*3.5. Overall Process Balances*

Finally, the glucose and xylose yields after EH of the fiber fraction and glucose and xylose yields after one-step acidic hydrolysis of the extract fraction were compared and calculated based on raw material. This consideration was made in order to evaluate the effect of alkaline extraction of the overall process balance and to monitor the carbohydrate losses during the whole process (Figure 11). Schütt [39] reports a consideration that a severity of around 4.5 is needed for gaining sufficient glucose rates after steam refining of poplar wood and enzymatic hydrolysis with Celluclast®.

**Figure 11.** Summed glucose and xylose yields after EH of the fiber fraction and after one-step acidic hydrolysis of the extracts—without (**a**) and with (**b**) alkaline extraction; all based on raw material input.

A different behavior was found for the steam refining of corn stover and enzymatic hydrolysis with Cellic® CTec2 in this study. It must be remarked that the following comparisons and differences are mainly due to the used enzymes. From previous experiments (data not shown here) it is clearly known that Cellic® CTec2 shows a much higher activity to the used biomass than Celluclast®, used in previous days. However, not shown comparison experiments with Celluclast® and steam-treated corn stover also show the highest glucose yields at severities around 4.5.

As shown in Figure 11a the glucose is also available at severities below 3.5 and there is no need for steam refining in higher regions. There is a slight increase in the glucose yield with increasing severity with no significant optimum. In contrast to those findings, there is a clear optimum for the xylose yields at a severity of 3.2. Regarding Schütt [39], the findings for the xylose yields are nearly equal and the optimum is also in severity regions below 3.5. Figure 11b represents the overall process balance after alkaline extraction of the fibers and following EH. In contrast to the prior findings, the optimum of all yields is now located at a severity of 2.77. Due to these findings, the optimal steaming conditions for steam-refined and steam-extracted samples are clearly visible at severities below 3.5, mainly due to the higher reactivity of the used enzymes.

### **4. Conclusions**

Due to the findings in the present study, several conclusions can be made. For steam refining experiments without subsequent alkaline extraction, the optimum of EH yields is located at a severity around 3.95. However, the optimum of total carbohydrate recovery from EH of the fibers and acidic hydrolysis of the extract fraction is at a severity around 3.4 with around 47.5% based on raw material.

For steam refining with subsequent alkaline extraction, different findings were made. For the EH yields around 100% of the theoretically available carbohydrates were found, even at severities below 3.5. However, also the total carbohydrate recovery shows the highest yields at these severities

and it is known that low contents of carbohydrate degradation products are beneficial for further process steps. Nonetheless, for lignin utilization severities around 3.95 might be optimal due to fewer carbohydrate contaminations.

Due to these findings, it can be concluded that steam refining pretreatment especially at severities below 3.5 seems to be interesting for corn stover. It could further be stated that steam refining at different severities and alkaline lignin removal is enhancing the enzymatic digestibility for the supply of monomeric carbohydrates significantly. Higher fiber yields and good digestibility after alkaline extraction giving high yields of fermentable carbohydrates for further value-added products, like ethanol or dicarboxylic acids. Therefore, further undervalued agricultural residues should be tested in the future at severities below 3.5 and with alkaline extraction. Afterward, the results can be compared with the results for corn stover presented in this study.

**Author Contributions:** Conceptualization, B.S.; funding acquisition, B.S.; investigation, M.J.K., M.B. and A.S.; supervision, B.S.; visualization, M.J.K.; writing—original draft, M.J.K.; writing—review and editing, M.J.K., M.B., A.S. and B.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Federal Ministry of Education and Research (BMBF) via Project Management Jülich (PtJ) in the PANDA project, grant number 031B0505B.

**Acknowledgments:** The Federal Ministry of Education and Research (BMBF) and Project Management Jülich (PtJ) are gratefully acknowledged for their financial support. The authors wish to thank Othar Kordsachia, who assisted in the proofreading of the manuscript. Martin Bellof, Autodisplay Biotech GmbH, is thanked for the raw material acquisition. Anna Knöpfle, Nicole Erasmy and Sascha Lebioda are also gratefully acknowledged for their technical support.

**Conflicts of Interest:** The authors declare no conflicts of interest.
