Tom Mullen and Brendan van Wyk

Company and address

Air Products South Africa, Private Bag X02, Kempton Park, 1620, South Africa



caustic, white liquor, oxidation, sodium thiosulphate, chlorides, potassium, extraction


Reducing losses in the liquor circuit of a bleached Kraft pulp mill provides economic and environmental benefits.  Sodium and sulphur makeup requirements are lower and the COD load to the effluent treatment system is reduced.

Unfortunately, these benefits come at a price. As losses are reduced, the concentration of non-process elements (NPE) in the liquor circuit rises. Chlorine and potassium are of particular concern. It is well known that the first melting point temperature of carryover in the recovery boiler is depressed as the concentration of these chemicals in the black liquor increases.  This can cause a dramatic rise in the fouling rate of the boiler and has the potential to accelerated corrosion.

The most common means of removing NPE from the liquor cycle is to purge ElectroStatic Precipitator (ESP) catch.  Other processes designed to control NPE include the Chlorine Removal Process (CRP), the Precipitator Dust Recovery Process (PDR, Eka Chemicals), and the Precipitator Dust Purification system (PDP, Paprican, and Prosep Technologies Inc.). 

This paper investigates the use of oxidized white liquor in the extraction stage as a means of purging NPE.  In particular, the use of "partially" oxidized white liquor OWL(T) where the Na2S is oxidized to thiosulphate (Na2S2O3).

WinGEMS analyses were conducted to determine the impact of reduced liquor losses on total caustic consumption and the concentration of NPE in the liquor cycle.  The data shows that replacing the Eo stage caustic with OWL(T) can reduce a mill's operating cost and is as effective at removing NPE from the liquor cycle as purging ESP catch.


One of the advantages of using chlorine dioxide in Kraft pulp bleaching is the ability to utilize the byproduct Na2SO4 (saltcake) from the chlorine dioxide (ClO2) generator to adjust for sodium and sulfur losses in the liquor circuit. In many cases the saltcake produced in the ClO2 generator exceeds the makeup requirement. The question is; "how can a mill take advantage of this excess saltcake"?

White Liquor Chemistry
The primary compounds in white liquor are NaOH and NaSH. NaSH is produced when Na2S in the smelt from the recovery boiler is mixed with water in the smelt-dissolving tank.  The chemical equation is:

Na2S + H2O   NaSH + NaOH

This production of caustic is reflected in the definition of white liquor "Effective Alkali" (NaOH + Na2S).  Na2SO4 added to black liquor is reduced in the recovery boiler to Na2S and then ultimately hydrolyzed to form NaSH + NaOH. One mole of NaOH is produced from one mole of saltcake. 

Unfortunately adding more saltcake to the black liquor will increase the sulphidity of the white liquor, which is unacceptable.  Adding caustic to the white liquor to control sulphidity will eliminate the problem but excess white liquor will be produced. This excess white liquor could be used in the bleach plant extraction stage to displace purchased caustic but it is well documented that the NaSH in white liquor will negatively impact the bleaching performance of the extraction stage. The NaSH can be oxidized, but should it be oxidized to thiosulphate (Na2S2O3) or sulphate (Na2SO4)?

When white liquor is oxidized to thiosulphate, OWL(T), the following equation applies:

2NaSH + 2 O2   Na2S2O3+ H2O

whereas, the equation:

NaSH + NaOH + 2O2   Na2SO4+ H2O

applies when the liquor is oxidized to sulphate, OWL(S)

From the above it can be seen that oxidizing the liquor to sulphate consumes caustic and ultimately brings us back to where we started. Clearly then, to take advantage of the caustic produced from excess saltcake, the NaSH in the white liquor should only be oxidized to thiosulphate.

Examples of the difference in caustic concentration between White Liquor (WL), OWL(T) and OWL(S) are presented in Table I.  The data on WL(1) and WL(2) is from the paper entitled: Sulfur Purge: Fully Oxidizing White Liquor For Use In the Bleach Plant Purges Excess Sulfur And Utilizes NaOH [1] and the data on WL(3) is from work conducted at Air Products and Chemicals Inc.

Table 1

Table 1. Typical white liquor analysis for W.L., OWL(T) and OWL(S)

The Q-OWL liquors were oxidized primarily to sulphate. In both Q-OWL cases the E.A., which is the true measure of white liquor NaOH concentration, was reduced by 17 and 14% respectively. In the case of WL(3), where the liquor is oxidized to thiosulphate, the E.A. remains constant at approximately 90g/l.  When this same liquor is oxidized to sulphate OWL(S) the E.A. drops by 17%. These data demonstrates that NaOH produced from the hydrolysis of Na2S is retained if the liquor is oxidized to thiosulphate and lost if the liquor is oxidized to sulphate.

Bleaching Studies

A review of studies on the suitability of OWL(T) in bleaching was presented in the paper entitled: Using White Liquor Oxidized to Thiosulphate in the Bleach Plant can Reduce Caustic Purchases by 20% [2]. It was concluded that OWL(T) is suitable for an E, and Eo stage or an oxygen bleaching stage (i.e., full oxygen stage after the first acid stage). The suitability of OWL(T) as an alkali source in Eop was inconclusive. In a further paper [3] it was found that synthetic oxidized white  liquor, where 36% of the sodium sulphide was converted to thiosulphate (i.e., 64% to sulphate), was as effective as NaOH in an Eop stage. In the same paper, the author investigated the rate of H2O2 decomposition in commercially oxidized white liquor. The decomposition was significant. The author attributed this to the presence of "residual sulphide".  Therefore in a commercial situation it will be important to ensure all of the Na2S is converted to thiosulphate.  The author also found that the presence of transition metals in commercial "soda" liquors had a pronounced effect on hydrogen peroxide decomposition and that the addition of magnesium sulfate into the liquor substantially reduced the rate of decomposition.


Mass balances were conducted using the WinGEMS simulation program. Figure 1 is a simplified schematic of the base case. The production rate was set at 1000 odmt/d (bleached). The chloride concentration in the wood was set at 150 ppm (on a dry basis) and the potassium concentration was set at 600 ppm.  The chloride concentration of the makeup caustic was 200 ppm.

Figure 1

Figure 1. Base case simulation configuration

The bleach plant sequence (not shown) was D0EoDED and the brownstock Kappa number was set at 25. The first stage Kappa factor was 0.2 and the total ClO2 requirement was 3% on pulp.  The caustic requirement for the Eo stage was 1.8%. The ClO2 was assumed to be produced in the R8 process, which produces 1.4 Kg Na2SO4/Kg ClO2. Therefore, the total available saltcake was 42 Kg/odmt.

The base case liquor losses were set at 3%.  Under these conditions, the white liquor chloride concentration was 1.8 g/l and the potassium concentration was 7.2 g/l. The Na2SO4 makeup requirement was 25.2 Kg/odmt, which meant that the mill had 16.8 Kg/odmt of excess saltcake. The caustic makeup was 12.1 Kg/odmt.

Once the base case was established, the liquor losses were reduced to a minimum of 1%. The influence of lower liquor losses on the liquor cycle Cl and K concentrations and the total mill caustic consumption was then determined for three cases:

No NPE purge
ESP catch purging
All of the Eo stage caustic is replace with OWL(T)

Figure 2 is a simplified schematic of the OWL(T) and ESP simulation.

Figure 2

Figure 2. OWL(T) in Eo and ESP Catch Purging

OWL vs ESP Catch

The purging of ash from the recovery boiler electrostatic precipitator is an established method of purging chloride and potassium. When black liquor is burned in a recovery boiler a portion of the inorganic chemicals are vaporized.  As the temperature of the flue gas decreases these vapors condense into microscopic particles. This "fume" is removed in the electrostatic precipitator and returned to the black liquor.  The fume composition is primarily Na2SO4 (80+%),  Na2CO3 (5-15%), KSO4 and NaCl.

Chlorine has a lower vaporization temperature than potassium and sodium and hence has a higher concentration (relative to Na and K) in the fume as compared to its concentration in black liquor. The change in the chloride concentration between the black liquor and the ESP catch is characterized by the "Chloride Enrichment Factor" which is defined as Cl/(Na+K) in the ESP catch divided by Cl/(Na+K) in the virgin black liquor (molar basis).  This value ranges from 1.5 to about 2.5 [4].  Similarly, there is a potassium enrichment factor "KEF", K/(Na+K). This value typically ranges between 1.2 and 2.0 reflecting the lower volatility of potassium [4].

The amount of ESP catch that may be economically purged is limited by the availability of sulphate, or more correctly , the availability of sulphur from the chlorine dioxide manufacturing process. The same limit applies when OWL(T) is used in the bleach plant. In this case it is the loss of the sulphur in the white liquor. 

For this reason, a new term has been introduced [2] called the "Sulphur Enrichment Factor" (SEF), which is defined as S/(Na +K) in the ESP catch divided by the S/(Na +K) in the black liquor.  Since the majority of the ESP catch is Na2SO4, the SEF will be greater than one.  Also, we can said that if the ratio of CEF/SEF is less than 1.0, then the ratio of Cl/S in the white liquor will be greater than the ratio of Cl/S in the ESP catch. As a result, more chloride will be purged from the system if OWL(T) is used in the bleach plant than through the purging of ESP catch.

In the ESP case it was assumed that the:

CEF factor was 2.5
KEF factor was 1.5
Na2CO3 concentration in the ESP catch was 8%

In the OWL(T) case:

Causticizing efficiency was 83%
Reduction efficiency was 95%


In the OWL(T) case, all of the Eo stage caustic was replaced with oxidized white liquor. At a liquor loss of 3%, the total saltcake makeup requirement was 40.7 Kg/odmt, which was less than the saltcake produced in the ClO2 generator. At a loss of 1%, the saltcake makeup was 33.9 Kg/odmt. For comparative purposes the purge of ESP catch was controlled to maintain the same saltcake makeup profile as the OWL(T) case.  This ensured that the liquor chloride and potassium concentrations were similar. 

Recovery Boiler Deposits

Graph 1 shows the change in chloride composition of the deposit in the generator bank of the recovery boiler.  In the No purge case the Cl/(Na+K) ratio goes from 2.36% to 5.56%.  In the OWL(T) and the ESP purging cases the chloride ratio dropped substantially and remains low as the liquor losses were reduced.

Graph 1

Graph 1. Generating bank deposit (CL/(Na + K) ratio (mole %) vs liquor losses (%), chlorine enrichment factor of 2.5

Graph 2 documents the change in potassium ratio. In the No Purge case the concentration rose from 5.15% to 12.1%.  At a liquor loss of 1% the K ratio in the OWL(T) case was 3.22% and in the ESP case it dropped to 4.65%. OWL(T) is a more effective K purge because the K/S ratio in white liquor is higher than the ratio in ESP dust.

Graph 2

Graph 2. Generating bank deposit K/(Na + K) ratio (mole %) vs liquor losses (%), potassium enrichment factor of 1.5

Graph 3 plots the change in generation bank deposit first melting point temperature. In the No purge case, the first melting point temperature would drop from 740C to 600C at a liquor loss of 1%. This reduction in temperature is very likely to cause a significant increase in boiler fouling rate.  In the OWL(T) case, the first melting temperature at a 1% liquor loss is 790C which is higher than the base case (i.e., 740C).  In the ESP case, the first melting temperature at a 1% liquor loss was approximately 770C.

Graph 3

Graph 3. Generating bank deposit first melting temperature vs liquor losses (%), makeup NaOH chloride concentration of 200 ppm (100% basis)

Caustic Makeup

Graph 4 documents the reduction in total caustic consumption when the liquor losses are reduced from 3% to 1%. Total caustic consumption includes the makeup requirement of the recovery cycle and the caustic used in the Eo stage of the bleach plant.  

Graph 4

Graph 4. Caustic savings vs liquor losses (%)

In the No Purge case, the caustic makeup is reduced by 9.8 Kg/odmt when the liquor losses are reduced from 3% to 1 %. From a caustic savings perspective an increase in white liquor K concentration is beneficial.  In the WinGEMS analysis it was assumed that KOH is equivalent to NaOH. Therefore, as the concentration of K increases, the white liquor sodium content is reduced and the sodium loss per Kg of liquor loss is lower. From a practical perspective, the high K concentration would have a negative impact on recovery operations.
At the base case conditions (i.e., 3% liquor loss), the replacement of the Eo stage caustic with OWL(T) reduced the total caustic demand by approximately 3Kg/odmt.  This reduction is due to the production of NaOH in the white liquor from Na2S:   Na2S + H2O NaSH + NaOH. At a liquor loss of 1%, the caustic savings in the OWL(T) case increased to 10.3 Kg/odmt.

In the ESP case, the caustic savings at 1% loss was 7.1 Kg/odmt  which is 3.2 Kg/odmt lower than the No Purge and the OWL(T) case. The difference is due to the loss of Na2CO3 in the purged ESP catch.


The WinGEMS simulations clearly establishes that the use of OWL(T) in the bleach plant removes Cl and K from the liquor cycle as effective as ESP catch purging.  The Cl/S and K/S ratio in white liquor are higher than the ratios in ESP catch. For a fixed availability of saltcake OWL(T) is a more effective means of purging K from the liquor cycle.

OWL(T)'s effectiveness at removing chlorine, relative to ESP depends upon the CEF and the concentration on chloride in the makeup caustic.  When the caustic makeup chloride concentration was increased to of  1% (10,000 ppm on a 100% basis), ESP catch purging was slightly better than the OWL(T) case.

At the cited operating conditions the replacement of Eo stage caustic with OWL(T) will reduce caustic consumption by 3 Kg/odmt as compared to the No Purge case at a liquor loss of 3%.  At a liquor loss of 1%, OWL(T) will save 3.2 Kg/odmt more caustic than the purging of ESP catch.

OWL(T) will also purging more heavy metals than ESP as the ratio of these metal to sulphur in white liquor is greater than the ratio in ESP catch. This purging action will further lower the liquor cycle deadload and may lower scaling rates (e.g., barium sulphate). The cleaner white liquor may also have a positive impact on the efficiency of OWL(T) in an Eop stage and/or reduce the need for the addition of magnesium sulphate.

Calcium carbonate scaling in the Eo stage may be an issue with OWT(T). Therefore it is important to have well clarified white liquor and to remove Ca from the pulp by operating the D0 stage at a pH of <2.5.

The reduction in black liquor Cl and K and the subsequent increase in the first melting point temperature of the carryover may provide an opportunity to reduce the recovery boiler sootblowing steam required [3].


In theory, 1 moles of caustic can be saved in the extraction stage for each mole of oxygen used in the white liquor oxidation process (i.e. 1.25 Kg/Kg). In practice, the ratio is lower due to the loss of Na2CO3 in the white liquor.  In the cases studied, the oxygen requirement was 3.0 Kg/odmt.

The economics of the process, relative to the alternatives (on a strict chemical cost basis) are simply:

$/day = Production (odmtpd) x (NaOH savings Kg/odmt x $/Kg NaOH- O2 consumed x $/Kg O2)

The NaOH savings relative to the alternatives is approximately 3Kg/odmt. From an absolute basis, replacing the bleach plant caustic with OWL(T) and lowering liquor losses from 3% to 1% will reduce caustic purchase by 11Kg/odmt while maintaining or reducing the concentration of Cl and K in the liquor cycle.


Excess saltcake from the production of ClO2 can be used to advantage by converting it to NaOH and then substituting the caustic in an Extraction stage with white liquor oxidized to thiosulphate.  WinGEMS simulations show that a caustic savings of 3 Kg/odmt can be achieved and that the chloride and potassium concentrations in the liquor cycle are reduced dramatically. Further caustic savings can be realized by lowering liquor losses.  In the case studied, a reduction from 3% to 1% lowered the caustic makeup by 11 Kg/odmt while maintaining the Cl & K concentrations in the liquor cycle below the base case levels.

A review of bleaching studies on the use of OWL(T) indicates that it is equivalent to caustic when used in an E, Eo or O bleaching stage.  Data on its equivalence in an Ep or Eop stage is mixed. The difference may be due to the presence of residual sulphide or transition metals in the commercially oxidized white liquor used in some of the studies.
The use of OWL(T) is a more economical and effective method of lowering chloride and potassium concentrations in white liquor than the purging of electrostatic precipitator catch.


1. Hurst M.M. , Yant R.E., Donovan J., Tseng J.K., Sulfur Purge: Fully Oxidizing White Liquor For Use In the Bleach Plant Purges Excess Sulfur And Utilizes NaOH., 2001 International Chemical Recovery Conference
2. Mullen W.T., Using White Liquor Oxidized to Thiosulphate in the Bleach Plant can Reduce Caustic Purchases by 20%, 2002 PAPTAC Annual Meeting
3. Parthasarathy V.R., Oxidative extraction of hardwood and softwood kraft pulps with sodium carbonate-sodium hydroxide mixtures or oxidized white liquor during multistage bleaching, Tappi J., 80(12): 253-261 (1997)
4. Backman R., Skrifvars B.J., Hupa M., Siisonen P. Mantyniemi J., Flue Gas and Dust Chemistry in Recovery Boilers and High Levels of Chlorine and Potassium, J. Pulp Paper Sci. 20(4), J119-J125 (1996)
5. Jemaa N., Paleologou M., Thompson R., Brown C, Sheedy M., Sodium chloride removal from oxidized white liquor using a fixed resin bed system, Pulp & Paper Canada, 103(3): 28-31 (2002)


[Home] [Title] [Author] [Organisation] [Keywords]