Industrial scale fouling of heat exchangers in isocyanate production

The fouling of a commercial stainless steel (AISI 316L) during the manufacture of polymeric methylene diphenyl diisocyanate (pMDI) has been studied using laboratory‐based fouling apparatus that simulates commercial production conditions. The goal of the work is to understand the mechanisms behind the corrosion and fouling during isocyanate production with a view to improving process efficiency, not only in this process, but also others using similar plant and processes. Steel coupons were exposed to a solution of pMDI and solid amine hydrochloride, with hydrogen chloride gas being bubbled through the reaction cell. A number of different conditions were investigated, the variables being pMDI concentration, HCl gas flow duration, immersion time and temperature. Following the fouling experiments the coupons were removed from the fouling rig, photographed, and examined by XPS and ToF‐SIMS; principal component analysis was used to extend the ToF‐SIMS analysis to identify organic fouling products. The extent of fouling is shown to be relatively insensitive to pMDI concentration, but significantly influenced by continual HCl flow and increased temperature, features which increase the extent of substrate corrosion thought to be a precursor to the fouling process itself. Both XPS and ToF‐SIMS confirm the formation of various nickel chlorides in the corrosion process. Urea and metal corrosion products are found to co‐exist on certain (random) areas of the coupon surface.


| INTRODUCTION
Polyurethanes are amongst the most versatile of polymeric materials.
Commercial applications of polyurethanes as adhesives, elastomers and coatings began in the 1940s.Flexible and rigid foams found commercial use from the 1950s.The synthesis process of polyurethanes allows great control over the chemical nature of the reaction system throughout the production process, enabling a wide range of uses, from low-density foams to adhesives.Polyurethanes are produced commercially through the exothermic polyaddition reaction between polyisocyanates and polyols.A relatively narrow range of polyisocyanates is coupled with a wide range of polyols to produce a variety of materials.Of the polyisocyanates used, methylene diphenyl diisocyanate (MDI) is one of the most common used in the production of polyurethanes.
The MDI production process is described in detail elsewhere 1 ; however, the main steps involve the polycondensation of aniline and formaldehyde.This intermediary product undergoes a phosgenation step, and crude MDI is subsequently obtained through a series of heat treatment steps used to remove excess reactants and solvents.Pure MDI and polymeric MDI (by pMDI, we are considering higher order/ oligomeric species) are obtained from crude MDI through a final distillation step.The heat exchangers used in these heat treatment steps are commonly made from stainless-steels such as AISI 316L or a Duplex version, chosen for their corrosion resistance.Fouling and degradation of the metal surface can still occur, reducing the efficiency of the heat exchanger.Fouling is a common process engineering phenomenon and can be described as the formation of unwanted material deposits on heat transfer surfaces during process heating and cooling.It occurs in all industries and most heat exchanger designs, with impacts ranging from heat transfer degradation to flow resistance and pressure drops.
Such fouling necessitates periodic cleaning and the disposal of contaminated water used for cleaning to restore the protective metal surface.Heat exchanger corrosion and fouling pose a significant financial and environmental impact not just for isocyanate production but the wider chemical production industry. 2Developing our understanding of the mechanisms behind corrosion and fouling under isocyanate relevant production conditions will not only benefit the isocyanate production industry but provide transferrable knowledge to other industries aiding in of other unique corrosion challenges.
The interface chemistry of isocyanates and metal surfaces has been investigated by a variety of research groups.The interface chemistry between steel alloys is widely believed to involve interactions between iron oxide from the oxide film and oxygen-and nitrogen-containing species from the polymer. 3,4Tardio et al. 6 investigated the interactions between MDI and 316L stainless steel with the aim of identifying specific interfacial interactions.It was concluded that a monolayer of MDI forms on the surface of the steel, bonding through a strong covalent bond between the isocyanate group of the MDI and the hydroxide group on the steel surface. 5Cycloaddition reactions between Fe=O and ÀN=CO were suggested based on time-of-flight secondary ion mass spectrometry (ToF-SIMS) data, and in other investigations, both the nucleophilic attack and the cycloaddition reaction were proposed, in keeping with the known reaction chemistry of isocyanates. 5,6e research presented herein builds upon previous studies of the initial fouling conditions of 316L stainless-steel and pMDI in isocyanate production conditions. 6Deposition mechanisms in the early stages of fouling were found to be strongly influenced by the solubility of organochloride reaction products formed in the production process.An increase in the equivalent overlayer thickness was observed with decreased pMDI solution concentrations and immersion durations.Various metal oxides were identified, and iron chloride was proposed to migrate through the deposited organic layer.This work will continue to explore the chemical aspects of the fouling process, examining more severe fouling conditions with the aim of replicating conditions representative of later stages of fouling in isocyanate production.The deposition mechanisms established in previous work 7 will be compared against the observations made in these more severe fouling conditions.

| METHODS AND MATERIALS
Stainless-steel samples were prepared in the same manner as the previous report, and a detailed preparation procedure can be found elsewhere. 7A schematic and an image of the fouling apparatus used for the following sample preparation are presented in Figure 1.A select number of experimental parameters were changed to replicate more severe fouling conditions.A summary of the sample preparation procedure is detailed below: • 316L stainless-steel coupons were cleaned in ultrasonic baths of acetone and hexane (supplied by Sigma-Aldrich and Fisher Scientific, respectively), followed by UV ozone cleaning (Hitachi High-Technologies, Zone Cleaner).
• Coupons were immersed in a solution of pMDI in chlorobenzene (supplied by Merck); 5 w/w% of solid amine hydrochloride was added to all solutions and heated to a range of temperatures detailed in Table 1.
• Independent variables changed include pMDI concentration, HCl gas flow duration, immersion duration and temperature.pMDI was specifically chosen over MDI, as it is the main isocyanate component that contributes to fouling in the production process.
F I G U R E 1 Schematic and image of fouling apparatus The concentration of pMDI was increased to a maximum of 100 v/ v% in these experiments.A complex interaction between solute solubility in chlorobenzene was proposed to influence the deposition mechanism. 7Removing the solvent in this experiment was chosen with the aim of removing the influence of this mechanism in these fouling conditions, as would be the case in late stages of the production process.
Increasing HCl gas flow duration aimed to maximise the acidity of the system and promote corrosion of the stainless-steel substrate.Temperatures were increased to a maximum of 180 C, representative of the hottest environments found in the production process.
Samples were subsequently analysed by the surface analytical techniques X-ray photoelectron spectroscopy (XPS) and time-of-flight secondary ion mass spectrometry (ToF-SIMS).Both are highly surface sensitive techniques selectively providing information from the top 10 nm of the surface.A K-Alpha + X-ray photoelectron spectrometer (Thermo Scientific, East Grinstead, United Kingdom) was employed in this work.The spectrometer was operated in the constant analyser energy mode.XPS survey and high-resolution spectra were collected with a pass energy of 200 and 50 eV and a step size of 0.4 and 0.1 eV, respectively.A monochromated Al Kα X-ray source was used with a radial spot size of 400 μm.The C 1s peak was used as the reference peak for charge correction for all spectra acquired.Charge compensation was achieved using an electron flood gun, and subsequent data processing was carried out using the manufacturer's Avantage v5.890 software.Peak areas of the high-resolution spectra were employed for quantification following the removal of a Shirley (S-type) background.For all peak fitting, a 30% Lorentzian-Voigt function was used.
Surfaces were analysed with the TOF.SIMS 5 (ION-TOF GmbH, Münster, Germany) instrument.Static SIMS conditions (ion dose <10 13 ions cm À2 ) were achieved using a 25 keV Bi 3 + primary ion beam rastered over an area of 100 Â 100 μm 2 , with a 9.0 kV reflectron extractor voltage.An electron flood gun was used to avoid electrostatic charging of the specimens.Positive and negative spectra were acquired in the high-spectral-resolution (high-current-bunched mode) mode over a mass range of 1-800 u.All analyses were carried out in triplicate.Spectral acquisition and processing were achieved using ION-TOF GmbH SurfaceLab v6.4 software.Mass calibration was achieved for all spectra as follows: A selection of approximately 20 peaks of known composition was used, with a variety of chemistries including hydrocarbon, organic oxygen containing peaks and inorganic (e.g., Fe + , Ni + and Mo + , when relevant) ions were used to ensure consistency of the mass scale within the spectral set.The Sur-faceLab software requires a variety of chemistries for calibration to ensure fidelity of peak assignment.ToF-SIMS spectra contain a vast amount of information with thousands of data points collected in a single spectrum.This makes univariate analysis of ToF-SIMS spectra challenging when comparing a few fragments and practically impossible for all fragments present in a spectrum.Multivariate analysis methods are frequently employed in the analysis of ToF-SIMS spectra to aid in the analysis of such large data sets.Principal component analysis is the most widely used multivariate analysis method for processing ToF-SIMS data.Here, the MATLAB-based software package simsMVA has been employed in the analysis of ToF-SIMS data sets. 8| RESULTS

| Visual analysis
Photographs of all samples prepared can be found in Figure 2. On their surface, 316L-1 and 316L-2 have minimal fouled material, and 316L-2 presents with slightly more visual fouling; however, the surface coverage is still minimal; 316 L-3 shows more fouling than 316L-1 or 316L-2; however, 316L-4 shows the most organic fouling with a thick yellow overlayer (identified as Zone C) covering most of the surface.Two visually distinct areas can be seen on 316L-3 and 316L-4, Zone A (light centre) and Zone B (dark edge) on 316L-3 and uncovered (Zone D) and overlayer (Zone C) regions on 316L-4.
These visually unique areas will be referred to in this manner from this point onwards.

| XPS
Isocyanate components were identified in the C1s, O1s and N1s spectra on all samples, confirming pMDI is present on the surface.6][7] Isocyanate components are also found in O1s spectra and are accompanied by an oxide component in samples 316L-1, 316L-2 and 316L-3.The N1s spectrum is uniquely positioned to provide a large amount of information regarding the isocyanate chemistry as the nitrogen atom is central to many of the reactions that isocyanates undergo.Isocyanate, amine, carbodiimide and N δ+ components were observed in N1s spectra.The surface concentration on each sample, including unique regions on 316L-3 (Zones A and B) and 316L-4 (Zones C and D), can be found in Table 2. Iron, chromium, nickel and molybdenum are present in the analysis of 316L-1, confirming a thin organic overlayer is present (<10 nm), and the bulk substrate is being sampled.This is confirmed by the slightly rising background to the higher binding energy after all transition metal peaks in the survey spectrum of 316L-1 (Figure 3A), which indicates these species are located beneath the organic overlayer.The presence of Mo, together with a higher concentration of the other metallic elements, in the spectrum from this specimen is related to the carbon assay, which is significantly lower than all other specimens, indicating a thinner carbonaceous overlayer.As the conditions seen by the individual specimens of Table 1 become more severe, so the organic overlayer grows in thickness at a rate higher than the diffusion of metal ions in the dynamically developing overlayer.F I G U R E 2 Photographs of all 316L stainless-steel samples following exposures described in Table 1.Visually distinct areas on 316L-3 and 316L-4 discussed are highlighted and identified as Zones A-D.

T A B L E 1 Sample codes and preparation conditions
T A B L E 2 Surface concentration for 316L samples, including visually unique regions The shape of the background in the survey spectrum of areas on 316L-3, the NCO:C-O ratios are 0.9 ± 0.1 and 2.9 ± 0.5, respectively.Based on these XPS results, Zone A is associated with urea type species and Zone B with isocyanate-based species.
Two visually distinct areas were identified on specimen 316L-2-04.Zone C (the overlayer regions) has an increased nickel and chlorine concentration (Table 2).The Ni2p satellite structure is in keeping with NiCl 2 , and the dominant chloride component in Cl2p spectrum is also in keeping with NiCl 2 .
The average ratios of NCO:C-O O1s components for Zones C and D were 1.4 ± 0.7 and 2.0 ± 0.5, respectively.This shows a similar trend as observed on 316L-3 where regions of higher nickel chloride concentration are associated with more urea-based pMDI species.
The lack of a rising background to the higher binding energy side of the Ni2p peak indicates nickel is located on the surface of the organic overlayer (Figure 3D).

| ToF-SIMS
All samples presented with pMDI characteristic fragments in both positive and negative spectra.7]13 Cr + , Fe + and Ni + elemental ions were found in the positive spectra of 316L-1, 316L-2 and 316L-4 but not in 316L-3, in keeping with the identified metal photoelectron peaks in their respective XPS spectra.
Groups of intense peaks were identified from 119-175 u in the negative spectra of all samples.All fragment masses in these highresolution spectra were below the nominal mass and are unlikely to be organic fragments from pMDI (as characteristic organic peaks invariably occur just above unit mass, as a result of the mass of hydrogen being 1.008 u).Various transition metal chloride corrosion products would have similar masses and would agree with the XPS analysis.
By comparing the measured intensity of these groups against the calculated isotopic distribution of different metal chloride species, one can reliably identify different transition metal species.The relative abundances used for the metal ions considered in this approach are provided in Table 3.The isotopic abundance of chlorine is 35 Cl = 75.77%and 37 Cl = 24.23%.These isotope patterns are readily identified by SIMS, and the sum of the different species present, to yield the calculated isotopic distribution of Figure 5, provides an added degree of confidence in the assignment of the individual fragment ions, when compared with the measured peak intensity, and are used in the assisgnments of the high resolution mass spectra in Figure 6.
To assign these fragment peaks, a database of molecular fragments (built within SurfaceLab v6.4 software) was generated.All peaks with a deviation of <150 ppm were examined as potential assignments.If a potential assignment meets the following four criteria, it is assigned with confidence.
1.It has a deviation of less than 100 ppm for all molecular isotopes.This is assessed by the so-called Δ (delta) parameter, which is a measure of the accuracy of the assignment attributed to a particular fragment ion: 2. There is good agreement between the theoretical and measured peak intensities (including minimal remaining measured counts not accounted for by the theoretical intensity).
3. Its main molecular isotopes cannot be accounted for by other plausible molecular isotopes.Various interface specific fragments identified in previous studies were identified on samples here. 5,6,13,14Fragments identified include CNO 2 Fe À , CNO 2 Cr À , C 3 N 3 Fe À and FeNH 2 + and are associated with the 316L samples thought to have the thinnest overlayers (316L-1 and 316L-2) and therefore suggests they are primarily located in the interfacial region.Three additional interface specific fragments not identified previously were found, NCONiCl À , NCOCrCl À and NCOCuCl À .The identification of NCOCuCl À was unexpected as copper is not an alloying element of the 316L stainless-steel used in this experiment.It is proposed that these species may have deposited from copper or brass pipework or fittings of the rig used during sample preparation.
To further investigate differences in the inorganic and organic chemistry, principal component analysis (PCA) was employed to establish correlations between sample type and these unique molecu- A separate positive polarity intra-sample PCA analysis was conducted on 316L-3 specifically examining pMDI characteristic fragments (see Figures 7 and 8).Due to the comparatively low intensity of pMDI characteristic fragments in the ToF-SIMS spectrum their contribution to the information contained within a principal component will be less, relative to more intense peaks such as the common hydrocarbons frequently seen in ToF-SIMS spectra.This is particularly true for pMDI fragments greater than 132 u, which have particularly low intensities but can provide valuable information regarding functionalisation and degree of cross-linking in the pMDI network.The sole use of pMDI characteristic fragments was to aid in a focused examination of subtle changes in pMDI chemistry without the pMDI fragment being drowned out by more intense, less relevant, fragments in the principal components.The same pMDI specific PCA comparison carried out for 316L-3 showed urea species to associate with Zone C of 316L-4 (see Figure 9).This follows a similar trend as observed on 316L-3, where urea species are associated with metal chloride corrosion products.
Although C 8 H 6 NO + was also found to associate with the thick overlayer, its value is considerably lower compared to C 7 H 8 N + and C 13 H 13 N 2 + and is not amongst the most significant observation of this principal component.This trend is seen in PC1 with a variance of 88.77% of 316L-4, increasing the confidence in this trend of urea and metal chloride species associating with one another.

| DISCUSSION
The first section of this discussion will compare the effects of increasing pMDI concentration and HCl gas flow in this work to the Increasing the concentration of pMDI from 50 to 100 v/v% has relatively minimal impact on the degree of adhered pMDI.Although pMDI concentration was found to be a major factor affecting pMDI adhesion in the previous stage, 7 it was concluded that this was due to the low concentrations of pMDI influencing the particular chemical relationship established by the experimental parameters.
Increasing the HCl gas flow had a much larger influence on the degree of deposited material and degree of corrosion.Carbon and nitrogen concentrations increased by approximately 12 and 3 at%, respectively, when HCl gas flow was increased and, along with a decreased iron and chromium concentration, confirms a thicker organic overlayer was deposited on the surface.The Ni2p and Cl2p XPS spectra show an increase in concentration for increased HCl gas flow and indicate nickel chloride is present on the surface, in agreement with the ToF-SIMS analysis.
The low iron and chromium concentration, in contrast to the higher nickel concentration, indicates more extensive corrosion for longer HCl gas flows.Previous studies have investigated the structure of the oxide passive film on stainless steels and can be summarised as an iron/chromium oxide outer layer enriched in chromium compared to the bulk composition.6][17][18] HCl reacts with the iron/chromium rich passive film until the iron and chromium oxide film is removed.HCl then reacts with nickel from the enriched nickel region, and these nickel chloride the foulant layer.One should be cautious of the influence the removal of these atoms from the bulk metal will have on the chemical and mechanical properties of stainless-steel components when reused for future MDI production.The exact chemical conditions have been shown to strongly influence the deposition mechanism and emphasises the importance of establishing chemical conditions that closely resemble production conditions when investigating industrial fouling on smaller scales.

| CONCLUSIONS
The following conclusions have been made: • Increasing pMDI concentration from 50-100 v/v% was shown to have a minimal effect on degree of deposited organic material compared to other variables such as the continuous inclusion of HCl and temperature.
• Continuous inclusion of HCl and increased temperature both led to signs of increased corrosion, and it was proposed that the corrosion of 316L stainless-steel influences the organic deposition process.
• Various nickel chlorides were identified in the Ni2p and Cl2p XPS spectra and the ToF-SIMS analysis.
• Urea-formation and metal corrosion products found to be associated with one another across unique areas of the sample surface.
• Increasing the temperature to 180 C had the strongest influence on the degree of fouling, and this could be due to the increased kinetics of corrosion and hence fouling based as described above.

The
Figure4Ais consistent with NiCl 2 . 9,10Three components were identified in the Cl2p spectrum, two chlorides and one organic component.The Cl2p spectrum (Figure4B) is dominated by the chloride components with Cl2p 3/2 binding energies of 198.0 and 199.0 eV, indicative of chromium/iron and nickel chlorides, respectively.
316L-2 compared to 316L-1 indicates 316L-2 has a thicker overlayer (Figure 3A,B).The rest of 316L-2 is representative of pMDI with a small chloride component from NiCl 2 .Trace concentrations of molybdenum are also identified on 316L-2.Two visually distinct areas were identified in the appearance of 316L-3.Zone B, the dark edge has elevated average carbon concentration and a distinct lack of any transition metal species.No oxide component was identified in the O1s spectrum either, further indicating the absence of any metallic species on the dark edge of 316L-3.Zone A, the light centre of the coupon, however, has a significant metal concentration, and an oxide component is identified in the O1s spectrum.The carbon concentration is slightly lower on the light centre, all indicating a slightly thicker organic overlayer on the dark edge.Based on previous work by Shimizu 11 and Tardio, 12 comparing the intensities of NCO:C-O O1s components can provide information relating to the characteristics of isocyanate chemistry on the surface.Values of approximately 3.0 are characteristic of urea-based species, and values of approximately 1.0 are indicative of isocyanate-based species in pMDI. 6,10,11Comparing the intensities of NCO and C-O F I G U R E 3 XPS survey spectra for all 316L samples.(A) 316L-1, (B) 316L-2, (C) 316L-3 and (D) 316L-4.Note: 316L-3/4 shows overlayer spectra of the visually distinct areas.F I G U R E 4 XPS (A) Ni2p and (B) Cl2p spectra of 316L-2 O1s components for Zone B (dark edge) and Zone A (light centre)

4 . 3 À
Figure5A.This is compared against the measured peak intensity (black bar) and shows good agreement with all peaks.Groups of metal-containing fragments were observed from 119-175 u on all samples, and the resultant spectra exhibiting the measured and theoretical peak intensities can be found in Figure5.Nickel chloride fragments were the most intense metal chloride species on samples 316L-2/4, in keeping with the XPS analysis.
lar species.When employed to investigate ToF-SIMS data, PCA analysis can highlight chemical differences between analysis areas and indicate how significantly they differ.The variance of Principal Component 1 (PC1) can be used as an indication of how much the surface chemistry differs between analysis areas.Higher variance values indicate the loadings plot has a higher significance to the whole data set and therefore a more significant difference in surface chemistry.Peak lists for positive and negative spectra were acquired using the peak search tool in SurfaceLab v6.4.All peaks in the m/z range of 1-800 u that showed significant intensities (>500 counts, S/N-ratio >2.0) were selected.In addition to this selection, any identified peaks that had low intensities and were not in the selection criteria, for example, low abundance molecular isotopes, were also added.With this peak selection method, a large, representative, sample of peak intensities was taken from the spectrum whilst ensuring all characteristic mass fragments are still included.The peak lists were then normalised to total ion signal, and Poisson Scaling and mean centring carried out in the usual way.Sample 316L-3 has a low variance value for PC1 for the negative polarity spectrum (52.11%), but a significant pattern is observed between analysis areas.The light centre area is associated with pMDI and corrosion material characteristic fragments.The dark edge is associated with C À , CH À , O À , OH À , C 2 H À and PO 3 À fragments.Saturated C x H y + fragments are associated with the dark area and are separated from pMDI characteristic fragments associated with light areas.

F I G U R E 5
Measured peak intensity and calculated isotopic distribution of fragments from ToF-SIMS spectrum (157-175 u) of (A) 316L-1, (B) 316L-2, (C) 316L-3 (light centre) and (D) 316L-4 (overlayer) PC1, shown in Figure 7, is characterised by aromatic hydrocarbon fragments such as C 6 H 5 + , C 7 H 7 + and C 13 H 9 + in the positive loadings axis and functionalised isocyanate-based fragments such as C 7 H 8 N + , C 8 H 6 NO + , C 13 H 10 N + and C 13 H 13 N 2 + in the negative.From inspection of the scores plot Zone A of 316L-3 associates with the functionalised pMDI fragments, with Zone B being associated with the aromatic hydrocarbons.This contrast is likely due to the differences in the fragmentation pattern of the pMDI material across the different areas of 316L-3.This could be a consequence of an increase in crosslink density demonstrated by the increase in aromatic hydrocarbon fragments, potentially resulting from a loss of functionality.Less cross-linked regions of pMDI material will tend to yield more functionalised fragments, which are not fragmented further upon ion bombardment.This would imply increased cross-linking across the dark edge region, Zone B, of 316L-3 compared to the light central region, Zone A. Investigations conducted by Tardio 12 and Shimizu 11 examined characteristic fragments of pMDI and a high-urea derivative of pMDI.Their studies concluded that C 8 H 6 NO + and C 14 H 8 NO + are characteristic of isocyanate dominated pMDI and C 13 H 11 N 2 + , C 13 H 13 N 2 + and C 14 H 11 N 2 O + are characteristic of urea-dominated pMDI.PC2 shown in Figure 8 separated these areas with the light area associating with F I G U R E 6 High-resolution ToF-SIMS spectra of all interface specific fragments identified on 316L-1 to 316L-4 urea characteristic fragments and dark area associating with isocyanate characteristic fragments.This provides evidence for the association of urea-based species with the interfacial region of pMDI and stainless steel.The confidence of this observation is, however, not high, as this trend is only observed in PC2 which has a variance of 15%, considerably smaller than the variance of PC1 at 82%.The PCA analysis of 316L-4 shows the thick overlayer of Zone C is associated with Cl À and various metal chloride species, specifically nickel chloride, in the negative polarity spectrum and Ni + , C 7 H 8 N + and C x H y O z + species in the positive polarity spectrum.The uncovered area, Zone D is associated with CN À and C 2 H À in the negative polarity spectrum and C x H y + fragments in the positive polarity spectrum.

F I G U R E 7
PCA analysis of ToF SIMS data from intrasample comparison on 316L-3 showing PC1 (A) score plot and (B) loadings.Note: only pMDI characteristic fragments have been examined in this specific analysis plot.Light centre (black) data points in (A) refers to Zone A and dark edge (red) to Zone B. F I G U R E 8 PCA analysis of ToF SIMS data from intrasample comparison on 316L-3 showing PC2 (A) score plot and (B) loading plot.Light centre (black) data points in Figure 7A refers to Zone A and dark edge (red) to Zone B. Note: only pMDI characteristic fragments have been examined in this specific analysis.parametersused in the previous stage detailed inBevas et al. 7 corrosion products migrate to the surface of the deposited organic material.Increased corrosion can lead to an increase in organic deposition by providing reactive sites left on the metal surface after chloride attack at which pMDI can react and adhere to before repassivation of the surface.The difference in ratio ofNCO to C-O components in O1s spectrum indicated a separation of urea-based and isocyanate-based species across the light centre and dark edge identified in the visual inspection.The surface composition established in the XPS analysis and the PCA analysis of ToF-SIMS data confirms metal corrosion products have a greater abundance across the light centre and ureabased species associate with this same region.This indicates both metal corrosion products and urea-based species are involved in the initial organic deposition process.The most significant organic fouling and corrosion were observed when increasing the temperature to 180 C on sample 316L-4.The high nickel chloride concentration found in the XPS and ToF-SIMS analysis confirms the advanced state of corrosion that had taken place on 316L-4.PCA analysis of ToF-SIMS data confirmed these nickel chloride species are associated with the overlayer region, and the shape of the XPS survey spectrum shows these species are primarily located on the surface.Migration of these nickel corrosion products could explain the presence of these species primarily at the outer surface.This work has shown the organic deposition process to be mainly influenced by HCl concentration and temperature, in contrast to the strong influence pMDI concentration had in the previous study by Bevas et al. 7 Urea-based species were again found to be an important component in the deposition process; however, their exact involvement is dependent on their interactions with the larger chemical environment.Regarding the implications for the commercial production of MDI, separating high HCl concentrations from high temperature stages of the production process will aid in reducing fouling.High temperatures are reached during the later stages of the production process and utilising highly corrosion resistant materials at these stages will maximise fouling resistance whilst minimising the high cost associated with these materials.Heat exchangers used in MDI production are commonly cleaned with jets of high-pressure water to remove fouled layers.This work has shown alloying elements of the stainless steel can be removed from the metal and migrate into F I G U R E 9 PCA analysis of ToF SIMS data from intrasample comparison on 316L-4 showing PC1 (A) score plot (B) loadings.Overlayer (black) data points in (A) refer to Zone C and uncovered (red) to Zone D. Note: only pMDI characteristic fragments have been examined in this specific analysis plot.
Relative abundances a of naturally occurring metal isotopes of relevance to calculations of Figure5