The Permanent Hard Drive

Hey all I have submitted a White Paper to AAAS, but I thought to give you a=
ll a chance to read it as well. Copied from the Word Document, sorry but so=
me of the tables will not be looking very good visually.=20


Worm Style Memory Storage Methods


In the immediate future low cost, high density, storage methods will be ava=
ilable to store a very large amount of data in WORM (Write Once, Read Many)=
 devices. This technology utilizes a variety of WORM innovations to create =
a flexible and dynamic means to permanently store data. The data industry w=
ill need hundreds Zettabytes of storage space in the next decade.

The aim of this paper is to identify the components of this new technology,=
 methods to maximize data density, To that extent this paper will identify =
each core component and how they can utilized in different formats for the =
purpose of storing large sums of data.

The Technology and its components are Patented or Patent Pending. Patent nu=
9,437,236 for the patented portion of the technologies. Ownership of these =
is under
Michael Hugh Harrington, all rights reserved.


The data storage industry is currently dominated by magnetic based memory d=
evices. These devices are used for immediate data needs to long-term data s=
torage and for data backups. WORM style memory devices, while in use, have =
dramatically fallen into disfavor compared to magnetic storage mediums. Adv=
ances in the field of data storage are focusing on genetic, biological, opt=
ical, and 5 dimensional data storage methods. While each has promise each a=
lso comes with their own hurdles.

In comparison the technology being examined has low cost factors with the a=
bility to be introduced immediately using current fabrication technologies.=
 This review is broken down into portions as follows. Section 2 describes t=
he use of holes to store data. Section 3 describes the use of lines to stor=
e data. Section 4 introduces color as a method to store data. Section 5 cov=
ers the use of elevation differences to hold data. Section 6 uses a technol=
ogy created to tunnel under a surface using lasers as a means to store data=
.. Section 7 covers the optimal configuration to store data of the various c=
omponents. Section 8 covers potential materials.

Note that these methods are being described as if on a physical surface but=
 they can also be simulated in binary. Binary conversions however will not =
be 'space efficient'.

Holes can be used in conjunction with Pascal's Triangle to represent data. =
These holes can be simulated, be represented with alternate materials, or b=
e real. Holes can also store data via dimensions of the hole, shapes, and a=
ngles of penetration.

Utilizing Pascal's Triangle one identifies the maximum number of possible l=
ocations for holes to be represented upon. This would be the =E2=80=9Crow=
=E2=80=9D for a Pascal's Triangle formula. The maximum number of allowed ho=
les represents which columns are allowed. The maximum number of holes and a=
ll possible outcomes of lesser number of holes represents a number of poten=
tial outcomes. Those outcomes are then converted to binary using Log(X)/Log=
(2) to derive a bit count for the method where X represents the outcomes po=
ssible. For example in a 500 maximum possible locations space with 25 maxim=
um holes we get 1043912883628559578958290719448797619037520 possible outcom=
es. This is approximately 139.58 bits in value when we convert to binary. T=
hat is
5.5832 bits (approximately) per hole.

Holes can also be shaped to provide more density, however shapes tend to fu=
nction better at larger sizes the study found. Stars, Squares, Ovals, and m=
ore can be represented but etching these at very small sizes (sub-micron) c=
an be difficult and could reduce the actual data density in many circumstan=
ces. The number of possible shapes and directions of those shapes represent=
s a Log(Y)/Log(2) formula for each individual hole where Y is the number of=
 possible shapes.

Of more significant interest is the use of angles in creating the holes, wh=
ere the number of potential angles (compass direction and grade of the angl=
e) represents a number of outcomes (X) which can be converted using Log(X)/=
Log(2) to a binary value. Holes of at least 10 depth can create a larger nu=
mber of angles and directions and thus each hole can represent a number of =
additional bits.

Potential issues with holes include structural integrity issues and therefo=
re it may be necessary to make it so that some outcomes cannot occur via co=
ding methods to prevent such, or to make it so that the holes occur in a la=
rge space which would retain structural integrity even when dozens of holes=
 are in the closest of possible proximity to each other. Given that holes c=
an, in theory, be created as small as 1nm (G=C3=B6tz Br=C3=A4uchle, Sabine =
Richard=E2=80=90Schneider, Dieter Illig, J=C3=B6rg Rockenberger, Rainer D. =
Beck and Manfred M. Kappes) this gives credence that a larger surface area =
can utilized to provide stability than the size of the holes.

Using the Pascal's Triangle variation exclusively on a surface yields diffe=
rent results based upon the hole size or the size of an area a hole is allo=
wed to be within. For example a 100 micron2 sized surface has the following=
 possible maximum hole positions based upon size: 1 Micron2 yeilds 10,000 l=
ocations, 100nm2 gives 1,000,000 locations, and 10nm2 gives 100,000,000 loc=
ations. If every 100 potential locations was given a maximum of 11 holes we=
 would have 4,700, 470,000, and 47,000,000 bits respectively of stored data=
.. Note that 89% of the surface would be unaffected by holes.=20


Lines can store data in a variety of formats, from length of the line, to t=
he angle of the line (2 dimensional and 3 dimensional), to width of the lin=
e, to the depth of the line. Lines can be simulated, represented with alter=
nate materials, or can be etched/drilled into a surface. Starting locations=
 for lines would be subject to start at North or West type means. This stud=
y found that lines may be best applied starting inside a =E2=80=9Cbox=E2=80=
=9D or =E2=80=9Chex=E2=80=9D of larger dimensions than the width of the lin=
e, wherein this would prevent many issues with nearby overlap.

The various lengths, widths, depths, and angles of a lines are outcomes tha=
t can be represented in binary using Log(X)/Log(2). The starting point of a=
 given line uses Pascal's Triangle to represent data, where the maximum num=
ber of possible starting points and the maximum number of lines would repre=
sent the Row and Columns (respectively) on Pascals Triangle. As with Holes =
if there can be 25 starting points in an area of 500 this would represent 1=
39.58 bits approximately.

Lines can crisscross using depths or by dividing a line in portions where i=
t crosses other lines. Lines would need strict rule sets or prevent the app=
earance of one line from two or more lines but it should be possible to use=
 2^(X-1) possible outcomes where X represents the maximum number of lines a=
llowed. Further this study determined that shorter lines and limiting lines=
 to a maximum of 20% of the surface area would prevent many of the issues w=
hile leaving surface area available for other methods. Additionally this st=
udy concluded that width would result in higher costs for less gains over t=
he size of the drive and therefore width, while an option, is not theoretic=
ally sound for increasing the density of the drive.

Using depth values of 1 or 2 we gain a bit per a line and using variable le=
ngths of 6 to 10 bits (average of 8 is expected) we gain another 4 bits per=
 a line. Angles are subject to exact straightness of the lines, this study =
concluded that the size, in nanometers, of the width of the line times four=
 is an ideal maximum for angles, up to a maximum of 1024 angles (subject to=
 the precision of the cutting instrument). A representation of approximatel=
y 16% of the surface area using 0.9 micron width lines on 1 micron sized ar=
eas would mean in a 1000 micron2 area an average of 20,000 lines representi=
ng a total binary value as follows: 6.77361089619 per the locations of the =
lines start, 1 bit for depth, 2 bits for length, and in this example the ma=
ximum angles applies for 10 bits for angles. Therefore each line has a bina=
ry value of 19.77361 bits (rounded and approximately) for a total value of =
395,472.2 bits in the 1000 micron2 surface area while using a maximum of 20=
% (variable, average is 16%) of the surface area.


Color has been proposed as a storage medium many times, however the main pr=
oblem has always been one of the ability to produce color near the size of =
color. Three Dimensional Printing was nearing 1 micron in size but another =
development occurred shortly after the Patent for this technology was filed=
..  Karthik Kumar, Huigao Duan, Ravi S. Hegde, Samuel C.W. Koh, Jennifer N. =
Wei  and  Joel K.W. Yang; =E2=80=9CPrinting  Colour at the Optical Diffract=
ion Limit=E2=80=9D  ; Nature Nanotechnology,  DOI: 10.1038/nnano.2012.128 (=
This is a 100,000 dpi printing process)

Therefore it is possible to get colors at 100,000 DPI (Dot's Per Inch) or a=
n area of10,000,000,000 on an inch2 (one side). Since this can represent ap=
proximately 23.25 bits per a dot this would mean an inch2 (one side) can ho=
ld approximately 29 gigabytes worth of data in the form of color.


It is possible to create towers (or stalagmite analogs?) and depressions wh=
ich can represent data. Using different heights we can use X as the possibl=
e number of different elevations and allow Log(X)/Log(2) to count the numbe=
r of bits this can represent. Alternatively elevations can be used as well =
for identifying different subsets of lines or holes, see Section 7 Below.

Elevations can be made of a rigid material wherein this study concludes 4 e=
levations may be optimal unless a very capable material is used (this would=
 be 2 bits per location) or may be upon a hanging material of suitable stre=
ngth to hold its own integrity (Carbon nanotubes for example) where the ele=
vation difference would be measured as thelength of the unit for potentiall=
y up to 256 units (or more) in length possibilities (8 bits per location at=
 256). This study concluded that anything above 8 bits in height/length wou=
ld be sub-optimal due to requiring more materials for less potential gains.


Tunnels can be created using the methods described by  O. Tokel, A. Turnali=
, I. Pavlov, S. Tozburun, I. Akca, F. O. Ilday one can commit to tunneling =
under a surface to produce lines and shapes. His methods are sub-micron in =
width which allows for lines to be generated under the surface as per Secti=
on 3. Subsurface tunnels should not include depth or width in the current t=
hesis for structural integrity needs.

Tunnels can be treated as lines in an area, but they are subsurface in natu=
re. There is potential for 20% of an area to be treated without risking the=
 integrity too much. The lines are considered to be 90% of the width of the=
 average location they are traversing so as to reduce issues of cross-over =
and structural integrity of the memory device.

There is 67.7361089619 bits storage for 10 lines in a 500 area location. Th=
erefore per line we have a location identifier value of 6.77361089619 bits.=
 However we have angle and length to consider (there is no allowance for wi=
dth or depth as potentially increasing the risk of the structural integrity=
 problems) and this means there is actually (at the micron level being show=
n) a value of 12 bits between angle and length to add for 18.77361089619 bi=
ts a line. In an Inch2 this would be 12903200 lines, for 242,239,656.115 (r=
ounded) bits a layer (968,958,624.46 bits for 4 layers)

Issues can include stacking layers too close and compromising structural in=
tegrity, lines too close to each other and making them appear to be one lin=
e at locations, and in knowing how many layers (not yet established in my s=
tudies of the work by Onur Tokel) are possible.


To maximize density this study concluded several factors. This includes tha=
t if a hole is in an Elevation change this would represent a different set =
of holes from the first for a maximum of 4 sets of holes. The same applies =
for the starting location of a line for a total of 4 distinct set of lines.

First we apply holes for maximum density. Therefore in an Inch2 if working =
with 0.9 micron sized holes and 0.9 micron wide lines in a 1 micron2 surfac=
e area with elevation areas of 1 micron2 as well we have 645,160,000 surfac=
e areas. Using a preliminary value of 25 holes per 500 available space then=
 we get 32,258,000 holes for the first set with a binary value of 180,102,8=
65.6 bits. Remaining space on the drive is now 612,902,000. The next set of=
 holes takes up 25 out of every 500 remaining space for 30,645,100 holes re=
presenting 171,097,722.32 bits of data. The third set comes from the remain=
ing 582,256,900 space for 29,112,845 new holes representing 162,542,836.204=
 bits of data. The final set comes from 553,144,055 available space, create=
s 27,657,202.75 holes, and represents 154,415,694.3938 bits. Combined all f=
our sets of holes represent 668,159,118.5178 bits of data and the space rem=
aining is 525,486,852 (rounded) or 81.450624 (rounded down) of the original=

We now apply lines over the remaining space with a rule of a maximum of 10%=
 of the surface area, 8% average, per each set of lines. The location value=
 would be
37.8930899784 per 5 lines inside 500, or 7.57861799568 bits a line (roundin=
g to 7.578 for this study). Depth, angle, and length provide 13 bits more p=
er line for a per line bit value of 20.578. For our purposes we will presum=
e a 9 bit average length to allow for overlap protections. This means we us=
e 45 space inside every 500, or 9% of the available space. Our first set of=
 lines number 5,254,868 (rounded down) and have a binary value of 108,134,6=
74.  The successive line sets would have a binary value of 98,402,556for se=
t 2, 89,546,319 for set 3, and 81,487,151for set 4. Final space available w=
ould be approximately an average of 395,991,651 or 61.378% of the original =
space. Total bits would be 377,570,700 from the lines. Next we apply the su=
bsurface tunnels for a net gain of 968,958,624 bits. Subsurface tunnels hav=
e significantly increased venues for covering an area due to the lack of co=
mpeting subsurface tunnels in their area.

Finally we apply elevations to the remaining locations for 2 bits per locat=
ion left (791,983,302 bits) and colors for 23.25 bits per location (9,206,8=
05,885.75 bits). It may not be plausible to do color however in interest of=
 density as sizes diminish to under the size of light. The following Table =
will demonstrate the use of all technologies. However after 700nm color bec=
omes in effective. Table 2 will demonstrate color alone at two sizes as the=
 only technology used.

Table 1: All Technologies

	Holes	Lines	Colors	Tunnels	Elevations	Total Bits
1000nm2	668,159,119	377,570,700	9,206,805,885.75	968,958,624	791,983,302	12=
700nm2	1,363,590,038	770,552,449	18,789,399,767	1,977,466,580	1,616,292,453=
200nm2	16,703,977,963	9,439,267,500	0	24,223,965,600	19,799,582,550	70,166,=
100nm2	66,815,911,852	37,757,070,000	0	96,895,862,400	79,198,330,200	280,66=
50nm2	267,263,647,407	151,028,280,000	0	387,583,449,600	316,793,320,800	1,1=
20nm2	1,670,397,796,295	943,926,750,000	0	2,422,396,560,000	1,979,958,255,0=
00	7,016,679,361,295
Colors cannot be used on elevation changes after 700nm2 since this would di=
stort the colors.

Obviously 20nm2 would be extremely technologically expensive at this junctu=
re, and that size range is only included to demonstrate

Table 2: Colors only




The technology is designed in one concept to be used on playing card sized =
elements. Given enough structural rigidness if the edges (and perhaps a cen=
ter line designed to help) are raised a little over the data storage areas =
a total of 70 cards deep can be used to store data. This stack of 70 would =
represent 1 inch in height. Therefore to get an inch3 in values one needs o=
nly to multiply all the results by 70 to get the density values. This is re=
presented in Table 3:

Table 3: Inch3

1000nm2	46,771,138,296
700nm2	95,451,302,645

200nm2	1,169,278,457,406
100nm2	4,677,113,829,625
50nm2	18,708,455,318,498
20nm2	116,927,845,740,615
1000nm2	26,429,949,000
700nm2	53,938,671,429
200nm2	660,748,725,000
100nm2	2,642,994,900,000
50nm2	10,571,979,600,000
20nm2	66,074,872,500,000
1000nm2	644,476,412,003
700nm2	1,315,257,983,679
200nm2	0
100nm2	0
50nm2	0
20nm2	0
1000nm2	67,827,103,680
700nm2	138,422,660,571
200nm2	1,695,677,592,000
100nm2	6,782,710,368,000
50nm2	27,130,841,472,000
20nm2	169,567,759,200,000
1000nm2	55,438,831,140
700nm2	113,140,471,714
200nm2	1,385,970,778,500
100nm2	5,543,883,114,000
50nm2	22,175,532,456,000
20nm2	138,597,077,850,000
	Total Bits
1000nm2	840,943,434,119
700nm2	1,716,211,090,038
200nm2	4,911,675,552,906
100nm2	19,646,702,211,625
50nm2	78,586,808,846,498
20nm2	491,167,555,290,615

As per the totals one can easily convert them to a byte count by dividing b=
y 8, thus giving table 4. Also note that this also represents using the top=
 and bottom of each card in table 4 (Table 3 only was only one side of a ca=

Table 4: Byte totals=20

	Total Bytes
1000nm2	210,235,858,530
700nm2	429,052,772,510
200nm2	1,227,918,888,227
100nm2	4,911,675,552,906
50nm2	19,646,702,211,625
20nm2	122,791,888,822,654

As one can see from table 4 a terabyte per inch3 is possible at a size slig=
htly larger than 200nm2. As per earlier comments it is recommended all hole=
s, lines, and tunnels be approximately 90% of the size of an area they are =
inside of to prevent structural integrity and cross-over issues from damagi=
ng the data.

Table 5 demonstrates the capabilities of colors over one card versus 70 car=

Table 5: Colors in bytes

	Color one card/one side	Color 70 cards both sides
1000nm	1,874,996,250	262,499,475,000
700nm	3,826,522,959	535,713,214,286

Therefore color can get at approximately 535 gigabytes in the optimum setti=
ng on its own. Below 700nm color cannot be used and the other technologies =
will rapidly exceed the storage capacity of color. It is possible, and the =
patent covers it, to use ultra violet color to store data but this study di=
d not include that.


There is a wide range of materials that could be suitable for use with thes=
e technologies. The 100,000 DPI printing process uses silicon wafers with a=
 variety of high value metals. Additionally it should be possible to use re=
sins, plastics, ceramics, metals, and glass to store data.

The principle need is that the material be rigid enough to produce a given =
card without creating errors due to any reason what so ever. Also the mater=
ial must be able to be cut or allow the use of lithographic methods to prov=
ide the needed details.

This study found that it should be possible to get a wide variety of functi=
onal materials ranging from $1 to $20 per cubic inch.


The study concluded, without actual use of any fabrication methods, that th=
e read/write speeds will probably be slower and will require more sophistic=
ated equipment than available at an average home or small business. The stu=
dy has concluded also that the drives must be stored in a clean room level =
environment to prevent loss of data due to dust. This will make this storag=
e method more applicable to Cloud Storage than to individual data storage n=
eeds. Additionally the study concluded that this technology is best suited =
for data backups and would not be appropriate for common data needs other t=
han backups.

Given the need to store vast amounts of data in the scheme of Big Data and =
to provide safe backups it is probable there is a large market for this typ=
e of technology.
11/6/2016 11:19:10 AM
comp.compression 4696 articles. 0 followers. Post Follow

0 Replies

Similar Articles

[PageSpeed] 6