the G production process but were directly related to the graphite spraying.As confirmed by the off-line SEM investigations on the airborne materials collected on the plates of the personal impactor worn by the worker, which do not indicate any particles attributable to the CVD graphene.As risk mitigation measure, it should be recommended to conduct this specific phase under an aspiration hood or in a glove box [28].

Thermal exfoliation and spray casting deposition of GNPs
In the process 2, during two FLG thermal expansions conducted at different temperatures (phase 2a) NMs release was identified, with personal PNC levels that increase reaching evident peaks greater than the significant values in correspondence of the opening of the furnace.By comparison between two thermal expansions, PNC levels when the phase was conducted at 1150 °C were about 2.5 times greater than the phase at 1050°C.Furthermore, the time of decay needed to PNC to return at the initial conditions was longer about 1 hour in the first case.SEM-EDS analysis confirmed that during both the thermal expansion phases (2a) there was a release of carbonaceous nanostructures with irregular geometry and characterized by sharp edges and wedges, with the same shape and dimensions of the produced material.These nanostructures could be parts of WEGs who have detached during the thermal expansion from the main structure.The liquid exfoliation phase (2b) showed no evidence of any NMs release since the PNC values measured during the tip sonication are lower than the corresponding significant values.Based on such results, the use of PPE as full face respirators, gloves and the lab coats, is justified during the phase 2a due to the recognized high exposure peaks in a short time lapse.Furthermore the design of thermal expansion in a closed system and the working time schedule taking into account the decay of risk conditions inside the room after the end of the process are recommended [29].
During the phase 2c, the PNC inside the laboratory show median values lower than those of the background due to the effectiveness of ventilation hoods in removing airborne submicrometric particles.In this case for a large part of the production process, the background contribution heavily influenced the PNC inside the production laboratory.Then, only after the removal of this background influence by statistical analysis, it was possible to highlight the PNC increase in correspondence of the sonication and the evaporation of the acetone in furnace (phase 2c), anyway lower than the PNC significant value.
During phase 2d PNC peaks, exceeding 4.5 times the PNC significant value, were observed in the worker's PBZ in correspondence to the single spraying activity.In any case, such values were lower than the NRV15min TWA proposed in literature.The finding of SEM (and EDS) analyses confirmed the evidence obtained from real-time measurements stated that on the filters of the personal impactor wore by the worker during the production, few-layers of carbon-based thin flakes with irregular geometry and shapes, were found (Figure 2).These structures, with lateral size of few microns, are characterized by two different types of geometry, a folded-like and a flatter one.The folded-like geometry can be probably ascribable to the out of plane flexibility and the high aspect ratio (large lateral size to thickness ratio) of GNPs.It may also be attributable to the mechanical stimulation during the spray coating procedure [34].A potential GNPs release during the spray-deposition has highlighted, representing the worst potential exposure conditions of the production process.However, the release occurs in a restricted area of the work environment, not involving other workers than the operator who performs the operation.Furthermore, the ventilated hood guarantees a rapid reduction of PNC, which almost instantly returns to safety levels, limiting the exposure time.Since the airborne GNPs resulted frequently linked to other structures of micrometric material, the filters of the full-face respirators PPE maintain their efficiency for this size of airborne particulate, as certified by the EN 14387 standard for the class of dust filters P3, with penetration requirements tested according to the EN 13274-7 standard.However it is recommended to perform the spray coating in an enclosed system or by using a glove box, or, alternatively, by introducing an automated device remotely controlled [30].

Wet Jet Mill of FLG
In the process 3, all the activities carried out during FLG production may be at potential risk of NMs (but not necessarily of FLG) emission due to the  values greater than 1 taking as a reference the FF background.During the phases 3a (Wet Jet Mill) and 3b (Rotovapor), the peaks of PNC time series were linked to the use of solvents (DMSO and NMP).During the phase 3a there was a generalized increase during the equipment cleaning operations with NMP (2.99E-01 ppm) in greater quantity than the DMSO (8.24E-3 ppm).Such intensity variations are also connected to the use by the workers during all the process, of such solvents plus isopropyl alcohol and acetone for cleaning the beakers.This was confirmed by VOCs analysis conducted in the workplace, in any case the mass concentration values of the solvents were below the OELs available in the literature (50 ppm for DMSO and 10 ppm for NMP respectively).Also in the phase 3b, the increase of PNC was associated with the use of solvents which are not present in the background of the laboratory and outside.
PNC values resulted strongly influenced by the environmental conditions (i.e. the forced ventilation system) during the phase 3d (Storage and Cleaning) in which the graphene is handheld in powder form.The attribution of the emission peaks to the production was not immediate due to the similar values measured by different instruments both inside the lab (NF and in the worker's PBZ) and in the prechamber (FF1), significantly different from the FF background in the hallway outside the lab (Figure 3a) [31].
Otherwise, HR-SEM analyses of the airborne materials sampled during the phase showed rare particles attributable in size and shape to those produced, but often linked to larger and visible structures in various overlapping layers (Figure 3b).Deepen characterization using analytical techniques such as, Raman spectroscopy and the Selected Area Electron Diffraction (SAED) with TEM, were functional to identify the characteristic parameters of the honeycomb lattice structure of graphene in the airborne samples [32].The integration of exposure assessment with a biomonitoring pilot study (no.6 exposed workers and same number of controls) was considered an added value for exposure characterization in this case [35].Results showed that the oxidative DNA damage in terms of positive subjects (with oxidative DNA damage higher than a cut-off value), was significantly higher in FLG workers compared to the controls [36].
Based on such evidences risk management measures, have been proposed to mitigate the potential risk for the workers.They should primarily include the possibility to contain the phases at higher risk through the implementation of structural and equipment interventions aimed at reducing the interface between the operator and the handled materials (e.g.closed systems, glove boxes, ventilated boxes).Furthermore it will be possible to organize the processes by reducing working times (e.g.implementation of specific procedures and shifts).Finally we recommend to enforce the maintenance programs in order to guarantee the efficiency of collective (e.g.aspiration hoods) and personal (e.g.gloves, glasses, masks and safety clothing) protective equipment.PPE should be worn by the workers during their presence in the production lab in all the phases in which the risk of contact with the material in powder form may happen.
The effectiveness evaluation after the introduction of a new closed system in which to perform the storage and cleaning phases, to reduce the interface between the operator and the handled materials in powder form, was conducted in the same laboratory after one year.The introduction of this containment system together with the enhancement of mechanical ventilation and the higher attention to the PPE use, seem to have significantly mitigated the PNC emissions and reduced the genotoxic and oxidative effects previously observed [33].

Prevention-through-Design in the graphene production scale up
According to the PtD approach, risk management measures must be adapted to the entire process or specific phases and taking into account the needs related to the different scaling up stages, from the laboratory, to the pilot plant and then to the industrial factory.First of all the amount of produced materials exponentially increases in this development chain.As an example in the wet jet mill FLG case the R&D laboratory process produced gram-scale quantities of graphene-based powder [27] while the pilot plant enables the massive production of kilograms per cycle [37].
As a general principle, the use of NM in powder form should be avoided.It is preferable to produce and store them suspended in liquid or bound to a solid matrix.When this is not feasible, closed systems with controlled ventilation should be used for handling.Listed below are some methods to prevent or reduce the emission [17]: -Wet NM synthesis: e.g.FLG synthesis by exfoliation of graphite in solvent/water with recirculation to reduce waste.-Encapsulation of powders in granules, by inclusion in a suitable matrix, coating them to reduce dustiness.-Granulation into micron-sized agglomerates, which can be re-dispersed to their original size, by spray drying/freeze-drying or pelletizing/self-agglomerating.-Optimization of operating conditions in order to reduce rejects and fugitive releases: e.g. the optimization of the furnace reaction temperature in the production of GNPs by thermal exfoliation, allows a reduction of the release of dust when the furnace was opened.-Dip coating or laminating are preferred over spraying.Alternatively, low pressure, high volume or airless spray systems produce less over-spray than conventional spraying.-Wet cutting techniques for machining processes.
Furthermore, local exhaust systems may act on the transmission from the source to the operator to reduce exposure after there has been an emission.Other recommended risk management measures to be implemented in the design of the H&S measures in the workplace will be: -Allow the powder production in a dedicated chamber, preferably with negative pressure to contain particles leakages.-Arrange air showers at the exit from the 'powders' chamber and develop specific storage procedures for dismissed PPE, to reduce particles transport to the other workplace areas.-Test the effectiveness of available respiratory PPE compared to the effective size distribution of airborne nanostructures (according to EN 14387 standard).-Promote biomonitoring longitudinal studies on exposed workers, including tailored protocols to investigate e.g.buccal micronucleus by cytome assay on exfoliated buccal cells, direct/oxidative DNA damage by blood-Fpg comet assay on venous blood samples and 8-oxodGuo urinary oxidized bases by urine samples [36].-Use of point-of-care sensors as described by Papadoupolou et al. 2021 [38] (Patent no. PCT/IB2022/053211) to rapidly detect NMs surfaces contaminations.

Conclusions
In the present study the PtD approach has been successfully applied in different graphene production cases at the R&D laboratory stage (TRL 1-4).
Main barriers are related to the lack of knowledge/data on the effectiveness of changes required in the production processes to obtain a cost-effective reduction of the exposure.Furthermore no single tool is available that can estimate the overall risks, offer solutions based on the risks level and estimate the impacts of such solutions.Technical drawbacks of monitoring instruments due to the specific aerodynamic properties of 2-D graphene-based materials are still a challenge to be faced by new technological solutions.Finally, high skills and costs related to instruments and human resources are needed to perform a comprehensive analysis.Future steps of this work will be the implementation the PtD approach to the pilot plant and the industrial stage of graphene production, by checking the effectiveness of the recommended H&S measures.In this view the development of easy and accessible methods to periodically monitor safety levels will be expected, such as through the integration of low-cost and smart sensors in the working phases at higher risk.
In conclusion, our methodology supports the complementary use of qualitative models and quantitative data to prevent risks in parallel with the industrial scale up of NMs production processes, by identifying tailored control measures giving also the opportunity to evaluate their effectiveness.As companies adopt the PtD approach, early by implementing the hierarchy of control measures, the possibility to minimize the potential for worker injuries and illnesses allows to mitigate the related costs and to improve safe and sustainable innovation processes.

Figure 1 .
Figure 1.PNC real-time time series during the phases 1b and 1c of CVD graphene production inside the laboratory: personal (CPC, continuous orange line) and near-field (FMPS, dotted blue line) at 1Hz resolution.

Figure 2 .
Figure 2. SEM images of GNP case study: a) drop-cast sample and b) sprayed sample of GNP waterbased PU paint; c) sample as collected on inertial impactor filter during the spray coating phase 2d of GNP water-based PU paint.

Figure 3 .
Figure 3. a) Box plot of PNC during phase 3d; b) SEM image of airborne graphene aggregates sampled in the worker's PBZ.