Wood Science and Technology Vol. 3 (1969) p. 232-238
Wood Chemical Composition as Related to Properties of
Handsheets Made from Loblolly Pine Refiner Groundwood
By CHARLES W. MCMILLIN*, Alexandria., La.
Summary
Bunt and tear strengths of handsh~tB made from 48 pulps disk-refined from chips of
varying chemical composition deo~ with inOre88ing extractive content after the independent
effects of fiber morphology were specified. Tbjs result was attributed to l~ed bond strength
caused by reduced surface tension forces and blocking of reactive sites on the fiber ~.
Zussmmenfassung
BerBtzahI und ReiBfestigkeit der Blatter yon 48 verschiedenen Zellstoffen aus soheiben-
gemahlenen Spa.nen mit unterschiedlichen chemischen Bestandteilen verrlngeMt sioh mit
ansteigendem ExtraktstoUgehalt; die davon unabhingigen Einfltisse des Faaeraufbau~ sind
dabei beriicksichtigt. Diese Tatsache wird der verminderten Bindungakraft ~ben, ver-
ursacht durch die herabgesetzte Oberflichensparmung und die BJookjerung yon BindungsstelJen
an den Faseroberflichen.
Introduction
Two previous papers [MOMn..LIN 1968b, 1969] have examined gross wood
characteristics, fiber morphology, and degree of refining in relation to sheet prop-
erties as part of a program to develop criteria useful in predicting and controlling
the papermaking potential of loblolly pine (PinU8 IaedaL.) refiner groundwood.
These studies revealed that most sheet properties could be improved and made
more uniform by isolation or selection of wood having the desired gross character-
iStics of high latewood content but relatively low density. It was further shown
that fiber prepared from wood having long, narrow-diameter traoheids with thick
walls yielded handsheets of improved properties. This was attributed to the ability
of such tracheids to unwind into highly deformable, ribbon-like particles as a
result of torsional stresses induced during the final phases of refining. Research
elsewhere [FORGAOS 1963] has shown such ribbon-like particles to provide the
coherence n~e88ary for strength development in mechanical pulps.
The present study was undertaken to determine whether wood chemical com-
position affects sheet properties after the effects of wood morphology have been
considered. A subsequent paper will examine pulp quality in relation to handsheet
strength.
* The author appreciatively acknowledges the assistance of the Roy O. Martin Lumber ~.,
Alexandria, La.; R. A. Le88k and J. Adams of Bauer Bros. ~., Springfield, Ohio; and D. Bower,
mathematical statistician at the Southern Forest Experiment Station, New Orleaos, La.
Wood Chemical Composition as ~lated to Handsheet Properti~
233
Procedure
The detailed procedures for selection of trees, wood preparation and classifica-
tion, refining, chip sampliilg, and handsheet testing have all been described previous-
ly [McMILLIN 1968 b]. All chemical constituents were determined from sample
chips used in the earlier studies and were correlated with the previously measured
sheet properties.
While pulps for the series of Studies were produced by both single- and double-
pass refining, this paper deals only with those from double-pass fiber. Specific
refining energy was 40 hp days per air-dry ton on the first pass and 30 hp days per
air-dry ton on the second.
Wood was selected and stratified into 12 categories. Two growth rates, two
specific gravities, and three radial positions in the tree were considered in a fac-
torial design. The wood in each category was chipped and the chips randomly
divided into four within-sample replications. Before refining, a subsample of
1000 chips was drawn from each replication for evaluation of wood properties.
Five chemical constituents were measured: hemicellulose, holocellulose,
alpha-cellulose, lignin, and extractives. They were correlated with four sheet
properties: sheet density, burst factor, tear factor, and breaking length.
Chemical constituents were determined from 100 chips randomly selected from
each subsample. The chips were air-dried and ground to meal in a Wiley mill.
The fraction passing a 4O-mesh screen and retained on a 6O-mesh screen was then
conditioned to equilibrium moisture content in a room maintained at a constant
50 percent relative humidity and 72° F.
Holocellulose and alpha-cellulose were determined by the method of ERICKSON
[1962], except that the amount of sodium chlorite solution per cycle was doubled
for the holocellulose determinations. Hemicellulose values were calculated as the
difference between holocellulose and alpha-cellulose values. Duplicate determina-
tions were made for each of the 48 samples, and the averaged result W&8 expressed
&8a percentage of the weight of ovendry extractive-free wood.
The alcohol-benzene extractive content W&8 determined in accordance with
TAPPI standard method T 6 os-59 and expressed in percent of ovendry weight.
Lignin content, expressed &8 a percentage of ovendry extractive-free wood, was
determined by the TAPPI method given in T 13 m-14.
Table 1 lists the results of the chemical determinations. The handsheet proper-
ties are those obtained in the earlier Study [McMILLIN 1968b]. These data, along
with the previously determined weighted average for latewood-earlywood morpho-
logical characteristics [MOMTT,T.TN 1969], were employed in multiple regression
analysis. The analysis followed a sequence of first considering the eHect of fiber
morphology [MCMILLIN 1969] and then introducing chemical factors by the usual
stepwise regression criteria. AIl equations are of the type 11 = bo + bl Xl + bl XI+" .
where 11 is a dependent variable, e.g., burst, tear; hi, a regression coefficient; and
Xi, an independent variable, e.g., a morphological characteristic or a chemical
constituent. Thus, the equations reveal the significant effects on sheet properties
of wood chemical constituents after the effects of fiber morphology have been
considered. Various transformations of the chemical data were considered, e.g.,
the square of extractive content and the ratio of alpha-cellulose content to extrac-
5 Wood Science and TechnolO8Y. Vol. 3
234
CHARLEs W. MCMILLIN
tive content. The equations were tested at the 95 percent level of probability, and
a.ll variables included were significant at that level.
Results
The 48 samples represented in Table 1 exhibited a range of chemical properties.
Holocellulose ranged from 68.44 to 75.63 percent; alpha-cellulose, from 43.90 to
53.24 percent; hemicellulose, from 17.59 to 28.81 peroent; lignin, from 27.11 to
33.16 percent; and extractives, from 2.49 to 14.20 percent.
Table 1. Ru.u. of OMmioal aM H ~ DelerrMJIaIiOII8l
Unex-
traot-
ed
ape-
cttkI
araT-
ttT
PosIUOIl
intree
(rineB
from
pith)
I How. .A1P
~RIDp oellu- oeUu- per ,I~ 1- Inch:
% %
~
I BUNt
Bttyfactor
Break.
tug
leDCth
Te&zo
faotor
,~
~
%
clam'
4.11
7.59
4.80
11.83
5.53
7.08
5.30
12.38
4.91
8.27
5.53
8.27
0...10
[ 0.427 1 44.82 I
I I eo..3
0...10 .457 46.82 .. f 720.5
0...10 .492 46.25 r 761.0
0...10 ;(,16 46.96 i 656.1
11...20
I .w, 4.5.60, 896.1
11...20 .459 47.23 J 996.1
11...20 .512 4,9.86 1,264,.0
11...20 .5U 48.20 I 724.2
21...30 .468 47.32 1,096.9
21...30 .438 46.09 1,312.4
21...30 .534 46.48 1,085.3
21...30 .511 00.37__)__-1--- 1,129.3
Table 2 lists multiple regression equations whioh most aocmrately deBOribe
handsheet properties in terms of weighted average morphological characteristics
and chemical constituents. Th~ cumulative RI values and the standard errors of
the estimates are &lBo given.
When the effects of fiber morphology had been accounted for (Eq. 1), no
chemical component was significantly related to sheet density. The effects of
fiber morphology on sheet density as well &B on the other handsheet properties
considered here have been diBCUB8ed previouBly [MoMn.LIN 1969].
After effects of fiber morphology had been considered, the ratio of alpha-
cellulose content to extractive content proved significantly related to burst factor
(Eq. 2). These two chemical constituents accounted for an additional 10 percent
of the variation in burst beyond the 48 percent accounted for by morphological
characteristics.
Figure 1 ch&rtB the effect on burst factor of extractives at two alpha-cellulose
contents. The graphed lines in this figure and in Fjg. 2 were obtained oy sub.
stituting a range of values for the chemical constituent on the X-axis and fixing
the remAining chemical variables in the regression equations at the indicated
levels. In all oases, mean values were used for the morphological functions.
1 Each numerical value is the average of four replications except ~t va1u~ for rings per
inch are baaed on one obeervation apiece.
26.00
25.45
25.59
24.72
24.74
24.33
23.35
23.14
1 23.08 25.20
23.30
23.00
29.24
28.99
1 29.32 30.26
28.99
28.81
28.76
29.14
30.21
29.84
30.45
27.42
6.31
7.74
10.42
1.2.84
5.75
4.07
4.12
5.73
3.38
3.27
2.91
3.19
0.289
.281
.266
.293
1.287
.293
.318
.292
.291
.319
.302
.299
3.34
3.25
3.82
2.96
4.41
4.87
6.42
3.23
5.61
7.00
5.49
5.67
41.7
42.2
48.9
35.5
54.3
58.4
82.2
41.3
67.0
76.5
61.3
61.7
1 70.81
72.53
71.84
70.76
70.34
71.43
73.19
71.46
70.40
71.28
69.78
73.37
