Bioprocess and Biosystems Engineering

, Volume 26, Issue 2, pp 93–101

Conversion of paper sludge to ethanol in a semicontinuous solids-fed reactor


  • Zhiliang Fan
    • Chemical and Biochemical Engineering Program, Thayer School of EngineeringDartmouth College
  • Colin South
    • Chemical and Biochemical Engineering Program, Thayer School of EngineeringDartmouth College
    • ViaLactia Biosciences
  • Kimberly Lyford
    • Chemical and Biochemical Engineering Program, Thayer School of EngineeringDartmouth College
  • Jeffery Munsie
    • Chemical and Biochemical Engineering Program, Thayer School of EngineeringDartmouth College
  • Peter van Walsum
    • Chemical and Biochemical Engineering Program, Thayer School of EngineeringDartmouth College
    • Department of Environmental StudiesBaylor University
    • Chemical and Biochemical Engineering Program, Thayer School of EngineeringDartmouth College
Original Paper

DOI: 10.1007/s00449-003-0337-x

Cite this article as:
Fan, Z., South, C., Lyford, K. et al. Bioprocess Biosyst Eng (2003) 26: 93. doi:10.1007/s00449-003-0337-x


Conversion of paper sludge to ethanol was investigated with the objective of operating under conditions approaching those expected of an industrial process. Major components of the bleached Kraft sludge studied were glucan (62 wt.%, dry basis), xylan (11.5%), and minerals (17%). Complete recovery of glucose during compositional analysis required two acid hydrolysis treatments rather than one. To avoid the difficulty of mixing unreacted paper sludge, a semicontinuous solids-fed laboratory bioreactor system was developed. The system featured feeding at 12-h intervals, a residence time of 4 days, and cellulase loading of 15 to 20 FPU/g cellulose. Sludge was converted to ethanol using simultaneous saccharification and fermentation (SSF) featuring a β-glucosidase-supplemented commercial cellulase preparation and glucose fermentation by Saccharomyces cerevisiea. SSF was carried out for a period of 4 months in a first-generation system, resulting in an average ethanol concentration of 35 g/L. However, steady state was not achieved and operational difficulties were encountered. These difficulties were avoided in a retrofitted design that was operated for two 1-month runs, achieving steady state with good material balance closure. Run 1 with the retrofitted reactor produced 50 g/L ethanol at a cellulose conversion of 74%. Run 2 produced 42 g/L ethanol at a conversion of 92%. For run 2, the ethanol yield was 0.466 g ethanol/g glucose equivalent fermented and >94% of the xylan fed to the reactor was solubilized to a mixture of xylan oligomers and xylose.


Paper sludgeEthanolSemicontinuous bioreactor

1 Introduction

Paper sludges are solid residues arising from wood pulping and papermaking operations. Sludge disposal is viewed as a substantial disposal problem for the paper industry [1, 2] and represents a significant cost at many mills. However, many paper sludges are potentially attractive raw materials for production of fermentation products such as ethanol [3, 4, 5, 6, 7, 8]. In contrast to essentially all native cellulosic materials [9], a significant fraction of sludges do not require pretreatment in order for near-theoretical yields to be achieved via enzymatic hydrolysis [8]. This feature combined with the availability of utilities and other infrastructural elements at many mills offers significant potential for simplified processing for a facility located at a paper mill as compared to dedicated facilities processing other cellulosic materials. Finally, enzymatic hydrolysis may facilitate recovery and reuse of minerals (e.g., titanium dioxide) from paper sludge. The combined value of ethanol production, avoided disposal, and mineral recovery can potentially exceed $200/dry ton paper sludge. Because of these features, paper sludge processing represents a potential point-of-entry and proving ground for emergent industrial processes featuring enzymatic hydrolysis [8].

In our previous report of detailed compositional analysis for 15 paper sludges [8], carbohydrate accounted for an average of 42% on a dry weight basis, with most of the balance consisting of ash (23.2 wt.% on average) and acid-insoluble volatile matter (17.4 wt.% on average). Individual sludges deviated substantially from these averages in some cases. The nature of this unexpectedly large fraction of presumably organic material was not elucidated but is of importance in the context of process design and economics.

Most prior studies of paper sludge processing via enzymatic hydrolysis and fermentation were undertaken primarily for the purpose of determining product yields and, in some cases, rates. Consistent with this objective, rather low substrate concentrations (<20 g/L cellulose in most cases) and batch vessels were employed in these studies, and little or no attention was given to producing ethanol at concentrations that are potentially recoverable in an economic operation. Fed-batch hydrolysis and fermentation of paper sludge was mentioned by Katzen et al. [6], however, no experimental data were presented. Lark et al. [7] reported production of 32 and 35 g/L ethanol from 180 and 190 g/L paper sludge, respectively, in 250-ml flasks operated in batch mode. Recovery of ethanol via distillation becomes more costly relative to that for process designs involving cellulosic substrates other than paper sludge at ethanol concentrations less than about 4 wt.% [10].

In this study we sought to produce ethanol from paper sludge under conditions more nearly representative of those expected in an industrial process and, in particular, at ethanol concentrations in excess of 4 wt.%. Pursuant to this, we developed and subsequently modified a semicontinuous, solids-fed reactor system. In order to close material balances for this system, it was necessary to refine our prior carbohydrate analysis techniques.

2 Materials and methods

2.1 Sources of sludge, yeast, and cellulase

Paper sludge was provided by Pulp and Paper of America from the Cascade mill (bleached Kraft process) in Gorham, NH. Saccharomyces cerevisiae D5A (provided by NREL) was used for simultaneous saccharification and fermentation (SSF). Cellulase (DP151) was from Iogen (Ottawa, Ontario) and β-glucosidase (Novozyme 188) was from Novo Nordisk (Wilton, CT). Sludge was stored at 4°C for 1 to 4 months before use.

2.2 Growth medium and solutions added to the reactor

Solid, moist paper sludge (moisture content about 70%) was packed into a 4-in. feed tube (described subsequently), autoclaved for 12 h at 121°C, and fed to the reactor as described subsequently. In addition, three solutions were fed to the reactor. The first solution was distilled water, which was sterilized for 60 min at 121°C. The second was a nutrient solution for either yeast extract–peptone “rich” medium or corn steep liquor–magnesium sulfate “lean” medium. The rich medium contained yeast extract and peptone as a fivefold concentrated solution to give a final concentration of peptone 20 g/L and yeast extract 10 g/L after combining with paper sludge and the other solutions fed to the reactor. The lean medium contained MgSO4 and corn steep liquor (CPC International, Summit-Argo, IL) as a fivefold concentrated solution to give final concentrations of 5 mM MgSO4 and 0.3% (volume/volume) corn steep liquor. The composition of the growth medium was based on that described by Kadam et al. [11]. The concentrated nutrient solution was prepared by adding MgSO4 to 9 L of distilled water in a 10 L glass carboy, adjusting to pH=4.5 with sulfuric acid, autoclaving for 1 h, and adding filter sterilized corn steep liquor after cooling. The third solution was prepared by adding β-glucosidase at a ratio of 3 units/filter paper unit (FPU) to undiluted Iogen cellulase and filter sterilizing. The Iogen cellulase was centrifuged prior to filter sterilization; the activity of the supernatant was measured to be 100 FPU/ml.

2.3 Analysis of feed and effluent samples

The dry weight of unreacted paper sludge and reactor effluents were determined from samples centrifuged at 4000 g for 15 min, washed in water, and centrifuged again. The resulting pellet was dried to constant weight over a period of several days at 70°C, cooled in a desiccator, and weighed. The carbohydrate content of dried cellulose-containing solid samples was determined via a first quantitative saccharification based on a 1-h, 30°C incubation in 72 wt.% H2SO4 followed by a second 1-h incubation in 4 wt.% H2SO4 at 121°C [12]. The solid residue from the first quantitative saccharification procedure was rinsed and dried, and a second quantitative saccharification was performed. Values for wt.% glucan in solid samples were calculated by adding the results from the two quantitative saccharifications
$$ \% \;glucan{\text{ = }}{\left[ {{\left( {\frac{{{\text{g}}\;glucan}} {{{\text{g}}\;solid}}} \right)}_{{{\text{QS1}}}} + {\left( {\frac{{{\text{g}}\;solid\;residue}} {{{\text{g}}\;solid}}} \right)}_{{{\text{QS1}}}} \times {\left( {\frac{{{\text{g}}\;glucan}} {{{\text{g}}\;solid\;residue}}} \right)}_{{{\text{QS2}}}} } \right]} \times 100\% . $$
Similar equations were used to calculate xylan and mannan concentrations, although the second quantitative saccharification was found to yield additional sugars only in the case of glucose.

Acid soluble lignin (wt.%) was determined by measuring the absorbance of the first quantitative saccharification hydrolyzate at 205 nM [12]. The acid-insoluble mineral content and acid-insoluble lignin content were determined by heating the dried solid residue from the second quantitative saccharification in a muffle furnace at 500°C.

2.4 Soluble carbohydrate and ethanol

For determination of ethanol and monomeric sugars, reactor samples were acidified with one volume 10 wt.% sulfuric acid per nine volumes sample, and centrifuged at 1500 g for 15 min. The supernatant was analyzed for soluble sugar and ethanol using an Aminex HPX-87H column (Biorad, Foster city, CA, refractive index detector) as described previously [13]. For determination of soluble sugar oligomers, a second reactor sample (or subsample) was acidified with one volume 10 wt.% sulfuric acid per three volumes of sample, and centrifuged at 1500 g for 15 min. The supernatant was transferred to a 30-mL serum bottle and autoclaved sealed at 121°C for 1 h. The hydrolyzate composition was determined via HPLC as described above, and oligomer concentrations (g/L) were calculated as follows:
$$ \begin{aligned} {\left[ {glucan oligomer} \right]}{\text{ = }}{\left( {{\left[ {glucose} \right]}{\text{ after hydrolysis}} - {\text{soluble}}\;{\left[ {glucose} \right]}{\text{ before hydrolysis}}} \right)} \times {{\text{162}}} \mathord{\left/ {\vphantom {{{\text{162}}} {{\text{180}}}}} \right. \kern-\nulldelimiterspace} {{\text{180}}}, & \\ {\left[ {xylan oligomer} \right]}{\text{ = }}{\left( {{\left[ {xylose} \right]}{\text{ after hydrolysis}} - {\text{soluble}}\;{\left[ {xylose} \right]}{\text{ before hydrolysis}}} \right)} \times {{\text{132}}} \mathord{\left/ {\vphantom {{{\text{132}}} {{\text{150}}}}} \right. \kern-\nulldelimiterspace} {{\text{150}}}{\text{.}} & \\ \end{aligned} $$

2.5 Power required for mixing

Power consumption for mixing was determined using a variable-speed motor adjusted to rotate at 100 rpm. The motor shaft was connected to a flat rectangular impeller blade of diameter 4 cm and height 1 cm, which was placed in a 600-ml vessel containing 350 ml of sludge. The current drawn by the motor was recorded when the motor was unloaded, turning in air, and when the motor was loaded, stirring sludge at varying concentrations and conversions. The difference in current drawn while loaded versus unloaded was used to determine the power requirement (amps×volts).

2.6 Reactor systems

Two bioreactor designs were investigated. One was called the “original” design (Fig. 1); the other was called the “retrofitted” design (Fig. 2). The solids feeder, the same for both designs, consisted of a motor (Model 4LL05, Dayton Electric, Niles, IL, USA) connected to a 3/4-in. threaded rod situated in a rod-housing. As depicted in Fig. 1, actuating the motor caused the threaded rod to rotate, which in turn caused a nut constrained from rotating by two welded “wings” to advance. The advancing nut moved a sleeve forward, which was fitted into a slot in a custom-machined Teflon plunger. This in turn caused the plug of paper sludge in the 4-in. diameter stainless steel feed tube to advance. The rod housing and feed tube assembly combined were 7 ft long. Both designs featured a sludge delivery mechanism featuring a 4-in. diameter stainless steel feed tube with a piston that advanced a solid plug of paper sludge into a rotary “chopper.” The chopper sheared the advancing paper sludge plug, causing fragments thus generated to fall into an 800-mL working volume jacketed stainless steel fermentor. The reactor temperature was maintained at 36°C by water circulated through a jacket on the fermentor.
Fig. 1

Diagram of the solids-fed reactor, original design (1, feed tube motor; 2, threaded rod; 3, plunger; 4, feed tube; 5, chopper motor; 6, chopper shaft; 7, chopper blade; 8, conical section (4-in. to 2-in. diameter); 9, liquid feed inlet; 10, overflow line; 11, fermentor; 12, stirrer shaft; 13, stirrer motor; 14, underflow line (with solenoid valve); 15, stirrer; 16, fermentor window; 17, fermentor jacket; 18, 2-in. isolation valve; 19, threaded rod; 20, nut; 21, rod housing; 22, sleeve tube; 23, wings)

Fig. 2

Diagram of the solids-fed reactor, retrofitted design. Reactor components are as for Fig. 1 except for those explicitly named

In the original design (Fig. 1), a 9-in. long conical section of process pipe with diameter starting at 4 in. and tapering to 2 in. was attached at one end to the bottom of the chopper and at the other end to a 2-in. isolation valve. The 2-in. isolation valve was in turn connected to a 2-in. diameter flanged opening in the feedplate to a 4-in. diameter, 800-ml working volume fermentor with a custom two-blade spiral agitator (blade width 16 mm, blade diameter 90 mm, and blade length 115 mm) connected to a shaft that penetrated the bottom of the reactor through a friction seal (Applikon, Foster City, CA). A steam fitting was placed at the top of the chopper assembly, and a drain tube for condensate was placed just above the ball valve. To replace a near-empty feed tube with a new tube filled with paper sludge, the isolation valve was closed to isolate the fermentor from the chopper assembly. The near-empty feed tube was then removed and replaced with a previously autoclaved full feed tube. The feed tube–chopper junction was steamed in place at 121°C for 30 min. Thereafter the 2-in. ball valve was opened and normal operation resumed.

The retrofitted reactor (Fig. 2) featured the following changes relative to the original design: (1) a 4-in. isolation valve was placed between the feed tube and the chopper assembly, the 2-in. isolation valve and conical section between the chopper and the fermentor in the original design were eliminated, and steam a steam fitting and condensate line were added on the feed-tube side of the 4-in. valve; (2) a common drive (Model 6Z822A, Dayton Electric, Niles, IL) mounted on top of the chopper was used to continuously turn the chopper and reactor agitator at 60 rpm, and the seal at the bottom of the fermentor was eliminated; (3) a 2-in. long section of 4-in. process pipe, added between the chopper and the fermentor, was used to anchor a bearing for the chopper/agitator shaft and for introducing nutrient, enzyme, and distilled water. A horizontal rod at the bottom of the chopper was placed just above the bearing supporter to knock off any sludge that landed there. With these changes, the feed tube replacement procedure featured retracting the plunger, closing the 4-in. ball valve, removing the old feed tube, connecting the new feed tube, steaming in place, opening the 4-in. ball valve, and advancing the paper sludge plug through the ball valve until material is added to the fermentor.

The feeding schedule for paper sludge and water, concentrated nutrient, and enzyme solutions (described above) was determined by a programmed logic controller (SLC-500 programmable controller, Rockwell Automation, Milwaukee, WI). This controller activated a motor (Model 4LL05, Dayton Electric) at the end of the feed tube for sludge delivery and peristaltic pumps for the concentrated nutrient solution, cellulase solution, and distilled water. The reactor system is presented in Fig. 3, including reservoirs, actuators, and interconnections between system elements.
Fig. 3

Diagram of system used for simultaneous saccharification and fermentation of paper sludge

2.7 Reactor operation

The reactor was first run in batch mode for 2 to 3 days before it was switched to semicontinuous mode. In semicontinuous mode, 1/8 of the reactor working volume was displaced every 12 h, corresponding to about 100 ml displaced every 12 h and a liquid residence time of 4 days. Effluent was withdrawn either from the top of the reactor or primarily from the bottom of the reactor with a minor fraction withdrawn from the top. Withdrawal from the top of the reactor was accomplished by opening a solenoid valve connected to an overflow line positioned at a height such that the volume of slurry in the reactor was 700 ml after the withdrawal was complete. Withdrawal from the bottom of the reactor was accomplished by opening a solenoid valve connected to an underflow line leading to an 80-ml closed sampling vessel equipped with a gas vent. After the sampling vessel was filled, the valve connected to the underflow line was closed and the valve connected to the overflow line was briefly opened in order to expel additional liquid leaving behind a slurry volume of 700 ml. As depicted in Fig. 4, the operating cycle of the reactor featured: (a) slurry withdrawn through the underflow line; (b) slurry withdrawn through the overflow line; (c) addition of sludge to the reactor; (d) closed-system operation at a 800-ml working volume.
Fig. 4

Reactor operating schedule

2.8 Calculation of flow rates, mass recoveries, and conversion

Material balance calculations for continuous paper sludge conversion were based on the flow rates of components into and out of the reactor at steady state. The rate of sludge fed to the reactor ( r sludge, g/day) was calculated using
$$ r_{{sludge}} = \frac{{\Delta V}} {{\Delta t}}\rho _{{sludge}} $$
  • ΔV=cumulative volume of sludge fed during the duration of the steady-state (L),

  • ρsludge=sludge density (g dry sludge/L),

  • Δt=duration of the steady state (days).

The density of the sludge in the feed tube was estimated by weighing the feed tube before and after an extended period of reactor feeding and by directly measuring the density of the sludge remaining in the feed tube after the reactor was shut down. Both methods agreed well (+2.8%) and were used to estimate the density of sludge in the feed tube at 185 g (dry weight)/L. The rate at which the i th insoluble component (cellulose, xylan, acid-insoluble minerals) entered the reactor ( r i,in, g/day) was calculated using
$$ r_{{i,{\text{in}}}} = r_{{{\text{sludge}}}} f_{{i,{\text{in}}}} , $$
  • f i,in =mass fraction of component i in unreacted paper sludge.

The rate at which the i th insoluble component left the reactor ( r i,out, g/day) was calculated using
$$ r_{{i,{\text{out}}}} = \frac{{{\mathop \Sigma \limits_{k = 1}^n }M_{k} f_{{i,k}} }} {{\Delta t}}, $$
  • M k=the weight of dry solids removed by the k th withdrawal of effluent (g),

  • f i, k =mass fraction of component i per mass solids in k th withdrawal of effluent.

The rate at which the j th soluble component (ethanol, xylan oligomer, xylose, glucan oligomer) left leaves the reactor ( r j,in, g/day) was calculated using
$$ r_{{j,{\text{out}}}} = \frac{{{\mathop \Sigma \limits_{k = 1}^n }V_{k} C_{{j,k}} }} {{\Delta t}}, $$
  • V k=the volume of the k th withdrawal of effluent (L),

  • C j, k =concentration of component j in k th liquid sample (g/L).

The % conversion of cellulose X was calculated using
$$ X = \frac{{r_{{{\text{cellulose}}{\text{,in}}}} - r_{{{\text{cellulose}}{\text{,out}}}} - r_{{{\text{glucose}}{\text{,oligomer}}{\text{,out}}}} }} {{r_{{{\text{cellulose}}{\text{,in}}}} }} \times 100\% . $$
The mass recovery of xylan Rxylan was calculated using
$$ R_{{{\text{xylan}}}} = \frac{{r_{{{\text{xylan,out}}}} + r_{{{\text{xylan,oligomer,out}}}} + r_{{{\text{xylose,out}}}} {\left( {{132} \mathord{\left/ {\vphantom {{132} {150}}} \right. \kern-\nulldelimiterspace} {150}} \right)}}} {{r_{{{\text{xylan,in}}}} }}. $$
The mass recovery of acid insoluble mineral (AIM) RAIM was calculated using
$$ R_{{{\text{AIM}}}} = \frac{{r_{{{\text{AIM,out}}}} }} {{r_{{{\text{AIM,in}}}} }}. $$
The yield of ethanol YEtOH/S was calculated using
$$ Y_{{{{\text{EtOH}}} \mathord{\left/ {\vphantom {{{\text{EtOH}}} {\text{S}}}} \right. \kern-\nulldelimiterspace} {\text{S}}}} = \frac{{r_{{{\text{EtOH,out}}}} }} {{r_{{{\text{cellulose}}}} \times \frac{{180}} {{162}} + r_{{{\text{glucose,in}}}} }}, $$
where rglucose,in arose from the small flow of glucose (<2.5% of rcellulose) originating from the cellulase solution.

3 Results

3.1 Compositional analysis

The composition of the paper sludge used in this study is presented in Table 1. The glucan content after a first quantitative saccharification was 57 wt.% relative to the dry mass of the original sample. Additional glucose representing 5 wt.% of the original sample was liberated after a second quantitative saccharification. Xylan and mannan totalled 11.5 wt.% and 2.5 wt.%, respectively, resulting in a total carbohydrate content (glucan+xylan+mannan) of 76 wt.%. Nonfermentable components (acid-soluble minerals and lignin, acid-insoluble minerals and lignin) made up an additional 20.5 wt.%. The overall mass recovery was 96.5% with acid-insoluble volatile matter representing 3%. Taking into account the water of hydrolysis, the mass of potential soluble sugars per mass dry sludge was 0.848.
Table 1

Paper sludge composition


  First quant.sacch.


  Second quant. sacch.




















3.2 Mixing characteristics

The power required to mix paper sludge (at 100 rpm) was investigated as a function of solids concentration and the extent of enzymatic hydrolysis expressed in terms of the fractional conversion, X ( X =0 for no hydrolysis, X =1 for complete hydrolysis). As presented in Fig. 5, power requirements increased sharply with increasing solids concentration for sludge with X =0 and X =0.15. At a solids content <6 wt.%, corresponding to a carbohydrate content <4.6 wt.% and a potential ethanol concentration <2.3 wt.%, unconverted sludge in the 600-ml beaker used in the mixing experiment rotated as a solid plug and cannot be said to have been mixed at all. By contrast, there was no discernable increase in the power required for mixing the residue resulting from sludge that had been completely hydrolyzed ( X =1) up to a concentration of 10 wt.%.
Fig. 5

Mixing characteristics of paper sludge

Our results (Fig. 5) indicate that it is exceedingly difficult to mix the unreacted paper sludge tested here at the solids concentration required in order to produce ethanol at the targeted concentration of 40 g/L. However, the mineral-rich residue present after complete enzymatic hydrolysis was no more difficult to mix than water. Other authors have also noted the pronounced liquification accompanying enzymatic hydrolysis of paper sludge [6]. In light of these results, a continuous or semicontinuous operating strategy would appear to be advantageous as compared to batch operation. In particular, it is desirable in order to have reasonable mixing requirements that the solids in the reactor at any particular time have a high mean conversion.

3.3 Semicontinuous SSF in the original bioreactor design

Semicontinuous SSF was performed in the reactor of the original design (Fig. 1) for up to 4 months. Data is presented in Fig. 6 for a reactor fed 75 g/L cellulose at 12-h intervals for an overall residence time (reactor volume/flow out) of 4 days. Rich medium was used and the cellulase loading was 20 FPU/g cellulose with supplemental β-glucosidase. The feed was changed at days 48 and 88. Effluent was removed from the top of the reactor up to day 76 and from the bottom of the reactor starting on day 77. For the period spanned by day 30 to day 118, the average ethanol concentration in the fermentation broth was 35 g/L, and the average solid out was about 10 g/L.
Fig. 6

Semicontinuous SSF, original reactor design

Using the original reactor design, the solids leaving the reactor increased over time following addition of a new feed tube. Accumulation of solids was less pronounced when reactor effluent was withdrawn from the bottom of the reactor rather than the top. The effluent solids concentration decreased markedly each time the feed tube was changed, which we attribute to compression of the paper sludge plug following changing of the feed tube.

In addition to solids accumulation, the reactor of the original design also had some operational difficulties. Several times during the run shown in Fig. 1 as well as other runs, solids would accumulate in the conical section above the 2-in. isolation valve (Fig. 1), resulting in inconsistent sludge delivery and, in some cases plugging, which required the reactor to be shut down. In addition, the seal for the bottom-mounted agitator (Fig. 1) failed on several occasions.

3.4 Development of a retrofitted reactor and consistency of solids feeding

A retrofitted reactor was developed (Fig. 2) with the objective of addressing the limitations of the original design in order to improve reliability. As detailed in the Methods section, features of the retrofitted reactor included no diameter reduction in the path of solids delivery and elimination of the seal in the bottom of the reactor. We operated the retrofitted reactor without mechanical failure for 3 months over the course of several runs, which represents a much higher degree of reliability than achieved in the original reactor. As may be seen from the data in Fig. 7, the consistency of feed delivery was much improved for the retrofitted design as compared to the original design.
Fig. 7

Comparison of solid delivery for the original and retrofitted reactor designs

3.5 Semicontinuous SSF using the retrofitted reactor

Semicontinuous SSF was carried out in the retrofitted reactor for two 1-month runs. For both runs, the reactor was operated at a residence time of 4 days with feeding at 12-h intervals, effluent was withdrawn from the bottom of the reactor, and lean medium was used. No glucose or other soluble glucan was detectable at any point during either of the runs.

Figure 8 presents data for a run using the retrofitted reactor with a cellulose feed of 120 g/L and cellulase loading of 15 FPU/g cellulose. After a transient of about ten days, steady-state was achieved with a mean ethanol concentration of 50 g/L and cellulose conversion of 74 wt.%.
Fig. 8

Semicontinuous SSF, retrofitted reactor, run 1

Figure 9 presents data for a second run, this time with a cellulose feed of 82 g/L and cellulase loading of 20 FPU/g glucan. Steady-state material balances for run 2 are presented in Table 2 for the period from day 15 to day 45. The ethanol produced at steady state averaged 42.1 g/L, and the conversion of cellulose averaged 92%. Mass recoveries of 98% and 93.3% were measured for xylose and acid insoluble minerals respectively, and the ethanol yield was 0.46 g ethanol/g glucose equivalent utilized.
Fig. 9

Semicontinuous SSF, retrofitted reactor, run 2

Table 2

Material balance for run 2

Flow rate (g/day)

Concentration (g/L)

Recovery, conversion or yield*

Cellulose in



Cellulose out



Soluble glucose oligomer



Cellulose conversion


Glucose in from enzyme



Xylan in



Soluble xylose out



Soluble oligomer xylan out



Xylan in the solid out



Xylan recovery


Acid-insoluble in



Acid-insoluble out



Acid-insoluble recovery





Ethanol yield


*See materials and methods for calculation details.

4 Discussion

A laboratory-scale solids-fed bioreactor capable of metered aseptic feeding of paper sludge was operated for periods of up to four months (Fig. 6). A retrofitted design displayed more consistent feed delivery, substantially better operational reliability, and allowed good material balance closure. Operation of these reactors involved feeding paper sludge at a solids content of about 30 wt.%, with the balance moisture. At this solids content, which is representative of the material produced by dewatering (e.g., via filter presses) prior to disposal in commercial paper mills, paper sludge is a solid that is slightly moist to the touch. When dilution by liquid feed streams (cellulase, nutrients, water) was factored in, the overall solids content fed to the reactor was about 12 wt.%. In comparison to prior work involving aseptic continuous fermentation of cellulosic feedstocks [9], this study employed much higher solids concentrations and also involved substantially more fibrous feedstock.

The semicontinuous operating mode used herein, which we found to be a satisfactory solution to the impracticality of mixing unreacted paper sludge at high feed solids concentrations, may be distinguished from fed-batch operation. As usually interpreted in the biochemical engineering literature and textbooks, fed-batch operation involves initiation of a reaction at low concentrations of cells and substrate with subsequent addition of further substrate (and in some cases catalyst) allowing development of high product concentrations in the fermentation broth, which is harvested in a batch-wise fashion. We suspect that higher reactor productivity (e.g., g ethanol/mh) can be achieved in semicontinuous mode as compared to fed-batch mode because the time-averaged concentration of cellulose–cellulase complex and cells is higher in the former than in the latter. The feeding schedule used here (1/8 reactor volume every 12 h) was not varied and is a target for future optimization.

In our previous report of compositional analysis of paper sludges [8], we found much higher acid-insoluble volatile matter content (averaging 17.43% for 11 sludges) than observed here (3%). We believe that the much lower acid insoluble organic matter content (presumably lignin) reported here is due to our inclusion of a second quantitative saccharification step in our analytical protocol as well as the properties of the particular sludge tested. We suspect that the buffering capacity (alkalinity) of the sludge is a contributing factor to the need for a second quantitative saccharification step.

An ethanol concentration of 50 g/L was observed in run 1, for which the cellulase loading was 15 IU/g and a conversion of 74% was achieved. In run 2, which featured a cellulase loading of 20 IU/g and achieved a cellulose conversion of 92%, the average ethanol concentration was 42 g/L. It is at this point unclear what factors are responsible for the lower conversion observed in run 1 as compared to run 2. The most likely explanations in our view are the lower cellulase loading and/or higher ethanol concentrations in run 1. The kinetic model of South et al. [14] suggests that the ethanol concentrations observed in this study are near an inhibitory threshhold above which conversion falls sharply for simultaneous saccharification and fermentation featuring yeast fermentation operated at 37°C. Although higher ethanol concentrations are attainable at lower temperatures, this has the undesirable effect of lowering the rate of cellulase-mediated cellulose hydrolysis. Ethanol concentrations in the range 40–45 g/L are typical of process designs for pretreated cellulosic feedstocks [15, 16, 17]. At this concentration, ethanol recovery is relatively inexpensive and does not prevent such processes from having a very favorable overall energy balance [10].

Although xylose was not converted by the yeast used in this study, several xylose-utilizing ethanol-producing organisms have been developed [18, 19, 20]. It would be of interest to try one or more such organisms for semicontinuous conversion of paper sludge as xylan conversion has the potential to increase the ethanol yield by nearly 20% as compared to conversion of glucan only for the sludge we studied. Over 94% of the xylan orginally present in paper sludge was solubilized, although not all to monomeric xylose. Recovery of the xylan present in the feed was high at 96%, which we take to be an indication of the absence of contamination. Over the course of many runs with the semicontinuous system described herein, contamination was occasionally observed and was accompanied by a drop in xylan recovery.

We continue to believe that paper sludge is one of the most promising feedstocks for near-term commercial application of technology for converting cellulosic feedstocks into commodity products. Several results of this study are seen as significant steps toward the realization of this potential. These include development of a system suitable for laboratory-scale process development and production of recoverable ethanol concentrations. Important tasks associated with the bioreactor system remain in order to more fully evaluate commercial viability. Such tasks include demonstration of high conversion at lower cellulase loadings, optimization of the feeding schedule, testing performance with one or more xylose-utilizing microorganisms, and design and testing of a scaleable feeding device.


This study was partially supported by the Consortium for Plant Biotechnology Research (grant no. OR22072-111) and the State of New Hampshire Governor’s Office of Energy and Community Services (grant no. 622188).

Copyright information

© Springer-Verlag 2003