GEOHORIZON

AUTHORS

Variation of organic matter

density with thermal maturity
Mark D. Rudnicki

ABSTRACT
An estimate or measurement of organic matter density is required for converting between the weight percent of total organic carbon (TOC) and the volume percent of organic matter
for wireline log calibration; it is therefore important to recognize
when signicant changes in organic matter density occur. A
method is presented for calculating organic matter density from
measurements of crushed-rock dry grain density and Soxhletextracted TOC. I have investigated the thermal evolution of
organic matter by tracking changes in the intrinsic density of
organic matter as a function of thermal maturity. Organic
matter density shows two step increases that correspond to the
generation of liquid hydrocarbons in the oil window (up to
~1.2% vitrinite reectance [Ro]) and the conversion of organic
matter to graphitelike carbon (more correctly, turbostratic
carbon) at high thermal maturity (>4% Ro). Profound structural changes of organic matter may, in part, determine the
maturity limits of source-rock tight liquids and shale-gas plays,
particularly at high thermal maturity, where gas is hosted within
the organic matterhosted pore system.

INTRODUCTION
The conversion of organic matter weight fraction to volume
fraction is an important step for the petrophysical modeling of
organic matterrich mudstones (Alfred and Vernik, 2013). In

Copyright 2016. The American Association of Petroleum Geologists. All rights reserved.
Manuscript received February 10, 2015; provisional acceptance August 3, 2015; revised manuscript received
August 11, 2015; nal acceptance August 18, 2015.
DOI:10.1306/08181515024

AAPG Bulletin, v. 100, no. 1 (January 2016), pp. 1722

17

Mark D. Rudnicki ~ ExxonMobil

Upstream Research Company, 22777
Springwoods Village Parkway, Spring, Texas
77389; mark.d.rudnicki@exxonmobil.com
Mark Rudnicki is a research scientist in the
Reservoir Quality and Unconventional
Resources Characterization section at
ExxonMobil Upstream Research Company. He
holds a B.Sc. degree in geology from the
University of Shefeld, United Kingdom, and a
Ph.D. in marine geochemistry from the
University of Cambridge, United Kingdom. His
research interests include understanding the
properties of mudstones that facilitate the
production of hydrocarbons.
ACKNOWLEDGMENTS
This work resulted from many discussions
with David Awwiller and Quinn Passey. I thank
e
Dmitry Daudin and Jean-Christophe Semper
for ongoing project and management support
for this research. I thank CoreLabs for
permission to publish derived data in Figure 2.
Senior Associate Editor Barry Katz and Irene
Arango provided thoughtful comments that
improved the clarity of this paper.
EDITORS NOTE
A color version of Figure 2 can be seen in the
online version of this paper.

this paper, I show how the density of organic

matter in organic-rich mudstones can be calculated using commonly measured total organic
carbon (TOC) and dry grain density data. I show
how changes in the structure of organic matter
can be diagnosed from the standpoint of changes
in organic matter density as a function of thermal
maturity, and how they can be used as a framework for understanding possible controls on unconventional mudstone play types as a function of
thermal maturity.
CALCULATION OF ORGANIC MATTER
DENSITY
The average density of organic matter may be
calculated on a porosity-free basis from the relationship between measurements of dry grain
density from the Gas Research Institute (GRI)
crushed-rock method for mudstone analysis
(Luffel and Guidry, 1992; Luffel et al., 1993;
Guidry et al., 1995) and weight-percent TOC
(e.g., from LECO measurements of powdered and
Soxhlet-extracted samples). To determine the
density of kerogen in the oil window, TOC
measurements can be complicated by the presence of residual oils; therefore, the measurements
must be made on a solvent-extracted sample.
Common solvents such as dichloromethane
(DCM) and 90% DCM:10% methanol have relatively low boiling points of approximately 37C
and 46C (98.6F and 114.8F), respectively, and,
therefore, they are not likely to signicantly affect
the structure of oil-window kerogen.
The dry grain density of a sample is calculated
as the volume-weighted sum of the organic and
mineral components:
rGrain = vrOM + 1 - vrMineral
(1)
where rGrain = dry grain density (g/cm3); rOM =
organic matter density (g/cm3) on a pore-free basis;
rMineral = mineral density (g/cm3); and n = volume
fraction of organic matter.
A second expression for rOM can be derived by
writing an equation for the weight fraction of organic matter as a function of the volume of organic
matter and the dry grain density:
18

Geohorizon

TOCf = Ck

vrOM
rGrain

(2)

where TOCf = weight-fraction TOC (= wt. % TOC/

100); and Ck = carbon weight fraction of organic
matter.
Rearranging equation 2 for n and substituting
into equation 1 yields
rMineral


rGrain =
(3)
TOC
rMineral
1 + Ck f
1
rOM
For a unit volume (1 cm3) of solid rock on a porefree basis, equation 2 expresses that the overall
weight of that volume will be equal to rGrain, and
the weight of the organic matter will be given by
the product nrOM. The term Ck is introduced to
express that the carbon weight fraction of organic
matter is not unity, and it arises because TOC is a
measure of carbon content and not kerogen content
(e.g., Tissot and Welte, 1984). In immature kerogen,
Ck 0.60.7 but increases during oil generation via
the same processes that reduce the hydrogen index
(Okiongbo et al., 2005) until vitrinite reectance
(Ro) is approximately 1% where Ck 1 (see, e.g.,
Durand, 1980). The same term was highlighted
by Alfred and Vernik (2013), who used a value
Ck 0.8 in the oil window. As an example, for
asphaltene precipitates, Ck 0.8 and rOM =
1.151.2 g/cm3 (e.g., Nomura et al., 2003; Rogel
and Carbognani, 2003). For the purposes of
this paper, I set Ck = 1, which is a good approximation in the gas window, but it should result in low
calculated organic matter densities below Ro 1%.
The relationship between TOC and the calculated
volume percent of organic matter is shown in
Figure 1. The curves in this gure were calculated
with rOM = 1.2 g/cm3; Ck = 0.8; and varying
n = 015 vol. %. Volume-percent TOC was then
recalculated for Ck = 0.6, 0.8, and 1.0 by tting
equation 3 to the resulting values for rGrain and
TOC. The plot illustrates that by setting Ck equal
to 1.0 in the oil window, the lower model value
for rOM = 1.052 g/cm3 does not entirely compensate for the reduction in kerogen weight
percent implied by the higher value for Ck. The
TOC volume percent is approximately 1% lower
than the model curve for Ck = 0.8. Likewise,

setting a lower value for Ck results in an overestimate of TOC volume percent.

By assuming a value for rOM and calculating
rGrain for each sample, equation 3 may be tted
using a least-squares approach to measured data
(e.g., the solver function of Microsoft Excel can be
applied to nd the best t value for rOM by minimizing the tting error). Errors in rOM resulting
from scatter in the rGrainTOC relationship may
also be calculated using commercial general-curvet software methods.
My method for calculating organic matter
density is based on the assumption that, for a suite
of samples, the mineral matrix composition, and
therefore rMineral , is more or less constant, so
changes in grain density are driven solely by
changes in the amount of organic matter. For this
reason, samples with bulk mineral compositions
that differ signicantly from the data set average
should be excluded from the analysis. Correlations
between mineral components and TOC may
result in the calculation of an apparent organic
matter density, because mineral changes to
rMineral will be compensated for by adjusting the
rOM. For example, in organic-rich mudstones,
TOC is frequently correlated with pyrite content
because of organic matterdriven sulfate reduction
during early diagenesis (e.g., Berner, 1964). In this
case, the decrease in grain density resulting from
low-density organic matter will be offset by an
increase in grain density resulting from highdensity pyrite.

THERMAL EVOLUTION OF ORGANIC

MATTER
Calculated organic matter densities are plotted
versus percent Ro for various tight liquids and shalegas plays in Figure 2. My results are also compared with a rst principles model for predicting
compositional yields from kerogen (chemical
structurechemical yields model [CSCYM]; Walters
et al., 2007). Error bars are calculated based on the
reported values for Ro where multiple measurements were made in a depth prole within the
reservoir interval (x-axis) and given the variability

Figure 1. Variation in calculated organic matter volume with

Ck. The curves were calculated using the parameters rMineral =
2.75 g/cm3; rOM = 1.2 g/cm3; Ck = 0.8; and n = 0%15%. Model
ts to the resulting calculated values of rGrain and weight-percent
TOC give rOM = 1.397 g/cm3 for Ck = 0.6 and rOM = 1.052 g/cm3
for Ck = 1.0. Ck = kerogen carbon fraction; TOC = total organic
carbon; n = volume of organic matter; rGrain = dry grain density;
rMineral = mineral density; rOM = organic matter density on a porefree basis.

in mineralogy between samples (y-axis). In the

calculated organic matter densities at Ro < 1.5%,
there is considerable scatter that may, in part, be
because of the inclusion of extractable hydrocarbons in TOC measurements made on nonextracted samples. In addition, the interpretation
of maturity based on Ro is subject to considerable
uncertainty because of the scarcity of true vitrinite
macerals in marine source rocks, the absence
of vitrinite before the radiation of vascular
land plants during the Silurian Period (e.g.,
Behrensmeyer et al., 1992), and the difculty of
maceral identication at very high thermal maturity
(e.g., Ro > 2.5%).
Despite these caveats, the plot illustrates that
organic matter density does not increase steadily
over the maturity range of Ro = 0.5%5.0% (e.g.,
Ward, 2010; Alfred and Vernik, 2013). Instead,
two steps are present in the calculated organic
matter densities that likely represent important
transitions in the structure of the organic matter.
The rst step, which occurs at Ro 1%, is represented
by an increase in organic matter density from
approximately 1.0 to 1.6 g/cm3. The calculated
densities are consistent with the CSCYM model
Rudnicki

19

Figure 2. Compilation of calculated organic matter densities

versus measured vitrinite reectances (Ro) for various tight liquids
and shale-gas unconventional plays. For comparison, the density
of type II kerogen (Upper Jurassic Draupne Formation, northern
North Sea, 3C/Ma [37.4F/Ma] heating rate) is also plotted,
calculated using the chemical structurechemical yields model
(CSCYM; Walters et al., 2007). The shaded region encompasses
most of the calculated density values and is drawn to illustrate key
changes in organic matter density. This gure includes data from
the Devonian Marcellus Formation of northern Pennsylvania,
provided by the CoreLabs Marcellus Consortium, and is used with
their permission.

of Walters et al. (2007), who also gave an empirical model for the density of kerogen:
Specific gravity
=

1
0:5129 + 0:298H=C + 23:131
mol: wt:

where H/C = atomic H/C ratio; and mol. wt. =

average molecular weight (g).
Because the average molecular weight of kerogen is large, the kerogen density is largely a
function of the H/C ratio. Therefore, the densication of immature kerogen will be dependent on
the generation and expulsion of hydrogen-rich
liquid hydrocarbons, thermal cracking releasing
gas, and the increasing aromatization of the
remaining solid organic matter. The calculated
values for organic matter density appear low in the
oil window compared with the CSCYM model
because of the assumption that the parameter
Ck = 1. A more detailed study of the density distribution of maceral components of type II marine
kerogen of samples of the Middle Devonian to
20

Geohorizon

Lower Mississippian New Albany Shale from Indiana (Mastalerz et al., 2012) has shown that alginite is
associated with the density range 1.11.15 g/cm3;
alginite and amorphinite lie in the range 1.151.2 g/cm3;
amorphinite dominates the range 1.21.4 g/cm3; and
amorphinite, vitrinite, and inertinite are associated
with densities greater than 1.4 g/cm3. This work
demonstrates that maceral composition is important
for determining the initial density of immature kerogen and, therefore, changes in density during maturation. For example, immature, alginite-rich type I
and II kerogens should have a lower initial density than
vitrinite-rich type III kerogen. These differences
should decrease with increasing thermal maturity.
The stepwise densication of organic matter is
one reason for the thermal-maturity dependence of
the Dlog-R relationship for the well-logbased estimation of TOC content (Passey et al., 1990); the
other reason is the generation of oil that replaces
pore water. A standard well-log overlay method,
Dlog-R relates the difference (D) between resistivity
(R) on a logarithmic scale (hence the name) and the
sonic log track. Separation between these two logs
arises from the resistivity of pore-lling hydrocarbons and the relatively slow sonic velocity of kerogen. As the density of kerogen increases through the
oil window, the sonic velocity of kerogen increases,
which changes the relationship to TOC content. The
plateau of organic-matter density versus maturity
after oil generation explains why the calibration of
the Dlog-R response with TOC content is constant
beyond Ro 1.2% (Sondergeld et al., 2010).
The second step in organic matter density, which
occurs at Ro 4%, is represented by an increase in
density from approximately 1.62.25 g/cm3. Between the two steps in organic matter density, there
is a wide maturity range from Ro = 1.2 4.0% over
which organic matter shows little variation in terms
of density and it is expected that pore systems will be
stable (i.e., unaffected by bulk changes in organic
matter structure). This wide maturity plateau is a
partial explanation for the extraordinary success of
shale-gas plays that generally rely on organic matterhosted pore systems for gas storage and transport.
Above Ro 4%, organic matter is converted
to carbon with a density approaching that of
graphite (r = 2.25 g/cm3). Walters et al. (2014)

used high-resolution transmission electron microscopy of samples of mudstones to identify

the increasing abundance of turbostratic carbon
(Oberlin, 1989) with thermal maturity increasing
from 0.6%4.5% Ro. In this paper, I familiarly
refer to turbostratic carbon as graphitelike carbon.
The carbon in high-maturity electrically conductive mudstones has a structure with layers that are
not perfectly arrayed as in metamorphic crystalline
graphite. Hexagonal graphite has a regular AB
stacking where the graphene layers are offset by a
half unit cell, whereas the stacking of turbostratic
carbon is less ordered (Li et al., 2007). The presence of graphitelike carbon can often be recognized by a signicant decrease in well-log resistivity
to approximately 0.2 ohm m for source rocks with
TOC greater than 4 wt. %. The narrow maturity
window over which this decrease occurs, Ro =
4.0%4.5%, explains why the transition to low
wireline log resistivity can occur within a relatively
short distance (1020 km [6.212.4 mi]) in a
producing play. The low observed wireline log
resistivity is a more reliable (and direct) indicator
of the change in organic matter structure caused by
graphitization than an interpreted value for vitrinite
maturity. Ultramature, low resistivity rocks are
present in North America in shale-gas plays such as
the Devonian Marcellus of northeastern Pennsylvania (e.g., Laughrey et al., 2011) and worldwide in
shale-gas plays such as the Cambrian Qiongzhusi
and Orodivcian-Silurian Wufeng-Longmaxi of the
Sichuan Basin, Peoples Republic of China (Yuman
et al., 2014). To date, there have been few well tests
in conductive (0.2 ohm m), ultramature shale-gas
plays, so it is not possible to dene this transition
denitively as an upper limit to shale-gas producibility.
If it turns out that the densication of organic matter at
the graphite transition is the upper limit of shale-gas
plays, then one implication would be that organic
matter forms the seal in shale-gas systems. If other
seals persist, the play may still work.

SUMMARY AND CONCLUSIONS

In this paper, I have shown how the intrinsic (porefree) density of organic matter can be calculated

using GRI crushed rock determination of dry grain

density (rGrain) and Soxhlet-extracted weightpercent TOC. Because an estimate or measurement of organic-matter density is required to
convert between weight-percent TOC and volumepercent organic matter for wireline log calibration,
it is important to recognize when signicant changes
in organic-matter density occur. For example, a key
component of the transition from tight liquids to
shale gas is the thermal maturation of organic matter
and expulsion of hydrocarbons, with associated
changes to the structure of the remaining kerogen. In
fact, organic matter increases in density in two steps
over the thermal-maturity range Ro = 0.5%5.0%,
indicating signicant changes in organic matter
structure. These structural changes represent the
maturation of organic matter during oil generation
over the range Ro = 0.8%1.2% and the transition to
ultramature, conductive mudstones over the range
Ro = 4.0%4.5%. The rst structural change is associated with oil generation and ultimately the
cracking of liquid hydrocarbons to gas with pyrobitumen as a residue, and therefore it encompasses
the liquids-rich shale play and the transition to shale
gas. The second structural change involves the increasing crystallization and densication of pyrobitumen and refractory kerogen. Because gas is borne
by organic matter pores (Loucks et al., 2009), rearrangement of the structure of the organic matter
may provide an opportunity to release gas that may,
in part, determine the upper maturity limit for
source-rock shale-gas plays.

REFERENCES CITED
Alfred, D., and L. Vernik, 2013, A new petrophysical model
for organic shales: Petrophysics, v. 54, p. 240247.
Behrensmeyer, A. K., J. D. Damuth, W. A. DiMichele,
R. Potts, H.-D. Sues, and S. L. Wing, 1992, Terrestrial
ecosystems through time: Evolutionary paleoecology of
terrestrial plants and animals: Chicago, University of
Chicago Press, 588 p.
Berner, R. A., 1964, An idealized model of dissolved sulfate
distribution in recent sediments: Geochimica et Cosmochimica Acta, v. 28, p. 14971503, doi:10.1016
/0016-7037(64)90164-4.
Durand, B., 1980, Quantitative importance, in B. Durand, ed.,
Kerogen: Insoluble organic matter from sedimentary

rocks: Paris, Editions

Technip, p. 2734.

Rudnicki

21

Guidry, K., D. Luffel, and J. Curtis, 1995, Development of

laboratory and petrophysical techniques for evaluating
shale reservoirs, nal report: Des Plaines, Illinois, Gas
Research Institute, 304 p.
Laughrey, C. D., T. E. Ruble, H. Lemmens, J. Kostelnik,
A. R. Butcher, G. Walker, and W. Knowles, 2011, Black
shale diagenesis: Insights from integrated high-denition
analyses of post-mature Marcellus Formation rocks,
Northeastern Pennsylvania: AAPG Search and Discovery
article 110150, accessed November 4, 2015, http://www
.searchanddiscovery.com/documents/2011/110150laughrey
/ndx_laughrey.pdf.
Li, Z. Q., C. J. Lu, Z. P. Xia, Y. Zhou, and Z. Luo, 2007, X-ray
diffraction patterns of graphite and turbostratic carbon:
Carbon, v. 45, p. 16861695, doi:10.1016/j.carbon.2007
.03.038.
Loucks, R. G., R. M. Reed, S. C. Ruppel, and D. M. Jarvie,
2009, Morphology, genesis, and distribution of nanometerscale pores in mudstones of the Mississippian Barnett Shale:
Journal of Sedimentary Research, v. 79, p. 848861,
doi:10.2110/jsr.2009.092.
Luffel, D. L., and F. K. Guidry, 1992, New core analysis
methods for measuring reservoir rock properties of Devonian shale: Journal of Petroleum Geology, v. 44,
p. 11841190, SPE-20571-PA, doi:10.2118/20571-PA.
Luffel, D. L., S. A. Hopkins, and P. D. Schettler, 1993,
Matrix permeability measurement of gas productive
shales: SPE Annual Technical Conference and Exhibition, Houston, Texas, October 36, 1993, SPE-26633-MS,
doi:10.2118/26633-MS, 10 p.
Mastalerz, M., A. Schimmelmann, G. P. Lis, A. Drobniak, and
A. Stankiewicz, 2012, Inuence of maceral composition
on geochemical characteristics of immature shale kerogen: Insight from density fraction analysis: International
Journal of Coal Geology, v. 103, p. 6069, doi:10.1016
/j.coal.2012.07.011.
Nomura, M., S. Murata, K. Kidena, T. Ookawa, and
H. Komoda, 2003, Consideration of the density evaluation of the model of heavy hydrocarbons by computer
simulation: Fuel Chemistry Division Preprints, v. 48,
no. 1, p. 4647.
Oberlin, A., 1989, High-resolution TEM studies of carbonization and graphitization, in P. A. Thrower, ed.,

22

Geohorizon

Chemistry and physics of carbon, v. 22: New York,

Marcel Dekker, p. 1143.
Okiongbo, K. S., A. C. Aplin, and S. R. Larter, 2005, Changes
in type II kerogen density as a function of maturity: Evidence from the Kimmeridge Clay Formation: Energy and
Fuels, v. 19, p. 24952499, doi:10.1021/ef050194+.
Passey, Q. R., S. Creaney, J. B. Kulla, F. J. Moretti, and
J. D. Stroud, 1990, A practical model for organic richness
from porosity and resistivity logs: AAPG Bulletin, v. 74,
no. 12, p. 17771794.
Rogel, E., and L. Carbognani, 2003, Density estimation of
asphaltenes using molecular dynamics simulations: Energy
and Fuels, v. 17, p. 378386, doi:10.1021/ef020200r.
Sondergeld, C. H., K. E. Newsham, J. T. Comisky, M. C. Rice,
and C. S. Rai, 2010, Petrophysical considerations in
evaluating and producing shale gas resources: SPE Unconventional Gas Conference, Pittsburgh, Pennsylvania,
February 2325, 2010, SPE-131768-MS, doi:10.2118
/131768-MS, 34 p.
Tissot, B. P., and D. H. Welte, 1984, Petroleum formation and
occurrence, 2nd ed.: New York, Springer-Verlag, 495 p.
Walters, C. C., H. Freund, S. R. Keleman, P. Peczak, and
D. J. Curry, 2007, Predicting oil and gas compositional
yields via chemical structurechemical yields modeling
(CSCYM): Part 2Application under laboratory and
geologic conditions: Organic Geochemistry, v. 38,
p. 306322, doi:10.1016/j.orggeochem.2006.09.010.
Walters, C. C., C. E. Kliewer, D. N. Awwiller, M. D. Rudnicki,
Q. R. Passey, and M. W. Lin, 2014, Inuence of turbostratic
carbon nanostructures on electrical conductivity in shales:
International Journal of Coal Geology, v. 122, p. 105109,
doi:10.1016/j.coal.2013.12.015.
Ward, J. A., 2010, Kerogen density in the Marcellus shale: SPE
Unconventional Gas Conference, Pittsburgh, Pennsylvania, February 2325, 2010, SPE-131767-MS, doi:
10.2118/131767-MS, 4 p.
Yuman, W., D. Dazhong, C. Xiangzhi, W. Shufang, and
W. Shiqian, 2014, Electrical property evidences of the
carbonication of organic matters in marine shales
and its geologic signicance: A case study of the
Lower Cambrian Qiongzhusi Shale in southern Sichuan
Basin: Natural Gas Industry, v. 34, p. 17, doi:10.3787
/j.issn.1000-0976.2014.08.001.