During 5-6-h long experiments the product gas composition did not change significantly; CO and CO2 were the major gas components with Ct-C4 hydrocar­bons accounting for approximately 20 vol% of the product gas. The product col­lected in the ESP that operated at the exit temperature of 80°C was a dark brown viscous liquid while that collected in the condensers was mostly water (90-95 wt%) with a thin layer of organics floating on the top. In every test significant amounts of carbon (coke and char) were produced which were collected in the reactor, the cyclone, and the filter. The carbon yields were determined by weight difference of the collected solids and of the initial catalysts. The yields of liquids corresponded to the weight of the fractions collected in the ESP and condensers, and the amount of gaseous product was calculated based on its volume and composition (excluding the carrier gas). Table 1 shows the results of two experiments carried out using the

Albemarle catalyst. The first experiment was run for 6 h and the final ratio of cata­lyst to biomass was 0.38 and the second experiment was run from 2 h with a final catalyst to biomass ratio of 1.2.

Mass balance closures were in the range 83-85% with 15-17% not accounted for. These losses most likely comprised of noncondensed water and volatile organic compounds that were not analyzed by gas chromatography. The amounts of carbon solids (char and coke) were in the range of 15-23% of the weight of the feedstock while the yields of total liquid were 35-40%, and those of gas were 25-28%. Similar product distribution was reported by Zhang et al. [32], Aho et al. [33], and Agblevor et al. [28] in bubbling fluidized beds. However, Lappas et al. [27] achieved significantly higher yields of liquids using a fluidized catalytic cracking system (circulating fluidized bed) with ZSM-5 catalyst. The most likely reason for that was maintaining the catalyst activity by continuous regeneration.

The liquid product included three phases: a very thin top organic layer, the middle most abundant water-rich fraction, and the bottom dark brown organic fraction. The yield of the heavy organic fraction was about 12% based on the feedstock. The light liquid hydrocarbon phase was 3-5% of the total condensate. The analysis of the organic liquid from the first test showed the following elemental composition: C 67-68 wt%, H 6.2-6.6 wt%, O 25-26 wt%. The oxygen content of this liquid was significantly less than for crude bio-oil from a non-catalytic process that contains 45-50% oxygen (37­40% on water-free basis). The organic liquid from the second test was even much more deoxygenated and included 83% C, 7.3% H, and 9.6% O, which is the lowest known to us oxygen content from bench-scale catalytic pyrolysis reported in the litera­ture ([27,28, 32, 33], all reporthigher oxygen contents of 13.5-22%).

In both tests, the biomass carbon conversion to organic liquid was 20-22% with most of carbon converted to gas and char/coke. It seems that in our experiments the lignin part of the feedstock was partly depolymerized (Mw 330-350) while the car­bohydrates were mostly converted to solids (char and coke), gas, and water. Agblevor et al. [29] had also concluded that the majority of the organic oil from catalytic pyrolysis in their experiments originated from lignin. The results showed that pyrol­ysis oil with highly reduced oxygen content can be produced by catalytic pyrolysis; however, the biomass carbon to liquid conversion was still low and more research is needed to improve the process performance.