Characterization of Limestone Surface Impurities and Resulting Quicklime Quality

: Quicklime, rich in CaO(s), is generated by calcining limestone at high temperatures. Parallel-flow regenerative lime kilns are the most energy-effective industrial method available today. To prevent major disruptions in such kilns, a high raw material quality is necessary. Under some conditions, impurity-enriched material may adhere to limestone pebbles and enter the kiln. In this study, limestone and corresponding quicklime were analyzed to evaluate the extent and composition of surface impurities and assess the effect on quicklime product quality, here defined as free CaO. This was performed by sampling and analyzing limestone, quarry clay, laboratory-produced quicklime, and industrially produced quicklime with XRF, SEM/EDX, and XRD; interpretations were supported by thermodynamic equilibrium calculations. In the laboratory-produced quicklime, the surface impurities reacted with calcium forming Larnite, Gehlenite, Åkermanite and Merwinite, reducing the quicklime quality. The results showed that the limestone surface layer comprised 1.2 wt.-% of the total mass but possessed 4 wt.-% of the total impurities. The effect on industrially produced quicklime quality was lower; this indicated that the limestone surface impurities were removed while the material moved through the kiln. Multicomponent chemical equilibrium calculations showed that the quarry clay was expected to be fully melted at 1170 ◦ C, possibly leading to operational problems.


Introduction
The primary source of quicklime (solid rich in CaO(s)) is limestone, a sedimentary rock extracted through quarrying or mining.Limestone is naturally rich in calcium carbonate (CaCO 3 ), predominantly in the mineral Calcite.The manufacturing of quicklime is energydemanding: Limestone is decomposed to quicklime at high temperatures in industrial lime kilns, typically shaft kilns or rotary kilns.The most energy-effective method available on the market is the twin-shaft parallel-flow regenerative kiln (PFR kiln); see a schematic representation in Figure 1.The energy requirement is around 3.2-4.2MJ/kg of quicklime with a maximum quicklime temperature reaching around 1030 • C [1,2].The shafts are filled, and the raw material enters the preheating zones at the top of the shafts, and quicklime is discharged at the bottom after passing the calcination and cooling zones.Around four to five times per hour, the burning alternates between shafts, and the hot gases from the burning shaft pass through a ring channel to the non-burning shaft.This setup enables high energy efficiency and the possibility to produce a uniform soft burnt quicklime with high quality.A disadvantage of this setup is that it requires a relatively high stone and fuel quality to be able to operate without major disruptions [3].Disruptions can be caused by block formations due to quicklime densification, agglomeration, and possibly melting.It can be related to non-uniform combustion or irregular energy consumption or caused by poor raw material quality, either as a high share of impurities or high thermal decrepitation of the limestone [4,5].
Minerals 2024, 14, 608 2 of 16 setup enables high energy efficiency and the possibility to produce a uniform soft burnt quicklime with high quality.A disadvantage of this setup is that it requires a relatively high stone and fuel quality to be able to operate without major disruptions [3].Disruptions can be caused by block formations due to quicklime densification, agglomeration, and possibly melting.It can be related to non-uniform combustion or irregular energy consumption or caused by poor raw material quality, either as a high share of impurities or high thermal decrepitation of the limestone [4,5].Impurities found in limestone deposits can be classified into two groups, homogenous and heterogeneous.Homogenous impurities co-deposit with the limestone and are dispersed throughout the formation.Heterogeneous impurities appear as inclusions or as loosely embedded materials and are formed during the sedimentation process; they are seen as layers in the deposits [7][8][9].Quicklime quality is influenced by factors like kiln configuration and fuel characteristics, but it is primarily influenced by the composition and characteristics of the raw material: the limestone feed rock [8,10].In quicklime production, all other compounds in the raw material except CaCO3(s) are considered impurities.Common impurities are clay and siliceous matter, and the major impurity elements are Si, Al, Fe, and Mg [8,11].Limestone can be classified based on the chemical purity of the rock, where a chemically purer limestone, under similar production parameters, results in a chemically purer quicklime product.The following limestone classification is used: very high purity equivalent CaO content > 55.2 wt.-%, high purity 54.3 < CaO < 55.2 wt.-%, medium purity 52.4 < CaO < 54.3 wt.-%, low purity 47.6 < CaO < 52.4 wt.-%, and impure limestone for CaO < 47.6 wt.-% [12][13][14].Table 1 presents typical ranges of impurities and trace elements found in commercial limestone used for quicklime production [9].During limestone processing such as blasting, crushing, and sieving, a fine quarry dust usually enriched in impurities is formed.The experience of lime kiln operators suggests that the fine particulate material can adhere to the surface of limestone pebbles fed to the kiln to such an extent that it affects kiln operational stability.This is reported to be more prominent in cold and wet conditions.In a previous study, it was presented that the surface impurities can reduce the quicklime quality (defined as the level of free lime, CaO(s)) up to 1.5 wt.-% according to multicomponent chemical equilibrium calculations [15].In addition, for every 1 wt.-% of impurities in the limestone kiln feed, a reduction in free CaO(s) up to 4 wt.-% can be expected as a result of both diluting and reacting with the CaO(s) in the quicklime.This effect is more prominent at high temperatures [8].
In the present study, the objective was to evaluate how surface impurities attached to limestone pebbles affect quicklime product quality, here defined as free CaO(s).This was Impurities found in limestone deposits can be classified into two groups, homogenous and heterogeneous.Homogenous impurities co-deposit with the limestone and are dispersed throughout the formation.Heterogeneous impurities appear as inclusions or as loosely embedded materials and are formed during the sedimentation process; they are seen as layers in the deposits [7][8][9].Quicklime quality is influenced by factors like kiln configuration and fuel characteristics, but it is primarily influenced by the composition and characteristics of the raw material: the limestone feed rock [8,10].In quicklime production, all other compounds in the raw material except CaCO 3 (s) are considered impurities.Common impurities are clay and siliceous matter, and the major impurity elements are Si, Al, Fe, and Mg [8,11].Limestone can be classified based on the chemical purity of the rock, where a chemically purer limestone, under similar production parameters, results in a chemically purer quicklime product.The following limestone classification is used: very high purity equivalent CaO content > 55.2 wt.-%, high purity 54.3 < CaO < 55.2 wt.-%, medium purity 52.4 < CaO < 54.3 wt.-%, low purity 47.6 < CaO < 52.4 wt.-%, and impure limestone for CaO < 47.6 wt.-% [12][13][14].Table 1 presents typical ranges of impurities and trace elements found in commercial limestone used for quicklime production [9].During limestone processing such as blasting, crushing, and sieving, a fine quarry dust usually enriched in impurities is formed.The experience of lime kiln operators suggests that the fine particulate material can adhere to the surface of limestone pebbles fed to the kiln to such an extent that it affects kiln operational stability.This is reported to be more prominent in cold and wet conditions.In a previous study, it was presented that the surface impurities can reduce the quicklime quality (defined as the level of free lime, CaO(s)) up to 1.5 wt.-% according to multicomponent chemical equilibrium calculations [15].In addition, for every 1 wt.-% of impurities in the limestone kiln feed, a reduction in free CaO(s) up to 4 wt.-% can be expected as a result of both diluting and reacting with the CaO(s) in the quicklime.This effect is more prominent at high temperatures [8].
Minerals 2024, 14, 608 3 of 16 In the present study, the objective was to evaluate how surface impurities attached to limestone pebbles affect quicklime product quality, here defined as free CaO(s).This was performed by analyzing the elemental composition of the raw materials with XRF, the thickness of the surface layer and morphology of the surface layer with SEM/EDX, and the phase composition in limestone, quarry clay, industrially produced quicklime, and laboratory-produced quicklime with XRD.

Materials and Methods
The limestone used in this study (LS) originates from a sedimentary deposit.The limestone deposit consists of 35-40 m thick strata on top of a layer of marlstone.The limestone stratum itself consists of different defined structures.In the bottom of the deposit, a layer of crinoid limestone is present, and the top layer consists of stromatoporoid limestone.In addition, elements of reef limestone that form defined bodies are present.Erosion from the reef limestone, present as fragment limestone, surrounds the defined bodies.Clay minerals are present as thin layers in of crinoid, stromatoporoid, and reef limestone, with a thickness ranging from mm scale up to 0.5-1 m.The clay minerals originate from weathered metamorphic bedrock that has been deposited during the formation of the limestone strata in deep, warm, and calm water.A schematic view of the deposit is presented in Figure 2. LS was collected during a period of moist weather and the elemental composition is presented in Table 2 (as received).It is classified as a medium-purity limestone [12][13][14].To investigate the amount and composition of surface impurities, a represen sample of 17 kg limestone was washed by mechanical surface cleaning through scru the 40-90 mm pebbles in de-ionized water in a 15 L cylindrical drum.The surface ma settled, and the excess water was decanted before drying at 105 °C.The procedure i described elsewhere [15].The washed limestone pebbles (LSwashed) and the washing due (LSwashing residue) were weighed before further analysis.Based on a nominal quic production rate of 450 tons quicklime/day (tpd), the total weight of material in each tion/month was calculated as To investigate the amount and composition of surface impurities, a representative sample of 17 kg limestone was washed by mechanical surface cleaning through scrubbing the 40-90 mm pebbles in de-ionized water in a 15 L cylindrical drum.The surface material settled, and the excess water was decanted before drying at 105 • C. The procedure is also described elsewhere [15].The washed limestone pebbles (LS washed ) and the washing residue (LS washing residue ) were weighed before further analysis.Based on a nominal quicklime production rate of 450 tons quicklime/day (tpd), the total weight of material in each fraction/month was calculated as assuming that everything except CaO(s) and loss on ignition (LOI) was impurities.
A representative sample of the quarry clay (QC), originating from the LS quarry was gathered manually from different heterogenous clay layers occurring in the vertical quarry wall at fresh bench fronts at an approximate depth of 6 m from the surface.The samples were then consolidated into one composite sample and served as a reference for the quarry clay composition.

Quicklime Samples
Laboratory-produced quicklime (LPQ) was produced by heating LS-pebbles to 1100 • C in a muffle furnace.The maximum heating rate was 10 • C/minute, and the samples were held for 4 h at the set temperature.The gas atmosphere was 25 vol.-%CO 2 and 75 vol.-%air with a flow rate of one furnace volume/minute.The samples were slowly cooled in the furnace in a flow of N 2 gas to avoid recarbonation and hydration.The total weight loss was 43-44 wt.-% for all the samples, indicating complete calcination.In Figure 3, a thick layer is visible on the surface of an LPQ pebble; this area was manually peeled off and further analyzed.
thick layer is visible on the surface of an LPQ pebble; this area was manually peeled off and further analyzed.
The industrially produced quicklime sample (IPQ) was gathered from a PFR kiln fed with LS as kiln feed.The nominal quicklime production rate was 450 tpd.The raw material pebble size was 40-90 mm.Flue gases typically consist of 12-13 vol-% CO2.The kiln was fired with fuel oil containing less than 1.0 wt.-% ash, and the ash content was assumed to be negligible.Since the ash content was assumed to be negligible, the influence of the ash on quicklime product quality was not investigated in this work.
Typical CaCO3 content in lime kiln feed was around 97 wt.-%, corresponding to an equivalent CaO of around 54.3 wt.-%.However, the limestone investigated in this study was collected during a period of moist weather, and the average CaCO3 content during the 30 days was 96.3 ± 0.5 wt.-% (n = 30), corresponding to an equivalent CaO of around 53.9 wt.-%.During the period, the operators reported a lower temperature in the preheating zone, an increased amount of lime kiln dust in the flue gas filter, and a high blocking tendency in the ring channel.The quicklime quality was determined by available CaO as 88.5 wt.-% and residual CO2-content as 1.2 wt.-% during the 30 days.

Analytical Methods
XRF analysis is a way of determining the elemental composition in solid materials and has been used extensively to characterize limestone materials previously; see, for example, [4,5,16].In this study, XRF was performed with an ARL Perform'x model GEN-X 2500 W, Thermo Fisher Scientific (Waltham, MA, USA) on LSas received, QC, LSwashed, and LSwashing residue for average elemental composition.The sulfur content was determined with Leco CS744 (LECO Europe B.V, Geleen, The Netherlands).LOI was gravimetrically determined at 950 °C.The industrially produced quicklime sample (IPQ) was gathered from a PFR kiln fed with LS as kiln feed.The nominal quicklime production rate was 450 tpd.The raw material pebble size was 40-90 mm.Flue gases typically consist of 12-13 vol-% CO 2. The kiln was fired with fuel oil containing less than 1.0 wt.-% ash, and the ash content was assumed to be negligible.Since the ash content was assumed to be negligible, the influence of the ash on quicklime product quality was not investigated in this work.
Typical CaCO 3 content in lime kiln feed was around 97 wt.-%, corresponding to an equivalent CaO of around 54.3 wt.-%.However, the limestone investigated in this study was collected during a period of moist weather, and the average CaCO 3 content during the 30 days was 96.3 ± 0.5 wt.-% (n = 30), corresponding to an equivalent CaO of around 53.9 wt.-%.During the period, the operators reported a lower temperature in the preheating zone, an increased amount of lime kiln dust in the flue gas filter, and a high blocking tendency in the ring channel.The quicklime quality was determined by available CaO as 88.5 wt.-% and residual CO 2 -content as 1.2 wt.-% during the 30 days.

Analytical Methods
XRF analysis is a way of determining the elemental composition in solid materials and has been used extensively to characterize limestone materials previously; see, for example, [4,5,16].In this study, XRF was performed with an ARL Perform'x model GEN-X 2500 W, Thermo Fisher Scientific (Waltham, MA, USA) on LS as received , QC, LS washed , and LS washing residue for average elemental composition.The sulfur content was determined with Leco CS744 (LECO Europe B.V, Geleen, The Netherlands).LOI was gravimetrically determined at 950 • C.
Scanning Electron Microscopy coupled with Energy-Dispersive X-ray spectroscopy (SEM/EDX, Zeiss EVO LS15) (Carl Zeiss AG, Baden-Wurttemberg, Germany) was used to compare the morphology and relative elemental composition of surface impurities in the LS, LPQ, and IPQ samples.The SEM/EDX used a LaB6 electron source operating at an extended pressure of 50-60 Pa with an acceleration voltage of 15 kV and a probe current of 500-700 pA.This method has been used previously to determine the morphology and elemental composition of both limestone and quicklime materials [4,5,17,18].
Crystalline phases in LS washing residue , QC, and the surface layers of IPQ and LPQ were analyzed with Powder X-ray diffraction (XRD).For LPQ, its visible surface layer, as pointed out in Figure 3, was manually peeled off.The surface layer of IPQ was removed with a steel brush.LPQ and IPQ were analyzed with Bruker AXS d8 advance (Bruker, Billerica, MA, USA) operating in a 2θ mode with Cu-Kα radiation and a spinning sample.LS washing residue and QC were analyzed with a Malvern PANalytical, X'pert 3 operating in 2θ mode with Cu-Kα radiation on a spinning sample (Malvern Panalytical Ltd, Malvern, UK).The diffractograms were matched against references in the PDF-4 2023 database [19] and semi-quantification of the identified phases was obtained with Rietveld refinement in Topas 4.2 provided by Bruker.The crystal structure references used for Rietveld refinement are listed in Appendix A.

Multicomponent Chemical Equilibrium Calculations
To estimate the share of melt in LS as received , LS washed , LS washing residue , and QC at high temperature, multicomponent chemical equilibrium calculations based on the XRF results were performed between 700 and 1400 • C. Factsage 8.2 with the commercially available databases FToxid and FactPS, a customized database for salt melt and a customized database for alite formation in systems with relatively high MgO concentrations, was utilized for this purpose [20][21][22].The models used for melt and solid solutions were as follows: A-slag liq (P and S excluded), Bred, bC2SC, aC2SB, OlivA, C2AF, C3AF, Maxi-alkH, Maxi-Salt, Maxi-Hexa, Maxi-Fair, and Maxi-glas.The elements used in the calculations were Al, Ca, Fe, K, Mg, Mn, Na, P, Si, Ti, and C, H, N, O, and S. The calculations were performed with methane and air (N 2 and O 2 ), with O 2 levels at around 3% in the gas phase.

Results and Discussion
Quicklime has many quality parameters, such as reactivity [23], available lime, and CO 2 content [24].In the present study, the chemical quality of the quicklime pebbles is discussed in terms of the amount of free CaO(s) as affected by the impurities attached to the surface of the limestone pebbles.

Elemental Composition of Raw Materials
The elemental composition of LS as received , LS washed , LS washing residue , and QC determined by XRF, IR-S analysis, and LOI, are presented in Table 2 together with the share of LS washed and LS washing residue .The elemental composition of LS as received represents the average composition of 24,000 tons of limestone consumed during the 30-day kiln period.The composition allows for chemical classification based on purity according to [12][13][14].LS as received and LS washed both represented medium-purity limestone.However, the impurities decreased the CaO content in the bulk feed by approximately 0.3 units of wt.-% from 53.13 wt.-% in LS washed to 52.79 wt.-% in LS as received .The analyzed surface layer, LS washing residue , represented impure limestone (CaO content 47.88 wt.-%), and the impurities comprised mainly MgO, SiO 2 , Al 2 O 3 , Fe 2 O 3, and K 2 O (expressed as oxides).
Based on the washing and weighing, it was determined that LS washed consisted of 98.8 wt.-% of the total limestone mass and that LS washing residue consisted of 1.2 wt.-% of the total limestone mass.Based on the LOI and CaO content from the XRF analysis, LS washed and LS washing residue were determined to contain a monthly load of 945 and 35 tons of impurities, respectively.
The CaO content in QC was 17.93 wt.-%, and the impurities consisted mainly of MgO, SiO 2 , Al 2 O 3 , Fe 2 O 3 , and K 2 O (expressed as oxides).The same elements were found in elevated levels in LS washing, residue, indicating that the origin of the impurities was the same.

Morphology, Elemental Distribution, and Phase Composition of Raw Materials
The relative average elemental distribution of the LS surface layer is presented in Table 3.It was determined with SEM/EDX on an oxygen, chlorine, and carbon-free basis and expressed as oxides.Areas 1, 2, and 3 in Figure 4 represent different morphologies: 1 represents bulk limestone, 2 represents coarse-grained surface impurities, and 3 represents fine-grained surface impurities.The morphology is visualized in Figure 4a (middle).The surface layer of LS had a different morphology and elemental composition as compared to the bulk of the sample (as can be seen in Figure 4 and Table 3, respectively).The thickness of the surface layer was on average 251 ± 106 µm (n = 24).The layer seldom covered the whole limestone surface, suggesting that the surface impurities only stuck on some parts of the surface, e.g., cavities, or that it was spalled off by abrasion or attrition during processing such as crushing, sieving, and transportation.
The SEM analysis showed that the surface layer comprised Ca-rich grains up to 300 µm, surrounded by a matrix of much smaller grains.In Figure 4a (bottom), the main surface layer impurities (Si, Al, Mg, Fe, and K) are highlighted in turquoise and Ca in red.As indicated by the coloring, a surface layer with different levels of Ca was present.In Table 3, the relative average elemental composition, expressed as oxides, of areas 1, 2, and 3 for LS is presented.In area 1, the CaO content was on average 95 wt.-%, in area 2 it was 79 wt.-% and in area 3 the CaO content was on average 55 wt.-%.
The phase composition determined by XRD and subsequent quantitative analysis of QC and LS washing residue is presented in Table 4.The diffractograms with main peaks marked are presented in Figure 5, and the list of ICSD references used for the quantification are presented in Appendix A.   The phase composition of QC consisted of a lower concentration of Calcite as compared to the LS washing residue (23 wt.-% compared to 86 wt.-%).Instead, QC contained Quartz (<1 wt.-%), Illite/Muscovite (38 wt.-%), Chlorite (8 wt.-%), Feldspar (9 wt.-%), Pyrite (2 wt.-%), Rutile (<1 wt.-%), and Ferroan Dolomite (20 wt.-%).In Ferroan Dolomite, up to 33 mol-% of Mg can be substituted with Fe.In LS washing residue , Calcite was determined as the main phase (86 wt.-%) with minor contributions from Dolomite (12 wt.-%), Illite/Muscovite (2 wt.-%), Quarts (<1 wt.-%), and Chlorite (<1 wt.-%).The identified phases show that the composition of the surface impurities was similar to that of the QC, suggesting that QC is the origin of the fine material attached to the surface of the pebbles.Some differences between the average elemental composition based on the XRF analysis and the XRD analysis of the QC were present.In the XRD analysis, back-calculations were used to evaluate the accuracy of the XRD analysis by recalculating the composition to pure oxides.According to back-calculations, the Al 2 O 3 content was overestimated, and the SiO 2 and Fe 2 O 3 content was underestimated compared to the XRF analysis for QC.However, some Al can be substituted with Fe and Si as a solid solution in Illite/Muscovite and Chlorite.In addition, components below the detection limit of XRD are another factor that can contribute to the differences seen in the XRD and XRF analysis.Furthermore, no attempt was made to distinguish between Illite and Muscovite, as they are closely related and produce similar diffractograms.Both Microcline and Albite were used to explain some minor peaks around 24 • , and possibly, a feldspar solid solution between the two endmembers was present.The diffractogram and main identified peaks are presented in Figure 5.

Morphology, Elemental Distribution, and Phase Composition of Quicklimes
The relative average elemental composition of the surface layer of LPQ and IPQ are presented in Table 3.It was determined with SEM/EDX on an oxygen, chlorine, and carbon-free basis and expressed as oxides.Area 1, 2, and 3 represent different morphologies: 1 represents bulk quicklime, 2 represents coarse-grained surface impurities and 3 represents fine-grained surface impurities.These morphologies are visualized in Figure 4b,c (middle).

LPQ
Figure 4b (top) exemplifies the morphology of the surface area of LPQ, and a surface layer with up to 700 µm thickness was determined.In Figure 4b (bottom), Ca is marked in red, and the main impurities Mg, Al, Si, and Fe are presented in turquoise.The average elemental composition expressed as oxides and exemplified by areas 1, 2, and 3 are presented in Table 3.In area 1 the CaO content was on average 97 wt.-%; in area 2, it was 83 wt.-%; and in area 3, it was 49 wt.-%.For area 3, the MgO content was 21 wt.-%,SiO 2 content was 18 wt.-%,and Al 2 O 3 content was 9 wt.-%.A zoomed-in section of area 3 is presented in Figure 6.It shows a semi-melted matrix consisting of mainly Ca, Si, and Al surrounding magnesium-rich crystals of an approximate size of 10-30 µm.

LPQ
Figure 4b (top) exemplifies the morphology of the surface area of LPQ, and a surface layer with up to 700 µm thickness was determined.In Figure 4b (bottom), Ca is marked in red, and the main impurities Mg, Al, Si, and Fe are presented in turquoise.The average elemental composition expressed as oxides and exemplified by areas 1, 2, and 3 are presented in Table 3.In area 1 the CaO content was on average 97 wt.-%; in area 2, it was 83 wt.-%; and in area 3, it was 49 wt.-%.For area 3, the MgO content was 21 wt.-%,SiO2 content was 18 wt.-%,and Al2O3 content was 9 wt.-%.A zoomed-in section of area 3 is presented in Figure 6.It shows a semi-melted matrix consisting of mainly Ca, Si, and Al surrounding magnesium-rich crystals of an approximate size of 10-30 µm.
The phase composition determined by XRD and subsequent qualitative and quantitative analyses of LPQ are summarized in Table 4, and the diffractogram with main peaks marked are presented in Figure 5.The lime (CaO) concentration of the surface layer of LPQ was 20 wt.-%, and the main impurities Si, Al, and Mg reacted with CaO, forming Larnite (Ca2SiO4), Melilite (Ca2MgSiO7/CaAl2Si2O7), and Merwinite (Ca3Mg(SiO4)2).All the above-mentioned phases are unwanted and reduce the free CaO and thus the quality of the product.The phase composition determined by XRD and subsequent qualitative and quantitative analyses of LPQ are summarized in Table 4, and the diffractogram with main peaks marked are presented in Figure 5.The lime (CaO) concentration of the surface layer of LPQ was 20 wt.-%, and the main impurities Si, Al, and Mg reacted with CaO, forming Larnite (Ca 2 SiO 4 ), Melilite (Ca 2 MgSiO 7 /CaAl 2 Si 2 O 7 ), and Merwinite (Ca 3 Mg(SiO 4 ) 2 ).All the above-mentioned phases are unwanted and reduce the free CaO and thus the quality of the product.

IPQ
Figure 4c (top) exemplifies the morphology of the surface layer of IPQ.No distinctive surface impurities are visible; however, coarser areas were present but not limited to the surface region.Figure 4c (bottom) shows the impurities (Si, Al, Mg, and S) in turquoise and Ca in red.Coarsening was noted and further highlighted as "2" in Figure 4c (middle).The average elemental composition expressed as oxides and exemplified by areas 1 and 2 are presented in Table 3.In area 1, the CaO content was on average 95 wt.-%, and in area 2, it was on average 76 wt.-%.
Impurities present in clay minerals, e.g., Si and Al, were found on the surface and dispersed within the quicklime matrix-see Figure 4c (middle)-indicating that it is not related to the heterogeneous clay layers found in the quarry but originated from the homogeneous distribution of impurities.However, it was not possible to determine if the impurities originated from heterogeneous or homogeneous dispersed material.
The phases determined by XRD and subsequent quantitative analysis of the surface layer of IPQ are presented in Table 4; the diffractogram with main peaks marked is presented in Figure 5. Lime was identified as the main phase, and Larnite, Merwinite, Periclase (MgO), and Portlandite (Ca(OH) 2 ) were also identified.Both Merwinite and Larnite lower the free CaO by diluting and reacting with CaO.Considering the operating temperatures and material flows of the kiln, it is feasible to assume that the Ca(OH) 2 is a reaction product of CaO hydration in the cooling zone of the kiln and during sampling, storage, transport, and preparation of the samples before the analysis.Furthermore, the absence of Hatrurite (Ca 3 SiO 5 ) indicates that the temperature in the shaft kiln was below that of Hatrurite formation [15].Some Illite/Muscovite was identified as the main clay mineral of the surface region of LS; see Table 4. Illite/Muscovite contains some K, also present in the XRF analysis of LS washing residue and the EDX analysis of the surface layer of LS.Illite decomposes at around 950 • C [25], and according to Aboufadil et al. [26], the presence of calcium can prohibit the formation of high-temperature K-feldspar, Sanidine, after decomposition.A possible explanation for the low K 2 O content in the EDX analysis of calcined material (LPQ area 3, 0.2 wt.-%K 2 O and IPQ area 2, 0.5 wt.-%K 2 O) is the decomposition of Illite followed by the evaporation of K.This needs to be further investigated in terms of possible effects on kiln operation.

Overall Quality Assessment
The amount of surface impurities was quantified based on the weight of the washed limestone and washing residue.Although the surface layer was estimated to consist of 1.2 wt.-% of the total input to the kiln feed, it represented 4% of the total load of impurities, expressed as oxides.However, the overall assessment of the LPQ and IPQ (SEM/EDX and XRD analysis) indicates that the surface layer was removed to a large extent from the IPQ during processing, for example, through attrition between pebbles and abrasion with refractory surfaces in the kiln.
The removed layer can form small particles, readily suspendable within the gas flow, and eventually, at least in part, exiting the kiln with the flue gases, thus not influencing the quality of the quicklime pebbles.
Another possible fate of the surface impurities is a contribution to build-ups on the walls and in the ring channel.The clay minerals Chlorite and Illite identified in the QC are expected to, at least to some extent, melt at the operation temperature of the shaft kiln [25,27,28].For example, Irassar et al. [25] calcined Illite/Chlorite shale at 800-1100 • C and estimated that 59% was amorphous at 1100 • C. Khalfaoui et al. [27] investigated the mineral transformation in chloritic-illitic clay at 900-1075 • C and observed a glassy phase consisting of SiO 2 -K 2 O-Al 2 O 3 above 900 • C. The effect of surface impurities on lime densification, agglomeration, and melting in shaft kilns can be discussed in terms of the "sticking tendency" (ST) as described by Vola et al. [5,29].A raw material with high ST can lead to build-ups in shaft kilns, with costly kiln shutdown periods.The ST mechanisms in lime shaft kilns are not yet fully understood.Earlier work has shown that some limestone samples exhibit melting, and some marlstones were shown to be completely melted at 1300 • C. Furthermore, a high presence of periclase can decrease the sticking tendency via counteracting lime densification [5].
To examine the first melt temperature and the share of melt for LS as received , LS washed , LS washing residue , and QC, multicomponent chemical equilibrium calculations were performed; the result is presented in Figure 7.For LS as received and LS washed , the share of salt melt was below 1 wt.-% and was thus excluded from Figure 7.For LS as received , oxide melt was predicted above 1350 • C, and for LS washed , oxide melt was predicted above 1330 • C. The difference is probably due to the TiO 2 and Mn 2 O 3 content in the elemental analysis of LS washed that was not reported in LS as received , although they were probably present to some extent.As can be seen in Figure 7, salt melt-levels above 1 wt.-% were predicted between 800 and 840 • C and between 1020 and 1270 • C, and oxide melt was predicted above 1320 • C for LS washing residue .For QC, salt melt was predicted between 870 and 1110 • C, and oxide melt was predicted above 1050 • C. QC was fully melted at 1170 • C. HaiDo et al. [30] have simulated the maximal gas temperature in PFR kilns resulting in T max = 1240 • C.This is above the temperature where QC is fully melted according to equilibrium calculations.Hence, a clay particle traveling with the gas flow can be expected to be partly or fully melted, possibly contributing to build-ups.tween burning cycles.Krause et al. [2] have measured that the material temperature varies between 850 and 1050 °C near the inner wall of the burn zone of a PFR kiln, and Schlegel et al. [31] have measured that average kiln bed temperature varies between 800 and 1100 °C near the burn zone.However, no attempt was made in the present study to quantitatively determine the extent to which the clay material exits the kiln attached to the surface of the quicklime pebbles, as dust with the gas flow, or leading to build-ups.It is expected to vary with kiln parameters, such as feed fraction, temperature, heating rate, and gas velocity.This should be further investigated, for example through sampling of the lime kiln dust, shaft block formations, and ring channel build-ups.
In addition, the results can serve as a base for further investigations within the quarries, for example, by locating the main sources of quarry fine particulates with high concentrations of impurities and applying actions to reduce limestone feed product exposure to the impurities, combined with follow-up of the amount and composition of surface particulates for verification.Within a quicklime process, melted QC is expected to dissolve Ca, and the first melt temperature will rise (i.e., the composition will approach LS washing residue ), meaning that at a maintained temperature, the melt will re-crystallize.If this occurs on the kiln wall, it can contribute to build-ups in the shafts and ring channel.This phenomenon can be reinforced by the cyclic nature of the PFR kiln, as the material temperature is expected to vary between burning cycles.Krause et al. [2] have measured that the material temperature varies between 850 and 1050 • C near the inner wall of the burn zone of a PFR kiln, and Schlegel et al. [31] have measured that average kiln bed temperature varies between 800 and 1100 • C near the burn zone.However, no attempt was made in the present study to quantitatively determine the extent to which the clay material exits the kiln attached to the surface of the quicklime pebbles, as dust with the gas flow, or leading to build-ups.It is expected to vary with kiln parameters, such as feed fraction, temperature, heating rate, and gas velocity.This should be further investigated, for example through sampling of the lime kiln dust, shaft block formations, and ring channel build-ups.
In addition, the results can serve as a base for further investigations within the quarries, for example, by locating the main sources of quarry fine particulates with high concentrations of impurities and applying actions to reduce limestone feed product exposure to the impurities, combined with follow-up of the amount and composition of surface particulates for verification.

Conclusions
In this study, limestone, quarry clay, and industrially produced quicklime were sampled during moist weather causing high levels of surface impurities to adhere to the limestone.The surface layer of limestone, industrially produced quicklime, and laboratoryproduced quicklime were studied in detail through XRF, SEM/EDX, and XRD analyses.Based on the results, it can be concluded that

•
For the tested limestone, an accumulation of impurities was found in the surface layer.The washing residue comprised 1.2 wt.-% of the total material but contained 4% of the total impurities in limestone kiln feed; • Elevated levels of reaction products were found on the surface of the quicklime and consisted primarily of the same elements found on the limestone surface.The XRD analysis identified Larnite, Gehlenite, Åkermanite, and Merwinite in laboratory produced quicklime, resulting in lower free CaO on the surface; • The amount of surface impurities was lower in the industrially produced quicklime.
This indicated that the limestone surface impurities were, at least to some extent, removed while the material moved through the kiln; • The multicomponent chemical equilibrium calculations showed that the quarry clay was expected to be fully melted at 1170 • C, possibly contributing to operational problems.

Figure 1 .
Figure 1.Schematic representation of the PFR kiln.Arrows denote air/gas directions.Color grading of arrows indicate air/gas temperature, where blue: cold air and red/yellow: hot gas.Adapted from [6].

Figure 1 .
Figure 1.Schematic representation of the PFR kiln.Arrows denote air/gas directions.Color grading of arrows indicate air/gas temperature, where blue: cold air and red/yellow: hot gas.Adapted from [6].

Figure 2 .
Figure 2. Schematic view of the limestone deposit.Clay minerals are present in all parts of t posit except the fragment limestone.MSL: Meters above Sea Level.

96 4 1
Performed by the limestone sample provider and corresponds to an average of 30 days.n.a analyzed.

Figure 2 .
Figure 2. Schematic view of the limestone deposit.Clay minerals are present in all parts of the deposit except the fragment limestone.MSL: Meters above Sea Level.
represents the weight-factor from quicklime to limestone and wt.-% washing residue/washed represents the share of each fraction from the washing procedure.The amount of impurities in each fraction was calculated as W Impurities in washing residue/washed ton months = W washing residue/washed ton months •(100 − LOI

Figure 3 .
Figure 3. Laboratory-produced quicklime sample after calcination.Surface layer partially covering the surface is marked with yellow arrow.

Figure 3 .
Figure 3. Laboratory-produced quicklime sample after calcination.Surface layer partially covering the surface is marked with yellow arrow.

Figure 5 .
Figure 5. XRD-diffractograms presented between 5 and 70 2θ° of LSwashing residue, bulk QC, and surface layers of IPQ and LPQ.The main peaks of the identified phases are marked.

Figure 5 .
Figure 5. XRD-diffractograms presented between 5 and 70 2θ • of LS washing residue , bulk QC, and surface layers of IPQ and LPQ.The main peaks of the identified phases are marked.

Figure 6 .
Figure 6.Upper left: SEM-BSE image of a detail in the surface layer of LPQ, area 3 in Figure 4. Others: elemental maps of Ca, Mg, Si, Al, and Fe.Field of view = 75 µm.3.3.2.IPQ Figure 4c (top) exemplifies the morphology of the surface layer of IPQ.No distinctive surface impurities are visible; however, coarser areas were present but not limited to the surface region.Figure 4c (bottom) shows the impurities (Si, Al, Mg, and S) in turquoise

Figure 7 .
Figure 7. Share of salt melt and oxide melt of total amount of condensed phases, expressed in wt.-%.

Figure 7 .
Figure 7. Share of salt melt and oxide melt of total amount of condensed phases, expressed in wt.-%.

Table 2 .
Analysis results of LS as received , LS washed , LS washing residue , and QC, in wt.-%, determined by XRF, IR-S, and LOI at 950 • C.

LS as received 1 LS washed LS washing residue QC
Performed by the limestone sample provider and corresponds to an average of 30 days.n.a.= not analyzed.

Table 3 .
Relative average elemental composition detected in bulk and surface areas analyzed with SEM/EDX, as exemplified in Figure4(wt.-%).Values reported on oxygen, chlorine, and carbon-free basis and expressed as wt.-% oxides.
n = number of analyzed areas.n.d.= not detected.